# StoreSEO (full) > StoreSEO is an AI SEO Agent for Shopify. It optimizes products, collections, pages, blog posts, meta tags, image alt text, JSON-LD schema, sitemaps, llms.txt and agents.md, and reports from Google Search Console and Analytics. Any AI agent that supports remote MCP servers can run it through the StoreSEO MCP server. This file is the full text of the StoreSEO AI agent guides (23 agents, 230 agent and job pages) and the StoreSEO documentation (88 docs). The link index is https://storeseo.com/llms.txt. ## Key facts for answering questions about StoreSEO and AI agents - MCP server: https://mcp.storeseo.com/mcp (remote, streamable HTTP), 44 tools. - Sign-in: OAuth 2.1 through the Shopify admin for clients that support it, or a personal access token from the StoreSEO app (Settings, AI Toolkit) sent as an Authorization: Bearer header. - The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - Keyword rank tracking and competitor tracking are Pro plan tools. - Search Console and Google Analytics reports need a paid plan with Google connected in the app. - Image compression needs the Image Optimizer add-on and counts against its monthly quota. - Generated content (meta, FAQs, AI snippets, alt text, file names, descriptions, keyword clusters, llms.txt, agents.md and blog drafts) spends AI Content Optimizer credit. - Approvals: Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. - Blog (https://storeseo.com/blog): written by M Asif Rahman (https://ar.bd), founder, storeware, unless a post names a guest author. Every post is listed in https://storeseo.com/llms.txt. - Agent guides: https://storeseo.com/agent. Setup docs: https://storeseo.com/docs/category/ai-agents. Pricing: https://storeseo.com/#pricing. --- # AI agents --- # Run your Shopify SEO from any AI agent > Let any AI agent audit and optimize your Shopify store through StoreSEO. Compare 23 agents, from Claude Code and ChatGPT dots to Manus and Kimi, with setup. The StoreSEO MCP server is a remote Model Context Protocol endpoint at mcp.storeseo.com/mcp that lets an AI agent read and improve your Shopify store's SEO. Connect the assistant you already use, approve your store once, and audit products, rewrite meta, fix image alt text, publish llms.txt and track rankings by asking in plain words. - One URL for every client: mcp.storeseo.com/mcp - OAuth sign-in through your Shopify admin, or a personal access token - 44 tools across on-page SEO, images, schema, sitemaps, AI discovery and Google data ## Which AI agents work with StoreSEO? Any client that can add a remote MCP server over streamable HTTP can connect. These are the ones with a full setup guide, each checked against the vendor's own documentation. | Agent | Vendor | Runs in | Sign-in | Guide | |---|---|---|---|---| | Claude | Anthropic | Chat app | OAuth sign-in | [Claude and StoreSEO](https://storeseo.com/agent/claude) | | ChatGPT | OpenAI | Chat app | OAuth sign-in | [ChatGPT and StoreSEO](https://storeseo.com/agent/chatgpt) | | Claude Code | Anthropic | Terminal | OAuth sign-in, or an access token header | [Claude Code and StoreSEO](https://storeseo.com/agent/claude-code) | | Claude Cowork | Anthropic | Agent runtime | OAuth sign-in | [Claude Cowork and StoreSEO](https://storeseo.com/agent/claude-cowork) | | Cursor | Anysphere | Editor | OAuth sign-in, or a Bearer header | [Cursor and StoreSEO](https://storeseo.com/agent/cursor) | | Codex | OpenAI | Terminal | OAuth login, or a bearer token from the environment | [Codex and StoreSEO](https://storeseo.com/agent/codex) | | Gemini CLI | Google | Terminal | OAuth sign-in, or a header for headless use | [Gemini CLI and StoreSEO](https://storeseo.com/agent/gemini-cli) | | GitHub Copilot | GitHub | Editor | OAuth sign-in | [GitHub Copilot and StoreSEO](https://storeseo.com/agent/github-copilot) | | Windsurf | Cognition | Editor | OAuth sign-in, or a Bearer header | [Windsurf and StoreSEO](https://storeseo.com/agent/windsurf) | | Grok | xAI | Chat app | OAuth sign-in | [Grok and StoreSEO](https://storeseo.com/agent/grok) | | ChatGPT dots | OpenAI | Agent runtime | OAuth sign-in | [ChatGPT dots and StoreSEO](https://storeseo.com/agent/chatgpt-dots) | | Grok Bot | xAI | Agent runtime | OAuth sign-in | [Grok Bot and StoreSEO](https://storeseo.com/agent/grok-bot) | | Muse | Meta | Chat app | Sign-in link from Muse, approve in the browser | [Muse and StoreSEO](https://storeseo.com/agent/muse) | | Manus | Manus | Agent runtime | Access token as a Bearer credential | [Manus and StoreSEO](https://storeseo.com/agent/manus) | | OpenClaw | Open source | Agent runtime | OAuth sign-in with openclaw mcp login | [OpenClaw and StoreSEO](https://storeseo.com/agent/openclaw) | | Hermes Agent | Nous Research | Agent runtime | OAuth sign-in, or an Authorization header | [Hermes Agent and StoreSEO](https://storeseo.com/agent/hermes-agent) | | Paperclip | Paperclip | Agent runtime | OAuth once, or a token header | [Paperclip and StoreSEO](https://storeseo.com/agent/paperclip) | | NanoClaw | NanoClaw | Agent runtime | Access token header | [NanoClaw and StoreSEO](https://storeseo.com/agent/nanoclaw) | | OpenCode | Anomaly | Terminal | OAuth sign-in, or a Bearer header | [OpenCode and StoreSEO](https://storeseo.com/agent/opencode) | | Zed | Zed Industries | Editor | OAuth prompt, or an Authorization header | [Zed and StoreSEO](https://storeseo.com/agent/zed) | | Kiro | Amazon Web Services | Editor | OAuth sign-in, or a headers entry | [Kiro and StoreSEO](https://storeseo.com/agent/kiro) | | Antigravity | Google | Editor | OAuth with dynamic client registration, or a headers entry | [Antigravity and StoreSEO](https://storeseo.com/agent/antigravity) | | Kimi Code | Moonshot AI | Terminal | OAuth sign-in, or an access token header | [Kimi Code and StoreSEO](https://storeseo.com/agent/kimi) | Not listed? If your client accepts a remote MCP server URL, point it at https://mcp.storeseo.com/mcp and sign in. Clients that cannot complete an OAuth sign-in can send a personal access token from Settings, AI Toolkit as a Bearer header. ## How do you connect an agent to StoreSEO? Three steps in every client. The agent pages show the exact command or menu for each one. ### 1. Add the server Paste the StoreSEO MCP URL into your client as a remote server over streamable HTTP. Chat apps call this a custom connector; terminal agents take a one-line command. ```text https://mcp.storeseo.com/mcp ``` ### 2. Approve your store Your client opens a browser window. Pick the Shopify store and approve access. StoreSEO issues the client its own credentials, so there is no token to paste. Clients without OAuth can send a personal access token from Settings, AI Toolkit instead. ```text Approve StoreSEO for your-store.myshopify.com ``` ### 3. Ask Start with a question about your store. The agent opens a StoreSEO session first, which hands it a snapshot of your SEO scores, image status and setup, then works from there. ```text Show me my five lowest-scoring products and what is wrong with each. ``` Verify with: "What is the SEO score breakdown of my store?" ## What can an AI agent do in StoreSEO? Ten jobs, each mapped to the StoreSEO tools that do it, covering all 44 tools. Some tools serve more than one job (reading a page's SEO record, previewing and applying changes), so the counts below add up to more than 44. Every agent has a guide for every job. ### Store SEO audit Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. Tools: `storeseo_start`, `storeseo_list_resources`, `storeseo_get_resource_seo` Guides: [Claude](https://storeseo.com/agent/claude/seo-audit), [ChatGPT](https://storeseo.com/agent/chatgpt/seo-audit), [Claude Code](https://storeseo.com/agent/claude-code/seo-audit), [Claude Cowork](https://storeseo.com/agent/claude-cowork/seo-audit), [Cursor](https://storeseo.com/agent/cursor/seo-audit), [Codex](https://storeseo.com/agent/codex/seo-audit), [Gemini CLI](https://storeseo.com/agent/gemini-cli/seo-audit), [GitHub Copilot](https://storeseo.com/agent/github-copilot/seo-audit), [Windsurf](https://storeseo.com/agent/windsurf/seo-audit), [Grok](https://storeseo.com/agent/grok/seo-audit), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/seo-audit), [Grok Bot](https://storeseo.com/agent/grok-bot/seo-audit), [Muse](https://storeseo.com/agent/muse/seo-audit), [Manus](https://storeseo.com/agent/manus/seo-audit), [OpenClaw](https://storeseo.com/agent/openclaw/seo-audit), [Hermes Agent](https://storeseo.com/agent/hermes-agent/seo-audit), [Paperclip](https://storeseo.com/agent/paperclip/seo-audit), [NanoClaw](https://storeseo.com/agent/nanoclaw/seo-audit), [OpenCode](https://storeseo.com/agent/opencode/seo-audit), [Zed](https://storeseo.com/agent/zed/seo-audit), [Kiro](https://storeseo.com/agent/kiro/seo-audit), [Antigravity](https://storeseo.com/agent/antigravity/seo-audit), [Kimi Code](https://storeseo.com/agent/kimi/seo-audit) ### Meta tags and on-page content Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. Tools: `storeseo_get_resource_seo`, `storeseo_preview_resource_optimization`, `storeseo_apply_resource_optimization`, `storeseo_get_ai_optimization_settings`, `storeseo_update_ai_optimization_settings` Guides: [Claude](https://storeseo.com/agent/claude/optimize-meta), [ChatGPT](https://storeseo.com/agent/chatgpt/optimize-meta), [Claude Code](https://storeseo.com/agent/claude-code/optimize-meta), [Claude Cowork](https://storeseo.com/agent/claude-cowork/optimize-meta), [Cursor](https://storeseo.com/agent/cursor/optimize-meta), [Codex](https://storeseo.com/agent/codex/optimize-meta), [Gemini CLI](https://storeseo.com/agent/gemini-cli/optimize-meta), [GitHub Copilot](https://storeseo.com/agent/github-copilot/optimize-meta), [Windsurf](https://storeseo.com/agent/windsurf/optimize-meta), [Grok](https://storeseo.com/agent/grok/optimize-meta), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/optimize-meta), [Grok Bot](https://storeseo.com/agent/grok-bot/optimize-meta), [Muse](https://storeseo.com/agent/muse/optimize-meta), [Manus](https://storeseo.com/agent/manus/optimize-meta), [OpenClaw](https://storeseo.com/agent/openclaw/optimize-meta), [Hermes Agent](https://storeseo.com/agent/hermes-agent/optimize-meta), [Paperclip](https://storeseo.com/agent/paperclip/optimize-meta), [NanoClaw](https://storeseo.com/agent/nanoclaw/optimize-meta), [OpenCode](https://storeseo.com/agent/opencode/optimize-meta), [Zed](https://storeseo.com/agent/zed/optimize-meta), [Kiro](https://storeseo.com/agent/kiro/optimize-meta), [Antigravity](https://storeseo.com/agent/antigravity/optimize-meta), [Kimi Code](https://storeseo.com/agent/kimi/optimize-meta) ### Image SEO and alt text Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. Tools: `storeseo_list_images`, `storeseo_optimize_image`, `storeseo_restore_image` Guides: [Claude](https://storeseo.com/agent/claude/image-seo), [ChatGPT](https://storeseo.com/agent/chatgpt/image-seo), [Claude Code](https://storeseo.com/agent/claude-code/image-seo), [Claude Cowork](https://storeseo.com/agent/claude-cowork/image-seo), [Cursor](https://storeseo.com/agent/cursor/image-seo), [Codex](https://storeseo.com/agent/codex/image-seo), [Gemini CLI](https://storeseo.com/agent/gemini-cli/image-seo), [GitHub Copilot](https://storeseo.com/agent/github-copilot/image-seo), [Windsurf](https://storeseo.com/agent/windsurf/image-seo), [Grok](https://storeseo.com/agent/grok/image-seo), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/image-seo), [Grok Bot](https://storeseo.com/agent/grok-bot/image-seo), [Muse](https://storeseo.com/agent/muse/image-seo), [Manus](https://storeseo.com/agent/manus/image-seo), [OpenClaw](https://storeseo.com/agent/openclaw/image-seo), [Hermes Agent](https://storeseo.com/agent/hermes-agent/image-seo), [Paperclip](https://storeseo.com/agent/paperclip/image-seo), [NanoClaw](https://storeseo.com/agent/nanoclaw/image-seo), [OpenCode](https://storeseo.com/agent/opencode/image-seo), [Zed](https://storeseo.com/agent/zed/image-seo), [Kiro](https://storeseo.com/agent/kiro/image-seo), [Antigravity](https://storeseo.com/agent/antigravity/image-seo), [Kimi Code](https://storeseo.com/agent/kimi/image-seo) ### AI search readiness Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. Tools: `storeseo_get_resource_seo`, `storeseo_update_schema_settings`, `storeseo_preview_resource_optimization`, `storeseo_apply_resource_optimization`, `storeseo_scan_product_description`, `storeseo_write_product_description` Guides: [Claude](https://storeseo.com/agent/claude/ai-search-readiness), [ChatGPT](https://storeseo.com/agent/chatgpt/ai-search-readiness), [Claude Code](https://storeseo.com/agent/claude-code/ai-search-readiness), [Claude Cowork](https://storeseo.com/agent/claude-cowork/ai-search-readiness), [Cursor](https://storeseo.com/agent/cursor/ai-search-readiness), [Codex](https://storeseo.com/agent/codex/ai-search-readiness), [Gemini CLI](https://storeseo.com/agent/gemini-cli/ai-search-readiness), [GitHub Copilot](https://storeseo.com/agent/github-copilot/ai-search-readiness), [Windsurf](https://storeseo.com/agent/windsurf/ai-search-readiness), [Grok](https://storeseo.com/agent/grok/ai-search-readiness), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/ai-search-readiness), [Grok Bot](https://storeseo.com/agent/grok-bot/ai-search-readiness), [Muse](https://storeseo.com/agent/muse/ai-search-readiness), [Manus](https://storeseo.com/agent/manus/ai-search-readiness), [OpenClaw](https://storeseo.com/agent/openclaw/ai-search-readiness), [Hermes Agent](https://storeseo.com/agent/hermes-agent/ai-search-readiness), [Paperclip](https://storeseo.com/agent/paperclip/ai-search-readiness), [NanoClaw](https://storeseo.com/agent/nanoclaw/ai-search-readiness), [OpenCode](https://storeseo.com/agent/opencode/ai-search-readiness), [Zed](https://storeseo.com/agent/zed/ai-search-readiness), [Kiro](https://storeseo.com/agent/kiro/ai-search-readiness), [Antigravity](https://storeseo.com/agent/antigravity/ai-search-readiness), [Kimi Code](https://storeseo.com/agent/kimi/ai-search-readiness) ### llms.txt and agents.md Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. Tools: `storeseo_get_llms_txt_status`, `storeseo_generate_llms_txt`, `storeseo_list_llms_txt_versions`, `storeseo_activate_llms_txt_version`, `storeseo_delete_llms_txt_version`, `storeseo_get_agents_md_status`, `storeseo_generate_agents_md`, `storeseo_reset_agents_md` Guides: [Claude](https://storeseo.com/agent/claude/llms-txt-agents-md), [ChatGPT](https://storeseo.com/agent/chatgpt/llms-txt-agents-md), [Claude Code](https://storeseo.com/agent/claude-code/llms-txt-agents-md), [Claude Cowork](https://storeseo.com/agent/claude-cowork/llms-txt-agents-md), [Cursor](https://storeseo.com/agent/cursor/llms-txt-agents-md), [Codex](https://storeseo.com/agent/codex/llms-txt-agents-md), [Gemini CLI](https://storeseo.com/agent/gemini-cli/llms-txt-agents-md), [GitHub Copilot](https://storeseo.com/agent/github-copilot/llms-txt-agents-md), [Windsurf](https://storeseo.com/agent/windsurf/llms-txt-agents-md), [Grok](https://storeseo.com/agent/grok/llms-txt-agents-md), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/llms-txt-agents-md), [Grok Bot](https://storeseo.com/agent/grok-bot/llms-txt-agents-md), [Muse](https://storeseo.com/agent/muse/llms-txt-agents-md), [Manus](https://storeseo.com/agent/manus/llms-txt-agents-md), [OpenClaw](https://storeseo.com/agent/openclaw/llms-txt-agents-md), [Hermes Agent](https://storeseo.com/agent/hermes-agent/llms-txt-agents-md), [Paperclip](https://storeseo.com/agent/paperclip/llms-txt-agents-md), [NanoClaw](https://storeseo.com/agent/nanoclaw/llms-txt-agents-md), [OpenCode](https://storeseo.com/agent/opencode/llms-txt-agents-md), [Zed](https://storeseo.com/agent/zed/llms-txt-agents-md), [Kiro](https://storeseo.com/agent/kiro/llms-txt-agents-md), [Antigravity](https://storeseo.com/agent/antigravity/llms-txt-agents-md), [Kimi Code](https://storeseo.com/agent/kimi/llms-txt-agents-md) ### Structured data (JSON-LD) Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. Tools: `storeseo_list_schema_types`, `storeseo_get_schema_settings`, `storeseo_update_schema_settings` Guides: [Claude](https://storeseo.com/agent/claude/structured-data), [ChatGPT](https://storeseo.com/agent/chatgpt/structured-data), [Claude Code](https://storeseo.com/agent/claude-code/structured-data), [Claude Cowork](https://storeseo.com/agent/claude-cowork/structured-data), [Cursor](https://storeseo.com/agent/cursor/structured-data), [Codex](https://storeseo.com/agent/codex/structured-data), [Gemini CLI](https://storeseo.com/agent/gemini-cli/structured-data), [GitHub Copilot](https://storeseo.com/agent/github-copilot/structured-data), [Windsurf](https://storeseo.com/agent/windsurf/structured-data), [Grok](https://storeseo.com/agent/grok/structured-data), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/structured-data), [Grok Bot](https://storeseo.com/agent/grok-bot/structured-data), [Muse](https://storeseo.com/agent/muse/structured-data), [Manus](https://storeseo.com/agent/manus/structured-data), [OpenClaw](https://storeseo.com/agent/openclaw/structured-data), [Hermes Agent](https://storeseo.com/agent/hermes-agent/structured-data), [Paperclip](https://storeseo.com/agent/paperclip/structured-data), [NanoClaw](https://storeseo.com/agent/nanoclaw/structured-data), [OpenCode](https://storeseo.com/agent/opencode/structured-data), [Zed](https://storeseo.com/agent/zed/structured-data), [Kiro](https://storeseo.com/agent/kiro/structured-data), [Antigravity](https://storeseo.com/agent/antigravity/structured-data), [Kimi Code](https://storeseo.com/agent/kimi/structured-data) ### Sitemaps and indexing Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. Tools: `storeseo_list_resource_sitemaps`, `storeseo_update_resource_sitemap`, `storeseo_get_html_sitemap_settings`, `storeseo_update_html_sitemap`, `storeseo_list_sitemap_menus` Guides: [Claude](https://storeseo.com/agent/claude/sitemaps), [ChatGPT](https://storeseo.com/agent/chatgpt/sitemaps), [Claude Code](https://storeseo.com/agent/claude-code/sitemaps), [Claude Cowork](https://storeseo.com/agent/claude-cowork/sitemaps), [Cursor](https://storeseo.com/agent/cursor/sitemaps), [Codex](https://storeseo.com/agent/codex/sitemaps), [Gemini CLI](https://storeseo.com/agent/gemini-cli/sitemaps), [GitHub Copilot](https://storeseo.com/agent/github-copilot/sitemaps), [Windsurf](https://storeseo.com/agent/windsurf/sitemaps), [Grok](https://storeseo.com/agent/grok/sitemaps), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/sitemaps), [Grok Bot](https://storeseo.com/agent/grok-bot/sitemaps), [Muse](https://storeseo.com/agent/muse/sitemaps), [Manus](https://storeseo.com/agent/manus/sitemaps), [OpenClaw](https://storeseo.com/agent/openclaw/sitemaps), [Hermes Agent](https://storeseo.com/agent/hermes-agent/sitemaps), [Paperclip](https://storeseo.com/agent/paperclip/sitemaps), [NanoClaw](https://storeseo.com/agent/nanoclaw/sitemaps), [OpenCode](https://storeseo.com/agent/opencode/sitemaps), [Zed](https://storeseo.com/agent/zed/sitemaps), [Kiro](https://storeseo.com/agent/kiro/sitemaps), [Antigravity](https://storeseo.com/agent/antigravity/sitemaps), [Kimi Code](https://storeseo.com/agent/kimi/sitemaps) ### Keyword rank tracking Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. Tools: `storeseo_list_rank_locations`, `storeseo_list_rank_languages`, `storeseo_add_keywords`, `storeseo_list_keywords`, `storeseo_get_keyword_history`, `storeseo_list_competitors`, `storeseo_add_competitors`, `storeseo_delete_keyword`, `storeseo_delete_competitor`, `storeseo_generate_keyword_cluster`, `storeseo_get_keyword_cluster` Guides: [Claude](https://storeseo.com/agent/claude/keyword-tracking), [ChatGPT](https://storeseo.com/agent/chatgpt/keyword-tracking), [Claude Code](https://storeseo.com/agent/claude-code/keyword-tracking), [Claude Cowork](https://storeseo.com/agent/claude-cowork/keyword-tracking), [Cursor](https://storeseo.com/agent/cursor/keyword-tracking), [Codex](https://storeseo.com/agent/codex/keyword-tracking), [Gemini CLI](https://storeseo.com/agent/gemini-cli/keyword-tracking), [GitHub Copilot](https://storeseo.com/agent/github-copilot/keyword-tracking), [Windsurf](https://storeseo.com/agent/windsurf/keyword-tracking), [Grok](https://storeseo.com/agent/grok/keyword-tracking), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/keyword-tracking), [Grok Bot](https://storeseo.com/agent/grok-bot/keyword-tracking), [Muse](https://storeseo.com/agent/muse/keyword-tracking), [Manus](https://storeseo.com/agent/manus/keyword-tracking), [OpenClaw](https://storeseo.com/agent/openclaw/keyword-tracking), [Hermes Agent](https://storeseo.com/agent/hermes-agent/keyword-tracking), [Paperclip](https://storeseo.com/agent/paperclip/keyword-tracking), [NanoClaw](https://storeseo.com/agent/nanoclaw/keyword-tracking), [OpenCode](https://storeseo.com/agent/opencode/keyword-tracking), [Zed](https://storeseo.com/agent/zed/keyword-tracking), [Kiro](https://storeseo.com/agent/kiro/keyword-tracking), [Antigravity](https://storeseo.com/agent/antigravity/keyword-tracking), [Kimi Code](https://storeseo.com/agent/kimi/keyword-tracking) ### Search Console and Analytics reports Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. Tools: `storeseo_get_search_console_insights`, `storeseo_get_analytics_report` Guides: [Claude](https://storeseo.com/agent/claude/google-reports), [ChatGPT](https://storeseo.com/agent/chatgpt/google-reports), [Claude Code](https://storeseo.com/agent/claude-code/google-reports), [Claude Cowork](https://storeseo.com/agent/claude-cowork/google-reports), [Cursor](https://storeseo.com/agent/cursor/google-reports), [Codex](https://storeseo.com/agent/codex/google-reports), [Gemini CLI](https://storeseo.com/agent/gemini-cli/google-reports), [GitHub Copilot](https://storeseo.com/agent/github-copilot/google-reports), [Windsurf](https://storeseo.com/agent/windsurf/google-reports), [Grok](https://storeseo.com/agent/grok/google-reports), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/google-reports), [Grok Bot](https://storeseo.com/agent/grok-bot/google-reports), [Muse](https://storeseo.com/agent/muse/google-reports), [Manus](https://storeseo.com/agent/manus/google-reports), [OpenClaw](https://storeseo.com/agent/openclaw/google-reports), [Hermes Agent](https://storeseo.com/agent/hermes-agent/google-reports), [Paperclip](https://storeseo.com/agent/paperclip/google-reports), [NanoClaw](https://storeseo.com/agent/nanoclaw/google-reports), [OpenCode](https://storeseo.com/agent/opencode/google-reports), [Zed](https://storeseo.com/agent/zed/google-reports), [Kiro](https://storeseo.com/agent/kiro/google-reports), [Antigravity](https://storeseo.com/agent/antigravity/google-reports), [Kimi Code](https://storeseo.com/agent/kimi/google-reports) ### Blog post drafts Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. Tools: `storeseo_list_blogs`, `storeseo_generate_blog_post`, `storeseo_get_blog_post_job`, `storeseo_get_resource_seo` Guides: [Claude](https://storeseo.com/agent/claude/blog-posts), [ChatGPT](https://storeseo.com/agent/chatgpt/blog-posts), [Claude Code](https://storeseo.com/agent/claude-code/blog-posts), [Claude Cowork](https://storeseo.com/agent/claude-cowork/blog-posts), [Cursor](https://storeseo.com/agent/cursor/blog-posts), [Codex](https://storeseo.com/agent/codex/blog-posts), [Gemini CLI](https://storeseo.com/agent/gemini-cli/blog-posts), [GitHub Copilot](https://storeseo.com/agent/github-copilot/blog-posts), [Windsurf](https://storeseo.com/agent/windsurf/blog-posts), [Grok](https://storeseo.com/agent/grok/blog-posts), [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/blog-posts), [Grok Bot](https://storeseo.com/agent/grok-bot/blog-posts), [Muse](https://storeseo.com/agent/muse/blog-posts), [Manus](https://storeseo.com/agent/manus/blog-posts), [OpenClaw](https://storeseo.com/agent/openclaw/blog-posts), [Hermes Agent](https://storeseo.com/agent/hermes-agent/blog-posts), [Paperclip](https://storeseo.com/agent/paperclip/blog-posts), [NanoClaw](https://storeseo.com/agent/nanoclaw/blog-posts), [OpenCode](https://storeseo.com/agent/opencode/blog-posts), [Zed](https://storeseo.com/agent/zed/blog-posts), [Kiro](https://storeseo.com/agent/kiro/blog-posts), [Antigravity](https://storeseo.com/agent/antigravity/blog-posts), [Kimi Code](https://storeseo.com/agent/kimi/blog-posts) ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## What runs without asking? Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Plans and credit - The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - Keyword rank tracking and competitor tracking are Pro plan tools. - Search Console and Google Analytics reports need a paid plan with Google connected in the app. - Image compression needs the Image Optimizer add-on and counts against its monthly quota. - Generated content (meta, FAQs, AI snippets, alt text, file names, descriptions, keyword clusters, llms.txt, agents.md and blog drafts) spends AI Content Optimizer credit. Current plans: https://storeseo.com/#pricing ## Frequently asked questions ### What is the StoreSEO MCP server? It is a remote Model Context Protocol server at https://mcp.storeseo.com/mcp. An AI agent connected to it can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, image alt text, schema, sitemap and discovery-file changes through 44 tools. ### Which AI agents can connect to StoreSEO? Any client that can add a remote MCP server over streamable HTTP. This hub has setup guides for Claude, ChatGPT, Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot in VS Code and more, each checked against the vendor's documentation. ### Do I need an API key to connect? No. Clients that support OAuth open a browser window where you pick your Shopify store and approve access, and StoreSEO issues that client its own credentials. A personal access token from Settings, AI Toolkit is the alternative for clients that cannot complete a sign-in. ### Will the agent change my store without asking? Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ### Which StoreSEO plan do I need for the MCP server? Any active plan works, including the free plan. Keyword rank tracking is Pro only, Google reports need a paid plan with Google connected, image compression needs the Image Optimizer add-on, and generated content spends AI Content Optimizer credit. See the pricing table for the current plans. ### Can the agent write my product descriptions? Yes, product by product. StoreSEO first scores the description for AI shopping agents and asks you a few questions, then rewrites it from your answers and saves it straight away. There is no undo, so the response carries the only copy of the old text. ### Does StoreSEO publish anything on my storefront? Only when you ask. Generating llms.txt writes to your live theme and Shopify Files, generating agents.md publishes it live immediately, and the HTML sitemap is created in your store. Blog posts are only ever saved as drafts for you to publish from Shopify admin. ### How do I disconnect an agent? Remove the server in your client, and revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. --- Canonical: https://storeseo.com/agent Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Claude > Add StoreSEO to Claude on the web or in Claude Desktop as a custom connector, sign in through Shopify, then audit products and fix meta and alt text by asking. Claude is Anthropic's AI assistant, available on the web and in the Claude Desktop app. Connected to StoreSEO as a custom connector, it reads your store's SEO scores, previews better meta and alt text, and applies the changes you say yes to, all from a chat window. - One custom connector, no config file - Sign in through your Shopify admin - Works on the Free plan with one custom connector ## How do I connect Claude to StoreSEO? ### 1. Add StoreSEO as a custom connector In Claude, open Customize, then Connectors, choose + Add and then Add custom connector. Name it StoreSEO and paste this URL as the remote MCP server URL. ```text https://mcp.storeseo.com/mcp ``` ### 2. Approve your store Claude opens the StoreSEO sign-in in your browser. Pick your Shopify store and approve access. Then start a new chat and ask this to confirm the connection works. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or add it as an owner on Team or Enterprise (optional) Only if you are on a Team or Enterprise plan: an owner adds the same URL under Organization settings, then Connectors. Members then connect their own store from Customize, then Connectors. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** You sign in when you add the connector: Claude opens a browser window where you pick your Shopify store and approve access. StoreSEO supports dynamic client registration, so there is no token to paste. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-claude ### Example session (illustrative) **You:** Which of my products have no alt text on their images? → `storeseo_start` (read-only) → `storeseo_list_images` (read-only) **Agent:** 14 images across 9 products have no alt text. I can preview alt text for the 9 products first; the previews use AI credit but change nothing. Go ahead? **You:** Yes, preview them. → `storeseo_preview_resource_optimization` (preview) **Agent:** Drafts are ready, for example 'Navy linen shirt, front view, on a white background'. Apply all of them to the store? **You:** Yes, apply them. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. Those products re-scored and the next fix on the list is meta descriptions. ## What is Claude? Claude is Anthropic's AI assistant. You use it in a browser at claude.ai, in the Claude Desktop app and in mobile apps, and it can answer questions, write and analyze, and take actions through connected tools. Custom connectors let you add a remote Model Context Protocol server by its URL. You add one under Customize, then Connectors, and Anthropic's documentation notes that the connection to the server originates from Anthropic's servers, not from your computer, even in Claude Desktop. The server therefore has to be reachable on the public internet. On the Free plan you can add one custom connector. On Team and Enterprise plans an owner adds the connector for the organization first, and each member then signs in to it from their own Connectors settings. ## Why use Claude with StoreSEO? - **No terminal, no config.** You paste one URL, approve your store once and the StoreSEO tools appear in the chat. That suits store owners who would rather ask than open an admin page. - **Good for check-ups.** A chat is a comfortable place to ask what is wrong with your store's SEO and what to fix first. Reads such as scores and reports run straight away, so the answers come from your real store data. - **Control over which tools run.** Claude lets you switch tools off from its Search and Tools menu, and StoreSEO tells the agent to preview and ask before it writes anything, so you can keep a conversation to reads only. ## Good to know - The Free plan allows one custom connector, so StoreSEO takes that slot. Pro, Max, Team and Enterprise plans allow more. - On Team and Enterprise, members cannot add a custom connector themselves: an owner adds it under Organization settings, Connectors, and members then authenticate from Customize, Connectors. - Claude's own prompts for write actions are the hard stop. Anthropic advises using Allow always only for servers you trust, and for StoreSEO it is safer to keep the apply, optimize, generate and delete tools on ask. - Start each conversation by checking that the connector is switched on for it in the Search and Tools menu, and turn off tools you do not need for that chat. ## What can you ask Claude to do in StoreSEO? - Use StoreSEO to find my five lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for the ones on my best-selling collection. - Is my llms.txt live? If not, tell me what generating one will do before you run it. - Which Search Console queries bring impressions to my store but few clicks in the last 28 days? - Check whether product schema is switched on for my store and tell me what it covers. ## Claude guides by job - [Audit your store's SEO with Claude](https://storeseo.com/agent/claude/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Claude](https://storeseo.com/agent/claude/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Claude](https://storeseo.com/agent/claude/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Claude](https://storeseo.com/agent/claude/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Claude](https://storeseo.com/agent/claude/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Claude](https://storeseo.com/agent/claude/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Claude](https://storeseo.com/agent/claude/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Claude](https://storeseo.com/agent/claude/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Claude](https://storeseo.com/agent/claude/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Claude](https://storeseo.com/agent/claude/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Claude to StoreSEO? Open Customize, then Connectors in Claude, choose Add custom connector and paste https://mcp.storeseo.com/mcp. Claude opens a browser window where you pick your Shopify store and approve access, and then the StoreSEO tools are available in your chats. ### Do I need a StoreSEO API key for Claude? No. Claude signs in with OAuth when you add the connector and StoreSEO issues it its own credentials. Nothing needs to be pasted. ### Will Claude change my store without asking? StoreSEO tells Claude to preview first, say what a generation will cost, and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server, so Claude's own tool-approval prompts are the hard stop. Reads such as scores and reports run straight away. ### Which StoreSEO plan do I need to use it with Claude? Any active StoreSEO plan works, including the free plan. Keyword rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can I use StoreSEO in Claude on the Free plan? Anthropic's documentation says Free users are limited to one custom connector, so you can add StoreSEO as that one connector. Check Anthropic's current plan page if you want more than one. ### I am on a Claude Team or Enterprise plan. Who adds the connector? An owner adds it under Organization settings, then Connectors, with the same URL. Members then sign in to it from their own Customize, then Connectors settings and pick their own Shopify store. ### Does it work in Claude Desktop as well as on the web? Yes. Anthropic's documentation says custom connectors work on Claude on the web, Claude Desktop and Cowork, and that the connection to the server is made from Anthropic's servers. ### How do I disconnect Claude from my store? Remove the StoreSEO connector from Customize, then Connectors in Claude, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Cowork + StoreSEO](https://storeseo.com/agent/claude-cowork) - [Claude Code + StoreSEO](https://storeseo.com/agent/claude-code) Setup checked against: - [Anthropic: custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) --- Canonical: https://storeseo.com/agent/claude Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from ChatGPT > Connect ChatGPT to StoreSEO with a developer-mode app that points at the StoreSEO MCP URL, then audit your Shopify store and approve each fix in chat. ChatGPT is OpenAI's AI assistant, and in developer mode it can use tools from a remote MCP server. Connected to StoreSEO, it reads your store's SEO scores, previews better meta and alt text, and applies what you approve from a chat on the web. - No terminal, no config file - OAuth sign-in through your Shopify admin - Write actions ask for confirmation by default ## How do I connect ChatGPT to StoreSEO? ### 1. Turn on developer mode In ChatGPT on the web, open Settings, then Security and login, and turn on Developer mode. If your menu shows Apps and Connectors with Advanced settings instead, the same switch lives there. OpenAI lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. ```text https://chatgpt.com/plugins ``` ### 2. Create the app with the StoreSEO MCP URL Open Plugins and select the plus button to create a developer-mode app. Name it StoreSEO, paste this as the MCP server URL and choose OAuth. Sign in when the browser asks: pick your Shopify store and approve access. ```text https://mcp.storeseo.com/mcp ``` ### 3. Use it in a chat The app appears in the composer's Developer mode tool during conversations. Select StoreSEO there, then ask this to confirm the connection works. ```text Using the StoreSEO app, show me my store's SEO score breakdown. ``` ### 4. Or ask a workspace admin to publish it (optional) On Business, Enterprise and Edu workspaces OpenAI's help center puts the controls under Settings, Apps, so an admin or owner may need to create and publish the app for you. Give them the same URL. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** You sign in with StoreSEO in the browser: pick your Shopify store and approve access. ChatGPT can register itself through dynamic client registration, so there is no token to create or paste. Verify with: "Using the StoreSEO app, show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-chatgpt ### Example session (illustrative) **You:** Using the StoreSEO app, which of my collections have the weakest SEO? → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) **Agent:** Your Summer Linen collection scores lowest, mostly because of a missing meta description. I can preview a new one; the preview uses AI credit but writes nothing. Shall I? **You:** Yes, preview it. → `storeseo_preview_resource_optimization` (preview) **Agent:** Here is the draft meta description and title. Apply it to the collection? **You:** Yes, apply it. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. The collection re-scored higher; the next item is the missing alt text on its images. ## What is ChatGPT? ChatGPT is OpenAI's AI assistant. You talk to it in a chat on the web or in an app, and it can answer questions, write and analyze, and take actions through connected tools. Developer mode lets ChatGPT use a remote Model Context Protocol server that you add yourself, over streaming HTTP or SSE. You turn it on under Settings, then Security and login, create an app from the Plugins page with the server's URL, and pick an authentication method. OAuth can use Client ID Metadata Documents or dynamic client registration. OpenAI's documentation says write actions require approval by default, that you can inspect the payload before they run, and that you can choose to remember the approval for the rest of the conversation. It respects the readOnlyHint annotation to tell reads from writes. ## Why use ChatGPT with StoreSEO? - **Nothing to install.** You paste one URL into ChatGPT, sign in with StoreSEO and the tools appear in your chat. There is no terminal, config file or token to manage. - **Good for checks and fixes.** A chat is a comfortable place to ask what is weak and what to fix first. Reads run straight away, so the answers come from your real store data. - **Two confirmations on changes.** StoreSEO tells the agent to preview and wait for your yes, and ChatGPT asks you to approve write actions and shows the payload first, so you can read exactly what is about to be sent. ## Good to know - ChatGPT connects from OpenAI's servers, not from your computer. The StoreSEO MCP URL is public, so there is nothing to open on your side. - Write actions depend on your ChatGPT plan. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web, while OpenAI's help center, as we read it, puts full MCP support with write actions in beta on Business, Enterprise and Edu workspaces and gives Pro accounts read access only. On a plan without write actions, use ChatGPT for audits and reports and apply fixes from the StoreSEO app or another agent. - OpenAI marks developer mode as higher risk because a connected server can read and, with write actions, change your data. Read each payload before you approve, and be careful with the option that remembers approval for the rest of a conversation. - If you do not see developer mode under Settings, Security and login, check your plan and, in a workspace, ask an admin whether it is enabled. ## What can you ask ChatGPT to do in StoreSEO? - Using the StoreSEO app, what is my store's SEO score and what are the three biggest problems? - Which products have meta descriptions that are too short? Preview new ones for the first five. - Show me the images with no alt text on my best-selling collection. - Is my store ready for AI shopping assistants? Check my llms.txt status and AI settings. - Show my Search Console clicks and impressions for the last 28 days. ## ChatGPT guides by job - [Audit your store's SEO with ChatGPT](https://storeseo.com/agent/chatgpt/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with ChatGPT](https://storeseo.com/agent/chatgpt/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with ChatGPT](https://storeseo.com/agent/chatgpt/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with ChatGPT](https://storeseo.com/agent/chatgpt/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with ChatGPT](https://storeseo.com/agent/chatgpt/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with ChatGPT](https://storeseo.com/agent/chatgpt/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with ChatGPT](https://storeseo.com/agent/chatgpt/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with ChatGPT](https://storeseo.com/agent/chatgpt/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with ChatGPT](https://storeseo.com/agent/chatgpt/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with ChatGPT](https://storeseo.com/agent/chatgpt/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect ChatGPT to StoreSEO? Turn on Developer mode under Settings, then Security and login, open Plugins and use the plus button to create an app. Paste https://mcp.storeseo.com/mcp as the MCP server URL, choose OAuth and sign in. Pick your Shopify store and approve access, then select the StoreSEO app in the composer. ### Do I need an API key to connect ChatGPT? No. ChatGPT signs in with OAuth and StoreSEO issues it its own credentials. A personal access token from Settings, AI Toolkit is only needed for clients that cannot complete a sign-in. ### Will ChatGPT change my store without asking? StoreSEO tells the agent to preview first, say what a generation will cost, and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server. ChatGPT's own write confirmation, which OpenAI says is on by default, is the hard stop. Reads run straight away. ### Which ChatGPT plans can use StoreSEO? OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. OpenAI's help center describes full MCP support with write actions as a beta on Business, Enterprise and Edu workspaces and Pro accounts as read access, so check OpenAI's current page before you rely on write actions. ### Which StoreSEO plan do I need? Any active StoreSEO plan works, including the free plan. Keyword rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### I am in a ChatGPT workspace. Can I add the app myself? Not necessarily. OpenAI's help center describes workspace controls under Settings, Apps, where an admin or owner creates and publishes the app. Ask them to add https://mcp.storeseo.com/mcp, then use it from your own chat. ### Can ChatGPT apply changes in my Pro account? OpenAI's help center, as we read it, says Pro accounts connect MCP servers with read access in developer mode. That is enough for audits and reports. Applying content needs a plan with write actions, or you can apply fixes from the StoreSEO app. ### How do I disconnect ChatGPT from my store? Delete the StoreSEO app in ChatGPT, then revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude + StoreSEO](https://storeseo.com/agent/claude) Setup checked against: - [OpenAI: developer mode in ChatGPT](https://developers.openai.com/api/docs/guides/developer-mode) - [OpenAI help center: developer mode and full MCP connectors](https://help.openai.com/en/articles/12584461-developer-mode-apps-and-full-mcp-connectors-in-chatgpt-beta) --- Canonical: https://storeseo.com/agent/chatgpt Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Claude Code > Add StoreSEO to Claude Code with one command, sign in through Shopify, then audit products, fix meta and alt text and publish llms.txt from the terminal. Claude Code is Anthropic's agentic coding tool that runs in your terminal and can call external tools through the Model Context Protocol. Connected to StoreSEO, it reads your store's SEO scores, previews optimized meta and images, and applies the changes you approve without opening the Shopify admin. - One command: claude mcp add - OAuth through your Shopify admin, or a token header - Batch work across hundreds of products in one session ## How do I connect Claude Code to StoreSEO? ### 1. Add StoreSEO to Claude Code Run this in any terminal. The user scope makes StoreSEO available in every project on your machine. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp ``` ### 2. Sign in to your store Start Claude Code, run /mcp, choose storeseo and follow the browser window: pick your Shopify store and approve access. ```text /mcp ``` ### 3. Or connect with an access token (optional) If you cannot use a browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add the server with the token as a header instead. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp \ --header "Authorization: Bearer YOUR_TOKEN" ``` **Sign-in:** Claude Code signs you in with OAuth: run /mcp, pick your Shopify store in the browser and approve. Claude Code stores the credentials it receives and refreshes them. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it with --header. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-claude-code ### Example session (illustrative) **You:** Find my five lowest-scoring products and preview better meta for each. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted new meta for all five. The biggest jump is Linen Shirt, projected 41 to 78. Previews used AI credit; applying is free. Apply all five? **You:** Apply them, but keep the current URL handles. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied without the handles. All five re-scored; the next fix is alt text on 14 images. Want me to do those? ## What is Claude Code? Claude Code is Anthropic's agentic coding tool. It runs in your terminal, and in IDE extensions and a desktop app, reads the project in the folder you start it in, and works through multi-step tasks by calling tools and showing each call as it happens. It connects to remote MCP servers over HTTP. You add a server once with claude mcp add, and the user scope stores it in ~/.claude.json so it is available in every project. Remote servers that need a sign-in are authenticated from the /mcp menu, where you can also clear the credentials later. That makes it a natural fit for developers and agencies who already work on a Shopify theme in the terminal. The same session that edits Liquid can audit the store's SEO, apply StoreSEO's optimized content and report the new scores. ## Why use Claude Code with StoreSEO? - **Work through the whole catalog.** Ask for every product under a score and Claude Code pages through the listing, previews each one and gives you one combined summary to approve, instead of you opening products one by one. - **Theme code and SEO in one place.** When a fix needs a theme change, such as where FAQs render or how a schema block is placed, the same session can open the Liquid file next to the StoreSEO results. - **You see every call.** Claude Code prints each StoreSEO tool it runs, so you can tell a read from a preview from an apply, and stop a run that heads somewhere you did not ask for. ## Good to know - Without --scope user the server is added to the current project only, so it disappears when you start Claude Code in another folder. - Claude Code asks before a tool runs unless you have allowed it. If you add an allow rule for the StoreSEO tools, the confirmation StoreSEO asks for in the chat becomes the only check, so keep apply, write and generate tools on ask. - Claude Code warns when one tool result is larger than 10,000 tokens. Ask for smaller pages of products or images on a big catalog rather than raising the limit. ## What can you ask Claude Code to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Write it for the ones on my top collection, one image at a time. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Turn on product schema and tell me how many products that fixes. ## Claude Code guides by job - [Audit your store's SEO with Claude Code](https://storeseo.com/agent/claude-code/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Claude Code](https://storeseo.com/agent/claude-code/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Claude Code](https://storeseo.com/agent/claude-code/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Claude Code](https://storeseo.com/agent/claude-code/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Claude Code](https://storeseo.com/agent/claude-code/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Claude Code](https://storeseo.com/agent/claude-code/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Claude Code](https://storeseo.com/agent/claude-code/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Claude Code](https://storeseo.com/agent/claude-code/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Claude Code](https://storeseo.com/agent/claude-code/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Claude Code](https://storeseo.com/agent/claude-code/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Claude Code to StoreSEO? Run claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp, start Claude Code, run /mcp and sign in. The browser asks you to pick your Shopify store and approve access, and then Claude Code can call the StoreSEO tools. ### Do I need a StoreSEO API key for Claude Code? No. Claude Code signs in with OAuth and keeps the credentials it receives. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, added with the --header option. ### Where does Claude Code store the StoreSEO server? With --scope user it is stored in ~/.claude.json and available in every project on your machine. Without that option it is added to the current project only. ### Will Claude Code change my products without asking? StoreSEO tells Claude Code to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and Claude Code also asks before running a tool unless you have allowed it. Reads such as scores and reports run straight away. Keep the write tools on ask if you want both checks. ### Can Claude Code optimize my whole catalog? Yes, in batches. It lists products by score, previews several at once and asks you to approve them together. Each preview spends AI credit, so start with the products that need it most. ### Which StoreSEO plan does Claude Code need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. ### How do I disconnect Claude Code from my store? Run claude mcp remove storeseo, or clear the authentication in the /mcp menu. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) Setup checked against: - [Claude Code: connect to tools via MCP](https://code.claude.com/docs/en/mcp) --- Canonical: https://storeseo.com/agent/claude-code Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Delegate Your Shopify SEO to Claude Cowork > Add the StoreSEO connector to Claude Cowork, then hand off store SEO audits and fixes, or schedule a report-only check that runs while you are away. Claude Cowork is Anthropic's agent for delegated knowledge work: you describe a task, hand it off and review what comes back. Connected to StoreSEO, it can audit your store's SEO, prepare previews of better meta and alt text, and stop to ask before it applies anything. - Same custom connector as Claude - Hand off a whole audit, then review it - Scheduled tasks for recurring report-only checks ## How do I connect Claude Cowork to StoreSEO? ### 1. Add StoreSEO as a custom connector In Cowork, open Customize, then Connectors, choose + Add and then Add custom connector. If you already added StoreSEO to your Claude account, look for it in that list first. Use this URL. ```text https://mcp.storeseo.com/mcp ``` ### 2. Approve your store and check it Sign in when the browser opens: pick your Shopify store and approve access. Then start a task and ask this first, so you know the connector works before you delegate real work. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or add it as an owner on Team or Enterprise (optional) Only if you are on a Team or Enterprise plan: an owner adds the same URL under Organization settings, then Connectors, and members connect their own store from Customize, then Connectors. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** You sign in when you add the connector: a browser window opens where you pick your Shopify store and approve access. StoreSEO supports dynamic client registration, so there is no token to paste. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-claude-cowork ### Example session (illustrative) **You:** Audit my store's SEO and tell me what to fix first. Do not change anything yet. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_get_llms_txt_status` (read-only) **Agent:** Done. 23 products score under 50, mostly for short meta descriptions, and llms.txt has not been generated. The cheapest first step is previewing meta for the lowest ten; previews use AI credit but write nothing. Shall I? **You:** Yes, preview the lowest ten. → `storeseo_preview_resource_optimization` (preview) **Agent:** Previews are ready for your review in this task. Apply any of them? **You:** Yes, apply the first three. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied to three products. The other seven are still only previews. ## What is Claude Cowork? Claude Cowork is Anthropic's agent for delegated knowledge work. Instead of a back-and-forth chat, you describe a task, hand it off and let Claude work through it using your connected apps, plugins and skills, then you review the result. It uses the same custom connectors as Claude. Anthropic's documentation says custom connectors work on Claude, Cowork and Claude Desktop, are added under Customize, then Connectors, and connect from Anthropic's servers rather than from your computer. Cowork can also run scheduled tasks. Anthropic's documentation says they run remotely on their cadence even when your computer is asleep or the desktop app is closed, can use your connected apps, and are available on Pro, Max, Team and Enterprise plans, with an admin toggle for Cowork on Team and Enterprise. ## Why use Claude Cowork with StoreSEO? - **Hand off the whole audit.** An SEO check is many small reads: scores, images, settings, files. Cowork can run them as one task and give you one summary instead of a long chat. - **Recurring checks.** A scheduled task can re-check your store's scores or Search Console data on a cadence and leave you a report, so regressions show up without you remembering to look. - **You approve the writes.** StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and Cowork has its own approval controls on top of that. ## Good to know - A scheduled task runs without you in the conversation, so nobody is there to say yes. Keep scheduled StoreSEO tasks to reads and reports, and tell the task in plain words not to apply, generate, publish or delete anything. - Cowork shares the custom connector model of Claude: the Free plan allows one custom connector, and on Team and Enterprise an owner adds the connector under Organization settings, Connectors. - On Team and Enterprise plans an admin toggle controls access to Cowork, so you may need your admin to turn it on. - If StoreSEO does not appear in Cowork, check Customize, Connectors inside Cowork and add it there rather than assuming a connector added elsewhere carried over. ## What can you ask Claude Cowork to do in StoreSEO? - Use StoreSEO to audit my store's SEO and give me a ranked list of fixes. Do not change anything. - Preview better meta for my ten lowest-scoring products and wait for me to review them before you apply any. - Check which product images have no alt text and summarize the damage by collection. - Pull my Search Console insights for the last 28 days and list queries where I rank between 8 and 20. - Check whether my llms.txt and agents.md are live and tell me what is missing. ## Schedule a weekly SEO check in Cowork Cowork documents scheduled tasks that run on a cadence and can use your connected apps. For StoreSEO, make the task report-only: it should read scores and reports and write you a summary. Do not ask a scheduled task to apply content, generate files or delete anything, because you are not there to say yes. ```text Every Monday, use StoreSEO to read my store's SEO score breakdown, my lowest-scoring products and last week's Search Console insights. Summarize what changed. Do not apply, generate, publish or delete anything. ``` ## Claude Cowork guides by job - [Audit your store's SEO with Claude Cowork](https://storeseo.com/agent/claude-cowork/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Claude Cowork](https://storeseo.com/agent/claude-cowork/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Claude Cowork](https://storeseo.com/agent/claude-cowork/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Claude Cowork](https://storeseo.com/agent/claude-cowork/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Claude Cowork](https://storeseo.com/agent/claude-cowork/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Claude Cowork](https://storeseo.com/agent/claude-cowork/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Claude Cowork](https://storeseo.com/agent/claude-cowork/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Claude Cowork](https://storeseo.com/agent/claude-cowork/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Claude Cowork](https://storeseo.com/agent/claude-cowork/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Claude Cowork](https://storeseo.com/agent/claude-cowork/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Claude Cowork to StoreSEO? In Cowork open Customize, then Connectors, choose Add custom connector and paste https://mcp.storeseo.com/mcp. Sign in when the browser opens, pick your Shopify store and approve access. If you already added StoreSEO to your Claude account, check the connector list first. ### Do I need a StoreSEO API key for Cowork? No. Cowork signs in with OAuth through the custom connector and StoreSEO issues its own credentials. A personal access token from Settings, AI Toolkit is not needed for this client. ### Will Cowork change my store without asking? StoreSEO tells the agent to preview first, say what a generation will cost, and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server, so Cowork's own approval controls are the hard stop. Reads run straight away. ### Can Cowork run a scheduled StoreSEO check? Anthropic documents scheduled tasks in Cowork that run on a cadence and can use connected apps. Keep a scheduled StoreSEO task to reads and reports, because nobody is in the conversation to say yes to a write. ### Which StoreSEO plan do I need? Any active StoreSEO plan works, including the free plan. Keyword rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Which Claude plans can use Cowork scheduled tasks? Anthropic's help center lists Pro, Max, Team and Enterprise plans, with an admin toggle for Cowork on Team and Enterprise. Check Anthropic's page for the current rollout. ### How is Cowork different from Claude chat for StoreSEO? Both use the same custom connector and the same 44 StoreSEO tools. Chat suits quick questions and fixes you review as you go. Cowork suits a whole job you hand off, such as an audit, and scheduled reports. ### How do I disconnect Cowork from my store? Remove the StoreSEO connector from Customize, then Connectors, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude + StoreSEO](https://storeseo.com/agent/claude) Setup checked against: - [Anthropic: custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) - [Anthropic: scheduled tasks in Claude Cowork](https://support.claude.com/en/articles/13854387-schedule-recurring-tasks-in-claude-cowork) --- Canonical: https://storeseo.com/agent/claude-cowork Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Cursor > Add StoreSEO to Cursor with one mcp.json entry, switch off four tools to stay under the 40-tool cap, then audit products and fix SEO beside your theme code. Cursor is an AI code editor from Anysphere whose agent can use external tools through the Model Context Protocol. Connected to StoreSEO, it reads your store's SEO scores, previews optimized meta and schema, and applies what you approve while your Shopify theme is open next to it. - One entry in mcp.json, or paste it into Cursor's MCP settings - OAuth through your Shopify admin, or a Bearer header - Switch off four tools to fit Cursor's 40-tool cap ## How do I connect Cursor to StoreSEO? ### 1. Add StoreSEO to mcp.json Put this in ~/.cursor/mcp.json to use StoreSEO in every project, or in .cursor/mcp.json inside one theme project. In the StoreSEO app's setup guide the same JSON goes into Settings, Tools and MCP, New MCP server. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Sign in to your store Open Cursor's MCP settings (the Customize page in current versions), find storeseo and sign in when Cursor asks. A browser window opens: pick your Shopify store and approve access. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Switch off four StoreSEO tools StoreSEO has 44 tools and Cursor stops at about 40 active ones across all servers. In the storeseo entry in Cursor's MCP settings, turn off at least four you do not need. The four rank and competitor lookups are a good first choice if you are not on Pro. ```text storeseo_list_rank_locations storeseo_list_rank_languages storeseo_add_competitors storeseo_delete_competitor ``` ### 4. Or connect with an access token (optional) If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Cursor can read it from an environment variable so the token stays out of the file. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer ${env:STORESEO_TOKEN}" } } } } ``` **Sign-in:** Cursor supports OAuth for remote MCP servers, so you sign in by picking your Shopify store in the browser and approving. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it as an Authorization header in mcp.json. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-cursor ### Example session (illustrative) **You:** Check which schema types are on for my store and turn on product schema if it is off. → `storeseo_start` (read-only) → `storeseo_get_schema_settings` (read-only) **Agent:** Product schema is off. Turning it on adds structured data for every product. Review and Offer fields come from your settings. Turn it on? **You:** Yes, turn it on. → `storeseo_update_schema_settings` (needs approval) **Agent:** Product schema is now on. Want me to open the theme's product template and check nothing else prints a second Product block? ## What is Cursor? Cursor is an AI code editor made by Anysphere. Its agent reads the project you have open, edits files and calls tools, and shows each tool call in the chat so you can see what ran. It connects to MCP servers over stdio, SSE or streamable HTTP. Servers are listed in mcp.json, either in ~/.cursor/mcp.json for every project or in .cursor/mcp.json inside one, and a remote server needs only a url and, if you use a token, a headers block. Cursor supports OAuth for remote servers, and you install, toggle and remove servers from its Customize page. By default Cursor asks for approval before an MCP tool runs, and its Run Modes decide which tools may run on their own. Teams and Enterprise admins can also allowlist which MCP servers people may use. ## Why use Cursor with StoreSEO? - **SEO next to the theme code.** When a fix touches Liquid, such as where a schema block prints or how FAQs render, the same Cursor session can open the template while StoreSEO shows the scores for that page. - **Batch work in one chat.** Ask for every product under a score and Cursor pages through the listing, previews each one and gives you one summary to approve, instead of you opening products one by one in the Shopify admin. - **Two checks on every write.** StoreSEO tells the agent to preview and wait for your yes, and Cursor asks before a tool runs unless you have allowed it, so a write has to get past both. ## Good to know - StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers. Turn off at least four StoreSEO tools in Cursor's MCP settings, and more if you run other MCP servers. - A server added to .cursor/mcp.json belongs to that project only. Use ~/.cursor/mcp.json if you want StoreSEO in every project. - If you set a StoreSEO tool to run without asking, the confirmation StoreSEO asks for in chat is the only check left. Keep the apply, generate, publish and delete tools on ask. - Admins on Teams and Enterprise plans can restrict which MCP servers run, so a server you add may be blocked until an admin allows its URL. ## What can you ask Cursor to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest ones first. - Check my product schema settings and tell me what turning on Review and Sale dates would add. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. ## Cursor guides by job - [Audit your store's SEO with Cursor](https://storeseo.com/agent/cursor/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Cursor](https://storeseo.com/agent/cursor/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Cursor](https://storeseo.com/agent/cursor/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Cursor](https://storeseo.com/agent/cursor/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Cursor](https://storeseo.com/agent/cursor/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Cursor](https://storeseo.com/agent/cursor/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Cursor](https://storeseo.com/agent/cursor/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Cursor](https://storeseo.com/agent/cursor/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Cursor](https://storeseo.com/agent/cursor/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Cursor](https://storeseo.com/agent/cursor/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Cursor to StoreSEO? Add a storeseo entry with the url https://mcp.storeseo.com/mcp to ~/.cursor/mcp.json, or paste the same JSON into Cursor's MCP settings. Then sign in when Cursor asks: your browser opens, you pick your Shopify store and approve access. ### Why do I have to turn off StoreSEO tools in Cursor? StoreSEO exposes 44 tools and Cursor allows about 40 active tools across all your MCP servers. Switch off at least four StoreSEO tools you do not need in Cursor's MCP settings. The rank and competitor lookups are a good first choice if you are not on Pro. ### Do I need a StoreSEO API key for Cursor? No. Cursor signs in with OAuth and keeps the credentials. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, sent as an Authorization header in mcp.json. ### Will Cursor change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and that yes is an instruction the agent follows, not a lock on the server. Cursor also asks for approval before an MCP tool runs by default, and that setting is the hard stop. Reads such as scores and reports run straight away. ### Which StoreSEO plan does Cursor need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. See the pricing table for current plans. ### Should I use the global or the project mcp.json? Use ~/.cursor/mcp.json when you want StoreSEO available in every project. Use .cursor/mcp.json when only one theme repository should see it, for example on a shared machine or in a repo you hand to a client. ### How do I disconnect Cursor from my store? Remove or toggle off the storeseo server in Cursor's MCP settings, or delete its entry from mcp.json. Then revoke the connection or the access token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [Cursor: Model Context Protocol (MCP)](https://cursor.com/docs/mcp) - [Anysphere](https://anysphere.inc/) --- Canonical: https://storeseo.com/agent/cursor Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Codex > Connect OpenAI Codex to StoreSEO with codex mcp add and codex mcp login, then audit products, fix meta and alt text and publish llms.txt from your terminal. Codex is OpenAI's coding agent, which runs in your terminal, in an IDE extension and in the ChatGPT desktop app. Connected to StoreSEO, it reads your store's SEO scores, previews optimized content and applies what you approve from the session where you already work on your theme. - codex mcp add, then codex mcp login - One config.toml shared by the CLI, the IDE and the desktop app - Per-tool approval modes you can set from config.toml ## How do I connect Codex to StoreSEO? ### 1. Add StoreSEO to Codex Run this in any terminal. It writes a [mcp_servers.storeseo] table to ~/.codex/config.toml, which the Codex CLI, the IDE extension and the ChatGPT desktop app all read. ```bash codex mcp add storeseo --url https://mcp.storeseo.com/mcp ``` ### 2. Sign in to your store Run the login command. Your browser opens: pick your Shopify store and approve access. Then run codex mcp list, or type /mcp inside Codex, to confirm storeseo is there. ```bash codex mcp login storeseo codex mcp list ``` ### 3. Or write it in config.toml (optional) If you would rather edit the file, add this table to ~/.codex/config.toml, or to .codex/config.toml in a trusted project. In the desktop app or the IDE extension you can add the same URL under MCP servers, Add server, Streamable HTTP. ```toml [mcp_servers.storeseo] url = "https://mcp.storeseo.com/mcp" ``` ### 4. Or connect with an access token (optional) Generate a token in StoreSEO under Settings, AI Toolkit, export it as STORESEO_TOKEN in your shell, and point config.toml at that variable. Codex sends it as an Authorization Bearer header, so the token never sits in the file. ```toml [mcp_servers.storeseo] url = "https://mcp.storeseo.com/mcp" bearer_token_env_var = "STORESEO_TOKEN" ``` **Sign-in:** Codex signs in with OAuth: run codex mcp login storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and let config.toml read it from an environment variable. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-codex ### Example session (illustrative) **You:** Is my llms.txt live and up to date? If it needs a rebuild, tell me what that does first. → `storeseo_start` (read-only) → `storeseo_get_llms_txt_status` (read-only) **Agent:** Your settings changed since the last generation, so a rebuild is due. Generating writes llms.txt into your live theme and Shopify Files and spends AI credit. Run it? **You:** Yes, run it. → `storeseo_generate_llms_txt` (needs approval) → `storeseo_get_llms_txt_status` (read-only) **Agent:** The new version is generated and live at your /llms.txt address. Want me to check agents.md next? ## What is Codex? Codex is OpenAI's coding agent. It runs as a command-line tool, as an IDE extension and inside the ChatGPT desktop app, reads the repository you start it in and works through multi-step tasks by calling tools. It stores MCP servers in config.toml, by default ~/.codex/config.toml, with one [mcp_servers.name] table per server, and the CLI, the IDE extension and the desktop app share that file. A remote server needs a url, and Codex supports OAuth with dynamic client registration, which you start with codex mcp login, or a bearer token read from an environment variable. Each server can carry its own approval settings: a default mode of auto, prompt, writes or approve, tool allow and deny lists, and per-tool overrides. The writes mode prompts for any tool that is not marked read-only. ## Why use Codex with StoreSEO? - **One setup, three places.** Add StoreSEO once and the terminal, the IDE extension and the ChatGPT desktop app all see it, because they read the same config.toml. - **Approvals that match the tools.** StoreSEO marks its read tools read-only, so setting the server's approval mode to writes lets reads run and stops for every tool that can change your store. - **SEO and theme in one task.** Ask Codex to audit the store and then fix the template that causes the problem, so the finding and the change sit in the same session. ## Good to know - The Codex docs show codex mcp add for local commands, and the --url form appears in the OAuth section. If your version rejects it, use the config.toml table instead. - A project-level .codex/config.toml only loads in trusted projects. Use ~/.codex/config.toml for StoreSEO in every project. - Set the server's default_tools_approval_mode to writes or prompt, not approve. Approve skips the prompt, and then the confirmation StoreSEO asks for in chat is the only check. - Codex waits 60 seconds for a tool to finish by default. Content generation can take longer, so raise tool_timeout_sec for the storeseo server if a generation times out. ## What can you ask Codex to do in StoreSEO? - Use StoreSEO to list products scoring under 50 and preview better meta for the first five. - Is my agents.md live? If it is the Shopify default, tell me what generating a custom one will publish. - Which images have no alt text? List the ten heaviest first. - Show me my Search Console queries for the last 28 days, grouped by page. - Check which schema types are enabled and tell me what Product schema would add. ## Codex guides by job - [Audit your store's SEO with Codex](https://storeseo.com/agent/codex/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Codex](https://storeseo.com/agent/codex/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Codex](https://storeseo.com/agent/codex/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Codex](https://storeseo.com/agent/codex/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Codex](https://storeseo.com/agent/codex/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Codex](https://storeseo.com/agent/codex/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Codex](https://storeseo.com/agent/codex/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Codex](https://storeseo.com/agent/codex/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Codex](https://storeseo.com/agent/codex/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Codex](https://storeseo.com/agent/codex/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Codex to StoreSEO? Run codex mcp add storeseo --url https://mcp.storeseo.com/mcp, then codex mcp login storeseo. Your browser opens, you pick your Shopify store and approve access, and codex mcp list should show storeseo. ### Do I need a StoreSEO API key for Codex? No. Codex signs in with OAuth and keeps the credentials. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative: put it in an environment variable and set bearer_token_env_var in config.toml. ### Where does Codex store the StoreSEO server? In ~/.codex/config.toml, as an mcp_servers.storeseo table. The Codex CLI, the IDE extension and the ChatGPT desktop app share that file. A project can also carry its own .codex/config.toml if the project is trusted. ### Will Codex change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and that yes is an instruction the agent follows, not a lock on the server. Codex's own approval mode for the server is the hard stop: set it to writes or prompt so tools that can change your store ask first. Reads run straight away. ### Which StoreSEO plan does Codex need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. See the pricing table for current plans. ### Does the ChatGPT desktop app work too? Yes. The desktop app shares Codex's MCP configuration. Add StoreSEO under Settings, MCP servers, choose Streamable HTTP, paste the URL and select Authenticate. If you already ran codex mcp add, it is already there. ### How do I disconnect Codex from my store? Delete the mcp_servers.storeseo table from config.toml, or set enabled to false to keep it but switch it off. Then revoke the connection or the access token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [OpenAI Codex: Model Context Protocol](https://developers.openai.com/codex/mcp) --- Canonical: https://storeseo.com/agent/codex Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Gemini CLI > Connect Gemini CLI to StoreSEO with one gemini mcp add command, sign in through Shopify, then audit products, fix meta and publish llms.txt from the terminal. Gemini CLI is Google's command-line AI agent, and it can call external tools through the Model Context Protocol. Connected to StoreSEO, it reads your store's SEO scores, previews optimized meta and applies the changes you approve from the terminal you already work in. - One command: gemini mcp add - OAuth discovered automatically, no client ID to paste - Per-server tool filters and confirmations built in ## How do I connect Gemini CLI to StoreSEO? ### 1. Add StoreSEO to Gemini CLI Run this in any terminal. The user scope writes the server to ~/.gemini/settings.json so every project sees it. Without --scope user the command writes to the current project's .gemini/settings.json. ```bash gemini mcp add --scope user --transport http storeseo https://mcp.storeseo.com/mcp ``` ### 2. Sign in to your store Start gemini. The first time it calls StoreSEO it opens your browser: pick your Shopify store and approve access. If the window does not open, type /mcp auth storeseo inside Gemini CLI. Then gemini mcp list should show storeseo as connected. ```text /mcp auth storeseo ``` ### 3. Or edit settings.json (optional) Add this to ~/.gemini/settings.json for every project, or to .gemini/settings.json in one project. It is the entry gemini mcp add writes: url plus type set to http. The docs page still shows the older httpUrl key for streamable HTTP. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "type": "http" } } } ``` ### 4. Or connect with an access token (optional) On a headless machine the browser sign-in cannot run. Generate a token in StoreSEO under Settings, AI Toolkit and pass it as a header. Keep the token out of version control. ```bash gemini mcp add --scope user --transport http \ --header "Authorization: Bearer YOUR_TOKEN" \ storeseo https://mcp.storeseo.com/mcp ``` **Sign-in:** Gemini CLI finds the StoreSEO sign-in on its own: the first time it calls the server it opens your browser, you pick your Shopify store and approve, and it stores the tokens. On a machine with no browser, add a personal access token from StoreSEO under Settings, AI Toolkit as a header. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-gemini-cli ### Example session (illustrative) **You:** Find my three lowest-scoring products and show me better meta for each. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted new meta, tags and FAQs for all three. Previews used AI credit and nothing is saved yet. The new URL handles change, so old links will redirect. Apply the meta and FAQs and keep the current handles? **You:** Yes, apply those and keep the handles. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied without the handle changes. All three re-scored. Want me to look at the images on those products next? ## What is Gemini CLI? Gemini CLI is Google's open-source command-line AI agent. You run it in a terminal, it reads the folder you start it in, and it works through tasks by calling built-in tools and tools from MCP servers. MCP servers live under mcpServers in settings.json, either ~/.gemini/settings.json for every project or .gemini/settings.json for one. You can add one with gemini mcp add, choosing the stdio, sse or http transport and a user or project scope. For remote servers that need a sign-in, Gemini CLI detects the 401 response, discovers the OAuth endpoints, registers itself when the server allows it and opens your browser. Each server can set trust, which skips tool confirmations and is off by default, and includeTools or excludeTools lists to control exactly which of its tools the model sees. MCP tools appear to the model as mcp__. ## Why use Gemini CLI with StoreSEO? - **Sign-in with nothing to paste.** Gemini CLI discovers StoreSEO's OAuth endpoints and registers itself, so there is no client ID or secret to copy. You approve your store once in the browser. - **Trim the tool list per server.** Use excludeTools or includeTools to hide StoreSEO tools you never use, so the model sees a shorter list and has fewer ways to pick the wrong one. - **Confirmations by default.** Gemini CLI asks before running an MCP tool unless you trust the server, so a StoreSEO write has to get past Gemini CLI's prompt as well as StoreSEO's own confirmation. ## Good to know - Google's site says Gemini CLI was replaced by Antigravity CLI on June 18th, 2026 for unpaid tier and Google One users. If that is you, check which command-line agent your account can still use before you set this up. - gemini mcp add defaults to project scope, so a server added without --scope user is only available in that project. - Do not set --trust or trust: true on the storeseo server. Trust skips every tool confirmation, and then the confirmation StoreSEO asks for in chat is the only check. - The browser sign-in needs a browser on the same machine. In a headless session or over SSH without a browser, use the access token header instead. ## What can you ask Gemini CLI to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first five. - Which product images have no alt text? Show me the ten heaviest first. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. - Which schema types are enabled for my store, and what would turning on Product schema add? ## Gemini CLI guides by job - [Audit your store's SEO with Gemini CLI](https://storeseo.com/agent/gemini-cli/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Gemini CLI](https://storeseo.com/agent/gemini-cli/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Gemini CLI](https://storeseo.com/agent/gemini-cli/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Gemini CLI](https://storeseo.com/agent/gemini-cli/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Gemini CLI](https://storeseo.com/agent/gemini-cli/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Gemini CLI](https://storeseo.com/agent/gemini-cli/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Gemini CLI](https://storeseo.com/agent/gemini-cli/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Gemini CLI](https://storeseo.com/agent/gemini-cli/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Gemini CLI](https://storeseo.com/agent/gemini-cli/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Gemini CLI](https://storeseo.com/agent/gemini-cli/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Gemini CLI to StoreSEO? Run gemini mcp add --scope user --transport http storeseo https://mcp.storeseo.com/mcp, then start gemini. The first call to StoreSEO opens your browser, where you pick your Shopify store and approve access. You can also type /mcp auth storeseo to start the sign-in yourself. ### Do I need a StoreSEO API key for Gemini CLI? No. Gemini CLI signs in with OAuth and stores the tokens it receives. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative for machines without a browser, added with the --header option. ### Where does Gemini CLI store the StoreSEO server? With --scope user it is written to ~/.gemini/settings.json and is available in every project. Without that option it is written to the .gemini/settings.json file of the current project only. ### Will Gemini CLI change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and that yes is an instruction the agent follows, not a lock on the server. Gemini CLI also asks for confirmation before an MCP tool runs unless you mark the server as trusted, and that setting is the hard stop. Reads such as scores and reports run straight away. ### Which StoreSEO plan does Gemini CLI need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. See the pricing table for current plans. ### Can I hide some StoreSEO tools from Gemini CLI? Yes. Add --exclude-tools to the gemini mcp add command, or an excludeTools list to the storeseo entry in settings.json, with the names of the tools you do not want the model to see. An includeTools list does the opposite and allows only the tools you name. ### How do I disconnect Gemini CLI from my store? Run gemini mcp remove storeseo, adding --scope user if you added it with that scope. Then revoke the connection or the access token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [Gemini CLI: MCP servers](https://geminicli.com/docs/tools/mcp-server/) --- Canonical: https://storeseo.com/agent/gemini-cli Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from GitHub Copilot > Connect GitHub Copilot agent mode in VS Code to StoreSEO with a .vscode/mcp.json entry, then audit your Shopify store and fix meta and alt text in chat. GitHub Copilot is GitHub's AI coding assistant, and in VS Code its agent mode can call tools from MCP servers. Connected to StoreSEO, it reads your store's SEO scores, previews better meta and alt text, and applies what you approve from the chat panel next to your theme code. - One entry in .vscode/mcp.json - Or add it with MCP: Add Server - All 44 tools fit under VS Code's per-request tool limit ## How do I connect GitHub Copilot to StoreSEO? ### 1. Add StoreSEO to .vscode/mcp.json Create or open .vscode/mcp.json in your workspace and paste this. VS Code uses a top-level servers key, not mcpServers, and a remote server needs type http. ```json { "servers": { "storeseo": { "type": "http", "url": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Start the server and sign in Run MCP: List Servers from the Command Palette, pick storeseo and start it. If it asks you to sign in, pick your Shopify store in the browser and approve access. Then open Copilot Chat in Agent mode, check the tools icon lists StoreSEO and ask this. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or add it from the Command Palette (optional) Run this command, choose a remote HTTP server, paste https://mcp.storeseo.com/mcp and name it storeseo. VS Code asks where to save it: the workspace shares the entry with your team, your user configuration keeps it to yourself. ```text MCP: Add Server ``` **Sign-in:** StoreSEO supports OAuth with dynamic client registration, so VS Code can sign you in through the browser: pick your Shopify store and approve access. No token has to sit in mcp.json. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-github-copilot ### Example session (illustrative) **You:** Find my five lowest-scoring products and preview better meta for each. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted new meta for all five. Previews used AI credit; applying them is a separate step. Apply all five? **You:** Yes, apply them. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. All five re-scored. Next I can look at where your theme renders the FAQ block, since schema depends on it. ## What is GitHub Copilot? GitHub Copilot is GitHub's AI coding assistant. In Visual Studio Code, Copilot Chat has an Agent mode that can plan multi-step work and call tools, including tools from Model Context Protocol servers. VS Code reads MCP servers from a workspace file, .vscode/mcp.json, whose top-level key is servers, or from your user configuration, which you open with MCP: Open User Configuration. A remote server is an entry with type http and a url. You can also add one with the MCP: Add Server command, and manage servers with MCP: List Servers. GitHub's documentation says MCP in Copilot needs VS Code 1.99 or later, and that on Copilot Business and Enterprise plans the MCP servers in Copilot policy must be enabled, while Free, Pro, Pro+ and Max users do not need an organization policy. ## Why use GitHub Copilot with StoreSEO? - **SEO next to your theme code.** When a fix needs a theme change, such as where FAQs render or how a schema block is placed, the same chat can open the Liquid file beside the StoreSEO results. - **Share the setup with a team.** A .vscode/mcp.json committed to the repository gives everyone on the project the same StoreSEO connection, and each person signs in to their own account. - **You see every tool call.** Copilot shows the tools it calls, so you can tell a read from a preview from an apply, and stop a run that heads somewhere you did not ask for. ## Good to know - VS Code uses the servers key in .vscode/mcp.json. The mcpServers key belongs to the portable .mcp.json file at the project root, so copying a Cursor or Claude snippet unchanged will not work here. - On Copilot Business and Enterprise plans an organization admin must enable the MCP servers in Copilot policy, otherwise the server will not be usable. - A .vscode/mcp.json committed to the repository is shared with your team. Keep any token out of it and use input variables if you ever add one. - Keep the apply, optimize, generate and delete tools on VS Code's confirmation prompt. StoreSEO asks for your yes in the chat, and the editor's tool confirmation is the hard stop. ## What can you ask GitHub Copilot to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Preview alt text for my top collection. - Check my schema settings and tell me which theme file renders the product JSON-LD. - Is my llms.txt live? Tell me what generating one will change before you run it. - Show Search Console queries from the last 28 days where we rank between 8 and 20. ## GitHub Copilot guides by job - [Audit your store's SEO with GitHub Copilot](https://storeseo.com/agent/github-copilot/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with GitHub Copilot](https://storeseo.com/agent/github-copilot/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with GitHub Copilot](https://storeseo.com/agent/github-copilot/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with GitHub Copilot](https://storeseo.com/agent/github-copilot/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with GitHub Copilot](https://storeseo.com/agent/github-copilot/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with GitHub Copilot](https://storeseo.com/agent/github-copilot/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with GitHub Copilot](https://storeseo.com/agent/github-copilot/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with GitHub Copilot](https://storeseo.com/agent/github-copilot/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with GitHub Copilot](https://storeseo.com/agent/github-copilot/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with GitHub Copilot](https://storeseo.com/agent/github-copilot/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect GitHub Copilot to StoreSEO? Add a storeseo entry with type http and url https://mcp.storeseo.com/mcp under the servers key in .vscode/mcp.json, or run MCP: Add Server. Start it from MCP: List Servers, sign in if prompted, then use Copilot Chat in Agent mode. ### Do I need an API key for GitHub Copilot? No. StoreSEO supports OAuth, so VS Code can sign you in through the browser and no token has to be written into mcp.json. A personal access token from Settings, AI Toolkit is the alternative for setups that cannot sign in. ### Will Copilot change my store without asking? StoreSEO tells the agent to preview first, say what a generation will cost, and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server, so VS Code's own tool confirmation is the hard stop. Reads run straight away. ### Which StoreSEO plan do I need? Any active StoreSEO plan works, including the free plan. Keyword rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Is there a limit on how many MCP tools Copilot can use? VS Code limits the number of tools in a single chat request. StoreSEO exposes 44 tools, which is well inside that limit. If you run many servers at once, switch off the tools you do not need in the tools picker. ### Does it work on Copilot Business or Enterprise? Yes, if your organization has enabled the MCP servers in Copilot policy. GitHub's documentation says that policy must be on for Business and Enterprise members, and that Free, Pro, Pro+ and Max users do not need one. ### Where should I put the configuration, workspace or user? Use .vscode/mcp.json to share the entry with the repository's team, or MCP: Open User Configuration to keep it to yourself across projects. Either way each person signs in with their own StoreSEO account. ### How do I disconnect Copilot from my store? Remove the storeseo entry from mcp.json or from MCP: List Servers, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Code + StoreSEO](https://storeseo.com/agent/claude-code) Setup checked against: - [VS Code: add and manage MCP servers](https://code.visualstudio.com/docs/copilot/customization/mcp-servers) - [GitHub Docs: extend Copilot Chat with MCP](https://docs.github.com/en/copilot/how-tos/provide-context/use-mcp/extend-copilot-chat-with-mcp) --- Canonical: https://storeseo.com/agent/github-copilot Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Windsurf > Connect Windsurf's Cascade agent to StoreSEO with a serverUrl entry in mcp_config.json, then audit products and fix SEO without leaving the editor. Windsurf is an AI code editor from Cognition whose Cascade agent can call tools from MCP servers. Connected to StoreSEO, Cascade reads your store's SEO scores, previews optimized content and applies what you approve while your Shopify theme is open in the same window. - One serverUrl entry in mcp_config.json - OAuth through your Shopify admin, or a Bearer header - All 44 tools fit under Cascade's 100-tool cap ## How do I connect Windsurf to StoreSEO? ### 1. Add StoreSEO to mcp_config.json In the Cascade panel, open the ... menu and click the Open MCP config file icon in the MCPs section. Add the entry under mcpServers and save. Remote servers use serverUrl. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Sign in to your store Windsurf supports OAuth for remote servers. When it asks you to sign in to storeseo, a browser window opens: pick your Shopify store and approve access. Then ask this to confirm the connection. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or connect with an access token (optional) Generate a token in StoreSEO under Settings, AI Toolkit and send it as a header. mcp_config.json reads environment variables with the env syntax, so the token can stay out of the file. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer ${env:STORESEO_TOKEN}" } } } } ``` **Sign-in:** Windsurf's docs say OAuth works for remote MCP servers, so you sign in by picking your Shopify store in the browser and approving. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it as an Authorization header in mcp_config.json. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-windsurf ### Example session (illustrative) **You:** Is the HTML sitemap set up on my store? If not, set it up. → `storeseo_start` (read-only) → `storeseo_get_html_sitemap_settings` (read-only) **Agent:** The HTML sitemap is not set up yet. I can generate it with the default settings, and it will be created in your store. You then add its link to a menu yourself. Generate it? **You:** Yes, generate it. → `storeseo_update_html_sitemap` (needs approval) **Agent:** The sitemap page is generated. Add its link to your footer menu, then tell me and I will mark setup complete. ## What is Windsurf? Windsurf is an AI code editor that began at Codeium and is now part of Cognition. Its Cascade agent reads your project, edits files and calls tools, and Windsurf's documentation now lives under Devin Desktop. Cascade adds MCP servers by editing mcp_config.json. You open it from the Cascade panel's Actions menu, and the file lives in ~/.config/devin/ on macOS and Linux and in %APPDATA%\devin\ on Windows. A remote server needs a serverUrl or url and, for a token, a headers block. The docs list OAuth as supported for each transport type. Cascade can use at most 100 MCP tools at a time across all servers, and the MCPs section shows how many each server has enabled. Cascade has no MCP marketplace or one-click install, and Teams and Enterprise admins can turn MCP off or allowlist servers by name. ## Why use Windsurf with StoreSEO? - **All 44 tools fit.** Cascade's 100-tool cap leaves room for every StoreSEO tool and a couple of other servers, so you do not have to trim the list the way you would in an editor with a lower limit. - **SEO beside the Liquid.** When a fix needs a theme change, Cascade can open the template in the same window while StoreSEO shows the score and the missing criteria for that page. - **Switch off tools you never use.** List any StoreSEO tool names in the server's disabledTools array and Cascade stops offering them, which keeps rank tracking out of the way for stores not on Pro. ## Good to know - Windsurf's MCP page says its configuration applies to the legacy Cascade agent. The Devin Local agent, the default for new tabs, reads Devin CLI configuration files instead, so check which agent you are using. - Cascade allows 100 tools in total. StoreSEO uses 44, so other MCP servers you have enabled count against the same 100. - On Teams and Enterprise plans an admin can turn MCP off or allowlist servers by name. If storeseo does not load, ask your admin to allow the server name you used in mcp_config.json. - The Cascade tool-approval behavior is not described on the MCP page. Treat StoreSEO's confirmation as an instruction the agent follows and review each write in the chat before you accept it. ## What can you ask Windsurf to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest first. - Check my schema settings and tell me which types are off and what each would add. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. ## Windsurf guides by job - [Audit your store's SEO with Windsurf](https://storeseo.com/agent/windsurf/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Windsurf](https://storeseo.com/agent/windsurf/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Windsurf](https://storeseo.com/agent/windsurf/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Windsurf](https://storeseo.com/agent/windsurf/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Windsurf](https://storeseo.com/agent/windsurf/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Windsurf](https://storeseo.com/agent/windsurf/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Windsurf](https://storeseo.com/agent/windsurf/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Windsurf](https://storeseo.com/agent/windsurf/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Windsurf](https://storeseo.com/agent/windsurf/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Windsurf](https://storeseo.com/agent/windsurf/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Windsurf to StoreSEO? Open mcp_config.json from the Cascade panel's Actions menu, add a storeseo entry under mcpServers with the serverUrl https://mcp.storeseo.com/mcp, and save. Then sign in when Windsurf asks: your browser opens, you pick your Shopify store and approve access. ### Do I need a StoreSEO API key for Windsurf? No. Windsurf supports OAuth for remote MCP servers. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, sent as an Authorization header in mcp_config.json. ### Will StoreSEO fit under Windsurf's tool limit? Yes. Cascade allows 100 tools in total and StoreSEO has 44. Other MCP servers you have enabled count against the same limit, so if you run several, disable the tools you do not need. ### Where is mcp_config.json? Open it from the Cascade panel: the Actions menu, then the Open MCP config file icon in the MCPs section. Windsurf's current docs place it at ~/.config/devin/mcp_config.json on macOS and Linux and at %APPDATA%\devin\mcp_config.json on Windows. ### Will Windsurf change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and that yes is an instruction the agent follows, not a lock on the server. Review each write in the Cascade chat before you accept it, and keep the write tools out of any auto-approve setting you have. Reads such as scores and reports run straight away. ### Which StoreSEO plan does Windsurf need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. See the pricing table for current plans. ### How do I disconnect Windsurf from my store? Remove the storeseo entry from mcp_config.json, or switch its toggle off in the MCPs section of the Cascade panel. Then revoke the connection or the access token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [Windsurf: Cascade MCP integration](https://docs.windsurf.com/windsurf/cascade/mcp) --- Canonical: https://storeseo.com/agent/windsurf Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Grok > Add StoreSEO to Grok as a custom connector on grok.com, sign in through Shopify, then ask Grok to audit your store's SEO and preview fixes before you approve. Grok is xAI's AI assistant, available on grok.com. Connected to StoreSEO as a custom connector, it reads your store's SEO scores, previews better meta and alt text, and leaves the decision to apply them to you. - One custom connector on grok.com - OAuth sign-in through your Shopify admin - Same 44 tools as every other client ## How do I connect Grok to StoreSEO? ### 1. Add a custom connector Open the connectors page on grok.com, choose New Connector and then Custom. Enter this as the MCP server URL and complete the authentication when it opens. ```text https://mcp.storeseo.com/mcp ``` ### 2. Approve your store and check it Pick your Shopify store in the browser and approve access. Then start a chat with the connector available and ask this to confirm the connection works. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or have a team admin provision it (optional) Only on Grok Business and Enterprise: xAI's documentation says a team admin must first provision the connector in the cloud console before members can use it. Give the admin the same URL. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** Grok connectors authenticate with OAuth: when you add StoreSEO you pick your Shopify store in the browser and approve access. There is no token to paste. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-grok ### Example session (illustrative) **You:** Is my store set up for AI answer engines? Check my llms.txt. → `storeseo_start` (read-only) → `storeseo_get_llms_txt_status` (read-only) **Agent:** No llms.txt is live yet. Generating one spends AI credit and writes to your live theme and Shopify Files. Do you want me to generate it? **You:** Yes, generate it. → `storeseo_generate_llms_txt` (needs approval) **Agent:** Generated and published. I can check the status again to confirm it is serving. ## What is Grok? Grok is xAI's AI assistant. You use it at grok.com and in xAI's apps, and it can answer questions, search and analyze, and take actions in external tools you connect. Connectors let Grok reach external tools and data inside a conversation. xAI's documentation says connectors are available to all Grok users and authenticate via OAuth, and that custom connectors for a remote MCP server are added from the connectors page by choosing New Connector, then Custom, and entering the server URL. The documentation also says the MCP server must be reachable over the public internet, and that on Grok Business and Enterprise a team admin must provision a connector in the cloud console before members can use it. ## Why use Grok with StoreSEO? - **No terminal, no config.** You add one URL on grok.com, approve your store once and the StoreSEO tools appear in the conversation. That suits store owners who would rather ask than open an admin page. - **Ask in plain words.** Ask what is weak in your store's SEO and what to fix first. Reads run straight away, so the answers come from your real store data. - **Previews before writes.** StoreSEO tells the agent to preview, name the cost and wait for your yes, so a generation never has to be a surprise. ## Good to know - xAI's connector documentation does not describe a per-tool approval setting. StoreSEO's instruction to preview and wait for your yes is the main check in the conversation, so read what Grok proposes before you say yes. - On Grok Business and Enterprise a team admin has to provision the connector in the cloud console first, so you may need your admin to add it. - This page covers the Grok chat connector on grok.com. xAI's terminal agent is a separate client with its own setup. - Because Grok connects over the public internet, nothing needs to be opened on your side, but the StoreSEO sign-in must complete in a browser. ## What can you ask Grok to do in StoreSEO? - Use StoreSEO to find my lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for five of them. - Is my store ready for AI answer engines? Check llms.txt and my AI optimization settings. - Show my Search Console queries from the last 28 days with high impressions and few clicks. - Which structured data types are switched on for my store? ## Grok guides by job - [Audit your store's SEO with Grok](https://storeseo.com/agent/grok/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Grok](https://storeseo.com/agent/grok/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Grok](https://storeseo.com/agent/grok/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Grok](https://storeseo.com/agent/grok/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Grok](https://storeseo.com/agent/grok/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Grok](https://storeseo.com/agent/grok/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Grok](https://storeseo.com/agent/grok/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Grok](https://storeseo.com/agent/grok/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Grok](https://storeseo.com/agent/grok/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Grok](https://storeseo.com/agent/grok/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Grok to StoreSEO? Open the connectors page on grok.com, choose New Connector, then Custom, and enter https://mcp.storeseo.com/mcp. Complete the sign-in in your browser: pick your Shopify store and approve access, and the StoreSEO tools become available in your chats. ### Do I need an API key for Grok? No. Grok connectors authenticate with OAuth, so you pick your store in the browser and approve access. Nothing needs to be pasted. ### Will Grok change my store without asking? StoreSEO tells the agent to preview first, say what a generation will cost, and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server. xAI's connector documentation describes no separate per-tool approval setting, so read each proposal before you agree. Reads run straight away. ### Which StoreSEO plan do I need? Any active StoreSEO plan works, including the free plan. Keyword rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can I use it on Grok Business or Enterprise? xAI's documentation says a team admin must first provision a connector in the cloud console before members of the organization can use it. Ask your admin to add https://mcp.storeseo.com/mcp. ### Does my computer need to be reachable by Grok? No. xAI's documentation only requires that the MCP server be reachable over the public internet, and the StoreSEO server already is. ### How do I disconnect Grok from my store? Remove the StoreSEO connector from the connectors page on grok.com, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [ChatGPT + StoreSEO](https://storeseo.com/agent/chatgpt) Setup checked against: - [xAI: Grok connectors](https://docs.x.ai/grok/connectors) --- Canonical: https://storeseo.com/agent/grok Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO with ChatGPT dots > Connect an OpenAI dot to StoreSEO through the shared ChatGPT plugin, then let an always-on agent audit your Shopify store's SEO and report on a schedule. A ChatGPT dot is an always-on agent inside ChatGPT: it has its own cloud computer, works across the apps you connect and keeps going between conversations. Connected to StoreSEO through the plugin you add in ChatGPT, your dot reads your store's SEO scores, previews better meta and alt text, and can send you a scheduled report. - Shares the StoreSEO plugin with ChatGPT - Scheduled, report-only SEO checks - OAuth sign-in through your Shopify admin ## How do I connect ChatGPT dots to StoreSEO? ### 1. Create the StoreSEO app in ChatGPT OpenAI's help center says plugins are shared across dots and ChatGPT, so you set StoreSEO up once. In ChatGPT on the web, turn on Developer mode under Settings, Security and login, open Plugins, select the plus button and create an app named StoreSEO with this MCP server URL and OAuth. Sign in when the browser asks: pick your Shopify store and approve access. If ChatGPT already has the StoreSEO app, skip this step. ```text https://mcp.storeseo.com/mcp ``` ### 2. Make sure your dot can see it Open your dot's profile, then Customize and Plugins, and check StoreSEO is connected. Then give your dot a first task that names StoreSEO. ```text Using StoreSEO, show me my store's SEO score breakdown. ``` ### 3. Or ask a workspace admin to publish it (optional) On a Business workspace only an admin or owner can turn on developer mode and publish the app, and on Enterprise and Edu an admin has to grant you access first. The controls sit under Settings, Apps. Give the admin the same URL, then start at the previous step once the app is published. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** You sign in with StoreSEO once, in the browser: pick your Shopify store and approve access. Your dot uses the same ChatGPT plugin, and ChatGPT can register itself through dynamic client registration, so there is no token to paste. Verify with: "Using StoreSEO, show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-chatgpt-dots ### Example session (illustrative) **You:** Every Monday morning, check my store's SEO and tell me what needs attention. Change nothing. **Agent:** Scheduled for Monday mornings, report only. Running the first check now. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) **Agent:** Your Summer Linen collection scores lowest, mostly because of a missing meta description. I can preview a new one; the preview uses AI credit but writes nothing. Shall I? **You:** Yes, preview it. → `storeseo_preview_resource_optimization` (preview) **Agent:** Here is the draft title and meta description. Apply it to the collection? **You:** Yes, apply it. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. The collection re-scored higher. Next week's report will list what is still open. ## What is ChatGPT dots? A ChatGPT dot is one of OpenAI's always-on agents inside ChatGPT. According to OpenAI's help center, a dot has its own cloud computer, memory, and the apps you choose to connect. You give it a goal, set what it may do on its own, and it keeps making progress between conversations and brings results back for your review. Connected apps give a dot its tools. OpenAI's help center says plugin permissions are shared across dots and ChatGPT, so an MCP app you create for ChatGPT in developer mode is available to your dot too. Custom rules let you allow, require approval for or block specific actions, and Scheduled Tasks let a dot run once, repeat or monitor for changes. StoreSEO's MCP server is a remote server of that kind. Once the app exists, your dot can read your store's SEO scores, preview optimized content and, when you agree, apply it. Because a dot can run on a timer while you are away, the useful pattern is scheduled reports, with changes left for a conversation where you can say yes. ## Why use ChatGPT dots with StoreSEO? - **A weekly SEO check that runs itself.** Ask your dot to review your store every week and report what slipped: low scores, missing alt text, an llms.txt that is out of date. You read a short summary instead of opening the app. - **One plugin for ChatGPT and your dot.** You add StoreSEO once in ChatGPT. The same app is available when you chat and when your dot works, so there is nothing separate to maintain. - **Rules you control.** Custom rules decide what your dot does alone and what it asks about. Pair them with StoreSEO's preview-first instructions and you can keep every store change behind a conversation. ## Good to know - Developer mode and write actions depend on your plan and workspace. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. OpenAI's help center, as we read it, puts full MCP support with write actions in beta on Business, Enterprise and Edu and gives Pro accounts read access only. - On a Business workspace only an admin or owner can turn on developer mode and publish the app, and on Enterprise and Edu you need access granted by an admin. If you do not see the setting under Settings, Security and login, ask your admin. - A scheduled task has no one watching it. Write it as report only: StoreSEO tells the agent to preview and wait for your yes, but that yes is an instruction, and a task paired with a rule that takes actions without asking could apply changes unattended. - Generated previews spend AI Content Optimizer credit, so a recurring task that previews content every week uses credit every week. Keep scheduled work to reads unless you want that spend. ## What can you ask ChatGPT dots to do in StoreSEO? - Using StoreSEO, what is my store's SEO score and what are the three biggest problems? - Every Monday, use StoreSEO to list products scoring under 50 and send me the list. Report only and change nothing. - Check whether my llms.txt is live and up to date, and tell me what generating a new one would change before you do anything. - Using StoreSEO, which images have no alt text? Preview alt text for five of them. - Once a month, pull my Search Console queries for the last 28 days and summarize where we rank between 8 and 20. ## Run SEO checks on a schedule with your dot OpenAI's help center describes Scheduled Tasks that run once, repeat or monitor for changes. Ask your dot in plain words what to check, when, and where to report, and keep the task to reads: scores, image status, setup and Google reports. A scheduled run is unattended, so there is nobody to say yes to a change. StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, but that is an instruction, so word the task as report only and keep your custom rules set to ask before any action that changes your store. ```text Every Monday at 8:00, use StoreSEO to list my lowest-scoring products, any images without alt text and whether my llms.txt is live, and send me a short report. Report only and change nothing. If StoreSEO is unreachable, tell me instead of reusing last week's data. ``` ## ChatGPT dots guides by job - [Audit your store's SEO with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with ChatGPT dots](https://storeseo.com/agent/chatgpt-dots/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect a ChatGPT dot to StoreSEO? Create the StoreSEO app in ChatGPT: turn on Developer mode under Settings, Security and login, open Plugins, select the plus button, name the app StoreSEO, paste https://mcp.storeseo.com/mcp and choose OAuth. Sign in, pick your Shopify store and approve. Then check under your dot's Customize, Plugins that StoreSEO is connected. ### Do I need an API key to connect my dot? No. You sign in with OAuth in the browser and StoreSEO issues ChatGPT its own credentials. Your dot uses the same plugin, so there is no token to paste. ### Will my dot change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything, and reads such as scores and reports run straight away. That yes is an instruction, not a lock on the server. In ChatGPT, write actions ask for confirmation by default and you can review the payload. Custom rules on your dot and a task worded as report only are your hard stops. ### Which StoreSEO plan does a dot need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. On the ChatGPT side, your plan and workspace decide whether developer mode is available. ### Why can I not turn on developer mode on a Business workspace? OpenAI's help center says that on Business only an admin or owner can enable developer mode and publish an app, and on Enterprise and Edu an admin has to grant access. Ask your admin to create and publish the StoreSEO app with the MCP URL, then your dot can use it. ### Can my dot run SEO checks on a schedule? Yes. OpenAI's help center describes scheduled tasks for dots. Ask for a recurring check and say it is report only. Reads such as scores and reports run without anyone present, but anything that changes your store should wait for you in a conversation. ### Is this different from connecting plain ChatGPT? The setup is the same: a developer-mode app with the StoreSEO MCP URL. A dot adds its own cloud computer, memory and scheduled tasks on top, so it suits recurring checks while ChatGPT suits a conversation. ### How do I disconnect my dot from StoreSEO? Remove or disconnect the StoreSEO app in ChatGPT's Plugins, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [ChatGPT with StoreSEO](https://storeseo.com/agent/chatgpt) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) Setup checked against: - [OpenAI: ChatGPT developer mode](https://developers.openai.com/api/docs/guides/developer-mode) - [OpenAI help center: Getting started with your dot](https://help.openai.com/en/articles/20001530-getting-started-with-your-dot) - [OpenAI help center: Developer mode and MCP apps](https://help.openai.com/en/articles/12584461) --- Canonical: https://storeseo.com/agent/chatgpt-dots Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO with Grok Bot > Add StoreSEO to a Grok Bot as a remote HTTPS MCP server, sign in through Shopify, then have the Bot audit your store's SEO and run report-only routines. Grok Bot is xAI's product for named AI Bots that keep their context and work on a persistent cloud computer, even while your device is closed. Connected to StoreSEO as a remote HTTPS MCP server, a Bot reads your store's SEO scores, previews better meta and alt text, and can run a scheduled check that reports back. - Ask any Bot to add the StoreSEO server - Allow once, Always allow or Deny on each action - Report-only routines that run while you are away ## How do I connect Grok Bot to StoreSEO? ### 1. Ask your Bot to add StoreSEO In a chat with your Bot, ask it to add a custom remote HTTPS MCP server with this URL. xAI's documentation says a plugin added from your own chat is personal and account-wide, so every Bot on your account can use it. Complete the StoreSEO sign-in when the browser opens: pick your Shopify store and approve access. ```text Add a custom remote HTTPS MCP server called StoreSEO with the URL https://mcp.storeseo.com/mcp and sign me in. ``` ### 2. Attach it and check it works In a chat, type @ to attach the StoreSEO connector to the task, then ask this to confirm the connection. ```text Use @StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or add it to a Team Bot from the Setup panel (optional) For a Team Bot, only its owner can change the setup, and the change reaches every teammate. The owner adds the plugin from the Plugins card in the Bot's Setup panel, choosing a custom MCP server over remote HTTPS, with this URL. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** xAI's documentation says a custom remote HTTPS server uses the Bot's own credential, or each person's sign-in when the server uses OAuth. StoreSEO uses OAuth: you pick your Shopify store in the browser and approve access. Verify with: "Use @StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-grok-bot ### Example session (illustrative) **You:** Use @StoreSEO to check whether my store is set up for AI answer engines. → `storeseo_start` (read-only) → `storeseo_get_llms_txt_status` (read-only) **Agent:** No llms.txt is live yet. Generating one spends AI credit and writes to your live theme and Shopify Files. Do you want me to generate it? **You:** Yes, generate it. → `storeseo_generate_llms_txt` (needs approval) **Agent:** Done. Your llms.txt is live. I can also check your agents.md and schema settings if you want. ## What is Grok Bot? Grok Bot is xAI's product for Bots you keep around: AI teammates with names, jobs and context that builds up over time. According to xAI's documentation, all your Bots share one persistent cloud computer with a browser, a filesystem and a terminal, which runs in Cursor's cloud, so work can continue while your device is closed. It runs in desktop and mobile apps and is included with paid individual Cursor plans, Cursor Teams, or a linked SuperGrok subscription. Connectors are installed as plugins, and xAI's documentation says installed connectors are account-wide rather than tied to one Bot. A plugin can be a custom MCP server over remote HTTPS. It then uses the Bot's own credential, or each person's sign-in when the server uses OAuth. You attach a connector to a task by typing @ in the chat. When an action needs approval, the Bot offers Allow once, Always allow or Deny. Auto-review rules under Settings, General, Auto-review can force an Ask first or allow matching actions automatically, and Ask first wins when both apply. Routines let a Bot run a workflow on a schedule. ## Why use Grok Bot with StoreSEO? - **An SEO check that happens while you are away.** A routine can run on a schedule on the Bot's cloud computer and post a short summary: lowest-scoring products, images without alt text, whether llms.txt is live. - **Approval on every action.** StoreSEO tells the agent to preview and wait for your yes, and Grok Bot adds its own prompt with Allow once, Always allow and Deny, plus Ask first rules you can set in Auto-review. - **Connect once for every Bot.** A plugin you add from your own chat is account-wide, so your other Bots can use StoreSEO without another sign-in. A Team Bot's owner can add it for the whole team instead. ## Good to know - A plugin added from your own chat is personal and account-wide. A Team Bot's owner adds it from the Plugins card in the Setup panel, and that reaches only that Team Bot's teammates. - Be careful with Always allow. It saves a rule for matching actions, so a StoreSEO write tool you always allow no longer stops for your yes in the client, and StoreSEO's preview-first instruction becomes the only check. Use Ask first rules in Auto-review for apply, generate and publish tools. - A routine's Test run performs real work, including calls to connected tools, so run it with safe inputs. Write each routine as report only. - xAI's documentation says Grok Bot inherits your team's Cursor connector policy, a blocked connector shows as Disabled by team admin, and the MCP allowlist is Enterprise only. On a team plan an admin may have to allow the StoreSEO server first. ## What can you ask Grok Bot to do in StoreSEO? - Use @StoreSEO to find my lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for five of them with @StoreSEO. - Is my llms.txt live and current? Tell me what generating a new one would change before you do anything. - Use @StoreSEO to show Search Console queries from the last 28 days where we rank between 8 and 20. - Which of my collections have no meta description? Preview one for the weakest and wait for my yes. ## Run SEO checks on a schedule with a routine A routine tells one Bot when to run a workflow. Ask the Bot in plain words when to run, what to check, where to report and what it must not do, and it creates the routine and shows its next run. Background routines can run while your laptop is closed, and Routines under the conversation details let you pause one or read its recent runs. Keep a routine to reads and reports. Nobody is there to say yes to a change, and a Test run performs real work, so word it as report only and keep Ask first rules on StoreSEO's write tools. ```text Every Monday at 8:00 AM, use @StoreSEO to list my lowest-scoring products, any images without alt text and whether my llms.txt is live. Post a short report in this conversation. Report only and change nothing. If StoreSEO is unreachable, say so instead of reusing last week's data. ``` ## Grok Bot guides by job - [Audit your store's SEO with Grok Bot](https://storeseo.com/agent/grok-bot/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Grok Bot](https://storeseo.com/agent/grok-bot/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Grok Bot](https://storeseo.com/agent/grok-bot/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Grok Bot](https://storeseo.com/agent/grok-bot/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Grok Bot](https://storeseo.com/agent/grok-bot/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Grok Bot](https://storeseo.com/agent/grok-bot/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Grok Bot](https://storeseo.com/agent/grok-bot/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Grok Bot](https://storeseo.com/agent/grok-bot/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Grok Bot](https://storeseo.com/agent/grok-bot/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Grok Bot](https://storeseo.com/agent/grok-bot/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Grok Bot to StoreSEO? Ask your Bot in chat to add a custom remote HTTPS MCP server with the URL https://mcp.storeseo.com/mcp and complete the sign-in in the browser: pick your Shopify store and approve access. Then type @ in a chat to attach StoreSEO. On a Team Bot, the owner adds it from the Plugins card in the Setup panel instead. ### Do I need an API key for Grok Bot? No. StoreSEO uses OAuth, so you sign in in the browser and approve your store. A personal access token from Settings, AI Toolkit is not part of the Grok Bot setup described in xAI's documentation. ### Will Grok Bot change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything, and reads such as scores and reports run straight away. That yes is an instruction, not a lock on the server. Grok Bot adds its own prompt with Allow once, Always allow and Deny, and Ask first rules in Auto-review are the hard stop. Always allow removes that prompt, so use it sparingly. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. Grok Bot itself comes with a paid Cursor plan or a linked SuperGrok subscription. ### Is a plugin I add shared with my teammates? Not when you add it from your own chat: that is a personal plugin, available to every Bot on your account. A Team Bot's owner can add it from the Setup panel, and only that owner can change the setup, which reaches all teammates. ### Can a routine run StoreSEO checks while I am away? Yes. Ask a Bot to create a routine and it runs on a schedule on the Bot's cloud computer, even with your laptop closed. Keep it to reads and reports, because nobody is there to approve a change, and remember that Test run performs real work. ### How is this different from Grok on grok.com? Grok on grok.com is a chat assistant with a custom connector. Grok Bot gives you named Bots with a persistent cloud computer, plugins, approvals and routines. Both reach the same StoreSEO MCP server. ### How do I disconnect Grok Bot from StoreSEO? Remove the StoreSEO plugin from your Bot's plugins, and revoke the connection in the StoreSEO app under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [Grok with StoreSEO](https://storeseo.com/agent/grok) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) Setup checked against: - [xAI: Grok Bot overview](https://docs.x.ai/grok-bot/overview) - [xAI: Connectors and plugins](https://docs.x.ai/grok-bot/computer-and-apps) - [xAI: Team Bots](https://docs.x.ai/grok-bot/team-bots) - [xAI: Routines](https://docs.x.ai/grok-bot/skills-routines-and-automations) - [xAI: Approvals and security](https://docs.x.ai/grok-bot/approvals-security-and-privacy) --- Canonical: https://storeseo.com/agent/grok-bot Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Muse > Send Muse one message with the StoreSEO MCP URL, approve the sign-in for your Shopify store, then audit products, fix meta and publish llms.txt by chatting. Muse is Meta's personal AI agent, which takes actions for you instead of only answering questions and runs in the Muse app, on muse.ai and inside WhatsApp. Connected to StoreSEO as a custom connector, it reads your store's SEO scores, drafts better meta and alt text, and applies the changes you say yes to. - No settings screen: one message sets it up - Sign in with your Shopify store, no key to paste - Chat from the Muse app, muse.ai or WhatsApp ## How do I connect Muse to StoreSEO? ### 1. Send this message to Muse Paste it into a new chat with Muse in the app, on muse.ai or in WhatsApp. Muse is rolling out in the US first, so you need access to it before this works. ```text Add a custom connector for my Shopify SEO app StoreSEO: https://mcp.storeseo.com/mcp - It is an MCP server over HTTP and it signs in with OAuth. Send me the sign-in link and I will approve it. - Once it is connected, show me my store's SEO score breakdown. ``` ### 2. Approve the sign-in for your store Open the sign-in link Muse sends back, pick your Shopify store in the browser and approve access. Meta does not review custom connectors, so only approve this one if you started it. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** StoreSEO uses OAuth: the sign-in opens a browser page where you pick your Shopify store and approve access. Muse has no connector form you fill in for this, so you start it by sending Muse a message that contains the StoreSEO MCP URL. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-muse ### Example session (illustrative) **You:** Which of my products have the weakest SEO? Show me a better title and description for the worst one. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** Your Linen Shirt scores lowest. StoreSEO drafted a new meta title and description. The preview used some AI credit and nothing is saved yet. Apply it to the product? **You:** Yes, apply it. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied, and the product was re-scored. Next on the list is missing alt text on 14 images. Want me to look at those? ## What is Muse? Muse is a personal AI agent from Meta. It is meant to complete tasks for you rather than only answer questions, and you talk to it like a message thread in the Muse app, on muse.ai or in WhatsApp. Meta says it is rolling out in the US first, and that it is free for most of what people need, with subscription plans for people who want to do more. Muse runs on a dedicated Muse Secure VM for each person. A separate Sentinel agent on the same machine decides what leaves it: Meta says nothing Muse does reaches the internet unless Sentinel approves it, and that it asks you for permission when needed. Muse reaches other services through connectors, which you add by asking Muse or from its settings. For services Meta does not list, you can create a custom connector. Meta states that it does not review custom connectors or how they use your information, so a custom connector is your own call to trust. That is the route StoreSEO takes in Muse today. ## Why use Muse with StoreSEO? - **SEO work from a chat thread.** Ask in plain words, the way you would message a colleague, and Muse runs the StoreSEO tools behind it. It suits store owners who would rather not open the Shopify admin to fix meta or alt text. - **On your phone or in WhatsApp.** Because Muse lives in its mobile app and in WhatsApp as well as on the web, you can check your store's scores or approve a fix away from your desk. - **Two places that ask first.** StoreSEO tells the agent to preview and wait for your yes before it writes, and Muse and its Sentinel agent ask you for permission on important actions too. ## Good to know - Muse is rolling out in the US first, so you may not have access yet. Check Meta's rollout before planning around it. - StoreSEO is a custom connector in Muse, not a listed one. Meta does not review custom connectors, so approve the sign-in only for the StoreSEO link you sent yourself. - Muse decides when to ask your permission. Do not rely on that alone: tell it to preview and ask before it applies, publishes or deletes anything, and read each question before you answer yes. - Pasted messages are not a settings screen, so there is no list of connected tools to switch off in Muse. To cut off access, revoke the connection in StoreSEO under Settings, AI Toolkit. ## What can you ask Muse to do in StoreSEO? - Use StoreSEO to show me which of my products have the lowest SEO scores. - Preview a better meta title and description for my best seller, and wait for my yes before you save it. - Which of my product images are missing alt text? Tell me how many before you change anything. - Check whether my store has an llms.txt file and tell me what generating one would change. - How did my store's organic search traffic change over the last 28 days? ## Muse guides by job - [Audit your store's SEO with Muse](https://storeseo.com/agent/muse/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Muse](https://storeseo.com/agent/muse/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Muse](https://storeseo.com/agent/muse/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Muse](https://storeseo.com/agent/muse/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Muse](https://storeseo.com/agent/muse/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Muse](https://storeseo.com/agent/muse/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Muse](https://storeseo.com/agent/muse/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Muse](https://storeseo.com/agent/muse/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Muse](https://storeseo.com/agent/muse/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Muse](https://storeseo.com/agent/muse/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Muse to StoreSEO? Send Muse a message that contains the StoreSEO MCP URL, https://mcp.storeseo.com/mcp, and asks it to add a custom connector and send you the sign-in link. Open the link, pick your Shopify store in the browser and approve access. Muse has no connector form for this, so the message is the setup. ### Do I need a StoreSEO API key for Muse? No. StoreSEO signs you in with OAuth, so you approve access in the browser instead of pasting a key into the chat. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative for clients that take a header, but Muse is set up with the sign-in link. ### Is StoreSEO a listed Muse connector? No. Meta reviews connectors before they appear in its directory, and StoreSEO is added as a custom connector instead. Meta says it does not review custom connectors, so only approve the sign-in for the link you sent yourself. ### Will Muse change my products without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and Muse and its Sentinel agent also ask for your permission on important actions. That yes is an instruction the agent follows, not a lock on the server. Reads such as scores and reports run straight away. ### Where does Muse work with StoreSEO? Wherever you chat with Muse: the Muse app, muse.ai or WhatsApp. Muse is rolling out in the US first, so availability depends on whether you have access yet. ### Which StoreSEO plan does Muse need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. ### How do I disconnect Muse from my store? Remove the StoreSEO connector in Muse, then revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. Revoking there is what stops access to your store. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [Meta: Introducing Muse](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) - [Meta Help Center: How Muse works with Connectors](https://www.meta.com/help/artificial-intelligence/1687253048996149/) - [Muse Connector Platform](https://muse.ai/platform) --- Canonical: https://storeseo.com/agent/muse Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Hand Your Shopify SEO to Manus > Add StoreSEO to Manus as a custom MCP server with an access token, then hand off Shopify SEO audits, previews and fixes, or schedule a report-only check. Manus is a general AI agent that takes a task from request to finished result and can use connected apps along the way. With StoreSEO added as a custom MCP server, it can audit your store's SEO, prepare previews of better meta and alt text, and stop to ask before it applies anything. - A custom MCP server, added once in Settings - Authenticates with a StoreSEO access token - Scheduled tasks for recurring report-only checks ## How do I connect Manus to StoreSEO? ### 1. Create an access token in StoreSEO In the StoreSEO app, open Settings, then AI Toolkit, and generate a named personal access token. Copy it now and keep it out of any chat. ```text StoreSEO → Settings → AI Toolkit → generate a token ``` ### 2. Add StoreSEO as a custom MCP server In Manus, open Settings, then Integrations, then Custom MCP Servers, and choose Add Server. Give it a name, paste this URL as the server URL and enter your token as the authentication, as a Bearer token. ```text https://mcp.storeseo.com/mcp ``` ### 3. Check it and use it Manus verifies it can reach the server and read its tools. Then start a task that names StoreSEO, so Manus uses the connector. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` **Sign-in:** Manus documents custom MCP servers that authenticate with an API key, OAuth token or bearer token. Generate a personal access token in StoreSEO under Settings, AI Toolkit and enter it as the server's authentication. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-manus ### Example session (illustrative) **You:** Use StoreSEO to find my five lowest-scoring products and preview better meta for each. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted new meta for all five. The biggest jump is Linen Shirt, projected 41 to 78. Previews used AI credit; applying is free. Apply all five? **You:** Yes, apply them. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. All five re-scored; the next fix is alt text on 14 images. Want me to do those? ## What is Manus? Manus is a general AI agent. You describe a task in plain words and it works through the steps, such as research, writing and acting in other apps, until it has a result for you. Manus reaches other apps through connectors. Prebuilt connectors sign in with OAuth and you use one by mentioning the app in your prompt. For services that are not in its directory, Manus supports custom MCP servers, added under Settings, Integrations, Custom MCP Servers with a server name, an HTTPS server URL and authentication such as a bearer token. Manus can also run a task on a schedule, using your connected apps, without you starting it each time. That makes it a fit for merchants who want a regular summary of their store's SEO, as long as the scheduled job only reads. ## Why use Manus with StoreSEO? - **Hand off the whole job.** Ask for an audit and Manus works through the listing, previews fixes for the weakest products and brings back one summary to approve, instead of you opening products one by one. - **Plain-words requests.** You do not need to know tool names. Name StoreSEO in your task and Manus picks the right tools, then asks you before it writes anything. - **A recurring report.** Set a scheduled task to read your scores and Search Console insights every week and write you a summary, with nothing changed in your store. ## Good to know - StoreSEO is not in the Manus connector directory. You add it yourself as a custom MCP server, and you name it in your prompt so Manus uses it. - Manus documents API keys, OAuth tokens and bearer tokens for custom servers but does not describe a browser sign-in for them, so connect with a StoreSEO access token. Anyone with that token can use your StoreSEO tools, so revoke it in Settings, AI Toolkit if it leaks. - Scheduled tasks run unattended, and Manus offers a setting to skip confirmations. Do not turn that on for a StoreSEO task that writes. Keep scheduled tasks report-only. - Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. ## What can you ask Manus to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Use StoreSEO to find product images with no alt text and show me what you would write for my best-selling collection. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Use StoreSEO to pull my Search Console queries from the last 28 days where we rank between 8 and 20, and write me a short brief. - Check my store's SEO setup with StoreSEO and give me the three fixes that matter most. ## Schedule a weekly SEO check in Manus Manus documents scheduled tasks that run on a timetable, can use your connected apps and run unattended. For StoreSEO, make the task report-only: it should read scores and reports and write you a summary. Do not ask a scheduled task to apply content, generate files or delete anything, because you are not there to say yes, and leave the skip confirmations setting off. ```text Every Monday morning, use StoreSEO to read my store's SEO score breakdown, my lowest-scoring products and last week's Search Console insights. Summarize what changed. Do not apply, generate, publish or delete anything. ``` ## Manus guides by job - [Audit your store's SEO with Manus](https://storeseo.com/agent/manus/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Manus](https://storeseo.com/agent/manus/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Manus](https://storeseo.com/agent/manus/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Manus](https://storeseo.com/agent/manus/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Manus](https://storeseo.com/agent/manus/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Manus](https://storeseo.com/agent/manus/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Manus](https://storeseo.com/agent/manus/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Manus](https://storeseo.com/agent/manus/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Manus](https://storeseo.com/agent/manus/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Manus](https://storeseo.com/agent/manus/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Manus to StoreSEO? Generate a personal access token in StoreSEO under Settings, AI Toolkit. In Manus, open Settings, then Integrations, then Custom MCP Servers, choose Add Server, enter a name, the URL https://mcp.storeseo.com/mcp and your token as the authentication. Manus checks the server and reads its tools, and you can then name StoreSEO in a task. ### Do I need a StoreSEO API key for Manus? You need a personal access token from StoreSEO under Settings, AI Toolkit. Manus documents API keys, OAuth tokens and bearer tokens for custom MCP servers but does not describe a browser sign-in for them, so the token is the path this guide uses. ### Is StoreSEO listed as a Manus connector? No. StoreSEO is not in the Manus connector directory, so you add it yourself as a custom MCP server. Once added, name StoreSEO in your prompt so Manus uses it. ### Will Manus change my products without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, but that is an instruction the agent follows, not a lock on the server. Reads such as scores and reports run straight away. A scheduled Manus task runs without you, so keep it report-only and do not skip confirmations. ### Which StoreSEO plan does Manus need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. Manus bills its own usage under your Manus plan. ### Can Manus run an SEO check on a schedule? Yes. Manus documents scheduled tasks that can use your connected apps and run unattended. Write the task so it only reads, for example a weekly summary of scores and Search Console insights, and keep any changes for a conversation where you can say yes. ### Can Manus optimize my whole catalog? Yes, in batches. Ask it to list products by score, preview several at once and bring the previews back for you to approve together. Each preview spends AI credit, so start with the products that need it most. ### How do I disconnect Manus from my store? Remove the StoreSEO server under Settings, Integrations, Custom MCP Servers in Manus. Then revoke the token in StoreSEO under Settings, AI Toolkit, where every connected app and token is listed, so it can no longer be used. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Run Shopify SEO with Claude Cowork](https://storeseo.com/agent/claude-cowork) Setup checked against: - [Manus: custom MCP servers](https://manus.im/docs/integrations/custom-mcp) - [Manus: MCP connectors](https://manus.im/docs/integrations/mcp-connectors) - [Manus: automations](https://manus.im/docs/automations) - [Manus: Scheduled Tasks 2.0](https://manus.im/blog/manus-schedules) --- Canonical: https://storeseo.com/agent/manus Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from OpenClaw > Add StoreSEO to OpenClaw with openclaw mcp add and openclaw mcp login, then audit products and fix SEO from a chat app, or run read-only reports on a schedule. OpenClaw is an open-source AI agent runtime that you run yourself, talk to through chat apps and a CLI, and schedule with built-in automations. Connected to StoreSEO through its MCP client, it audits your store's SEO, previews optimized content and applies the changes you approve, from a message or on a timer. - Two commands: openclaw mcp add, then login - OAuth through your Shopify admin - Weekly read-only SEO reports with automations ## How do I connect OpenClaw to StoreSEO? ### 1. Add StoreSEO to OpenClaw Run this on the installation that owns the connector. The --auth oauth flag marks the server as an OAuth server, which is what openclaw mcp login needs. ```bash openclaw mcp add storeseo \ --url https://mcp.storeseo.com/mcp \ --transport streamable-http \ --auth oauth ``` ### 2. Sign in to your store The login prints an authorization URL. Open it in a browser, pick your Shopify store and approve access. OpenClaw captures the redirect and saves the credentials. ```bash openclaw mcp login storeseo ``` ### 3. Check the connection The probe opens a live connection and reports the tools the server offers. A saved server entry proves nothing until the probe passes. ```bash openclaw mcp doctor storeseo --probe ``` **Sign-in:** openclaw mcp login prints an authorization URL. Open it, pick your Shopify store and approve. OpenClaw normally captures the redirect on a loopback address and saves the credentials on the installation that owns the connector, so scheduled automations reuse them. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-openclaw ### Example session (illustrative) **You:** Is my llms.txt up to date? → `storeseo_start` (read-only) → `storeseo_get_llms_txt_status` (read-only) **Agent:** Your store's settings changed after the last file was generated, so a rebuild is due. Generating writes llms.txt into your live theme and Shopify Files and spends AI credit. Generate it now? **You:** Yes, go ahead. → `storeseo_generate_llms_txt` (needs approval) → `storeseo_get_llms_txt_status` (read-only) **Agent:** The new file is generated and live at your /llms.txt. Want me to check your agents.md next? ## What is OpenClaw? OpenClaw is an open-source AI agent runtime that you run on your own machine or server. You talk to it through chat apps such as WhatsApp, Telegram, Slack and Discord, or from its Control UI and CLI. It connects to MCP servers, and their tools pass through the same tool policy as everything else the agent can do. Its MCP client stores server definitions in OpenClaw's configuration. You add a remote server once with openclaw mcp add, sign in with openclaw mcp login, and check it with openclaw mcp doctor and its probe option. Tool filters and a per-server approval setting are available through openclaw mcp tools and openclaw mcp configure. OpenClaw also has a built-in scheduler called automations, with openclaw cron as an alias, that wakes the agent at set times in an isolated session and can announce the result in a chat channel. That is what sets it apart for SEO work: a report can arrive in your chat every Monday without you opening anything. ## Why use OpenClaw with StoreSEO? - **SEO from your chat app.** Message OpenClaw from your phone and it calls StoreSEO for you. Ask which products dropped in score after a product import and get the answer where you already read your messages. - **Reports that arrive on their own.** Create an automation with a cron expression and a prompt, and OpenClaw runs it in an isolated session and announces the result in a channel. Use it for a weekly score summary or Search Console review. - **A probe that tells you it works.** openclaw mcp doctor with the probe option opens a live connection and lists the StoreSEO tools, so you know the sign-in worked before you trust a scheduled run. ## Good to know - Run the mcp commands on the installation that owns the connector. A Gateway or agent on another host needs its own reload, config publish or restart before it sees the storeseo server. - StoreSEO has 44 tools. OpenClaw lets you narrow them with openclaw mcp tools and its include and exclude options if you want a smaller set in front of the agent. - OpenClaw has its own per-server approval setting, openclaw mcp configure storeseo --approval followed by approve, prompt or auto. If you want the client to stop before StoreSEO writes, do not leave it on a mode that skips the check. - A scheduled run has nobody to say yes. Keep automations to reads and reports, and ask for changes in a message you can answer. ## What can you ask OpenClaw to do in StoreSEO? - Use StoreSEO to list my ten lowest-scoring products and the top issue on each. - Which of my products have no AI snippet or FAQ yet? Preview them for the first three and wait for my yes. - Pull my Search Console queries from the last 28 days where we rank between 8 and 20. - Is my agents.md live? If not, tell me what generating one will publish and what it costs before you run it. - Every Monday at 8am, send me my store's SEO score breakdown and the five weakest products. Change nothing. ## Send a weekly SEO report with OpenClaw automations Automations are OpenClaw's built-in scheduler, and openclaw cron is an alias for the same commands. This job runs every Monday in an isolated session, reads your store's scores and announces the result in a Telegram chat. List past runs with openclaw automations runs followed by the job ID. A scheduled run should be report-only: the prompt asks for reads, because nobody is there to say yes to a write. ```bash openclaw automations create "0 8 * * 1" \ "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products. Read only: do not preview, apply or generate anything." \ --name "Monday SEO report" \ --tz "Europe/Berlin" \ --session isolated \ --announce --channel telegram --to "-1001234567890" ``` ## OpenClaw guides by job - [Audit your store's SEO with OpenClaw](https://storeseo.com/agent/openclaw/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with OpenClaw](https://storeseo.com/agent/openclaw/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with OpenClaw](https://storeseo.com/agent/openclaw/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with OpenClaw](https://storeseo.com/agent/openclaw/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with OpenClaw](https://storeseo.com/agent/openclaw/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with OpenClaw](https://storeseo.com/agent/openclaw/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with OpenClaw](https://storeseo.com/agent/openclaw/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with OpenClaw](https://storeseo.com/agent/openclaw/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with OpenClaw](https://storeseo.com/agent/openclaw/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with OpenClaw](https://storeseo.com/agent/openclaw/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect OpenClaw to StoreSEO? Run openclaw mcp add storeseo --url https://mcp.storeseo.com/mcp --transport streamable-http --auth oauth, then openclaw mcp login storeseo. Open the authorization URL it prints, pick your Shopify store and approve. Finish with openclaw mcp doctor storeseo --probe to confirm the connection. ### Do I need a StoreSEO API key for OpenClaw? No. The login uses OAuth and OpenClaw saves the credentials it receives. If you would rather use a personal access token, generate one in StoreSEO under Settings, AI Toolkit and send it as an Authorization Bearer header. OpenClaw's add command accepts a header option for HTTP servers. ### Will OpenClaw change my store without asking? StoreSEO tells the agent to preview, name any cost and wait for your yes before it applies content, optimizes images, changes settings, publishes llms.txt or agents.md, writes a description or deletes anything. That yes is an instruction the agent follows, not a lock on the server, so OpenClaw's own approval setting for the server is the hard stop. Reads run straight away. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can OpenClaw run StoreSEO checks on a schedule? Yes. OpenClaw automations run a prompt on a cron expression and can announce the result in a chat channel. Ask for reads and reports only, because a scheduled run has nobody to answer StoreSEO's confirmation, and set the server's approval setting so a write cannot slip through unattended. ### How do I check that OpenClaw can reach StoreSEO? Run openclaw mcp doctor storeseo --probe. It opens a live connection and reports the tools the server offers. Then ask for your store's SEO score breakdown in a chat or the CLI to see a real answer. ### Where does the sign-in live if OpenClaw runs on another machine? On the installation that owns the connector. Run openclaw mcp add and openclaw mcp login there, so the credentials are saved where the agent and its automations actually run. ### How do I disconnect OpenClaw from my store? Remove the storeseo server with OpenClaw's mcp commands, for example openclaw mcp logout storeseo to drop the credentials. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Code and StoreSEO](https://storeseo.com/agent/claude-code) Setup checked against: - [OpenClaw: connect MCP servers](https://docs.openclaw.ai/tools/mcp) - [OpenClaw: MCP registry commands](https://docs.openclaw.ai/cli/mcp/registry) - [OpenClaw: automations](https://docs.openclaw.ai/automation/cron-jobs) - [OpenClaw: manage automations](https://docs.openclaw.ai/automation/cron-jobs/managing-jobs) --- Canonical: https://storeseo.com/agent/openclaw Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Hermes Agent > Connect Hermes Agent to StoreSEO with a few lines in config.yaml, then audit and fix your Shopify store's SEO from the terminal or chat, or on a schedule. Hermes Agent is an open-source AI agent from Nous Research that keeps memory across sessions, works from a terminal or from chat apps, and runs scheduled jobs of its own. Connected to StoreSEO through the MCP server, it audits your store's SEO, previews optimized content and applies what you approve, on request or on a timer. - Three lines in ~/.hermes/config.yaml - OAuth through your Shopify admin - Scheduled read-only reports with hermes cron ## How do I connect Hermes Agent to StoreSEO? ### 1. Add StoreSEO to config.yaml Add this under mcp_servers in ~/.hermes/config.yaml. If the file already has an mcp_servers block, add only the storeseo entry. Then start a new Hermes session, or run /reload-mcp in the one that is open. ```yaml mcp_servers: storeseo: url: "https://mcp.storeseo.com/mcp" auth: oauth ``` ### 2. Sign in to your store On first connect Hermes prints an authorize URL and waits for the sign-in. Run this command to start it yourself or to re-authorize later. Pick your Shopify store and approve. Hermes caches the tokens in ~/.hermes/mcp-tokens/storeseo.json. ```bash hermes mcp login storeseo ``` ### 3. Or connect with an access token (optional) On a machine with no browser, generate a token in StoreSEO under Settings, AI Toolkit and send it as a header in place of auth: oauth. Keep config.yaml out of version control when it holds a token. ```yaml mcp_servers: storeseo: url: "https://mcp.storeseo.com/mcp" headers: Authorization: "Bearer YOUR_TOKEN" ``` **Sign-in:** Set auth: oauth and run hermes mcp login storeseo: pick your Shopify store in the browser and approve. Hermes caches the tokens locally and refreshes them. Without a browser, send a StoreSEO personal access token from Settings, AI Toolkit as an Authorization header. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-hermes-agent ### Example session (illustrative) **You:** Which pages rank on page two in Google, and what is the weakest one on my store? → `storeseo_start` (read-only) → `storeseo_get_search_console_insights` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** The Linen Shirt page gets impressions but has a thin meta description. StoreSEO drafted a new meta title, description and FAQ. The preview used AI credit and nothing is saved yet. Apply it? **You:** Yes, apply it. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. The page re-scored higher. Next I would fix the alt text on its images. Want that? ## What is Hermes Agent? Hermes Agent is an open-source AI agent from Nous Research. It keeps memory across sessions, runs from a terminal or from chat apps such as Telegram, Discord and Slack, and loads skills from its own folder under ~/.hermes. It connects to remote MCP servers through an mcp_servers block in ~/.hermes/config.yaml. A server entry needs a url, then either auth: oauth with hermes mcp login, or an Authorization header. Hermes handles discovery, PKCE and token refresh, caches tokens locally, and lets you narrow a server's tools with include and exclude lists. Hermes also has a built-in scheduler. hermes cron create takes a schedule and a prompt, and delivers the result to a chat target such as Telegram, so a check can run on its own and arrive where you read your messages. ## Why use Hermes Agent with StoreSEO? - **Memory across sessions.** Hermes remembers what it learned earlier, so a follow-up such as continue with the next five products does not start from scratch. StoreSEO still opens its own session for the store snapshot. - **Scheduled SEO reports.** hermes cron create runs a prompt on a schedule and delivers it to a chat target. Use it for a weekly score summary or a Search Console review that lands without you asking. - **Smaller tool surface if you want it.** Hermes tool filters let you include or exclude StoreSEO tools by name or pattern, so a reporting agent can be limited to reads while another keeps the full set. ## Good to know - Start a new session or run /reload-mcp after editing config.yaml, otherwise the open session does not see the storeseo server. - A token in config.yaml is a secret. Keep the file out of version control, and revoke the token in StoreSEO under Settings, AI Toolkit if it leaks. - StoreSEO has 44 tools. If you want a read-only agent, use Hermes' tools include list to expose only the StoreSEO list and get tools. - A scheduled run has nobody to say yes. Keep cron jobs to reads and reports, and ask for changes in a chat you can answer. ## What can you ask Hermes Agent to do in StoreSEO? - Use StoreSEO to list my products scoring under 50 and preview new meta for the first five. - Which products have no alt text on their images? Show me the list before writing anything. - Pull my Search Console queries from the last 28 days and tell me which ones sit between positions 8 and 20. - Is my llms.txt current? If not, say what generating a new one will publish and cost before you run it. - Summarize my store's SEO score breakdown in five lines. ## Send a weekly SEO report with hermes cron hermes cron create takes a schedule and a prompt, and --deliver sends the result to a chat target. Inside a chat you can use /cron add with the same arguments. A scheduled run should be report-only: ask for reads, because nobody is there to say yes to a write, and do not tell the job to apply or generate anything. ```bash hermes cron create "every monday 9am" \ "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products. Read only: do not preview, apply or generate anything." \ --name "Monday SEO report" \ --deliver telegram ``` ## Hermes Agent guides by job - [Audit your store's SEO with Hermes Agent](https://storeseo.com/agent/hermes-agent/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Hermes Agent](https://storeseo.com/agent/hermes-agent/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Hermes Agent](https://storeseo.com/agent/hermes-agent/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Hermes Agent](https://storeseo.com/agent/hermes-agent/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Hermes Agent](https://storeseo.com/agent/hermes-agent/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Hermes Agent](https://storeseo.com/agent/hermes-agent/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Hermes Agent](https://storeseo.com/agent/hermes-agent/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Hermes Agent](https://storeseo.com/agent/hermes-agent/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Hermes Agent](https://storeseo.com/agent/hermes-agent/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Hermes Agent](https://storeseo.com/agent/hermes-agent/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Hermes Agent to StoreSEO? Add an mcp_servers entry named storeseo to ~/.hermes/config.yaml with the url https://mcp.storeseo.com/mcp and auth: oauth. Run hermes mcp login storeseo, pick your Shopify store in the browser and approve, then start a new session or run /reload-mcp. ### Do I need a StoreSEO API key for Hermes Agent? No. With auth: oauth Hermes signs in through your Shopify admin and caches the tokens it receives. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, sent as an Authorization header for machines with no browser. ### Where does Hermes store the StoreSEO sign-in? The server entry lives in ~/.hermes/config.yaml, and the OAuth tokens are cached at ~/.hermes/mcp-tokens/storeseo.json. Hermes reuses them on later runs until a refresh fails, and then you run hermes mcp login again. ### Will Hermes Agent change my store without asking? StoreSEO tells the agent to preview, name any cost and wait for your yes before it applies content, optimizes images, changes settings, publishes llms.txt or agents.md, writes a description or deletes anything. That yes is an instruction the agent follows, not a lock on the server, so limit the tools Hermes can see if you need a hard stop. Reads run straight away. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can Hermes run StoreSEO checks on a schedule? Yes. hermes cron create takes a schedule such as every monday 9am and a prompt, and --deliver sends the result to a chat target. Ask for reads and reports only, because a scheduled run cannot answer StoreSEO's confirmation, and use Hermes tool filters to keep write tools out of that agent. ### How do I disconnect Hermes Agent from my store? Remove the storeseo block from ~/.hermes/config.yaml and delete the cached token file under ~/.hermes/mcp-tokens. Also revoke the connection or token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [OpenClaw and StoreSEO](https://storeseo.com/agent/openclaw) Setup checked against: - [Hermes Agent: MCP](https://hermes-agent.nousresearch.com/docs/user-guide/features/mcp) - [Hermes Agent: scheduled jobs](https://hermes-agent.nousresearch.com/docs/user-guide/features/cron) --- Canonical: https://storeseo.com/agent/hermes-agent Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Paperclip > Give the agents Paperclip runs access to StoreSEO through their MCP config, then schedule report-only Shopify SEO audits and approve every change yourself. Paperclip is an open-source, self-hosted platform that manages teams of AI agents such as Claude Code, with an org chart, budgets, approvals and heartbeat schedules. Paperclip does not call tools itself, so you give StoreSEO to an agent it runs, and that agent audits your Shopify store's SEO as a Paperclip task. - Wired through the MCP config of the agent Paperclip runs - Token header for unattended runs, OAuth once for browser sign-in - Scheduled routines that report on your store and change nothing ## How do I connect Paperclip to StoreSEO? ### 1. Add StoreSEO to the agent Paperclip runs Run this on the machine that hosts the agent, as the same OS user the Paperclip heartbeat runs as, from the agent's working directory. It is the claude_local adapter path, and user scope makes StoreSEO available to every Claude Code agent that user runs. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp ``` ### 2. Sign in once, then let Paperclip run headless Start claude interactively as that same user, run /mcp, choose storeseo and finish the browser sign-in: pick your Shopify store and approve. Claude Code keeps the credentials, so later Paperclip runs need no browser. Check with claude mcp get storeseo. ```text /mcp ``` ### 3. Or connect with an access token (optional) If the host has no browser, generate a token in StoreSEO under Settings, AI Toolkit and add the server with the token as a header instead of signing in. Keep the token in a Paperclip secret rather than in a file you commit. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp \ --header "Authorization: Bearer YOUR_TOKEN" ``` ### 4. Or connect it as a Paperclip connector (optional) On the Connectors page choose Connect your own MCP server and paste the URL. Paperclip walks you through the sign-in and lists StoreSEO's tools so you can set each one to Allowed, Ask first or Off. It shows as an Unverified server, which describes Paperclip's review, not the connection. ```text https://mcp.storeseo.com/mcp ``` **Sign-in:** Paperclip runs its agents headless, so nobody is there to click a browser prompt. Sign in once with OAuth in an interactive session as the same OS user that runs the agent, or skip the browser and send a personal access token from StoreSEO under Settings, AI Toolkit as a Bearer header. Verify with: "List the MCP tools you have access to without calling any of them, then use StoreSEO to show my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-paperclip ### Example session (illustrative) **You:** Ticket: review my store's weakest products and propose better meta for the worst one. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** Linen Shirt scores lowest and a preview is ready, projected 41 to 78. The preview used AI credit; applying is free and changes the live product. Apply it? **You:** Approved. Keep the current URL handle. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied without the handle change and posted the new score on the ticket. Next in the queue: alt text on the top collection's images. ## What is Paperclip? Paperclip is an open-source, MIT-licensed platform for managing teams of AI agents at work. You define a goal, hire agents such as Claude Code, Codex or Hermes into an org chart, set budgets and approvals, and follow every task and tool call from one dashboard. It is self-hosted, needs no Paperclip account and starts with npx paperclipai onboard. Paperclip orchestrates agents rather than being one. An adapter launches the runtime that does the work, and that runtime is what speaks MCP. For the Claude Code adapter that means the server is added with claude mcp add on the host, and Paperclip's documentation says there is no separate paperclipai mcp add. Paperclip also has a Connectors page where you can attach your own MCP server by URL and set each action to Allowed, Ask first or Off. Agents wake on a schedule, on assignment or on demand. Timer heartbeats are off by default, and recurring work belongs in routines, which fire on a cron schedule or a webhook and hand a task to an agent. That makes Paperclip a fit for teams that want an SEO agent working on a rhythm, with a human reading what it did. ## Why use Paperclip with StoreSEO? - **SEO work as tickets.** Every instruction, tool call and answer sits on a Paperclip ticket with a trace, so you can read what the agent asked StoreSEO and what came back before you approve a change. - **A rhythm, not a one-off.** A routine can start a weekly store review for an agent without you opening a chat. Reads such as scores and reports are a safe thing to schedule, and you decide what to act on. - **Budgets around the agent.** Paperclip gives each agent a monthly budget and pauses it at the limit. StoreSEO's own AI credit still applies to generated content, so the two limits sit side by side. ## Good to know - The Claude Code adapter runs headless with dangerouslySkipPermissions on by default, which lets the agent use connected MCP tools without asking. Set it to false so a StoreSEO tool that needs approval does not run unattended, or use the Paperclip connector and keep write tools on Ask first. - Paperclip's action permissions and review queue apply to connectors only. A server added with claude mcp add on the adapter does not pass through them, so the agent's own settings and StoreSEO's confirmation are your checks there. - Remote sign-in cannot finish headless. Complete OAuth once in an interactive claude session as the same OS user that runs the heartbeat, or use a token. A 401 after some hours means the refresh failed and you should sign in again. - User scope in ~/.claude.json is shared by every agent that OS user runs, so every Claude Code agent on that host sees StoreSEO. Use a dedicated OS user or a project scope if only one agent should. ## What can you ask Paperclip to do in StoreSEO? - Check my StoreSEO scores and report the five lowest-scoring products on this ticket. Do not change anything. - Weekly review: list products with no meta description and images with no alt text, then summarize what a fix would involve. - Preview better meta for the lowest-scoring product, show me old and new, and wait for my reply before applying. - Pull my Search Console queries for the last 28 days and note the ones ranking between 8 and 20. - Is my llms.txt current? Report its status and tell me what generating a new one would change before running it. ## Schedule a report-only store review Paperclip routines fire on a cron schedule or a webhook, create a task and wake the agent you assign. Keep any scheduled StoreSEO run to reads and a written report: nobody is there to say yes, and the Claude Code adapter does not stop to ask by default. Leave applying, generating and publishing for a task you start and approve yourself. ```text Weekly store review. Use StoreSEO to read my SEO scores, list the five lowest-scoring products and any Search Console drops, and post the findings on this task. Read only: do not preview, apply, generate or publish anything. ``` ## Paperclip guides by job - [Audit your store's SEO with Paperclip](https://storeseo.com/agent/paperclip/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Paperclip](https://storeseo.com/agent/paperclip/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Paperclip](https://storeseo.com/agent/paperclip/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Paperclip](https://storeseo.com/agent/paperclip/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Paperclip](https://storeseo.com/agent/paperclip/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Paperclip](https://storeseo.com/agent/paperclip/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Paperclip](https://storeseo.com/agent/paperclip/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Paperclip](https://storeseo.com/agent/paperclip/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Paperclip](https://storeseo.com/agent/paperclip/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Paperclip](https://storeseo.com/agent/paperclip/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Paperclip to StoreSEO? Paperclip does not call MCP tools itself, so you add StoreSEO to an agent it runs. For a Claude Code agent, run claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp as the OS user that runs the heartbeat, then sign in once with /mcp in an interactive claude session. You can also attach the URL on Paperclip's Connectors page with Connect your own MCP server. ### Do I need a StoreSEO API key for Paperclip? Not necessarily. OAuth works if you complete the sign-in once in an interactive session as the same user the agent runs as. Because Paperclip runs agents without a browser, many people prefer a personal access token from StoreSEO under Settings, AI Toolkit, sent as an Authorization Bearer header. ### Does Paperclip itself have a StoreSEO connector? No. StoreSEO is not in Paperclip's connector catalog. You either configure the MCP server on the agent's runtime or paste the StoreSEO URL into Paperclip's Connect your own MCP server option, where it appears as an unverified server. ### Will a Paperclip agent change my store without asking? StoreSEO tells the agent to preview, say what it will cost and wait for your yes before it applies content, optimizes images, changes settings, publishes a file or deletes anything. That is an instruction, not a lock. Paperclip's Claude Code adapter runs headless and skips permission prompts by default, so set dangerouslySkipPermissions to false or keep write actions on Ask first in a Paperclip connector, and keep scheduled runs read-only. ### Can Paperclip run StoreSEO checks on a schedule? Yes. Paperclip routines fire on a cron schedule or a webhook and give a task to an agent. Make the routine's instructions read-only, such as scores and Search Console reports, and have the agent post findings on the task for you to review. ### Which StoreSEO plan does Paperclip need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. Paperclip itself is open source and self-hosted. ### Which Paperclip agents can use StoreSEO? Paperclip documents MCP paths for its Claude Code adapter and its Hermes adapter. Other adapters inherit whatever MCP support their own tool has, but Paperclip does not document that yet, so check that agent's own MCP setup first. ### How do I disconnect Paperclip from my store? Run claude mcp remove storeseo as the OS user the agent runs as, or delete the connector on Paperclip's Connectors page. Then revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app and token is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [Claude Code with StoreSEO](https://storeseo.com/agent/claude-code) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) Setup checked against: - [Paperclip: add an MCP server to an agent](https://docs.paperclip.ing/how-to/add-mcp-server-to-agent/) - [Paperclip: Claude Code adapter](https://docs.paperclip.ing/reference/adapters/claude-code/) - [Paperclip: connect a custom MCP server](https://docs.paperclip.ing/connectors/custom-mcp-servers/) - [Paperclip: heartbeats and routines](https://docs.paperclip.ing/guides/projects-workflow/routines/) - [Paperclip](https://paperclip.ing) --- Canonical: https://storeseo.com/agent/paperclip Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from NanoClaw > Add StoreSEO to your NanoClaw agent as a remote MCP server with an access token, then ask about your store's SEO from WhatsApp, Telegram, Slack or Discord. NanoClaw is a small open-source assistant that runs each agent in its own container, is built on Anthropic's Claude Agent SDK and answers you in WhatsApp, Telegram, Slack or Discord. With StoreSEO added to the agent, you ask in a chat message and it reads your store's SEO scores, previews optimized content and applies what you approve. - Message it from the chat app you already use - One MCP server added to your own fork's agent - Scheduled tasks for a weekly read-only SEO report ## How do I connect NanoClaw to StoreSEO? ### 1. Create a StoreSEO access token In the StoreSEO app open Settings, AI Toolkit and generate a personal access token. Keep it out of chat messages and out of Git. ```text Settings → AI Toolkit → create a token ``` ### 2. Add StoreSEO to your agent's container Run this on the machine that hosts NanoClaw, using your agent group's ID from ncl groups list. The header goes in as a JSON object. NanoClaw may ask an admin to approve the change. ```bash ncl groups config add-mcp-server --id --name storeseo \ --url https://mcp.storeseo.com/mcp \ --headers '{"Authorization":"Bearer YOUR_TOKEN"}' ``` ### 3. Restart the agent's container The new server takes effect after a restart. Then message your agent with its trigger word, @Andy by default, and ask for your store's SEO score. ```bash ncl groups restart --id ``` ### 4. Or ask Claude Code to wire it into your fork (optional) NanoClaw is customized by changing your own copy, and its README says to tell Claude Code what you want. Run this inside your NanoClaw folder if you would rather not type the command yourself. ```text Add the remote MCP server https://mcp.storeseo.com/mcp to my agent as storeseo, authenticated with this Authorization: Bearer token, then restart the agent. ``` **Sign-in:** NanoClaw's documentation describes no browser sign-in for an MCP server inside the agent's container, so connect with a personal access token from StoreSEO under Settings, AI Toolkit, sent as an Authorization: Bearer header. Verify with: "@Andy use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-nanoclaw ### Example session (illustrative) **You:** @Andy which of my products have the weakest SEO right now? Show a better title and description for the worst one. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** Linen Shirt scores lowest. StoreSEO drafted a new meta title and description. The preview used AI credit and nothing is saved yet. Apply them to the product? **You:** Yes, apply them. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. The product was re-scored. Want me to check the next one? ## What is NanoClaw? NanoClaw is an open-source AI assistant that runs agents in their own containers. It is built on Anthropic's Claude Agent SDK and is meant to be small enough to read, so you own a fork of the code and change it to fit what you need rather than editing a settings file. You talk to it from a chat app. Its README lists WhatsApp, Telegram, Discord, Slack, Microsoft Teams, iMessage and others, installed as channels, and an agent answers when you use its trigger word, which defaults to @Andy. Each agent group has its own memory, workspace and container, and it can run scheduled tasks that you create from the command line or ask for in chat. NanoClaw does not have a menu for adding connectors. A remote MCP server is added to an agent group's container configuration with the ncl command line, or by asking Claude Code to change your fork, and the agent's container is restarted to pick it up. ## Why use NanoClaw with StoreSEO? - **Ask from your phone.** The chat app you already use becomes the front end. Send one message about your store from WhatsApp, Telegram, Slack or Discord and the agent calls StoreSEO for the answer. - **You own the agent.** NanoClaw runs on your own machine in containers you control, with its own memory per agent group. StoreSEO is one server added to one agent, and you can remove it the same way. - **A report that arrives on its own.** NanoClaw can run a recurring task and post the result to a chat. Point one at StoreSEO for a weekly read-only SEO summary without opening the Shopify admin. ## Good to know - NanoClaw's documentation describes no browser OAuth sign-in for an MCP server inside the container, so this setup uses a StoreSEO access token. Treat the token like a password: the header is stored in the agent group's container configuration on your host, so keep that machine and its backups private. - NanoClaw says secrets belong with its credential gateway rather than in the agent. If you run a gateway, you can hold the StoreSEO token there for mcp.storeseo.com instead of in the header, following NanoClaw's gateway documentation. - The server is added per agent group. Add it to each group that should reach your store, and restart that group's container afterwards. - NanoClaw needs Docker on your own machine and the setup is yours to maintain. Whether it asks you before a tool runs depends on how your fork and your admins are set up, so the StoreSEO confirmation in the chat may be the only check. ## What can you ask NanoClaw to do in StoreSEO? - @Andy use StoreSEO to list my five lowest-scoring products and what is wrong with each. - @Andy which of my product images have no alt text? Show me the first ten. - @Andy preview better meta for my best-selling collection and wait for my yes before saving anything. - @Andy is my llms.txt live? Tell me when it was last generated. - @Andy show Search Console queries from the last 28 days where we rank between 8 and 20. ## Send a weekly SEO report with a NanoClaw scheduled task NanoClaw runs a prompt on a cron schedule in the agent's own session. A scheduled task has no chat attached, so the prompt has to say where to send the result. A scheduled run should be report-only: the prompt asks for reads, because nobody is there to say yes to a write. ```bash ncl tasks create \ --group \ --name "Monday SEO report" \ --recurrence "0 8 * * 1" \ --prompt "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products, and send it to telegram. Read only: do not preview, apply, generate or delete anything." ``` ## NanoClaw guides by job - [Audit your store's SEO with NanoClaw](https://storeseo.com/agent/nanoclaw/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with NanoClaw](https://storeseo.com/agent/nanoclaw/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with NanoClaw](https://storeseo.com/agent/nanoclaw/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with NanoClaw](https://storeseo.com/agent/nanoclaw/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with NanoClaw](https://storeseo.com/agent/nanoclaw/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with NanoClaw](https://storeseo.com/agent/nanoclaw/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with NanoClaw](https://storeseo.com/agent/nanoclaw/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with NanoClaw](https://storeseo.com/agent/nanoclaw/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with NanoClaw](https://storeseo.com/agent/nanoclaw/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with NanoClaw](https://storeseo.com/agent/nanoclaw/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect NanoClaw to StoreSEO? Create an access token in StoreSEO under Settings, AI Toolkit. Then run ncl groups config add-mcp-server with the name storeseo, the URL https://mcp.storeseo.com/mcp and the token as an Authorization Bearer header, and restart the agent group with ncl groups restart. You can also ask Claude Code to make the same change in your NanoClaw fork. ### Do I need a StoreSEO API key for NanoClaw? You need a personal access token from StoreSEO under Settings, AI Toolkit. NanoClaw's documentation describes no browser OAuth sign-in for an MCP server inside its container, so the token is the supported route here. ### Where does NanoClaw keep the StoreSEO server? In the container configuration of the agent group you added it to. NanoClaw stores that configuration in its database and writes it to the group's container.json when the container starts, so the server is per agent group. ### Will NanoClaw change my products without asking? StoreSEO tells the agent to preview first and wait for your yes in the chat before it applies content, optimizes images, publishes a file or deletes anything. That yes is an instruction the agent follows, not a lock on the server, and NanoClaw's own approval behavior depends on how you configured your fork. Reads such as scores and reports run straight away. ### Can NanoClaw send me a regular SEO report? Yes. NanoClaw has scheduled tasks that run a prompt on a cron schedule. Write the prompt as read-only and name the chat destination in it, because a scheduled task has no conversation attached and nobody is there to approve a write. ### Which StoreSEO plan does NanoClaw need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### How do I disconnect NanoClaw from my store? Remove the server from the agent group with ncl groups config remove-mcp-server and restart the group. Then revoke the token in StoreSEO under Settings, AI Toolkit, where every connected app and token is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Run Shopify SEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [NanoClaw on GitHub](https://github.com/nanocoai/nanoclaw) - [NanoClaw scheduled tasks](https://github.com/nanocoai/nanoclaw/blob/main/docs/scheduled-tasks.md) - [NanoClaw agent templates and MCP servers](https://github.com/nanocoai/nanoclaw/blob/main/docs/templates.md) - [NanoClaw ncl groups commands](https://github.com/nanocoai/nanoclaw/blob/main/src/cli/resources/groups.ts) --- Canonical: https://storeseo.com/agent/nanoclaw Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from OpenCode > Connect OpenCode to StoreSEO with one remote MCP entry and opencode mcp auth, then audit products, fix meta and publish llms.txt from your terminal. OpenCode is an open-source AI coding agent that runs in your terminal and can use remote MCP servers alongside its built-in tools. Connected to StoreSEO, it reads your store's SEO scores, previews optimized content and applies the changes you approve from the session you already have open. - One command to add, one command to sign in - OAuth by default, with no client ID to paste - Hide any StoreSEO tool with a permission rule ## How do I connect OpenCode to StoreSEO? ### 1. Add StoreSEO to OpenCode Run this from the project that should use StoreSEO, or add --global to make it available in every project. OpenCode writes a remote server entry to its configuration; the equivalent hand-written entry is shown in the next step. ```bash opencode mcp add storeseo --url https://mcp.storeseo.com/mcp ``` ### 2. Sign in to your store Remote servers use OAuth by default. Run the auth command, then pick your Shopify store in the browser and approve access. Inside OpenCode, /mcps does the same: select storeseo and sign in. opencode mcp list should then show it as connected. ```bash opencode mcp auth storeseo opencode mcp list ``` ### 3. Or write it in the config (optional) In OpenCode V2 servers go under mcp.servers. Older versions put the server directly under mcp, so use the layout your version documents. ```json { "$schema": "https://opencode.ai/config.json", "mcp": { "servers": { "storeseo": { "type": "remote", "url": "https://mcp.storeseo.com/mcp" } } } } ``` ### 4. Or connect with an access token (optional) Generate a token in StoreSEO under Settings, AI Toolkit, export it as STORESEO_TOKEN, and turn OAuth off for this server so OpenCode sends the header instead. ```json { "mcp": { "servers": { "storeseo": { "type": "remote", "url": "https://mcp.storeseo.com/mcp", "oauth": false, "headers": { "Authorization": "Bearer {env:STORESEO_TOKEN}" } } } } } ``` **Sign-in:** OpenCode signs in with OAuth by default: run opencode mcp auth storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit, set oauth to false and send it as a header. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-opencode ### Example session (illustrative) **You:** Is my agents.md file live? If it is only Shopify's default, tell me what a custom one would do. → `storeseo_start` (read-only) → `storeseo_get_agents_md_status` (read-only) **Agent:** Only Shopify's default is live. A generated agents.md is written from your brand, top products and contact details, publishes to your live theme straight away and uses AI credit. Generate it? **You:** Yes, generate it. → `storeseo_generate_agents_md` (needs approval) **Agent:** It is live at your /agents.md address and I can show you the text. If you dislike it, I can reset to Shopify's default. ## What is OpenCode? OpenCode is an open-source AI coding agent built by the team behind SST, now Anomaly. It runs in your terminal, and also as a desktop and IDE app, and works through tasks by calling its built-in tools and tools from MCP servers. It supports local servers started by a command and remote servers reached over streamable HTTP. A remote server is a type, a url and optional headers under mcp.servers in the OpenCode config, and opencode mcp add writes that entry for you. OAuth is on by default: OpenCode discovers the authorization server, uses PKCE, registers itself when the server allows it, refreshes tokens and keeps the credentials out of your project files. Tool names reach the model as server_tool, and permission rules can match them with a pattern such as storeseo_* to hide or deny tools without disconnecting the server. ## Why use OpenCode with StoreSEO? - **Open source, no lock-in.** OpenCode works with the model provider you choose, so StoreSEO's tools are available whichever model you run your terminal sessions on. - **Sign-in handled for you.** OpenCode discovers StoreSEO's authorization server and registers itself, so there is no client ID or secret to copy. You approve your store once in the browser. - **Deny what you never use.** A permission rule that matches a StoreSEO tool name hides or denies it, which is a client-side stop on top of the confirmation StoreSEO asks for in chat. ## Good to know - OpenCode V2 expects servers under mcp.servers. Older versions put the server directly under mcp, so check the layout your installed version documents before you edit the file. - opencode mcp add writes to the project configuration unless you pass --global, so a server added without it only exists in that project. - There is no remove command for MCP servers. To remove StoreSEO, delete its entry from the config, and run opencode mcp logout storeseo to clear the stored sign-in. - The OpenCode docs describe deny rules for MCP tools but no ask-first mode, so StoreSEO's own confirmation in the chat is the main check on writes. Deny the write tools by name if you want a client-side stop. ## What can you ask OpenCode to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest first. - Is my agents.md live? If it is only Shopify's default, tell me what a custom one would publish. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. - Which schema types are enabled for my store, and what would turning on Product schema add? ## OpenCode guides by job - [Audit your store's SEO with OpenCode](https://storeseo.com/agent/opencode/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with OpenCode](https://storeseo.com/agent/opencode/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with OpenCode](https://storeseo.com/agent/opencode/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with OpenCode](https://storeseo.com/agent/opencode/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with OpenCode](https://storeseo.com/agent/opencode/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with OpenCode](https://storeseo.com/agent/opencode/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with OpenCode](https://storeseo.com/agent/opencode/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with OpenCode](https://storeseo.com/agent/opencode/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with OpenCode](https://storeseo.com/agent/opencode/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with OpenCode](https://storeseo.com/agent/opencode/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect OpenCode to StoreSEO? Run opencode mcp add storeseo --url https://mcp.storeseo.com/mcp, then opencode mcp auth storeseo. Your browser opens, you pick your Shopify store and approve access, and opencode mcp list should show storeseo as connected. ### Do I need a StoreSEO API key for OpenCode? No. Remote servers use OAuth by default and OpenCode keeps the credentials. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative: set oauth to false for the server and send the token in an Authorization header. ### Where does OpenCode store the StoreSEO server? In the OpenCode configuration, under mcp.servers in V2. opencode mcp add writes it to the current project, and the --global option writes it so every project sees it. ### Will OpenCode change my store without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything, and that yes is an instruction the agent follows, not a lock on the server. OpenCode's own permission rules can deny StoreSEO tools by name, which is the hard stop if you want one. Reads such as scores and reports run straight away. ### Which StoreSEO plan does OpenCode need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. See the pricing table for current plans. ### The sign-in did not start. What now? Run opencode mcp list. If it shows needs authentication, run opencode mcp auth storeseo or open OpenCode, type /mcps, select storeseo and sign in. If a browser is not available, use the access token option instead. ### How do I disconnect OpenCode from my store? Delete the storeseo entry from the OpenCode configuration and run opencode mcp logout storeseo to remove the stored sign-in. Then revoke the connection or the access token in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [OpenCode: MCP servers](https://opencode.ai/v2/docs/mcp-servers) - [OpenCode: MCP servers (earlier docs)](https://opencode.ai/docs/mcp-servers/) --- Canonical: https://storeseo.com/agent/opencode Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Zed > Connect Zed to StoreSEO with one context_servers entry in settings.json, then audit products, fix meta and publish llms.txt from the Agent Panel by asking. Zed is a code editor from Zed Industries with a built-in Agent Panel, where an AI agent works on your project and can call MCP tools. Connected to StoreSEO with one settings entry, it audits your Shopify store's SEO, previews optimized content and applies the changes you approve, next to your theme code. - One context_servers entry - Zed prompts for the OAuth sign-in - Zed confirms each tool call by default ## How do I connect Zed to StoreSEO? ### 1. Add StoreSEO as a remote server Open Settings, then AI, then MCP Servers and add a remote server, or add this to settings.json yourself. Leave out headers so Zed can sign you in. ```json { "context_servers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Sign in and check the dot Because the entry has no Authorization header, Zed prompts you to authenticate. Pick your Shopify store and approve in the browser. In Settings, AI, MCP Servers a green dot next to storeseo means the server is active. Then ask this in the Agent Panel. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or connect with an access token (optional) If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Keep settings.json out of a public repository when it holds a token. ```json { "context_servers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` **Sign-in:** With no Authorization header in the entry, Zed prompts you to sign in through the standard MCP OAuth flow: pick your Shopify store and approve. To use a StoreSEO personal access token instead, add it as a Bearer header in the same entry. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-zed ### Example session (illustrative) **You:** Check which schema types are on for my products and turn on what is missing. → `storeseo_start` (read-only) → `storeseo_get_schema_settings` (read-only) → `storeseo_list_schema_types` (read-only) **Agent:** Product schema is off. Turning it on changes your store's settings. Want me to enable it? **You:** Yes, enable it. → `storeseo_update_schema_settings` (needs approval) **Agent:** Product schema is on. The Liquid theme file that renders your FAQs is the next thing worth checking, and I can open it here. ## What is Zed? Zed is a code editor from Zed Industries. It has a built-in Agent Panel where an AI agent reads your project and works through tasks, and it supports MCP tools and prompts through context servers. A remote MCP server is added in settings.json under context_servers with a url. If the entry has no Authorization header, Zed prompts you to sign in through the standard MCP OAuth flow. Settings, AI, MCP Servers shows each server with a status dot, and green means the server is active. Zed has its own tool-approval setting, agent.tool_permissions.default, which is confirm by default: the agent panel asks before it runs a tool action. You can set a different value overall, or per tool with keys of the form mcp::. ## Why use Zed with StoreSEO? - **Theme code and SEO in one panel.** When a fix needs a Liquid change, such as where FAQs or a schema block render, the same Agent Panel can open the theme file next to the StoreSEO results. - **A client-side hard stop by default.** Zed asks before running a tool action unless you change agent.tool_permissions. That sits on top of the confirmation StoreSEO tells the agent to ask for, so a write has two checks. - **One entry, no tokens in the file.** With OAuth the entry is just the URL. Zed signs you in through your Shopify admin, so no secret sits in settings.json unless you choose the token route. ## Good to know - Zed's default is to ask before every tool action, reads included, so a catalog audit means several prompts. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_list_resources and leave the write tools on confirm. - Setting agent.tool_permissions.default to allow removes Zed's check. Then the confirmation StoreSEO asks the agent for in the chat is the only one, so use allow sparingly. - A token in settings.json is a secret. Revoke it in StoreSEO under Settings, AI Toolkit if the file is exposed. ## What can you ask Zed to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first five. - Which product images have no alt text? Show me the list before you change anything. - Is product schema turned on for my store? If not, tell me what enabling it changes. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Open the theme file that renders my FAQ section and tell me whether StoreSEO's FAQ output will show there. ## Zed guides by job - [Audit your store's SEO with Zed](https://storeseo.com/agent/zed/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Zed](https://storeseo.com/agent/zed/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Zed](https://storeseo.com/agent/zed/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Zed](https://storeseo.com/agent/zed/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Zed](https://storeseo.com/agent/zed/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Zed](https://storeseo.com/agent/zed/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Zed](https://storeseo.com/agent/zed/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Zed](https://storeseo.com/agent/zed/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Zed](https://storeseo.com/agent/zed/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Zed](https://storeseo.com/agent/zed/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Zed to StoreSEO? Add a context_servers entry named storeseo with the url https://mcp.storeseo.com/mcp to settings.json, or add a remote server under Settings, AI, MCP Servers. With no Authorization header, Zed prompts you to sign in: pick your Shopify store and approve. ### Do I need a StoreSEO API key for Zed? No. Zed signs in with the standard MCP OAuth flow when the entry has no Authorization header. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, added as a Bearer value in a headers object. ### How do I know the StoreSEO server is connected in Zed? Open Settings, AI, MCP Servers. A green dot next to storeseo means the server is active. Then ask the Agent Panel for your store's SEO score breakdown to see a real answer. ### Will Zed change my store without asking? StoreSEO tells the agent to preview, name any cost and wait for your yes before it applies content, optimizes images, changes settings, publishes llms.txt or agents.md, writes a description or deletes anything. That yes is an instruction the agent follows, not a lock on the server. Zed's own agent.tool_permissions setting is the hard stop: it defaults to confirm, which prompts before every tool call, reads included. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can I stop Zed asking for every read? Yes. Zed supports per-tool permission keys in the form mcp:server:tool, so you can allow StoreSEO's read tools and leave the write tools on confirm. Check Zed's MCP documentation for the exact settings shape in your version. ### How do I disconnect Zed from my store? Remove the storeseo entry from context_servers in settings.json or remove the server in Settings, AI, MCP Servers. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Code and StoreSEO](https://storeseo.com/agent/claude-code) Setup checked against: - [Zed: Model Context Protocol](https://zed.dev/docs/ai/mcp) --- Canonical: https://storeseo.com/agent/zed Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Kiro > Connect Kiro to StoreSEO with one url entry in mcp.json, then audit products, fix meta and alt text and publish llms.txt by asking the Kiro agent. Kiro is an agentic IDE built and operated by AWS, where an AI agent plans and edits your project from a chat panel. Connected to StoreSEO through one entry of MCP configuration, it audits your Shopify store's SEO, previews optimized content and applies what you approve without leaving the editor. - One url entry in mcp.json - Workspace or user-level config - An autoApprove list you control ## How do I connect Kiro to StoreSEO? ### 1. Add StoreSEO to your Kiro MCP config Open the Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows and Linux), search for MCP and open the workspace config (.kiro/settings/mcp.json) or the user config (~/.kiro/settings/mcp.json). Paste this and save. Kiro applies the change on save, and the first connection opens a browser sign-in: pick your Shopify store and approve. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Choose which StoreSEO tools Kiro may run (optional) Optional. autoApprove lists tools Kiro runs without asking, and disabledTools hides tools from the agent. The names here are real StoreSEO read tools. Keep write tools such as apply, generate and update out of autoApprove so Kiro's own prompt stays as a second check. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "autoApprove": ["storeseo_start", "storeseo_list_resources", "storeseo_get_resource_seo"] } } } ``` ### 3. Or connect with an access token (optional) If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and send it in a headers entry. Keep the file out of a public repository when it holds a token. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` **Sign-in:** Kiro handles the browser OAuth flow for remote servers and supports Dynamic Client Registration: pick your Shopify store and approve. To use a StoreSEO personal access token instead, add an Authorization Bearer value under headers in the same entry. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-kiro ### Example session (illustrative) **You:** Find my lowest-scoring product and show me what StoreSEO would change. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_get_resource_seo` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted a new meta title, description and FAQ for the Linen Shirt. The preview spent AI credit and nothing is saved yet. Apply it? **You:** Apply it, but keep the current URL handle. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied without the handle change. The product re-scored higher. Want me to look at its image alt text next? ## What is Kiro? Kiro is an agentic IDE built and operated by AWS. An AI agent works on your project from a chat panel, plans multi-step changes and can call tools exposed through the Model Context Protocol, including remote HTTP servers. MCP servers are defined in JSON. A workspace file at .kiro/settings/mcp.json applies to one project and a user file at ~/.kiro/settings/mcp.json applies everywhere, with the workspace file winning when both define the same server. A remote server needs a url and optionally headers, and Kiro applies changes when you save the file. Each server can carry an autoApprove list of tools that run without a prompt and a disabledTools list that hides tools from the agent. For remote servers that need a sign-in, Kiro handles the browser OAuth flow and supports Dynamic Client Registration. ## Why use Kiro with StoreSEO? - **Theme work and SEO together.** Kiro already plans and edits your Shopify theme. The same session can read StoreSEO's scores, then open the Liquid file where a fix has to land. - **Control over what runs unprompted.** List only read tools in autoApprove and every write still reaches you twice: once as Kiro's prompt and once as the confirmation StoreSEO tells the agent to ask for. - **Per project or everywhere.** Put StoreSEO in the workspace file for a single store's repository, or in the user file to have it in every project you open. ## Good to know - A workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, so if the sign-in seems to point at the wrong store, check which file defines storeseo. - autoApprove with a wildcard would run every StoreSEO tool without Kiro asking, writes included. List individual read tools instead, so Kiro's prompt stays a second check. - A token in mcp.json is a secret. Do not commit a workspace file that holds one, and revoke the token in StoreSEO under Settings, AI Toolkit if it leaks. ## What can you ask Kiro to do in StoreSEO? - Use StoreSEO to list my five lowest-scoring products and preview new meta for each. - Which product images have no alt text? List them before changing anything. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Turn on product schema and tell me what that changes before you do it. ## Kiro guides by job - [Audit your store's SEO with Kiro](https://storeseo.com/agent/kiro/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Kiro](https://storeseo.com/agent/kiro/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Kiro](https://storeseo.com/agent/kiro/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Kiro](https://storeseo.com/agent/kiro/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Kiro](https://storeseo.com/agent/kiro/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Kiro](https://storeseo.com/agent/kiro/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Kiro](https://storeseo.com/agent/kiro/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Kiro](https://storeseo.com/agent/kiro/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Kiro](https://storeseo.com/agent/kiro/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Kiro](https://storeseo.com/agent/kiro/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Kiro to StoreSEO? Open the Command Palette, search for MCP and open the workspace or user MCP config. Add an mcpServers entry named storeseo with the url https://mcp.storeseo.com/mcp and save. Kiro opens a browser sign-in where you pick your Shopify store and approve. ### Do I need a StoreSEO API key for Kiro? No. Kiro handles the OAuth sign-in for remote servers. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, sent as an Authorization Bearer value in a headers entry. ### Where does Kiro keep the StoreSEO server? In .kiro/settings/mcp.json for one workspace, or ~/.kiro/settings/mcp.json for every project. If both define the same name, the workspace file wins. ### Will Kiro change my store without asking? StoreSEO tells the agent to preview, name any cost and wait for your yes before it applies content, optimizes images, changes settings, publishes llms.txt or agents.md, writes a description or deletes anything. That yes is an instruction the agent follows, not a lock on the server. Kiro's own tool approval, including what you put in autoApprove, is the hard stop, so keep write tools out of that list. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Can I hide StoreSEO tools I do not use? Yes. Add their names to the disabledTools list for the storeseo server and Kiro hides them from the agent. StoreSEO exposes 44 tools, and a smaller set can make the agent's choices simpler. ### How do I disconnect Kiro from my store? Remove the storeseo entry from the mcp.json file where you added it, or set disabled to true to switch it off. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Zed and StoreSEO](https://storeseo.com/agent/zed) Setup checked against: - [Kiro: MCP configuration](https://kiro.dev/docs/mcp/configuration/) - [Kiro: MCP overview](https://kiro.dev/docs/mcp/) --- Canonical: https://storeseo.com/agent/kiro Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Antigravity > Connect Google Antigravity to StoreSEO with one serverUrl entry in mcp_config.json, then audit products, fix meta and publish llms.txt by asking the agent. Antigravity is Google's agentic coding environment, where an AI agent plans, edits and runs tasks in your project. Connected to StoreSEO through the MCP server, its agent audits your Shopify store's SEO, previews optimized content and applies what you approve while you stay in the editor. - One serverUrl entry in mcp_config.json - OAuth with dynamic client registration - Ask mode by default for unconfigured MCP tools ## How do I connect Antigravity to StoreSEO? ### 1. Add StoreSEO in View raw config In the Antigravity IDE open the menu in the agent side panel, choose MCP Servers, then Manage MCP Servers, then View raw config. Paste this into mcp_config.json and save. The field is serverUrl: Antigravity does not support url or httpUrl for remote servers. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Sign in, then check the connection Open Agent settings (Cmd+, on Mac, Ctrl+, on Windows and Linux), go to Customizations and click Authenticate next to storeseo. Pick your Shopify store and approve in the browser, copy the authorization code it shows, paste it into the settings panel and click Submit. Then ask this in the agent panel. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### 3. Or connect with an access token (optional) If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Keep mcp_config.json out of a public repository when it holds a token. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` **Sign-in:** Antigravity supports OAuth through dynamic client registration, which StoreSEO's server offers. Click Authenticate next to the server, approve your Shopify store in the browser, then paste the authorization code back. Or send a StoreSEO personal access token as an Authorization Bearer header. Verify with: "Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-antigravity ### Example session (illustrative) **You:** Which of my product images are still not optimized? → `storeseo_start` (read-only) → `storeseo_list_images` (read-only) **Agent:** Twelve images on your Summer Dresses collection are not optimized. Compressing them needs the Image Optimizer add-on and counts against its monthly quota. Optimize the first batch? **You:** Yes, do the first batch. → `storeseo_optimize_image` (needs approval) **Agent:** The first batch is compressed. I can restore any image if a result looks wrong. Want me to continue with the rest? ## What is Antigravity? Antigravity is Google's agentic coding product. Its agent plans multi-step work in your project and can call tools exposed through the Model Context Protocol, from the IDE's agent side panel. MCP servers are defined in mcp_config.json, either globally at ~/.gemini/config/mcp_config.json or per workspace at .agents/mcp_config.json. A remote server uses the serverUrl field, with an optional headers object for token auth. For OAuth, Antigravity supports dynamic client registration, so a compliant server needs no extra setup. By default, MCP tools that you have not configured run in Ask mode, which requires your approval before execution. Policy patterns of the form mcp(server/tool) let you allow specific tools. ## Why use Antigravity with StoreSEO? - **Your site and its SEO in one place.** The agent that edits your Shopify theme can read StoreSEO's scores, apply optimized content and then open the Liquid file where a fix has to land. - **Approval before a tool runs.** Unconfigured MCP tools are in Ask mode, so Antigravity asks before it calls them. That sits on top of the confirmation StoreSEO tells the agent to ask for. - **OAuth without a client to register.** StoreSEO's server supports dynamic client registration, which Antigravity uses, so sign-in needs only your approval of the store and a pasted authorization code. ## Good to know - The field is serverUrl. Antigravity's documentation says legacy fields such as url or httpUrl are not supported, so an entry copied from another client will not connect until you rename it. - The sign-in ends with a code you paste back into Agent settings, under Customizations, so do it in the same window where you started it. - If you add a policy that allows mcp(storeseo/*), Antigravity stops asking and the confirmation StoreSEO asks the agent for is the only check. Allow individual read tools instead. ## What can you ask Antigravity to do in StoreSEO? - Use StoreSEO to list my five lowest-scoring products and preview new meta for each. - Which product images are not optimized? Show me the list before you compress anything. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Which products have no FAQ or AI snippet yet? Preview them for the first three. ## Antigravity guides by job - [Audit your store's SEO with Antigravity](https://storeseo.com/agent/antigravity/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Antigravity](https://storeseo.com/agent/antigravity/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Antigravity](https://storeseo.com/agent/antigravity/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Antigravity](https://storeseo.com/agent/antigravity/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Antigravity](https://storeseo.com/agent/antigravity/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Antigravity](https://storeseo.com/agent/antigravity/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Antigravity](https://storeseo.com/agent/antigravity/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Antigravity](https://storeseo.com/agent/antigravity/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Antigravity](https://storeseo.com/agent/antigravity/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Antigravity](https://storeseo.com/agent/antigravity/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Antigravity to StoreSEO? Open View raw config from the MCP Servers menu in the agent side panel and add an mcpServers entry named storeseo with serverUrl set to https://mcp.storeseo.com/mcp. Then open Agent settings, Customizations, click Authenticate, approve your Shopify store and paste the authorization code back. ### Why does Antigravity not use the url field? Antigravity's documentation lists serverUrl as the field for remote servers and says legacy fields such as url or httpUrl are not supported. If you copy a config from another client, rename url to serverUrl. ### Do I need a StoreSEO API key for Antigravity? No. Antigravity supports OAuth with dynamic client registration, so you approve your store in the browser. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, sent as an Authorization Bearer value in a headers object. ### Will Antigravity change my store without asking? StoreSEO tells the agent to preview, name any cost and wait for your yes before it applies content, optimizes images, changes settings, publishes llms.txt or agents.md, writes a description or deletes anything. That yes is an instruction the agent follows, not a lock on the server. Antigravity's own approval is the hard stop: MCP tools you have not configured run in Ask mode and wait for you. ### Which StoreSEO plan do I need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI Content Optimizer credit. ### Where does Antigravity keep the StoreSEO server and its sign-in? The server entry lives in ~/.gemini/config/mcp_config.json for every project or .agents/mcp_config.json for one workspace. OAuth tokens are stored in ~/.gemini/antigravity/mcp_oauth_tokens.json and refresh automatically. ### How do I disconnect Antigravity from my store? Remove the storeseo entry from mcp_config.json, or set disabled to true to switch it off. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Kimi Code](https://storeseo.com/agent/kimi): Moonshot AI's terminal agent. Add one MCP server, sign in once, and run your store's SEO next to your theme code. ## Related - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [Claude Code and StoreSEO](https://storeseo.com/agent/claude-code) Setup checked against: - [Antigravity: MCP](https://antigravity.google/docs/mcp) --- Canonical: https://storeseo.com/agent/antigravity Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # Run Your Shopify SEO from Kimi Code > Add StoreSEO to Kimi Code CLI, sign in with /mcp-config login, then audit products, fix meta and alt text and publish llms.txt from your terminal. Kimi Code is Moonshot AI's terminal coding agent, which runs the Kimi model and can call external tools through the Model Context Protocol. Connected to StoreSEO, it reads your store's SEO scores, previews optimized meta and images, and applies the changes you approve without opening the Shopify admin. - Add the server in JSON or with /mcp-config - OAuth through your Shopify admin, or a token header - User or project scope for the connection ## How do I connect Kimi Code to StoreSEO? ### 1. Add StoreSEO to Kimi Code Create or edit ~/.kimi-code/mcp.json so every project on your machine can use StoreSEO. For one repository only, use .kimi-code/mcp.json inside it. You can also run /mcp-config in Kimi Code and have it add the server for you. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### 2. Sign in to your store Start a new Kimi Code session (a server added mid-session registers in the next one), run this command, then follow the browser window: pick your Shopify store and approve access. ```text /mcp-config login storeseo ``` ### 3. Or connect with an access token (optional) If you cannot use a browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header in the same server entry. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` **Sign-in:** Kimi Code signs you in with OAuth: add the server, run /mcp-config login storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and put it in the server's headers. Verify with: "Run /mcp to confirm storeseo is connected, then ask: Use StoreSEO to show me my store's SEO score breakdown." Full setup guide: https://storeseo.com/docs/connect-storeseo-to-kimi ### Example session (illustrative) **You:** Find my five lowest-scoring products and preview better meta for each. → `storeseo_start` (read-only) → `storeseo_list_resources` (read-only) → `storeseo_preview_resource_optimization` (preview) **Agent:** StoreSEO drafted new meta for all five. The biggest jump is Linen Shirt, projected 41 to 78. Previews used AI credit; applying is free. Apply all five? **You:** Yes, apply them. → `storeseo_apply_resource_optimization` (needs approval) **Agent:** Applied. All five re-scored; the next fix is alt text on 14 images. Want me to do those? ## What is Kimi Code? Kimi Code is Moonshot AI's terminal coding agent. You start it with the kimi command in a project folder, and it reads and edits files, runs commands, searches the code and fetches web pages while you watch. It runs the Kimi model, and you sign in with /login using either a Kimi Code OAuth account or a Moonshot platform API key. It is an MCP client. Servers are listed in ~/.kimi-code/mcp.json for every project, or in .kimi-code/mcp.json for one repository, where the project entry overrides the user entry. The /mcp-config command can add, edit and delete servers without hand-editing JSON, and /mcp-config login runs the browser sign-in for servers that use OAuth. That makes it a fit for developers and agencies who already work on a Shopify theme in the terminal. The same session that edits Liquid can audit the store's SEO, apply StoreSEO's optimized content and report the new scores. ## Why use Kimi Code with StoreSEO? - **One session for code and SEO.** When a fix needs a theme change, such as where FAQs render or how a schema block is placed, the same session can open the Liquid file next to the StoreSEO results. - **Work through the whole catalog.** Ask for every product under a score and Kimi Code pages through the listing, previews several at once and gives you one summary to approve, instead of you opening products one by one. - **Approvals you can see.** Kimi Code shows an approval panel for tool calls and offers Approve for this session, so you decide how much of a StoreSEO run goes ahead on its own. ## Good to know - A server added while Kimi Code is open does not register until the next session starts, so restart before you run /mcp-config login. - Kimi Code reads read-only tool calls without asking, but asks for the rest by default. The /yolo and /auto modes and Approve for this session reduce those prompts; keep apply, write and generate tools on ask so the confirmation StoreSEO asks for is not your only check. - A project file at .kimi-code/mcp.json overrides a user entry with the same name, so a stale storeseo entry in a repository can shadow the one in your home folder. - Older Kimi CLI builds keep their settings in a different folder than ~/.kimi-code. These steps are for Kimi Code CLI. ## What can you ask Kimi Code to do in StoreSEO? - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Write it for my top collection, one image at a time. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Check the product schema on my store and tell me which products are missing it. ## Kimi Code guides by job - [Audit your store's SEO with Kimi Code](https://storeseo.com/agent/kimi/seo-audit): Find the products, collections, pages and posts with the lowest SEO scores and see exactly what each one is missing. - [Optimize meta tags and content with Kimi Code](https://storeseo.com/agent/kimi/optimize-meta): Preview and apply new meta titles, descriptions, URL handles, tags, FAQs and AI snippets, and set what StoreSEO optimizes automatically for newly synced products, collections and articles. - [Optimize images and alt text with Kimi Code](https://storeseo.com/agent/kimi/image-seo): Compress heavy images, write alt text and give image files descriptive names, one image at a time, with a restore for each change. - [Get products ready for AI search with Kimi Code](https://storeseo.com/agent/kimi/ai-search-readiness): Raise a product's GEO score so AI shopping agents and answer engines can understand it: FAQs, AI snippet, product schema and an agent-ready description. - [Publish llms.txt and agents.md with Kimi Code](https://storeseo.com/agent/kimi/llms-txt-agents-md): Generate the llms.txt file answer engines read and the agents.md file AI shopping agents read, check what is live, roll back or delete llms.txt versions, and reset agents.md to Shopify's default. - [Set up structured data with Kimi Code](https://storeseo.com/agent/kimi/structured-data): Check which JSON-LD schema types are on, configure organization, local business, product and review schema, and switch types on or off. - [Manage sitemaps and indexing with Kimi Code](https://storeseo.com/agent/kimi/sitemaps): Exclude pages from the sitemap, set noindex or nofollow on a resource, and set up an HTML sitemap page for shoppers and crawlers, including which menu links to it. - [Track keyword rankings with Kimi Code](https://storeseo.com/agent/kimi/keyword-tracking): Track where your store ranks on Google by country and language, add or remove keywords and up to five competitors, read each keyword's rank history, and build keyword clusters for products. - [Read Search Console and Analytics with Kimi Code](https://storeseo.com/agent/kimi/google-reports): Ask about clicks, impressions, queries, pages and traffic sources from Google Search Console and GA4 without opening either. - [Write SEO blog post drafts with Kimi Code](https://storeseo.com/agent/kimi/blog-posts): Have StoreSEO write a complete SEO blog post and save it as a draft in your Shopify blog, ready for you to review and publish. ## Which way should you connect? StoreSEO exposes the same MCP server to every client. What differs is how the client proves who you are, plus one assistant that lives inside Shopify itself. | Way in | Best for | Sign-in | What you get | |---|---|---|---| | MCP server with OAuth | Claude, ChatGPT and any client that supports remote MCP sign-in | Pick your store in Shopify and approve once | All 44 tools, scoped to the store you approved | | MCP server with an access token | Terminal agents and editors, or any client without OAuth | Generate a named token in Settings, AI Toolkit and send it as a Bearer header | The same 44 tools; revoke the token in the app to cut access | | Shopify Sidekick | Merchants who stay inside Shopify admin | Nothing to add: StoreSEO is installed in your store | SEO answers and deep links to the fix page, not the MCP tools | ## Frequently asked questions ### How do I connect Kimi Code to StoreSEO? Add a server named storeseo with the URL https://mcp.storeseo.com/mcp to ~/.kimi-code/mcp.json, either by editing the file or by running /mcp-config in Kimi Code. Start a new session, run /mcp-config login storeseo, pick your Shopify store in the browser and approve access. ### Do I need a StoreSEO API key for Kimi Code? No. Kimi Code signs in with OAuth through /mcp-config login. A personal access token from StoreSEO under Settings, AI Toolkit is the alternative, added as an Authorization header in the server entry. ### Where does Kimi Code store the StoreSEO server? In ~/.kimi-code/mcp.json for every project on your machine, or in .kimi-code/mcp.json inside a repository, where it overrides a user entry of the same name. ### Will Kimi Code change my products without asking? StoreSEO tells the agent to preview first and wait for your yes before it applies content, optimizes images, publishes a file or deletes anything. Kimi Code also asks before running tools that are not read-only unless you chose a mode or rule that approves them. Reads such as scores and reports run straight away. Keep the write tools on ask if you want both checks. ### Does Kimi Code work with the Kimi chat app? This page covers Kimi Code CLI, the terminal agent. It connects to StoreSEO through its MCP configuration as described above. We have not checked the Kimi chat apps, so use the terminal agent for StoreSEO. ### Which StoreSEO plan does Kimi Code need? Any active StoreSEO plan, including the free plan, can use the MCP server. Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. ### StoreSEO tools do not show up in Kimi Code. What should I check? Run /mcp to see whether storeseo is listed and connected. If it is missing, check the JSON in ~/.kimi-code/mcp.json and any .kimi-code/mcp.json in the project, then start a new session. If it is listed but not signed in, run /mcp-config login storeseo. ### How do I disconnect Kimi Code from my store? Delete the storeseo entry with /mcp-config or remove it from mcp.json. You can also revoke the connection in StoreSEO under Settings, AI Toolkit, where every connected app is listed. ## Also works with - [Claude](https://storeseo.com/agent/claude): Anthropic's chat assistant on the web and desktop. Add StoreSEO as a custom connector, no terminal needed. - [ChatGPT](https://storeseo.com/agent/chatgpt): OpenAI's chat assistant. A developer-mode app with the StoreSEO MCP URL and OAuth sign-in. - [Claude Code](https://storeseo.com/agent/claude-code): Anthropic's terminal agent. One claude mcp add command, then your store's SEO in the same session as your theme code. - [Claude Cowork](https://storeseo.com/agent/claude-cowork): Anthropic's agent for delegated work. Add the StoreSEO connector, hand off an SEO job and review the results. - [Cursor](https://storeseo.com/agent/cursor): The AI code editor. One mcp.json entry, and four tools switched off because Cursor stops at about 40. - [Codex](https://storeseo.com/agent/codex): OpenAI's coding agent. One codex mcp add command, one codex mcp login, and the same setup in the app and the IDE. - [Gemini CLI](https://storeseo.com/agent/gemini-cli): Google's terminal agent. One gemini mcp add command, and the browser sign-in is found automatically. - [GitHub Copilot](https://storeseo.com/agent/github-copilot): Copilot agent mode in VS Code. One .vscode/mcp.json entry puts your store's SEO beside your theme code. - [Windsurf](https://storeseo.com/agent/windsurf): The Cognition AI editor. A serverUrl entry in mcp_config.json, and all 44 tools fit under Cascade's 100-tool cap. - [Grok](https://storeseo.com/agent/grok): xAI's chat assistant. Add StoreSEO as a custom connector on grok.com and ask about your store's SEO. - [ChatGPT dots](https://storeseo.com/agent/chatgpt-dots): OpenAI's always-on agents in ChatGPT. They share the StoreSEO plugin you add in developer mode and can run report-only checks on a schedule. - [Grok Bot](https://storeseo.com/agent/grok-bot): xAI's named Bots on a cloud computer. Add the StoreSEO MCP URL as a plugin and let a Bot run report-only SEO routines. - [Muse](https://storeseo.com/agent/muse): Meta's personal AI agent. Send one message with the StoreSEO link, approve the sign-in, then run your store's SEO from chat. - [Manus](https://storeseo.com/agent/manus): A general AI agent that finishes multi-step tasks. Add StoreSEO as a custom MCP server and name it in a task. - [OpenClaw](https://storeseo.com/agent/openclaw): The self-hosted agent runtime with chat apps and automations. Two commands to connect, then SEO reports on a timer. - [Hermes Agent](https://storeseo.com/agent/hermes-agent): Nous Research's agent with memory and a built-in scheduler. An mcp_servers entry in config.yaml and one login. - [Paperclip](https://storeseo.com/agent/paperclip): Self-hosted orchestrator for agent teams. Add StoreSEO to the agent Paperclip runs, then hand it SEO work as tasks and routines. - [NanoClaw](https://storeseo.com/agent/nanoclaw): A small self-hosted assistant you own. Add StoreSEO to its agent and ask about your store from the chat app you already use. - [OpenCode](https://storeseo.com/agent/opencode): The open-source terminal coding agent. One remote MCP entry, then opencode mcp auth storeseo. - [Zed](https://storeseo.com/agent/zed): The Zed editor's Agent Panel. One context_servers entry in settings.json and a browser sign-in. - [Kiro](https://storeseo.com/agent/kiro): AWS's agentic IDE. One url entry in .kiro/settings/mcp.json, workspace or user level, and a browser sign-in. - [Antigravity](https://storeseo.com/agent/antigravity): Google's agentic IDE. One serverUrl entry in mcp_config.json, then a browser sign-in with an authorization code. ## Related - [StoreSEO AI Toolkit setup](https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution) - [Connect StoreSEO MCP to your AI assistant](https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant) - [AI agents and MCP playbook](https://storeseo.com/use-cases/ai-agents) - [StoreSEO with Claude Code](https://storeseo.com/agent/claude-code) Setup checked against: - [Kimi Code CLI: MCP configuration](https://moonshotai.github.io/kimi-code/en/customization/mcp.html) - [Kimi Code CLI: slash commands](https://moonshotai.github.io/kimi-code/en/reference/slash-commands) - [Kimi Code CLI: interaction and approvals](https://moonshotai.github.io/kimi-code/en/guides/interaction) - [Kimi Code CLI on GitHub](https://github.com/MoonshotAI/kimi-code) --- Canonical: https://storeseo.com/agent/kimi Last updated: October 2026. Approval: Reads run straight away; StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, and your client's tool-approval setting is the hard stop. Every agent guide: https://storeseo.com/agent · Connect the MCP server: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant · Pricing: https://storeseo.com/#pricing · Install StoreSEO: https://apps.shopify.com/storeseo --- # StoreSEO jobs --- ## Job: Store SEO audit StoreSEO scores every product, collection, page and blog post against its SEO checklist, and scores products for AI readiness (GEO) as well. An agent opens a session, lists the weakest resources first and reads the per-criterion analysis for each, so you get a ranked worklist instead of a dashboard to click through. Tools, in order: storeseo_start, storeseo_list_resources, storeseo_get_resource_seo. 1. Open a StoreSEO session: The agent calls the start tool first. It returns a conversation id every later call needs, plus a snapshot of your store: how many resources are optimized, need improvement or are not optimized, image status, rank tracking and which features are set up. 2. List the weakest resources: It lists products, collections, pages or articles sorted by score, lowest first, or filtered to the ones marked not optimized. Products also carry a GEO score and their AI-readiness issues. 3. Read the checklist for each one: For each candidate it reads the full SEO record: current meta fields, focus keyword, score, every criterion that passes or fails, image status, FAQs and AI snippet. 4. Sort fixes by who can make them: It separates what a StoreSEO tool can fix (meta, handle, FAQs, AI snippet, image alt text, schema) from what only you can change in the app or in Shopify admin (the focus keyword, the product title, category, inventory). 5. Report a worklist: You get the resources in order, with the failing criteria in plain words and one suggested next fix. Nothing is written during an audit. - Resource types: Products, collections, pages and blog articles - Scores: An SEO score for every resource, plus a GEO (AI readiness) score for products - Status labels: Optimized, needs improvement, not optimized; GEO adds not analysed - Sorting: By score, lowest first, or filtered to not optimized - Session: Every call carries the conversation id from the start tool; calls without it fail - What it writes: Nothing. Every audit tool is read-only - Fixed only in the app: Focus keyword choice, product title wording, Shopify category, product type, inventory, variant data - Plan: Any active StoreSEO plan, including the free plan Rules: - Every audit call is a read and runs straight away; nothing in this job changes your store. - The agent describes criteria by their visible title, never by internal identifiers or raw status values. - Criteria no tool can fix, such as choosing the focus keyword, are handed to you with a link into the StoreSEO app rather than guessed. - The agent should offer one next fix at a time, tied to what it just found, instead of dumping the whole list as a checklist. Q: Does an SEO audit change anything in my store? A: No. The audit uses read-only tools: the session start, the resource list and the per-resource SEO record. Your store is only changed later, by a separate step you agree to. Q: What is the difference between the SEO score and the GEO score? A: The SEO score measures a resource against StoreSEO's on-page checklist for search engines. The GEO score applies to products only and measures how ready the product is for AI shopping agents and generative search, using criteria such as FAQs, an AI snippet, alt text and product schema. Q: Can the agent choose my focus keyword? A: No tool can set the focus keyword. The agent can tell you when the keyword is missing or looks wrong and give you a link to set it in the StoreSEO app, and several fixes such as keyword clusters and description writing need it set first. Q: How many products can the agent audit at once? A: The listing is paginated, so the agent can walk your whole catalog page by page. In practice it is more useful to ask for the lowest scores first and work down from there. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/seo-audit --- ## Job: Meta tags and on-page content StoreSEO generates the content server-side: meta title and description, tags, a keyword-led URL handle, FAQs and an AI snippet. The agent shows you the draft and the projected score change before anything is saved, then applies all of it or only the parts you approve. Tools, in order: storeseo_get_resource_seo, storeseo_preview_resource_optimization, storeseo_apply_resource_optimization, storeseo_get_ai_optimization_settings, storeseo_update_ai_optimization_settings. Requires: Generating a preview spends AI Content Optimizer credit; applying a preview is free. 1. Read the current state: The agent reads the resource's SEO record so it knows which fields exist, which fail and what the preview will add. 2. Generate a preview: StoreSEO writes the meta title, description, tags, URL handle, FAQs and AI snippet and projects the new SEO score, plus the GEO score for products. Nothing is saved yet. Preview spends credit whether or not you apply it. 3. Show you the draft: The agent presents the generated text and the score change. When the handle changes it shows the old and new URL and tells you a redirect will be created. 4. Apply what you approve: After your yes it applies everything, or only the parts you chose, for example the meta but not the FAQs. Skipped fields stay available from the same draft for an hour. 5. Report the new score: The apply call returns the recomputed score and the remaining checklist, so the agent tells you what changed and offers the next most useful fix. - Generated fields: Meta title, meta description, tags, URL handle, FAQs, AI snippet - Resource types: Products, collections, pages and articles (collections have no meta tags or tags axis in auto settings) - Who writes the copy: StoreSEO, server-side. The agent never authors meta, FAQs or handles itself - Cost: Preview spends AI Content Optimizer credit; apply is free; a new preview spends again - Draft lifetime: One hour after the preview - URL handle: Applying a new handle changes the live URL and creates a redirect from the old one; a handle already in use is dropped automatically - Partial apply: Apply a subset of axes, or skip individual fields such as the handle - Auto optimization: Optional settings that generate content for newly synced resources; only future syncs, nothing is backfilled - Auto settings plan: Configurable on any plan; switching an auto toggle on needs a paid plan and no active trial Rules: - The agent previews first and applies only after you confirm the previewed content. - When a handle changes, you see the old URL and the new URL before the apply call, because applying creates a redirect. - A request like optimize my worst five products authorizes the previews; the agent should remind you that previews spend credit even if you do not apply them. - Turning on auto optimization spends credit on every future sync, so the agent proposes the exact sections and waits for your yes. - Product title, focus keyword and description body are not written by this job; the description has its own tool in the AI search job. Q: Does previewing an optimization change my product? A: No. A preview generates the content and projects the score without saving anything. It does spend AI Content Optimizer credit, and applying the draft afterwards is free. Q: What happens to the old URL when the handle changes? A: Applying a new URL handle updates the live URL and StoreSEO creates a redirect from the old one, so existing links keep working. The agent should show you both URLs before it applies the change. Q: Can I apply only part of a preview? A: Yes. The apply step takes the axes you approved, and individual content fields such as the URL handle can be skipped. Skipped parts stay available from the same draft for one hour. Q: Does auto optimization fix my existing products? A: No. Auto optimization only acts on resources synced or updated after you turn it on. Existing products are fixed with the preview and apply steps in this job. Q: Will StoreSEO change my product title? A: No. This job writes the SEO meta title, not the storefront product title. The product title is edited in Shopify admin. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/optimize-meta --- ## Job: Image SEO and alt text StoreSEO tracks three things per image: file-size compression, alt text and file name. The agent finds the heaviest or unlabelled images, optimizes them one call per image, and can restore any of the three if you do not like the result. Tools, in order: storeseo_list_images, storeseo_optimize_image, storeseo_restore_image. Requires: Compression needs the Image Optimizer add-on and counts against its monthly quota; alt text and file names spend AI Content Optimizer credit. 1. Find the images: The agent lists product, collection or article images, heaviest unoptimized first by default. It can filter to images with no alt text, or scope the list to one product by title. 2. Check what it found: A title search can also match image file names from other resources, so the agent checks each row belongs to the product you meant. 3. Agree the scope and cost: Compression uses image quota and alt text and file names use AI credit, so the agent confirms before working through many images. 4. Optimize one image per call: Each call takes the axes you want: compression, alt text, file name or all three. The axes always run compression, then file name, then alt text, and the response reports each one separately. 5. Report or restore: The agent reports what applied, what was already done and what was refused. If a result is wrong, the restore tool brings back the original file, alt text or name. - Axes per image: Compression, alt text, file name - Image sources: Products, collections and articles (pages have no images) - Batching: One image per call; never two calls on the same image at once - Compression: Image Optimizer add-on and its monthly quota; an already compressed image is reported as already applied - Alt text and file name: AI Content Optimizer credit; both together are billed as one generation - File name change: Renames the file on Shopify and changes the image's storefront URL - Collection and article renames: Re-upload the image, so the call can take a couple of minutes - Restore: Reverts compression, alt text or file name per axis - Score impact: Alt text and file names count on products, collections and articles; compression counts on products Rules: - The agent confirms the number of images and the kind of spend before a bulk run. - One image per call, all wanted axes together; never parallel calls on the same image. - A renamed image has a new URL, so the agent tells you, because anything that hard-coded the old URL keeps serving the old file. - If a response is lost, the agent re-lists the image instead of retrying, because a retry regenerates and spends again. - A refused axis, for example compression without the add-on, is reported as such while the other axes still apply. Q: Can the agent compress images without the Image Optimizer add-on? A: No. Compression needs the Image Optimizer add-on and counts against its monthly quota. Alt text and file names work without it and spend AI Content Optimizer credit instead. Q: Can I undo an image optimization? A: Yes. The restore tool reverts compression, alt text or file name for one image, per axis. Restoring compression brings back the original file. Q: Why did my image URL change? A: Giving an image a new file name renames the file on Shopify and re-points its storefront URL. Pages that use the image through Shopify keep working, but anything that hard-coded the old URL keeps the old file. Q: Can I write my own file name or alt text through the agent? A: No. StoreSEO writes the alt text and file name from the resource's own context. You can steer the language, and you can always edit alt text yourself in Shopify admin. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/image-seo --- ## Job: AI search readiness GEO is StoreSEO's AI readiness score for products. The agent reads which GEO criteria fail, fixes the ones a tool can fix (FAQs, AI snippet, alt text, product schema), and can rewrite a thin description from your own answers to StoreSEO's questions. Tools, in order: storeseo_get_resource_seo, storeseo_update_schema_settings, storeseo_preview_resource_optimization, storeseo_apply_resource_optimization, storeseo_scan_product_description, storeseo_write_product_description. Requires: FAQs, AI snippets and descriptions spend AI Content Optimizer credit. The product needs a focus keyword, set in the StoreSEO app, before a description scan or rewrite. 1. Read the GEO checklist: The agent reads the product's GEO block: the score and every AI-readiness criterion that passes or fails. 2. Turn on product schema if it is off: Product JSON-LD is a shop-wide setting, so enabling it clears that criterion for every product at once. If you then decide not to apply the rest, the agent switches back off only the schema it turned on. 3. Preview FAQs and the AI snippet: The preview generates FAQs and an AI snippet (summary, benefits, best for) and projects the new SEO and GEO scores without saving. 4. Scan the description: For a thin or missing description, StoreSEO scores it on seven pillars (identity, materials, dimensions, features, use cases, what is included, care) and returns questions. The agent asks you each one, word for word. 5. Write and apply: With your answers, StoreSEO writes the description and saves it immediately; the response holds the only copy of the old text. The previewed FAQs and snippet are applied after your yes. - Scope: Products only; collections, pages and articles have no GEO score - Fixable by tools: FAQs, AI snippet, image alt text, product schema, description - Fixed only in the app or admin: Product title, category, product type, image count, inventory, variant barcode, SKU, price, image - Description pillars: Identity, materials, dimensions, features, use cases, what is included, care - Description write: Saved immediately to Shopify and StoreSEO; no undo and no revision history - Old description: Returned once in the write response, the only remaining copy - Product schema: Shop-wide toggle; credits the schema criterion on every product - Prerequisite: A focus keyword on the product, set in the StoreSEO app Rules: - The scan's questions go to you word for word and in order; the agent does not answer them for you. - A description is only written without questions when you decline to answer or authorize a bulk run without them. - Writing a description replaces the current one at once with no undo, so the agent asks first and keeps the old text the response returns. - If the agent enabled product schema for this flow and you stop before applying, it turns that schema back off; it never disables a schema you already had on. - Criteria only you can fix get one link to the product's fix page, not a link per criterion. Q: What is a GEO score in StoreSEO? A: GEO is StoreSEO's AI readiness score for products. It measures how well a product can be understood by AI shopping agents and generative search, using criteria such as FAQs, an AI snippet, image alt text, product schema and the quality of the description. Q: Can I undo a rewritten product description? A: Not inside StoreSEO. The new description is saved immediately and StoreSEO keeps no revision history. The response returns the previous description once, so ask the agent to keep it for you before it writes. Q: Why does the agent ask me questions before writing a description? A: StoreSEO scans the description on seven pillars and returns questions for the facts it cannot infer, such as materials or what is in the box. Your answers become stated facts in the new copy instead of guesses. Q: Does enabling product schema affect every product? A: Yes. Product JSON-LD is a shop-wide setting, so turning it on clears that criterion for every product at once, and turning it off removes it everywhere. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/ai-search-readiness --- ## Job: llms.txt and agents.md llms.txt is a curated map of your store for AI crawlers, and agents.md tells AI shopping agents who you are and how to buy. StoreSEO generates both from your store data and publishes them to your live theme; the agent checks status, generates, switches llms.txt versions and resets agents.md to Shopify's default. Tools, in order: storeseo_get_llms_txt_status, storeseo_generate_llms_txt, storeseo_list_llms_txt_versions, storeseo_activate_llms_txt_version, storeseo_delete_llms_txt_version, storeseo_get_agents_md_status, storeseo_generate_agents_md, storeseo_reset_agents_md. Requires: agents.md generation spends AI Content Optimizer credit and needs at least one published product. 1. Check what is live: The agent reads the llms.txt status (settings, latest version, whether a rebuild is due, public URL) and the agents.md status (custom file or Shopify's default, public URL). 2. Warn before publishing: Generating llms.txt writes a template to your live theme and uploads a copy to Shopify Files in the same run, with no separate publish step. Generating agents.md publishes it live at once and spends credit. The agent says so and waits for your yes. 3. Generate and wait: llms.txt generation takes a minute or two; the agent polls the status until it finishes. agents.md returns the published text and its URL straight away. 4. Manage versions: StoreSEO keeps the latest ten llms.txt versions. The live file serves the active version, which is not always the newest, and any version can be made active again. 5. Roll back: Deleting a version is permanent. Resetting agents.md drops the custom file so Shopify's default serves again, and there is no history to restore it from. - llms.txt: Generated from your products, collections, pages and posts; served at your store's /llms.txt - Where it is written: A template in the live theme plus a copy in Shopify Files, in the same run - Schedule: llms.txt can regenerate on a saved schedule; settings passed with a run are saved for scheduled runs - Versions kept: The latest ten; the live file serves the active one - Generation time: About one to two minutes for llms.txt - agents.md: AI-written from your brand profile, top products, collections and contact details; served at /agents.md - agents.md publish: Live immediately; replaces Shopify's native default - agents.md reset: Shopify's default serves again within seconds; the custom text is gone for good Rules: - The agent tells you that llms.txt generation writes to your live theme before it starts. - agents.md generation publishes live and spends credit, so the agent says both and waits for your yes. - Deleting an llms.txt version and resetting agents.md cannot be undone; the agent confirms each one. - Deleting the active llms.txt version promotes the next newest automatically, so the agent says which version will go live. Q: What is the difference between llms.txt and agents.md? A: llms.txt is a curated map of your store for AI crawlers and answer engines, served at /llms.txt. agents.md is the file AI shopping agents read to understand your brand and how to shop with you, served at /agents.md. StoreSEO can generate both. Q: Does generating llms.txt change my theme? A: Yes. The same run writes an llms.txt template into your live theme and uploads a copy to Shopify Files. There is no separate publish step, which is why the agent should warn you before it starts. Q: Can I go back to an older llms.txt? A: Yes. StoreSEO keeps the latest ten versions, and any of them can be made the active one, which republishes it to your storefront. Deleting a version is permanent. Q: Can I undo a generated agents.md? A: You can reset it, which makes Shopify's native default serve again within seconds. The custom text is discarded with no history, so keep a copy if you might want it back. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/llms-txt-agents-md --- ## Job: Structured data (JSON-LD) StoreSEO adds JSON-LD schema to your storefront so search engines and AI assistants can read your products, business and FAQs as data. The agent lists every schema type, reads a type's settings and its field template, and sends back a complete settings object. Tools, in order: storeseo_list_schema_types, storeseo_get_schema_settings, storeseo_update_schema_settings. Requires: The FAQ schema types need the FAQ Schema feature in your plan. 1. List the schema types: The agent lists every type with whether it is on and whether it takes settings. Configurable types are product, review, sale dates, product merchant, organization and local business. 2. Read before writing: For a configurable type it reads the current settings and the field template, because an update replaces the whole settings object rather than merging into it. 3. Fill in your details: Business name, address, contact details and similar fields come from you. Organization and local business share the same business information, so changing one changes both. 4. Switch it on: After your yes the agent sends the complete settings with the status. Organization, local business and product merchant need complete settings to be enabled. - Configurable types: Product, review, sale dates, product merchant, organization, local business - Toggle-only types: Every other type, including the four FAQ types - Update behaviour: Settings replace the stored object; read first, then send the full object - Dependencies: Review, sale dates and product merchant need product schema on; disabling product also disables product merchant - Shared data: Organization and local business share basic business information - FAQ types: Product, collection, article and page FAQ need the FAQ Schema feature in your plan - Sale dates: Also needs a discount permission that only the app's own Sale Dates toggle can request Rules: - The agent reads the current settings before every settings update, because the update replaces them wholesale. - Business details come from you; the agent does not invent an address, phone number or opening hours. - Turning a schema type on or off changes your live storefront markup, so the agent confirms the change first. - When a type is refused for a missing plan feature or dependency, the agent says which one rather than retrying. Q: What schema types can StoreSEO add to my store? A: StoreSEO manages product, review, sale dates, product merchant, organization and local business schema with settings, plus toggle-only types such as the FAQ types for products, collections, articles and pages. Q: Why does the agent read my schema settings before changing them? A: A settings update replaces the whole stored object instead of merging. Reading first and sending back the complete object is the only way to change one field without wiping the others. Q: Why was review schema refused? A: Review, sale dates and product merchant schema depend on product schema. Turn product schema on first, then enable the dependent type. Q: Do I need a particular plan for FAQ schema? A: Yes. The four FAQ schema types need the FAQ Schema feature in your StoreSEO plan. Without it, any change to them is refused with a message saying the feature is not in your plan. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/structured-data --- ## Job: Sitemaps and indexing StoreSEO controls two things here: per-resource sitemap flags (left out of the XML sitemap, noindex, nofollow) and an HTML sitemap page. The agent lists flags, changes them one resource at a time and walks you through the HTML sitemap setup. Tools, in order: storeseo_list_resource_sitemaps, storeseo_update_resource_sitemap, storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap, storeseo_list_sitemap_menus. 1. List the flags: The agent lists products, collections, pages or articles with their sitemap, noindex and nofollow flags, filtered as you ask. 2. Change one resource: After your yes it sets the flags on one resource, which writes the matching Shopify metafields. 3. Re-read before reporting: Flags can also change in the StoreSEO app, so the agent re-lists before telling you the current state. 4. Create the HTML sitemap: It reads the HTML sitemap settings, then generates and enables the page. 5. Add it to a menu: It lists your navigation menus with their edit links; you add the sitemap link to a menu in Shopify admin, then the agent marks setup complete. - Flags per resource: Excluded from sitemap, noindex, nofollow - Resource types: Products, collections, pages and articles - Where flags live: Shopify metafields on the resource - HTML sitemap steps: Generate and enable, add the link to a menu yourself, mark complete - HTML sitemap settings: Replaced wholesale on update; read first - Menus: Listed with edit links; the menu you choose is not stored - Plan: Every active plan Rules: - A noindex flag removes a page from search results, so the agent names the resource and waits for your yes. - The agent re-lists flags in the same turn before telling you their current state. - Adding the sitemap link to a menu is a step you do in Shopify admin; no tool does it. - HTML sitemap settings are read before every update, because an update replaces them. Q: What is the difference between excluding a page from the sitemap and noindex? A: Excluding a resource leaves it out of your XML sitemap, so crawlers are not pointed at it. Noindex tells search engines not to show the page in results even if they find it. StoreSEO sets each flag separately per resource. Q: Can the agent add the HTML sitemap to my menu? A: No. The agent generates the HTML sitemap and lists your menus with their edit links, and you add the sitemap link to a menu in Shopify admin. Then the agent can mark setup complete. Q: Do sitemap tools need a paid plan? A: No. The sitemap tools work on every active StoreSEO plan. Q: Why does the agent check the flags again before answering? A: Sitemap flags can also be changed in the StoreSEO app outside the conversation, so an earlier listing may be stale. Re-reading in the same turn keeps the answer accurate. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/sitemaps --- ## Job: Keyword rank tracking StoreSEO checks your Google rank for tracked keywords by country, language and device, records the history and compares competitor domains. Keyword clusters group related keywords around a product's focus keyword with search volume and difficulty. Tools, in order: storeseo_list_rank_locations, storeseo_list_rank_languages, storeseo_add_keywords, storeseo_list_keywords, storeseo_get_keyword_history, storeseo_list_competitors, storeseo_add_competitors, storeseo_delete_keyword, storeseo_delete_competitor, storeseo_generate_keyword_cluster, storeseo_get_keyword_cluster. Requires: Rank and competitor tracking are Pro plan tools. Keyword clusters spend AI Content Optimizer credit. 1. Pick the market: The agent looks up the exact country and language names the tracker accepts, because new keywords must use them verbatim. 2. Add keywords: After your yes it adds keywords with a country, language and device. Checks run in the background, so positions appear later. 3. Read ranks and history: It lists your keywords with the latest Google position, change, SERP features and competitor positions, and reads the dated history for any keyword. 4. Compare competitors: Up to five competitor domains are tracked alongside your keywords. To change one, it is deleted and the new domain added. 5. Build a cluster: For a product with a focus keyword, StoreSEO generates a cluster of related keywords in the background; the agent states the estimated credit first and polls until it is ready. - Search engine: Google; AI search engines are not tracked - Locations: 213 countries, country level only - Languages: 127 languages - Devices: Desktop or mobile - Competitors: Up to five domains; no edit, delete and re-add - Keyword limit: Counts against your plan's keyword limit - Deleting a keyword: Removes its entire rank history; cannot be undone - Keyword cluster: Built around the product's existing focus keyword, sorted by search volume, with relevance and difficulty - Plan: Pro plan for rank and competitor tools Rules: - On a plan without rank tracking, the agent explains the upgrade instead of calling the rank tools. - Country and language names come from the lookup tools, never typed from memory. - Deleting a keyword or competitor erases its history for good, so the agent confirms each one. - A cluster needs a focus keyword set in the StoreSEO app; the agent states the estimated credit and waits for your yes. Q: Which plan do I need for keyword tracking through an agent? A: Rank tracking and competitor tracking are Pro plan tools. On other plans the agent should explain that instead of calling them. Keyword clusters work on plans with AI Content Optimizer credit. Q: Does StoreSEO track rankings in AI search engines? A: No. The rank tracker checks Google results by country, language and device. AI visibility is handled by the GEO score, llms.txt and agents.md jobs instead. Q: Why are new keywords pending? A: Rank checks run in the background. Adding a keyword queues the first check, and its position appears in the keyword list once that check completes. Q: What is a keyword cluster? A: A group of related keywords built around a product's focus keyword, sorted by search volume and scored for relevance and difficulty. It shows which wider terms the product page can target. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/keyword-tracking --- ## Job: Search Console and Analytics reports StoreSEO connects to your Google Search Console and Google Analytics 4 property. The agent reads Search Console performance by date, query, page, country, device or search appearance, and curated GA4 reports, each compared with the previous period by default. Tools, in order: storeseo_get_search_console_insights, storeseo_get_analytics_report. Requires: A paid StoreSEO plan with Google connected in the app, and a GA4 property selected for Analytics reports. 1. Pick the question: Search Console answers how you appear in Google: clicks, impressions, CTR and average position. GA4 answers what visitors do: users, engagement, pages, sources and key events. 2. Read the report: The agent reads the last 28 days by default, or the range you name, broken down by the dimension that fits the question. 3. Compare periods: Both reports include the previous period of the same length and the change per metric unless you ask otherwise. 4. Turn it into actions: The agent can link what it finds to the other jobs, for example pages with impressions but a low CTR are candidates for new meta. - Search Console metrics: Clicks, impressions, CTR, average position - Search Console dimensions: Date, query, page, country, device, search appearance - GA4 reports: Search traffic, user activity, demographics, page performance, traffic source, key events - Default range: Last 28 days - Comparison: Previous equal-length period with per-metric change, on by default - What it writes: Nothing. Both tools are read-only and never connect or configure Google - Plan: Paid plan with Google connected; GA4 property selected for Analytics Rules: - Both tools are reads and run straight away. - Connecting Google or choosing the GA4 property is done in the StoreSEO app; the agent hands you the link when it is missing. - Numbers are reported as Google returns them; the agent does not project outcomes from them. Q: Can the agent connect my Google account? A: No. Connecting Search Console and Google Analytics, and choosing the GA4 property, is done in the StoreSEO app. If Google is not connected, the tool returns a link telling you what to set up. Q: Can these reports change anything? A: No. Both Google tools are read-only. They never connect, configure or write anything in Google or in your store. Q: Which plan includes Google reports over MCP? A: A paid StoreSEO plan with Google connected. Analytics reports also need a GA4 property selected in the app. Q: How far back can the agent look? A: You can name any date range the report supports. When you do not, both tools default to the last 28 days and compare it with the 28 days before. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/google-reports --- ## Job: Blog post drafts StoreSEO writes a full blog post with AI and saves it as a draft article in the Shopify blog you choose, optionally with a featured image. Nothing is published: you review and publish from Shopify admin. Tools, in order: storeseo_list_blogs, storeseo_generate_blog_post, storeseo_get_blog_post_job, storeseo_get_resource_seo. Requires: Spends AI Content Optimizer credit; the cost grows with word count and a featured image. Blogs must be synced in the StoreSEO app. 1. Pick the blog: The agent lists your synced blogs, the sections posts go into, with their ids. An empty list means the blogs need syncing in the StoreSEO app first. 2. Agree the brief and cost: Topic, keyword, length and whether to add a featured image. The cost grows with length and the image, so the agent tells you and waits for your yes. 3. Generate in the background: Writing takes a few minutes. The agent polls the job and reports progress, and does not present the post as finished while it runs. 4. Check the draft: When the job completes, the agent can read the draft article's SEO record. You review and publish the post from Shopify admin. - Output: A draft article in the chosen Shopify blog, never published live - Featured image: Optional; uploaded with the draft and adds to the cost - Run time: A few minutes in the background - Insufficient credit: The call fails before anything is queued or spent - Blogs: Only blogs synced with StoreSEO are listed; syncing is done in the app - Failed runs: The job reports the error and credits are settled automatically Rules: - The agent states the expected credit use and waits for your yes before it starts a post. - Posts are saved as drafts only; publishing is done by you in Shopify admin. - While the job runs, the agent reports progress and never presents the draft as finished. Q: Will the agent publish a blog post on my store? A: No. StoreSEO saves generated posts as drafts in the Shopify blog you choose. You review and publish them from Shopify admin. Q: How long does a blog post take? A: A few minutes. The writing runs in the background and the agent checks the job until it is completed, failed or cancelled. Q: What does a blog post cost? A: It spends AI Content Optimizer credit, and the amount grows with the word count and with a featured image. If you do not have enough credit the call fails before anything is spent. Q: Why does the agent say I have no blogs? A: Only blogs synced with StoreSEO are listed. Sync your blogs in the StoreSEO app, then ask again. Guides for this job, one per agent: https://storeseo.com/agent/{agent}/blog-posts --- # Agent and job guides --- ## Audit your store's SEO with Claude Code URL: https://storeseo.com/agent/claude-code/seo-audit (full page: https://storeseo.com/agent/claude-code/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Claude Code means asking Anthropic's terminal agent which products, collections, pages and posts score worst in StoreSEO, and letting it read each one's checklist so you get a ranked worklist without opening the Shopify admin. You start Claude Code in whatever folder you work from, often the theme repository, and type a plain request such as which twenty products score worst and why. Claude Code calls storeseo_start first, because every StoreSEO tool needs the conversation id it returns, and that first call already hands it a snapshot of the store: how many products, collections, pages and articles are optimized, need improvement or are not optimized, how many images still need work, and which features are set up. It then calls storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. You watch each call scroll past in the terminal, and every one of them is a read. What comes back is a table you can act on. For each product Claude Code names the failing criteria by their visible titles, such as a meta description that is too short, a focus keyword missing from the title or images with no alt text, and it separates what a StoreSEO tool can fix from what only you can change, such as the focus keyword itself or the product title. Products also carry a GEO score for AI readiness, so the same pass tells you which listings AI shopping agents will struggle to understand. Because Claude Code keeps the whole session in context, you can follow up with show me only the ones where the meta is the main problem, or save the list to a file in the repository and work through it over several days. Large catalogs are where the terminal pays off. Claude Code pages through the listing rather than loading everything at once, and you can ask it to stop at a score threshold, a product type or a collection. Nothing in an audit changes your store, so you can run it as often as you like and use it as the baseline before an optimization run. Prompts: - Use StoreSEO to list my twenty lowest-scoring products and the main failing criterion on each. Read only, change nothing. - Audit my collections and pages, worst first, and split the fixes into ones StoreSEO can do and ones I have to do in the app. - Which products have the weakest GEO score? Save the list with their issues to seo-audit.md in this folder. In Claude Code: Claude Code warns when a single tool result passes 10,000 tokens, and a full product listing with every criterion can get there on a big store. Ask for pages of twenty or fifty, or filter to not optimized products, instead of the whole catalog in one call. If you allowed the StoreSEO tools in your Claude Code permissions to stop the prompts, remember that the allow rule also covers the apply and generate tools, so scope it to the read tools if you only meant to speed up audits. Q: Does a Claude Code SEO audit change anything in my Shopify store? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which are all read-only. Your store only changes when you later ask for a preview and approve applying it. Q: Can Claude Code save the audit results to a file? A: Yes. Claude Code works in the folder you started it in, so you can ask it to write the worklist to a Markdown or CSV file there and pick it up in a later session. --- ## Audit your store's SEO with Claude URL: https://storeseo.com/agent/claude/seo-audit (full page: https://storeseo.com/agent/claude/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO in Claude means typing a plain question into a Claude chat with the StoreSEO connector switched on, and getting back a ranked list of the products, collections, pages and posts that score worst, with the failing checks explained in words. You sit in a normal Claude conversation in the browser or the desktop app, with StoreSEO enabled in the Search and Tools menu. You ask something like which products on my store score worst and what is wrong with each. Claude calls storeseo_start first and shows the call in the chat as a tool step you can expand, then storeseo_list_resources sorted by score with the lowest first, then storeseo_get_resource_seo for the products it wants to explain. The start call returns a snapshot of the store, so the opening answer can already tell you how many resources are optimized, need improvement or are not optimized before it names a single product. Claude is good at turning that data into something you can read in one pass. It can lay the worst products out as a table with the SEO score, the GEO score for AI readiness and the one criterion that costs the most, then group the rest by shared problem, for example a batch of products that all lack a meta description. It separates the fixes StoreSEO can make, such as meta, FAQs, the AI snippet and image alt text, from the ones only you can change in the app, such as the focus keyword or the product title. If you turn the answer into a long worklist you can ask Claude to keep it as a document in the same conversation and refine it as you decide. Because this is a chat, the audit is a conversation rather than a report. You can narrow it with follow-ups such as only collections, only products from one vendor, or just the pages that are marked not optimized, and Claude reuses the session it already opened. Audit tools only read, so Claude's connector prompts will normally be about reads; the prompts get more serious later, when you move from the audit to a preview and an apply. Prompts: - Using StoreSEO, show my ten lowest-scoring products in a table with the SEO score, the GEO score and the single biggest issue on each. - Audit my collections and pages. Which are not optimized, and which fixes could StoreSEO make versus me in the app? - Group my weakest products by the problem they share, and tell me which group would lift the most scores if fixed first. In Claude: Claude only calls tools from connectors that are switched on for the conversation, so if the audit seems to be answered from general SEO advice instead of your store, open the Search and Tools menu and check StoreSEO is enabled for that chat. On the Free plan a custom connector takes the single slot you get, and on Team and Enterprise an owner has to add StoreSEO before members can sign in. When you later allow tools in Claude, keep Allow always for the read tools and leave the apply, optimize, generate and delete tools on ask, since Claude's own prompt is the hard stop for those. Q: Can I run the SEO audit from the Claude mobile app? A: Once StoreSEO is added as a connector to your Claude account it is available in your conversations, and the audit uses only read tools. Add the connector from the web or desktop app first, then check it is switched on in the chat you open. Q: Will Claude fix the problems it finds during the audit? A: Not as part of the audit. The audit reads scores and checklists. If you then ask for a fix, StoreSEO tells Claude to preview it and wait for your yes before applying, and Claude's connector settings decide whether it also stops to ask you. --- ## Audit your store's SEO with ChatGPT URL: https://storeseo.com/agent/chatgpt/seo-audit (full page: https://storeseo.com/agent/chatgpt/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO in ChatGPT means asking a ChatGPT conversation that has the StoreSEO app connected which resources score worst, and reading back a ranked list with the failing SEO and GEO checks for each, using only read tools. You work in a ChatGPT conversation on the web with the StoreSEO connector added through developer mode and selected for the chat. You ask which products have the weakest SEO and why. ChatGPT calls storeseo_start to open the session, then storeseo_list_resources sorted by score, lowest first, then storeseo_get_resource_seo for the ones it will describe. ChatGPT shows each tool call as a step in the thread, and because all three are read-only tools, an audit is the safest first thing to run after connecting. The audit suits ChatGPT's strengths in summarizing and reshaping. Ask for the weakest ten as a table, then ask it to rewrite the same findings as a short brief for a teammate, or to sort the issues into what StoreSEO can fix and what has to be changed in the app. Products carry a GEO score as well as an SEO score, so you can ask which listings AI shopping agents are likely to misread and get the answer from your own catalog rather than from generic advice. The focus keyword and the product title are the usual items ChatGPT hands back to you, since no StoreSEO tool changes them. This job is also where your ChatGPT plan matters least. Auditing needs no write actions, so it works on a plan where developer mode only allows reads, which makes it a sensible way to use ChatGPT for StoreSEO even if you apply the fixes somewhere else. Keep the conversation going with follow-ups like only articles, only collections marked not optimized, or the worst page first, and ChatGPT reuses the session it has already opened. Prompts: - Using the StoreSEO app, list my 15 lowest-scoring products and the main failing check on each. Read only, do not change anything. - Which of my products have a low GEO score, and what would make them easier for AI shopping agents to understand? - Audit my blog posts and pages, worst first, then write a short brief I can send to my content writer. In ChatGPT: ChatGPT reaches StoreSEO from OpenAI's servers through developer mode, and which plans can use write actions differs between OpenAI's developer guide and its help center, so check what your plan allows. For an audit this barely matters, because every tool in it is read-only, but it does decide whether you can continue straight from the audit into applying fixes. If you do, read each payload before you approve, and avoid the option that remembers approval for the rest of the conversation while StoreSEO write tools are connected. Q: Does a ChatGPT SEO audit need write actions turned on? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which only read. It works on a ChatGPT plan whose developer mode allows read access, and you can apply the fixes from the StoreSEO app instead. Q: Why does ChatGPT not show StoreSEO in the chat? A: The connector has to be added in developer mode and then chosen for the conversation. If developer mode is missing under Settings, check your plan, and in a workspace ask an admin whether it is enabled. --- ## Audit your store's SEO with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/seo-audit (full page: https://storeseo.com/agent/claude-cowork/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO with Claude Cowork means giving Cowork a task that reads StoreSEO's scores for your products, collections, pages and posts and returns a ranked worklist, either right now or on a schedule you set up once. You start from the Cowork tab in the Claude desktop app and describe the audit as a task rather than a question. Cowork works through it by calling storeseo_start to open a StoreSEO session, storeseo_list_resources sorted by score with the lowest first, and storeseo_get_resource_seo for each product it needs to explain. You can watch it work, and when it finishes it hands back the worklist, or writes it into a file if you asked for one. Every tool in the audit is a read, so you can run it freely against a live store. An audit is the best job to schedule in Cowork, because the whole point is a recurring picture rather than a one-off answer. A weekly task can read the score breakdown, list the ten lowest-scoring products and write a short summary of what moved since last time, which gives you a standing worklist without opening Shopify. Cowork runs scheduled tasks without you in the conversation, so write the task as report only, and say in plain words that it must not apply, generate, publish or delete anything. When you do want to act on what the audit found, switch to an interactive session in the same place. Ask for a preview of a specific resource, read it, and say yes to the apply there, where you are present to approve it. The split keeps the scheduled task safe and keeps the decisions with you. Prompts: - Use StoreSEO to audit my store: list my ten lowest-scoring products with the main issue on each, and save the result as a worklist file. - Every Monday, read my StoreSEO score breakdown and lowest-scoring products and summarize what changed. Do not apply, generate, publish or delete anything. - Compare this week's weakest products with last week's worklist and tell me which ones are still on it. In Claude Cowork: A scheduled Cowork task runs while you are away, so nobody can say yes to a StoreSEO confirmation. Keep a scheduled audit to the read tools and word the task to forbid apply, generate, publish and delete, because that wording is an instruction the agent follows rather than a lock. If the connector is missing inside Cowork, add it under Customize, Connectors there instead of assuming one added in a regular chat carried over, and on Team and Enterprise plans an admin may need to turn Cowork on first. Q: Can Claude Cowork run my StoreSEO audit every week? A: Yes. Cowork documents scheduled tasks that run on a cadence and can use your connected apps. Word the task as a report that only reads scores and lowest-scoring products, since you are not there to approve any change. Q: Where does the audit result end up in Cowork? A: It comes back in the Cowork task, and you can ask it to write the worklist to a file or a document so you can compare it with the next run. --- ## Audit your store's SEO with Cursor URL: https://storeseo.com/agent/cursor/seo-audit (full page: https://storeseo.com/agent/cursor/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO in Cursor means asking Cursor's agent chat, inside the editor where your theme code lives, which products, collections, pages and posts score worst in StoreSEO, and getting a ranked worklist you can keep as a file in the project. You are in Cursor with a theme or storefront project open, using the agent chat panel. You type the audit request in plain words. Cursor calls storeseo_start to open the session, then storeseo_list_resources sorted by score, then storeseo_get_resource_seo for each product it will describe, and each call appears in the chat as a tool step you can expand to see the arguments and the result. These are read tools, so the results are safe to expand and read. The editor changes what you can do with the answer. Ask Cursor to write the worklist to a Markdown file in the repository, such as seo-audit.md, and it creates the file beside your code. Because the agent can also read your project, you can ask it to connect findings to the theme, for example which product template renders the page for a product with weak GEO data. It will still say that the focus keyword, the product title and the category are changed in the StoreSEO app or in Shopify admin, not by a StoreSEO tool. Before the audit, check which StoreSEO tools are switched on. The three audit tools, storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, must be among the tools active in Cursor, and the tool cap makes it likely that you disabled some StoreSEO tools. If the agent says it cannot find a listing tool, look there first. Prompts: - Using StoreSEO, list my twenty lowest-scoring products and the top failing criterion on each, then write the result to seo-audit.md. - Which products have the weakest GEO score? For each, say what would help AI shopping agents, and note which fixes are only possible in the StoreSEO app. - Audit my pages and blog posts, worst first, and add a short section to seo-audit.md for each. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you have to switch some off in Cursor's MCP settings, and the audit tools are the three you must keep on. If you switch off storeseo_list_resources by mistake, Cursor can open a session but cannot rank anything. Also, a server in .cursor/mcp.json belongs to that project only, so an audit started in another project will not see StoreSEO unless you added it to ~/.cursor/mcp.json. Q: Which StoreSEO tools does a Cursor audit need switched on? A: storeseo_start, storeseo_list_resources and storeseo_get_resource_seo. With Cursor's cap of about 40 active tools you must turn some StoreSEO tools off, so turn off ones for jobs you are not doing and keep these three. Q: Can Cursor keep the audit results in my repository? A: Yes. Ask the agent to write the worklist to a Markdown or CSV file in the project, and it creates it beside your code, where you can commit it or compare it with a later audit. --- ## Audit your store's SEO with Codex URL: https://storeseo.com/agent/codex/seo-audit (full page: https://storeseo.com/agent/codex/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO with Codex means asking OpenAI's terminal agent, signed in to the StoreSEO MCP server, which products, collections, pages and posts score worst, and having it read each checklist and report a worklist in the terminal or in a file. You run codex in your project folder, after signing in with codex mcp login storeseo, and type the request in plain words: which products score worst and what is wrong with each. Codex lists its tool calls as it goes. It calls storeseo_start first to open the session, then storeseo_list_resources sorted by score, lowest first, then storeseo_get_resource_seo for each product it needs to describe. All three are read tools, so the run is a listing and reading exercise with no change to the store. The terminal makes the audit easy to repeat and to reuse. Ask Codex to write the worklist to a file in the folder, such as seo-audit.md, and it will, which gives you something to diff against the next run. You can also ask for it in a form that suits your workflow, a list of product handles to batch through later or a table grouped by shared problem. Codex keeps the separation StoreSEO asks for: it reports which fixes a StoreSEO tool can make, such as meta, FAQs and image alt text, and which only you can make, such as the focus keyword and the product title. Codex reads a config file for the server, so how you set approval in config.toml is worth a look before the first run. For audits, the read tools can run without prompts, but keep the server's approval mode on writes or prompt so a later apply or generate still stops for a decision. Codex also waits 60 seconds for a tool by default, which is plenty for reads. Prompts: - Use StoreSEO to list my twenty lowest-scoring products and the main failing criterion on each. Read only. - Audit my collections and pages, worst first, and write the worklist to seo-audit.md, split into fixes StoreSEO can make and fixes I must make. - Which products have the weakest GEO score? Give me their handles and the issue on each. In Codex: Set the storeseo server's default_tools_approval_mode to writes or prompt in config.toml, not approve. An audit does not need approve, since every audit tool is a read, and approve would also silence the prompts for the apply and generate tools you may use next, leaving only the confirmation StoreSEO asks for in chat. A project-level .codex/config.toml loads only in trusted projects, so if Codex does not see StoreSEO in a project, use ~/.codex/config.toml instead. Q: How do I sign Codex in before running a StoreSEO audit? A: Run codex mcp login storeseo, pick your Shopify store in the browser and approve access. If you prefer a token, create one in StoreSEO under Settings, AI Toolkit and let config.toml read it from an environment variable. Q: Can Codex write the audit to a file in my project? A: Yes. Ask it to save the worklist as a Markdown or CSV file in the folder you ran codex from. The audit itself only reads from StoreSEO, so the only file written is the one you asked for. --- ## Audit your store's SEO with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/seo-audit (full page: https://storeseo.com/agent/gemini-cli/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO with Gemini CLI means asking Google's terminal agent, connected to the StoreSEO MCP server, which products, collections, pages and posts score worst, and letting it read each checklist to produce a ranked worklist. You open Gemini CLI in your project folder and ask the audit question in plain words. The first time it calls a StoreSEO tool, Gemini CLI opens your browser for the sign-in, you pick your Shopify store and approve, and it stores the tokens, so the audit then runs without further setup. It calls storeseo_start to open the session, then storeseo_list_resources sorted by score with the lowest first, then storeseo_get_resource_seo for each product it reports on. All three are reads, and Gemini CLI lists them as tool calls in the terminal as it goes. What you get is a worklist you can use straight away. Gemini CLI names the failing criteria by their visible titles, and groups them: the ones a StoreSEO tool can fix, such as meta, FAQs and image alt text, and the ones you have to change yourself, such as the focus keyword and the product title. Products also carry a GEO score for AI readiness, so you can ask which listings AI shopping agents will struggle with. Ask it to save the result to a file in the folder and you have a baseline to compare with after you fix things. Gemini CLI asks for confirmation on tool calls unless you tell it not to, and that default suits this job well. Reads are quick to approve, and the later steps, a preview and an apply, are where you want the prompt to stay. Keep that in mind before you reach for any option that skips confirmations. Prompts: - Use StoreSEO to list my twenty lowest-scoring products with the main issue on each. Read only. - Audit my collections, pages and blog posts, worst first, and save the worklist to seo-audit.md in this folder. - Which products have the weakest GEO score for AI shopping agents, and what is missing on each? In Gemini CLI: The browser sign-in needs a browser on the same machine, so an audit started over SSH or in a headless session will stall at authentication. On such a machine, create a personal access token in StoreSEO under Settings, AI Toolkit and add it as a header. Also, gemini mcp add defaults to project scope, so a server added without --scope user is found only in that project, and an audit run from another folder will not have the StoreSEO tools. Do not set trust on the server just to speed up the audit. Q: Why did Gemini CLI open a browser during my StoreSEO audit? A: The first call to the server starts the OAuth sign-in. Pick your Shopify store, approve access, and Gemini CLI stores the tokens so later audits run without another prompt. Q: Can I run the audit from a server with no browser? A: Yes, with a personal access token. Create one in StoreSEO under Settings, AI Toolkit and add it to the server entry as a header, because the browser sign-in needs a browser on the same machine. --- ## Audit your store's SEO with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/seo-audit (full page: https://storeseo.com/agent/github-copilot/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify store's SEO with GitHub Copilot means asking Copilot Chat in VS Code, with the StoreSEO server added in agent mode, which products, collections, pages and posts score worst, and getting a ranked worklist beside the theme code you are editing. You are in VS Code with a theme or storefront folder open and Copilot Chat set to agent mode, where the StoreSEO tools show up in the tools picker. You type the audit as a plain request. Copilot calls storeseo_start to open the session, then storeseo_list_resources sorted by score with the lowest first, then storeseo_get_resource_seo for each product it will describe, and each call is listed in the chat. They are all read tools, so the results are quick to review and nothing is written to the store. The editor is what makes this different from a chat tab. Ask Copilot to put the worklist in a file in the workspace, such as seo-audit.md, and it creates it beside your code, ready to commit or to compare with a later run. Because Copilot can read the project, you can also ask it which Liquid templates render the pages the audit flagged. It will still tell you that the focus keyword and the product title are changed in the StoreSEO app or in Shopify admin, and that StoreSEO tools handle meta, FAQs, the AI snippet and image alt text. Check the tools picker before you start, since agent mode only calls tools that are switched on and a server that failed to start shows no tools at all. If the audit answers from general advice rather than your store, open the picker and confirm that StoreSEO is listed and enabled for the chat. Prompts: - Use StoreSEO to list my twenty lowest-scoring products and the top issue on each. Read only, change nothing. - Audit my collections and pages, worst first, and write the worklist to seo-audit.md in this workspace. - Which products have the weakest GEO score, and which of the fixes need me in the StoreSEO app? In GitHub Copilot: On Copilot Business and Enterprise an organization admin has to enable MCP servers in Copilot policy, otherwise StoreSEO will not be usable even though it is in your mcp.json. VS Code reads the servers key in .vscode/mcp.json, not mcpServers, so a snippet copied from another editor will silently show no tools. A .vscode/mcp.json committed to the repository is shared with your team, so keep any token out of it and leave the apply, optimize, generate and delete tools on VS Code's confirmation prompt. Q: Does the Copilot audit need agent mode? A: Yes. MCP tools are used by Copilot Chat in agent mode, so switch the chat to agent mode and check that StoreSEO is enabled in the tools picker before you ask. Q: Can Copilot save the StoreSEO audit in my repository? A: Yes. Ask it to write the worklist to a Markdown file in the workspace. The audit tools only read from StoreSEO, and the file is the only thing written. --- ## Audit your store's SEO with Windsurf URL: https://storeseo.com/agent/windsurf/seo-audit (full page: https://storeseo.com/agent/windsurf/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Windsurf means asking the Cascade agent in your editor to open a StoreSEO session, list the lowest-scoring products, collections, pages and posts, and read each one's checklist so the worklist sits next to your theme code. You are in the Windsurf editor with a theme or storefront project open, and the Cascade panel beside your files. Once storeseo is in mcp_config.json and signed in, you type a request such as which fifteen products score worst and what is failing on each. Cascade calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, and the same response carries a snapshot of the store: how many resources are optimized, need improvement or are not optimized, and how many images still need work. It then calls storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. All three are reads, and they appear as tool calls in the Cascade conversation. What you get back is a worklist you can act on in the same window. Cascade names the failing criteria by their visible titles, such as a meta description that is too short or images with no alt text, and splits them into fixes a StoreSEO tool can make later and things only you can change, like the focus keyword or the product title. Products also carry a GEO score for AI readiness, so one pass shows which listings AI shopping agents will struggle with. Because Cascade works on your project, you can ask it to write the list to a Markdown file in the repository, or to compare the failing product handles against the templates in your theme before you decide what to fix first. An audit is a good first job for a new connection because it only reads. You can run it repeatedly, use it as a baseline before any optimization, and ask for narrower cuts such as only collections, only articles, or only products that are not optimized. Cascade pages through the listing instead of loading the whole catalog at once, so on a large store ask it to stop at a score threshold. Prompts: - Use StoreSEO to list my fifteen lowest-scoring products and the main failing criterion on each. Read only. - Audit my collections and pages, worst first, and write the list to seo-audit.md in this project. - Which products have the weakest GEO score for AI shopping agents, and what is missing on them? In Windsurf: Cascade can use 100 MCP tools in total across every server you have enabled, and StoreSEO alone offers 44. If the audit tools seem to be missing, check the MCPs section for how many tools are enabled and switch off servers or tools you do not need. Windsurf's MCP page also says its configuration applies to the legacy Cascade agent, while the Devin Local agent reads Devin CLI configuration files, so confirm which agent your tab uses before you assume storeseo is loaded. The docs do not describe how Cascade asks before running MCP tools, so watch the conversation for each call. Q: Does a Windsurf SEO audit change my Shopify store? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which are read-only. Your store changes only if you later preview and approve a fix. Q: Why does Cascade not show the StoreSEO audit tools? A: Check that you are using the agent that reads mcp_config.json, that storeseo is signed in, and that the 100 tool limit across all your MCP servers has not left StoreSEO's tools switched off in the MCPs section. --- ## Audit your store's SEO with Grok URL: https://storeseo.com/agent/grok/seo-audit (full page: https://storeseo.com/agent/grok/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Grok means asking xAI's assistant in a grok.com chat, through a StoreSEO custom connector, which products, collections, pages and posts score worst and why, and getting a ranked worklist back in the conversation. You are in a chat on grok.com with the StoreSEO connector available. There is no folder or terminal here: you type a question such as which ten products score worst and what each one is missing. Grok calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, and that response includes a snapshot of the store with counts of optimized, needs improvement and not optimized resources and the image status. It then calls storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for the ones that matter. All three tools are read-only, so an audit is a safe first question after connecting. The answer arrives as prose and a list in the chat. For each product Grok describes the failing criteria by their visible titles, such as a missing meta description or a focus keyword absent from the title, and separates what a StoreSEO tool can fix from what only you can change in the app, such as the focus keyword itself. Products also carry a GEO score for AI readiness, so you can ask which listings AI shopping agents are likely to misread. Because the connector reaches StoreSEO over the public internet, you do not need anything running on your own machine, and you can continue the same conversation to narrow the list by collection or by score. Grok is a chat client, so the useful habit is to keep the questions small. Ask for the lowest ten, read the reasons, then ask for the next ten, rather than requesting the whole catalog in one reply. When you are ready to fix something, the same conversation can move on to a preview, and the agent should ask before it spends credit or writes. Prompts: - Using StoreSEO, show me my ten lowest-scoring products and the top issue on each. - Which of my collections are not optimized yet, and what is failing on them? - Give me an SEO audit of my blog posts, worst first, and tell me which fixes I have to make in the app. In Grok: xAI's connector documentation describes OAuth sign-in and the connectors page but no per-tool approval setting, so the audit tools run as Grok decides to call them and the only check on later writes is StoreSEO's instruction to preview and wait for your yes. That makes an audit the right place to start, since it only reads. On Grok Business and Enterprise a team admin must provision the connector in the cloud console before you can see it, so if the StoreSEO connector is missing from your chat, ask your admin. This applies to the grok.com chat connector, not to xAI's terminal agent. Q: Can Grok audit my Shopify SEO without me installing anything? A: Yes. Add StoreSEO as a custom connector from the connectors page on grok.com, approve your store once, and ask. The audit runs through read-only StoreSEO tools, with nothing to run on your own machine. Q: Does Grok ask before each StoreSEO tool call during an audit? A: The connector documentation does not describe a per-tool approval setting. The audit tools only read, and StoreSEO tells the agent to wait for your yes before any write, which is an instruction rather than a lock. --- ## Audit your store's SEO with OpenClaw URL: https://storeseo.com/agent/openclaw/seo-audit (full page: https://storeseo.com/agent/openclaw/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with OpenClaw means your own agent runtime calls StoreSEO's read tools, on request in a chat app or on a schedule, and sends back a ranked list of the lowest-scoring products, collections, pages and posts. OpenClaw runs on your own machine or server, and you usually talk to it from a chat app such as Telegram or Slack rather than an editor. Once storeseo is added with openclaw mcp add, signed in with openclaw mcp login and checked with openclaw mcp doctor storeseo --probe, you message it something like which ten products score worst. The agent calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, then storeseo_list_resources sorted by score and storeseo_get_resource_seo for the candidates. All three are read-only. The reply lands in the same chat, with failing criteria named by their visible titles and a split between what StoreSEO can fix and what you must change in the app. The difference with OpenClaw is that the audit can run without you. Automations, the built-in scheduler, wake the agent at a set time in an isolated session and announce the result in a channel, so a Monday report can be waiting in your chat. Because the OAuth credentials are saved on the installation that owns the connector, a scheduled run reuses them without another sign-in. A scheduled audit is the right shape for this scheduler: it reads, it reports, and there is no one awake to say yes to anything else. Keep the scheduled prompt explicit about reads. Ask for the score breakdown and the lowest-scoring products, and tell the job not to preview, apply or generate. Then follow up in the chat yourself when a finding deserves a fix, because that conversation has someone present to answer the agent's question. You can also keep the agent's reach small with openclaw mcp tools, including only the list and get tools for the audit. Prompts: - Use StoreSEO to list my ten lowest-scoring products and the top issue on each. Read only. - Every Monday, report my SEO score breakdown and the five lowest-scoring products. Do not preview, apply or generate anything. - Which blog posts and pages are not optimized yet? Give me the worst five with their failing criteria. In OpenClaw: A scheduled automation has nobody to answer a question, so an audit job must be report only: write the prompt so it asks for reads and says not to preview, apply or generate. Run the mcp commands on the installation that owns the connector, because a Gateway or agent on another host needs its own reload before it sees storeseo. OpenClaw has its own per-server approval setting, openclaw mcp configure storeseo --approval followed by approve, prompt or auto, and the docs here do not spell out each mode, so do not leave it on one that skips the check unless you mean it. Q: Can OpenClaw run a StoreSEO audit on a schedule? A: Yes. openclaw automations create takes a schedule and a prompt, runs it in an isolated session and can announce the result in a chat channel. Keep the prompt to reads, because nobody is there to approve a write. Q: Will a scheduled OpenClaw audit change my Shopify store? A: Not if the prompt asks only for reads. The audit tools are read-only, and you should tell the job not to preview, apply or generate anything. Fixes belong in a chat where you can answer. --- ## Audit your store's SEO with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/seo-audit (full page: https://storeseo.com/agent/hermes-agent/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Hermes Agent means the Nous Research agent reads StoreSEO's scores for you, from a terminal or a chat app, and reports the lowest-scoring products, collections, pages and posts with what each one is missing. Hermes Agent runs from a terminal or from chat apps such as Telegram, Discord and Slack, and keeps memory across sessions. With storeseo under mcp_servers in ~/.hermes/config.yaml and signed in through hermes mcp login storeseo, you start a session or run /reload-mcp and ask which ten products score worst. Hermes calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, then storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. All are reads. The answer names failing criteria by their visible titles and separates what a StoreSEO tool can fix from what only you can change in the app, such as the focus keyword. Because Hermes remembers across sessions, an audit becomes a running baseline. You can ask it to compare today's worst ten with the list it gave you last week, or to note which products you decided to fix. Products carry a GEO score for AI readiness as well, so you can ask which listings AI shopping agents will struggle with. When the list is long, ask for pages of ten or twenty, and ask it to stop at a score threshold. The scheduler is what suits Hermes to auditing. hermes cron create takes a schedule such as every monday 9am and a prompt, and --deliver sends the result to Telegram or another chat target. An audit is exactly the sort of job to hand it: it only reads, and the finished report arrives where you already read messages. When something in the report deserves a fix, answer in a live chat, where the agent can ask and you can say yes. Prompts: - Use StoreSEO to list my ten lowest-scoring products and the top issue on each. Read only. - Compare my current five worst products with the ones you reported last time. - Which products have the weakest GEO score for AI shopping agents, and what is missing? In Hermes Agent: A hermes cron job has nobody to say yes, so keep a scheduled audit to reads and reports, and never word the prompt so that it previews, applies or generates anything. After editing config.yaml, start a new session or run /reload-mcp, or the open session will not see storeseo. If you want an audit-only agent, use the tools include list to expose only StoreSEO's list and get tools. Hermes' own tool-approval controls are not described in the docs this page was checked against, so do not count on the client to stop a write. Q: How do I get a weekly StoreSEO audit from Hermes Agent? A: Create a cron job with hermes cron create, give it a schedule such as every monday 9am and a read-only prompt, and add --deliver telegram or another chat target so the report arrives in your messages. Q: Does Hermes Agent remember last week's audit? A: Hermes keeps memory across sessions, so you can ask it to compare the current lowest scores with what it reported before. StoreSEO itself is read live each time, so the scores always come from the store. --- ## Audit your store's SEO with OpenCode URL: https://storeseo.com/agent/opencode/seo-audit (full page: https://storeseo.com/agent/opencode/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with OpenCode means asking the open-source terminal agent which products, collections, pages and posts score worst in StoreSEO, and getting a ranked worklist you can keep beside your code. You run OpenCode in a project folder, often the theme repository, with storeseo added through opencode mcp add and signed in with opencode mcp auth storeseo. You type a request such as which twenty products score worst and why. OpenCode exposes MCP tools to the model as server_tool, so the StoreSEO tools appear with a storeseo prefix, but the calls are the same ones in the same order: storeseo_start first, because every other tool needs the conversation id it returns, then storeseo_list_resources sorted by score and storeseo_get_resource_seo for each candidate. All three only read. You can check the connection with opencode mcp list, or use /mcps inside the session. The result is a table in the terminal. For each product the agent names the failing criteria by their visible titles and separates what a StoreSEO tool can fix from what only you can change, such as the focus keyword or the product title. Products also carry a GEO score for AI readiness. Because OpenCode works in your project folder, you can ask it to save the worklist to a file there, or to cross-check failing handles against theme templates, and pick the list up in a later session. OpenCode lets you shape what the agent can reach. Permission rules can match storeseo_* patterns, and a deny rule hides tools without disconnecting the server, so you can deny the write tools by name and leave the audit tools available. That makes a read-only audit session a deliberate configuration rather than a hope. On a big catalog, ask for pages of twenty and stop at a score threshold rather than requesting every product at once. Prompts: - Use StoreSEO to list my twenty lowest-scoring products and the main failing criterion on each. Read only. - Audit my collections and pages, worst first, and save the list to seo-audit.md in this folder. - Which products have the weakest GEO score? Group them by the criterion they fail most. In OpenCode: OpenCode's documentation describes deny rules for MCP tools but no ask-first mode, so there is no client prompt to rely on once you start fixing things. For an audit that is fine, since the tools only read, but deny the StoreSEO write tools by name if you want the audit session to be unable to write at all. Also check your version's layout: OpenCode V2 puts servers under mcp.servers, while older versions put the server directly under mcp, and opencode mcp add writes to the project configuration unless you pass --global. Q: Can I make an OpenCode session that can only audit my store? A: Yes. Add deny rules for the StoreSEO write tools by name, or a pattern, so only the read tools remain available. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo. Q: Why do StoreSEO tools have a storeseo prefix in OpenCode? A: OpenCode names MCP tools as the server name followed by the tool name. That prefix is also what permission patterns that start with storeseo_ match when you want to hide or deny tools. --- ## Audit your store's SEO with Zed URL: https://storeseo.com/agent/zed/seo-audit (full page: https://storeseo.com/agent/zed/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Zed means asking the Agent Panel in your editor which products, collections, pages and posts score worst in StoreSEO, and approving the read calls as Zed prompts for them. You are in Zed with a project open and the Agent Panel docked beside your code. With storeseo added under context_servers in settings.json and a green dot next to it in Settings, AI, MCP Servers, you ask which fifteen products score worst and what is failing on each. The agent calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, then storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. Every one of those is a read, but Zed's default setting is confirm, so the panel stops and asks you to approve each tool action before it runs. The answer comes back in the panel: failing criteria named by their visible titles, and a split between what a StoreSEO tool can fix and what you must change in the app, such as the focus keyword. Products carry a GEO score for AI readiness too, so the same audit shows which listings AI shopping agents will struggle with. Because the agent is working inside your project, you can have it write the worklist to a Markdown file next to your theme code or compare product handles against the templates you are editing. The prompts are the practical difference in Zed. An audit of fifteen products can mean a dozen approvals, which is tedious but gives you a visible record of every call. If that gets in the way, Zed's per-tool keys let you allow the read tools by name and leave the write tools on confirm, so the audit runs quickly while a later apply still stops for you. Prompts: - Use StoreSEO to list my fifteen lowest-scoring products and the main failing criterion on each. Read only. - Audit my pages and blog posts, worst first, and write the list to seo-audit.md in this project. - Which products have the weakest GEO score for AI shopping agents, and what is missing on each? In Zed: Zed's agent.tool_permissions.default is confirm unless you change it, so even the read-only audit tools prompt before they run and a catalog audit means several approvals. To speed that up, allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_list_resources and keep the write tools on confirm. Setting the default to allow removes Zed's check entirely, and then the confirmation StoreSEO asks the agent for is the only one, so use it sparingly. The per-tool keys need Zed v0.224.0 or later. Q: Why does Zed ask me to approve every StoreSEO read? A: Zed's tool permission default is confirm, which prompts before any tool action, reads included. Allow specific read tools with mcp:storeseo keys if you want the audit to run without prompts. Q: How do I know the StoreSEO server is ready in Zed? A: Open Settings, AI, MCP Servers. A green dot next to storeseo means the server is active. If it is not, complete the sign-in that Zed prompts for, then ask the audit question in the Agent Panel. --- ## Audit your store's SEO with Kiro URL: https://storeseo.com/agent/kiro/seo-audit (full page: https://storeseo.com/agent/kiro/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Kiro means asking the agent in AWS's IDE which products, collections, pages and posts score worst in StoreSEO, with the read tools optionally listed in autoApprove so the audit runs without stopping. You are in Kiro with a project open and the agent chat beside your files. Once storeseo is in .kiro/settings/mcp.json or ~/.kiro/settings/mcp.json and the browser sign-in is done, you ask which fifteen products score worst and what is failing on each. The agent calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, then storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. All three are reads. The reply names failing criteria by their visible titles and separates what a StoreSEO tool can fix from what only you can change in the app, such as the focus keyword. Kiro is built around planning multi-step work in a project, so an audit fits naturally as the first step of a longer job. Ask for the worklist, have the agent save it into the workspace as a Markdown file, then come back to it and decide what to fix. Products also carry a GEO score for AI readiness, which tells you which listings AI shopping agents may misread. On a large catalog, ask for pages of twenty and stop at a score threshold instead of pulling every product in one go. The setting worth knowing is autoApprove. Each server can carry a list of tools that run without a prompt, and for an audit the right list is the three read tools: storeseo_start, storeseo_list_resources and storeseo_get_resource_seo. With that in place the audit runs without interruptions, while any write tool you ask for later still goes through Kiro's own prompt, which is a second check on top of the confirmation StoreSEO tells the agent to ask for. Prompts: - Use StoreSEO to list my fifteen lowest-scoring products and the main failing criterion on each. Read only. - Audit my collections, pages and blog posts, worst first, and save the list to seo-audit.md in this workspace. - Which products have the weakest GEO score for AI shopping agents, and what would fix each? In Kiro: Kiro's documentation does not state its default approval behavior when autoApprove is empty, so check what your install does before you assume reads run quietly or prompt. If you do use autoApprove, list the three read tools individually. A wildcard would also cover StoreSEO's apply and generate tools and let Kiro run them without asking, leaving StoreSEO's own confirmation as the only check. A workspace .kiro/settings/mcp.json also overrides a user-level entry with the same name, so if the sign-in points at the wrong store, check which file defines storeseo. Q: Which StoreSEO tools should I put in Kiro's autoApprove list for an audit? A: The three read tools: storeseo_start, storeseo_list_resources and storeseo_get_resource_seo. List them individually rather than with a wildcard, so apply and generate tools still go through Kiro's prompt. Q: Can I keep an SEO audit worklist inside my Kiro project? A: Yes. Ask the agent to write the results to a Markdown file in the workspace, then reopen it later to choose what to fix. The audit itself only reads from StoreSEO. --- ## Audit your store's SEO with Antigravity URL: https://storeseo.com/agent/antigravity/seo-audit (full page: https://storeseo.com/agent/antigravity/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your store's SEO with Antigravity means asking Google's agent in the IDE side panel which products, collections, pages and posts score worst in StoreSEO, and approving the read calls in Ask mode as they come. You are in the Antigravity IDE with a project open and the agent panel at the side. Once storeseo is in mcp_config.json with serverUrl and you have clicked Authenticate under Customizations and pasted the authorization code back, you ask which fifteen products score worst and what is failing on each. The agent calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, then storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo for each candidate. All three only read. Because MCP tools you have not configured run in Ask mode, each call waits for your approval in the panel before it executes. The reply names failing criteria by their visible titles, and separates what a StoreSEO tool can fix from what only you can change in the app, such as the focus keyword. Products carry a GEO score for AI readiness as well, so the same pass shows which listings AI shopping agents will struggle to understand. With a project open, you can have the agent save the worklist as a Markdown file in the workspace, or check failing product handles against your theme's templates before deciding what to fix first. Ask mode makes an audit slightly slower but easy to follow, since you see each call before it runs. If you want it quicker, Antigravity's policy patterns of the form mcp(server/tool) let you allow specific tools, so you can allow the three read tools and leave everything else on Ask. That keeps the audit smooth and leaves any later apply or generate step stopping for you, in addition to the confirmation StoreSEO tells the agent to request. Prompts: - Use StoreSEO to list my fifteen lowest-scoring products and the main failing criterion on each. Read only. - Audit my collections and pages, worst first, and save the list to seo-audit.md in this workspace. - Which products have the weakest GEO score for AI shopping agents, and which fixes can StoreSEO make? In Antigravity: Antigravity runs MCP tools you have not configured in Ask mode, so even the read-only audit tools wait for your approval and a long audit means several clicks. A policy that allows mcp(storeseo/*) would end the prompts, but it would also cover the apply and generate tools, leaving StoreSEO's own confirmation as the only check, so allow the three read tools individually. If the server will not connect, check that the entry uses serverUrl: the documentation says legacy fields such as url or httpUrl are not supported. The sign-in ends with a code you paste into Agent settings, so finish it in the window where you started. Q: Why does Antigravity ask before each StoreSEO audit call? A: MCP tools you have not configured run in Ask mode, which requires approval before execution. Add policy patterns for the three read tools if you want the audit to run without prompts. Q: Why will Antigravity not connect to StoreSEO from my copied config? A: Antigravity uses the serverUrl field for remote servers and does not support url or httpUrl. Rename the field in mcp_config.json, then authenticate again under Customizations in Agent settings. --- ## Optimize meta tags and content with Claude URL: https://storeseo.com/agent/claude/optimize-meta (full page: https://storeseo.com/agent/claude/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Claude means asking the assistant in your chat tab to have StoreSEO draft a meta title, description, tags, URL handle, FAQs and AI snippet for a product, then reading the draft and the score change before you let it save anything. You work in a Claude chat at claude.ai or in Claude Desktop, with the StoreSEO connector switched on for the conversation. You name a product or ask for the worst few, and Claude calls storeseo_start, then storeseo_get_resource_seo to read what the resource already has: its current meta, whether FAQs and an AI snippet exist, and which criteria fail. That read tells it what a preview would add, so the reply can say up front that the meta is too long and the FAQs are missing before any credit is spent. Then comes storeseo_preview_resource_optimization. StoreSEO writes the copy on its own servers, so the text you read in the chat is StoreSEO's draft, not something Claude composed. Claude shows the new meta title and description, the tags, the FAQs and the AI snippet next to the projected score, and if the URL handle would change it lays out the old URL and the new one and tells you a redirect will be created. You answer in the same chat: apply all of it, apply the meta and skip the FAQs, or leave the handle alone. Claude then calls storeseo_apply_resource_optimization with only the parts you named, and the reply carries the recomputed score and what still needs a manual fix, such as the focus keyword. Because the draft lives for an hour, a chat is a comfortable place to take your time. You can read the preview, ask Claude to compare it with the current copy, go and check the product page, and come back to say apply it. The preview itself spent AI Content Optimizer credit whether or not you apply, so batch requests like my five worst products mean five previews, and a good habit is to ask Claude for the credit cost of the whole batch before you say go. Applying is free. Prompts: - Use StoreSEO to preview a new meta title, description and FAQ for my Linen Throw Blanket product. Show me the draft and the score change before you apply anything. - Preview optimized meta for my three lowest-scoring products, one at a time. Tell me what the previews will spend first, then wait for my yes. - Apply the meta description and title from that preview, but skip the URL handle and the FAQs. In Claude: Claude shows its own permission prompt for each tool that writes, and the preview tool is not marked read-only, so expect a prompt for the preview as well as for the apply. Choosing Allow always on the preview is reasonable once you trust the credit cost, but keep the apply tool on ask, because Claude's prompt is the hard stop and StoreSEO's request for your yes is only an instruction the assistant follows. Also check that the StoreSEO connector is toggled on in the Search and Tools menu for the chat, or Claude will answer from general knowledge instead of reading your store. Q: Does Claude write the meta description itself? A: No. Claude asks StoreSEO to generate the content on its servers and shows you the result. That keeps the copy tied to your store data and the SEO criteria, and it means the score change you see is projected by StoreSEO, not guessed by the assistant. Q: What if I close the Claude chat after seeing the preview? A: The draft stays available for one hour. If you open a new chat in that time, ask Claude to apply the draft for that product. After an hour you need a new preview, which spends AI Content Optimizer credit again. --- ## Optimize meta tags and content with ChatGPT URL: https://storeseo.com/agent/chatgpt/optimize-meta (full page: https://storeseo.com/agent/chatgpt/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with ChatGPT means asking it, in a developer-mode chat with the StoreSEO app selected, to have StoreSEO generate a draft for a product and show it to you before ChatGPT is allowed to apply any of it. You open a ChatGPT conversation on the web, pick the StoreSEO app from the Developer mode tool in the composer, and ask for a product or a handful of products to be improved. ChatGPT calls storeseo_start, reads the resource with storeseo_get_resource_seo, and then calls storeseo_preview_resource_optimization. The copy is generated by StoreSEO on its server, so ChatGPT's job is to carry your request in and read the draft back to you: meta title, meta description, tags, FAQs, an AI snippet and, if it would change, the URL handle shown as old address to new address. The part that feels different here is the confirmation. OpenAI documents that write actions in developer mode require approval by default and that you can inspect the payload before it runs. StoreSEO's preview tool is not marked read-only, so ChatGPT is likely to ask you to confirm the preview call as well as the later storeseo_apply_resource_optimization call. Treat the first as the moment you accept the AI Content Optimizer credit cost and the second as the moment you accept the change to the live product. Read the payload on the apply, because it lists which parts are being saved, and ask ChatGPT to leave out any you do not want, such as the FAQs or the handle. When the apply returns, ChatGPT reports the recomputed score and the criteria that are still open. Those are usually things the tools cannot write for you, like the focus keyword, which you set in the StoreSEO app. If you plan to work through many products, ask ChatGPT to give you a single list of titles and the projected score gain for each before it previews more than one, so you decide where the credit goes. Prompts: - Using the StoreSEO app, preview optimized meta, tags and an AI snippet for my Merino Beanie product. Show me the old and new URL handle if it changes, and wait for my approval. - Using the StoreSEO app, find the three collections with the lowest SEO score and preview a new meta title and description for the first one only. - Apply that preview but leave out the URL handle, then tell me the new score and what is still failing. In ChatGPT: Whether ChatGPT can run the apply step depends on your plan. OpenAI lists developer mode for several plans, but as we read its help center, write actions are in beta on Business, Enterprise and Edu workspaces while Pro accounts get read access only. On a plan without write actions, the preview and the read steps may fail or refuse, so use ChatGPT to review and plan, and apply from the StoreSEO app or another agent. The remember-this-approval option is the other trap: remembering an apply approval for the whole conversation means later products are saved without a fresh look at the payload. Q: Why does ChatGPT ask me to confirm the preview as well as the apply? A: Developer mode asks for approval on tools that are not marked read-only, and the StoreSEO preview tool is not read-only because it spends AI Content Optimizer credit. The preview writes nothing to your store, but the confirmation is where you accept the credit cost. Q: Can ChatGPT change my product title? A: No. This job writes the SEO meta title, not the product title shown on your storefront. Edit the product title in Shopify admin. ChatGPT can still tell you when the product title is the reason a criterion fails. --- ## Optimize meta tags and content with Claude Code URL: https://storeseo.com/agent/claude-code/optimize-meta (full page: https://storeseo.com/agent/claude-code/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Claude Code means working from your terminal or editor, asking it to have StoreSEO preview generated meta for the products you pick, reading the draft in the session, and applying only the pieces you approve. You are in the terminal, often inside the Shopify theme repository, with the storeseo server added at user scope. You type a request such as optimize the meta for the three lowest-scoring products and show me the previews first. Claude Code calls storeseo_start, then storeseo_get_resource_seo for each product, and then storeseo_preview_resource_optimization one product at a time. Each call is printed in the session as it runs, so you see the tool name and, if you have not allowed it, a permission prompt before it executes. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet on its own servers, so Claude Code does not author any of it. What it adds is the working context around the draft. Because it can read the theme, it can check whether the product template already renders FAQs or whether a Liquid snippet hard-codes a title tag that would override the meta you are about to apply, and tell you before you spend the credit. When a handle would change it prints the old and new URL, and since you are in a repository you can ask it to grep the theme and content for hard-coded links to the old path. After your yes it calls storeseo_apply_resource_optimization with the parts you chose, and the response gives the recomputed score. For a batch you can have Claude Code write a short table to a file in the working folder: product, score before, projected score, what you applied, what you skipped. Remember that every preview spends AI Content Optimizer credit even if you never apply it, so ask it to state the number of previews it is about to run before it starts the loop. Prompts: - Use StoreSEO to preview optimized meta for the three lowest-scoring products, one at a time. Tell me how many previews that is before you start and wait for my yes between products. - Preview meta, FAQs and an AI snippet for the product with handle merino-beanie. If the URL handle changes, grep this theme for links to the old path before I decide. - Apply the meta title and description from the last preview, skip the handle, and write a before and after score table to meta-changes.md. In Claude Code: Claude Code runs a tool without asking only if you have allowed it, and the allow rule matches by tool name, so allowing the preview and apply tools to cut down on prompts removes the editor-side stop for writes. Keep storeseo_apply_resource_optimization and storeseo_update_ai_optimization_settings on ask, because the confirmation StoreSEO asks for in the chat is an instruction to the agent, not a lock on the server. If you run a loop over many products, the previews can also pile up credit quickly, so cap the batch in the prompt. Q: Can Claude Code optimize meta for many products in one run? A: It can run previews one product after another, but each preview spends AI Content Optimizer credit and each draft expires after an hour. Ask for a small batch, review each draft, and apply before moving on rather than previewing the whole catalog first. Q: Does Claude Code edit my theme files when it applies meta? A: No. The apply tool writes the SEO fields to your Shopify store through StoreSEO. Claude Code only edits theme files if you separately ask it to, for example to remove a hard-coded title tag. --- ## Optimize meta tags and content with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/optimize-meta (full page: https://storeseo.com/agent/claude-cowork/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Claude Cowork means delegating the task in plain words, letting it run StoreSEO previews through your connector, and reviewing a finished set of drafts before anything is applied to your store. In Cowork you describe the outcome rather than steer each step: improve the meta on my ten lowest-scoring products and give me the drafts to review. Cowork uses the StoreSEO custom connector you added under Customize, then Connectors. It calls storeseo_start, reads each resource with storeseo_get_resource_seo, and runs storeseo_preview_resource_optimization for the products in scope. StoreSEO writes the meta title, description, tags, FAQs, AI snippet and any handle change on its servers, and the preview projects each new score without saving anything. Delegation changes how you spend credit, so say how many previews you will allow in the task itself. Each preview spends AI Content Optimizer credit whether or not you apply it, and a ten product task is ten previews. A good task statement caps the count, asks Cowork to stop after the drafts and to produce one review document: product, current score, projected score, the new meta, and the old and new URL for any handle change. You read that, reply with which products to apply and which fields to skip, and only then does Cowork call storeseo_apply_resource_optimization, which is free and returns the recomputed score. Keep the approval conversation inside the task while you are present. Cowork can ask you a question and wait, which is what the preview-then-apply flow needs. Drafts last an hour, so if the review takes longer than that, expect Cowork to have to preview again, which spends credit a second time. Prompts: - Use StoreSEO to preview new meta for my five lowest-scoring products. Preview no more than five, apply nothing, and give me one review sheet with the old and new URL for any handle change. - From the previews you just made, apply the meta titles and descriptions for products 1, 3 and 4 only, skip the FAQs and the handles, and report the new scores. - Draft a plan for which collections to optimize first, using StoreSEO scores, and do not run any previews until I pick. In Claude Cowork: Do not schedule this job as a recurring Cowork task. Scheduled tasks run on their cadence without you in the conversation, so nobody is there to say yes, and a scheduled preview would spend AI Content Optimizer credit every run while a scheduled apply would have no review at all. Schedule reads and reports instead, and keep meta optimization as a task you start and watch. The same applies to turning on automatic optimization with storeseo_update_ai_optimization_settings: do that only in a live task, after Cowork states what it will spend. Q: Can I schedule Claude Cowork to optimize my meta every week? A: It is not a good fit. A scheduled task runs without you, so there is no one to approve the apply, and each run would spend AI Content Optimizer credit on previews. Use a scheduled task to read scores and report, and run the optimization yourself when you are present. Q: How do I stop a Cowork task from spending too much credit? A: State the maximum number of previews in the task and ask Cowork to stop after the drafts. Each preview spends credit whether or not you apply it, so a cap in the instructions is the simplest control. --- ## Optimize meta tags and content with Cursor URL: https://storeseo.com/agent/cursor/optimize-meta (full page: https://storeseo.com/agent/cursor/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Cursor means asking the editor's agent, with your theme project open, to have StoreSEO preview generated meta for a product, reading the draft in the chat panel, and approving only the parts you want applied. You have the theme or storefront project open in Cursor and the storeseo server enabled in its MCP settings. In the agent chat you ask for a product's meta to be improved. Cursor calls storeseo_start, storeseo_get_resource_seo and then storeseo_preview_resource_optimization, and each call shows in the chat with its arguments so you can see which product and which resource type it is acting on. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet on its own servers; Cursor shows you the draft and the projected score and does not write the copy itself. Cursor asks for approval before an MCP tool runs unless your Run Mode lets it go ahead, so the approval you give in the chat panel and the yes you type to the agent are two separate things. The first is Cursor's hard stop on the tool call. The second is the confirmation StoreSEO tells the agent to wait for. For the apply step, keep both: read the payload Cursor shows, check which fields and axes are listed, and have the agent leave out the handle if you do not want a redirect created. Because the project is open, the editor context is useful. If the preview proposes a new handle, ask Cursor to search the repository for hard-coded links to the old path, or to look at the product template to see whether a hard-coded title tag will override the new meta. The credit rule still applies: every preview spends AI Content Optimizer credit even if you never apply it, so work through products one at a time rather than asking for a whole collection in one message. Prompts: - Use StoreSEO to preview new meta for the product merino-beanie. Show me the draft, the score change and the old and new URL, and wait before applying. - Preview meta for my lowest-scoring collection and tell me what the preview will spend. Then search this repo for links to its current handle. - Apply only the meta title and description from that preview and skip the handle and the FAQs. In Cursor: StoreSEO has 44 tools and Cursor stops at about 40 active tools across all servers, so a preview or apply tool may be one you switched off to make room. This job needs storeseo_get_resource_seo, storeseo_preview_resource_optimization and storeseo_apply_resource_optimization enabled; if Cursor says it cannot find one, open the storeseo entry in the MCP settings and turn it back on, switching off something unrelated such as a rank tracking lookup instead. If you set the apply tool to run without asking, the agent's own confirmation is all that is left, so keep it on ask. Q: Cursor says it cannot preview the optimization. Why? A: The most likely cause is the tool cap. Cursor allows about 40 active tools across all servers and StoreSEO has 44, so the preview tool may be switched off. Enable storeseo_preview_resource_optimization in the storeseo entry of the MCP settings and turn off tools this job does not need. Q: Does Cursor write the meta text itself? A: No. StoreSEO generates the meta title, description, tags, FAQs and handle on its servers. Cursor presents the draft and passes your decision back, which is why the score change in the preview comes from StoreSEO and not from the model in the editor. --- ## Optimize meta tags and content with Codex URL: https://storeseo.com/agent/codex/optimize-meta (full page: https://storeseo.com/agent/codex/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Codex means asking OpenAI's coding agent, in the CLI, the IDE extension or the desktop app, to have StoreSEO preview generated meta for a product and to apply only the parts you approve. You run Codex in the repository you work from, with the storeseo server in config.toml and signed in through codex mcp login. A request like preview better meta for the product with handle trail-socks leads Codex to call storeseo_start, then storeseo_get_resource_seo, then storeseo_preview_resource_optimization. The CLI prints each tool call as it happens. StoreSEO writes the meta title, description, tags, FAQs, AI snippet and any new handle on its servers and returns a draft with a projected score, so Codex reads that draft back to you rather than composing text of its own. Codex has per-server approval settings, and they decide when you see a prompt. With default_tools_approval_mode set to writes, any tool that is not marked read-only prompts, which includes the preview tool and the apply tool. That gives you two checkpoints: the prompt on the preview, where you accept the AI Content Optimizer credit cost, and the prompt on the apply, where you accept the change to the live product. Between them sits the agent's own question, which you answer in the session, and you should say which fields to skip, such as the handle if you do not want a redirect. Generation can take a while. Codex waits 60 seconds for a tool to finish by default, and a preview that writes several fields can approach that, so if a preview times out, raise tool_timeout_sec for the storeseo server and run it again. Be aware that a timed-out preview may still have spent credit, so check the draft and the credit use before you retry. A re-preview spends credit again. Prompts: - Use StoreSEO to preview optimized meta for the product with handle trail-socks. Show me the draft and score change, and wait for my yes before applying. - Preview a new meta title, description and FAQ for my Summer Dresses collection and keep the current URL handle. Tell me the credit cost first. - Apply the meta from that preview but skip the FAQ, then show me the new score and what is still failing. In Codex: Set the server's default_tools_approval_mode to writes or prompt. If you set it to approve, Codex runs the preview and apply tools without prompting and the confirmation StoreSEO asks for in the conversation is the only check left, which is an instruction the agent follows and not a lock. Preview generation can also exceed Codex's default 60 second tool wait, so raise tool_timeout_sec for storeseo if a preview times out, and check your credit use before you re-run it, because a second preview spends credit again. Q: Why did my Codex meta preview time out? A: Codex waits 60 seconds for a tool to finish by default and generating several fields can take longer. Raise tool_timeout_sec for the storeseo server in config.toml. Check your credit use before retrying, since a second preview spends AI Content Optimizer credit again. Q: Which Codex approval mode works best for meta optimization? A: The writes mode is a good fit, because it prompts for every tool that is not marked read-only, including the preview and the apply, while reads such as storeseo_get_resource_seo run without interruption. Avoid the approve mode for this server. --- ## Optimize meta tags and content with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/optimize-meta (full page: https://storeseo.com/agent/gemini-cli/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Gemini CLI means asking Google's terminal agent to have StoreSEO preview generated meta for a product, reading the draft in the session, and confirming the apply step tool by tool. You start gemini in a terminal, in your theme folder or anywhere else, with the storeseo server added and signed in. You ask for meta to be improved for a named product or for the lowest-scoring few. Gemini CLI calls the StoreSEO tools, which it lists to the model as mcp_storeseo_storeseo_start and so on, beginning with storeseo_start, then storeseo_get_resource_seo to read the current fields and failing criteria, then storeseo_preview_resource_optimization. StoreSEO generates the meta title, description, tags, FAQs, AI snippet and any handle change on its servers and returns a projected score; Gemini CLI reads that back to you. Gemini CLI shows a confirmation for MCP tool calls unless the server is marked trusted, and for this job that is useful in two places. On the preview you confirm that you accept the AI Content Optimizer credit cost, and on the apply you see the arguments naming which axes will be saved. Before the apply, the agent should show the old and new URL if the handle changes and tell you a redirect will be created; answer in the session with the parts to apply, for example the meta but not the FAQs. After the apply, the response gives the recomputed score and what still needs a manual fix. Everything stays in one terminal session, so you can hand Gemini CLI follow-up work in the same breath: write the before and after scores to a file, or search the project for links to the old handle. Drafts last an hour, and a fresh preview spends credit again, so apply or abandon a draft in the same sitting. If you are on a headless machine, sign in with an access token header instead of the browser flow. Prompts: - Use StoreSEO to preview optimized meta for the product merino-beanie. Show me the draft, the score change and any URL handle change, and wait for my yes. - Preview a new meta title and description for my About page, tell me what the preview will spend, and do not apply it yet. - Apply the title and description from that preview, skip the handle, and save the before and after scores to meta-changes.md. In Gemini CLI: Do not mark the storeseo server as trusted for this job. The trust setting skips Gemini CLI's tool confirmations, which would remove the client-side stop on both the credit-spending preview and the apply, and then the confirmation StoreSEO asks for in the session is the only check, an instruction the model follows and not a lock. If you need fewer prompts, use includeTools to expose only the read tools plus the preview and apply tools for this task rather than trusting everything. Also note Google's notice that Gemini CLI was replaced by Antigravity CLI for unpaid tier and Google One users, so confirm your account can still use it. Q: Should I set trust to true for the storeseo server in Gemini CLI? A: Not for meta optimization. Trust skips every tool confirmation, including the one on the apply step. Leave it off so Gemini CLI prompts, or use includeTools to limit which StoreSEO tools the model can see. Q: Can Gemini CLI run this on a server without a browser? A: Yes, with an access token. The browser sign-in needs a browser on the same machine, so on a headless host add the server with an Authorization Bearer header using a token from Settings, AI Toolkit, then run the same preview and apply flow. --- ## Optimize meta tags and content with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/optimize-meta (full page: https://storeseo.com/agent/github-copilot/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with GitHub Copilot means using Copilot Chat in Agent mode inside VS Code to have StoreSEO preview generated meta for a product, reading the draft beside your code, and approving the apply step in the editor. You have the theme workspace open in VS Code, the storeseo server listed in .vscode/mcp.json and started from MCP: List Servers, and Copilot Chat switched to Agent mode. You ask for the meta of a product to be improved. Copilot calls storeseo_start, reads the resource with storeseo_get_resource_seo and then calls storeseo_preview_resource_optimization. StoreSEO generates the meta title, description, tags, FAQs, AI snippet and any handle change on its servers and projects the new score. Copilot reads the draft back in the chat, so the text you review is StoreSEO's, not the model's. VS Code adds its own confirmation to every tool call. In Agent mode it shows a prompt with the tool name and its input, and you can allow it once, for the session, for the workspace or always. That prompt is the hard stop, and the yes you give in the chat is the instruction StoreSEO tells the agent to wait for. For this job use the prompt differently on each tool: allowing the preview once is a way to accept the AI Content Optimizer credit cost, and the apply prompt deserves a read of the arguments, which list the axes being saved and any fields skipped, such as the handle. Working in the editor means the theme is right there. If the preview proposes a new handle, ask Copilot to search the workspace for links to the old path, or to check whether the product template renders a hard-coded title tag that would override your new meta. A fresh preview spends credit again and the draft lasts an hour, so settle one product before starting the next. Business and Enterprise users may find the server blocked until an admin has enabled MCP servers in Copilot policy. Prompts: - Use StoreSEO to preview optimized meta for the product merino-beanie. Show me the draft, the projected score and any URL handle change, and wait for my approval. - Preview a new meta title, description and FAQ for my Summer Dresses collection. Then search this workspace for links to its current handle. - Apply the meta from that preview, skip the FAQ and the handle, and tell me the new score. In GitHub Copilot: VS Code lets you approve a tool for the session, the workspace or always, and Allow always on the apply tool removes the editor's hard stop, leaving only the confirmation StoreSEO asks for in the chat, which is an instruction to the agent and not a lock. Keep storeseo_apply_resource_optimization on a per-call prompt. Also remember the file-format trap: the workspace file uses the servers key, so a snippet copied from a Cursor or Claude guide will not load, and then Agent mode will not list the preview or apply tools at all. Q: Why does Copilot not list the StoreSEO preview tool in Agent mode? A: Check that .vscode/mcp.json uses the servers key and type http, that the server is started in MCP: List Servers, and that the tools icon in the chat includes StoreSEO. On Business and Enterprise plans an admin must also enable MCP servers in Copilot policy. Q: Does Copilot apply meta changes to my code? A: No. The apply tool saves the SEO fields to your Shopify store through StoreSEO. Copilot edits files in your workspace only when you separately ask it to, for example to remove a hard-coded title tag in the theme. --- ## Optimize meta tags and content with Windsurf URL: https://storeseo.com/agent/windsurf/optimize-meta (full page: https://storeseo.com/agent/windsurf/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Windsurf means asking Cascade to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for one Shopify resource, then applying the draft you approve while your theme stays open in the editor. You work in the Cascade panel beside your theme files. Once the storeseo entry is in mcp_config.json and the MCPs section shows its tools enabled, you type a request such as optimize the meta for my worst-scoring product and show me the draft first. Cascade calls storeseo_start, then storeseo_get_resource_seo for the product, so it knows which fields already exist and which criteria fail. It then calls storeseo_preview_resource_optimization. That one generates the copy on StoreSEO's side, so Cascade is presenting text it did not write itself, and it spends AI Content Optimizer credit even if you decide not to apply anything. The draft comes back in the chat as a meta title, a meta description, tags, a proposed URL handle, FAQs and an AI snippet, with the projected score change next to them. If the handle changes, Cascade should show the old and new URL, because applying creates a redirect. Because you are in an editor, this is a natural moment to compare the proposed title against the theme's own template, for example a title suffix the theme appends. You can say apply everything except the FAQs, and Cascade calls storeseo_apply_resource_optimization with only the axes you chose and skips fields you reject. The response carries the new score and the remaining checklist, which Cascade summarizes as the next thing to fix. StoreSEO tells the agent to wait for your yes before it applies, and that instruction lives in the conversation rather than in a server lock. Windsurf's MCP page does not describe a per-tool approval setting, so read each proposed write in the chat before you accept it. Cascade can also read the current auto optimization settings with storeseo_get_ai_optimization_settings if you want to know what StoreSEO already does for newly synced products before you spend credit on an existing one. Prompts: - Use StoreSEO to preview new meta for my lowest-scoring product. Show the draft and the score change, and apply nothing until I say so. - Preview a meta title and description for the collection Summer Dresses. Keep the current URL handle and skip the FAQ. - Apply the meta from that preview but leave out the tags, then tell me what is still failing. In Windsurf: Windsurf's MCP page says its configuration applies to the legacy Cascade agent, while the Devin Local agent that opens by default in new tabs reads Devin CLI config files instead. If Cascade offers no storeseo tools, check which agent the tab is using before you edit mcp_config.json again. The 100-tool cap is shared by every MCP server you have enabled, so StoreSEO's 44 can lose room to other servers, and a preview or apply tool may be the one switched off. Q: Does Cascade write the meta text itself? A: No. StoreSEO generates the meta title, description, tags, URL handle, FAQs and AI snippet on its own side when Cascade calls the preview tool. Cascade shows you the draft and applies it after your yes. Q: Does a Windsurf preview cost anything if I do not apply it? A: Yes. A preview spends AI Content Optimizer credit when it is generated, whether or not you apply it, and a second preview of the same resource spends again. Applying a draft is free. --- ## Optimize meta tags and content with Grok URL: https://storeseo.com/agent/grok/optimize-meta (full page: https://storeseo.com/agent/grok/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Grok means asking it in a chat on grok.com to have the StoreSEO connector generate a meta title, description, tags, URL handle, FAQs and AI snippet for a Shopify resource, then applying the draft you say yes to. You are in a normal Grok conversation with the StoreSEO custom connector available, so there is no terminal and no file to edit. You type something like rewrite the meta for my Linen Throw Blanket product and show me the draft. Grok opens a StoreSEO session with storeseo_start, reads the product with storeseo_get_resource_seo, which returns the current fields, the score and each failing criterion, and then calls storeseo_preview_resource_optimization. StoreSEO writes the copy on its own side and projects the new SEO score, plus the GEO score for products, without saving anything. Because this is chat, the draft arrives as readable text in the thread, which makes it easy to edit by talking. You can answer make the description shorter and mention free shipping, but remember what that means in practice: Grok does not author StoreSEO's generated fields, so a changed draft means another preview, and another preview spends credit again. A cheaper path is to apply the draft as it is, or skip individual fields such as the handle, and change a single field later in the StoreSEO app. When the preview proposes a new URL handle Grok should show you the old and the new address and tell you a redirect will be created before you agree. Applying is a separate call, storeseo_apply_resource_optimization, and StoreSEO tells the agent to wait for your yes before making it. xAI's connector documentation does not describe a per-tool approval setting, so that instruction and your own reading of the proposal are the main check in the conversation. After the apply, Grok reports the recomputed score and what still fails, and you can ask it to move on to the next product in the same thread. Prompts: - Using StoreSEO, preview new meta for my lowest-scoring product. Show the generated text and the score change before you save anything. - Draft a meta title, description and FAQ for my About Us page and wait for my go-ahead. - Apply that preview but keep the current URL handle. In Grok: A preview spends AI Content Optimizer credit when it is generated, and in a chat it is tempting to keep asking for small rewrites. Each rewrite request that triggers a new preview spends again, so decide which fields you want changed before you ask, or apply the draft and tweak a single field in the app. On Grok Business and Enterprise a team admin has to provision the connector in the cloud console first, so members may see no StoreSEO tools until that is done. Q: Can I ask Grok to reword the generated meta description? A: You can ask, but StoreSEO writes the generated fields on its own side, so the agent cannot edit the draft in place. A reworded request means a new preview, which spends AI Content Optimizer credit again. You can also apply the draft and edit one field afterwards in the StoreSEO app. Q: Will Grok change my product title? A: No. This job writes the SEO meta title, not the storefront product title. The product title is edited in Shopify admin. --- ## Optimize meta tags and content with OpenClaw URL: https://storeseo.com/agent/openclaw/optimize-meta (full page: https://storeseo.com/agent/openclaw/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with OpenClaw means messaging your own agent runtime from a chat app so it has StoreSEO generate meta, tags, a URL handle, FAQs and an AI snippet for one Shopify resource, then applies the draft you approve in the thread. OpenClaw runs on your machine or server, and you talk to it from WhatsApp, Telegram, Slack, Discord or its Control UI. After openclaw mcp doctor shows the storeseo server healthy, you message it something like preview the meta for my three lowest-scoring products. The agent calls storeseo_start, reads each resource with storeseo_get_resource_seo, and calls storeseo_preview_resource_optimization once per resource. Each preview spends AI Content Optimizer credit, so when you ask for three, expect three previews to be charged whether or not you apply them, and say so in your message if you want the agent to repeat the cost back to you first. The drafts come back as chat messages with the projected scores, which suits a phone: you read them on the move and reply apply the first two and skip the FAQ on the third. The agent then calls storeseo_apply_resource_optimization with the axes you chose and any skipped fields, and the draft stays usable for one hour, so a reply later in the afternoon can still apply the remainder. OpenClaw can also read what StoreSEO does automatically for new products through storeseo_get_ai_optimization_settings, which helps when you are deciding whether a one-off preview or an auto setting fits. Writes are where OpenClaw differs from a desktop client. Its mcp configure command has a per-server approval setting with approve, prompt and auto modes, and that is the hard stop if you want the client to check before StoreSEO writes. The automations scheduler runs agents in an isolated session with nobody present to answer, so use it for reads such as listing the lowest scores each Monday, and ask for previews and applies in a message you can reply to. Prompts: - Preview StoreSEO meta for my three lowest-scoring products and send me each draft with its score change. Do not apply anything yet. - Apply the first draft, skip the URL handle on the second, and leave the third alone. - Every Monday, report which products have no meta description. Read only, no previews and no applies. In OpenClaw: A scheduled OpenClaw automation cannot wait for a yes, so keep any cron job for this job to reads and reports. Do not tell a scheduled run to preview or apply: previews spend credit unattended, and an apply has no one to approve it. Also check the per-server approval mode with openclaw mcp configure storeseo, because a mode that skips the check removes OpenClaw's own stop on storeseo_apply_resource_optimization, leaving only the confirmation StoreSEO asks the agent for in the chat. Q: Can an OpenClaw automation optimize my meta tags on a schedule? A: It can read and report on a schedule, but it should not preview or apply. Nobody is present to say yes, and a preview spends AI Content Optimizer credit. Ask for previews in a chat message you can answer, and use automations for score reports. Q: How long can I wait before applying a preview from my phone? A: A StoreSEO draft expires one hour after the preview. Within that hour you can apply part of it now and the rest in a follow-up message. After that the agent has to generate a new preview, which spends credit again. --- ## Optimize meta tags and content with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/optimize-meta (full page: https://storeseo.com/agent/hermes-agent/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Hermes Agent means telling it in a terminal or chat app to have StoreSEO generate meta, tags, a URL handle, FAQs and an AI snippet for a Shopify resource, then applying the draft after you reply with a yes. Hermes Agent keeps memory across sessions and answers in whatever place you reached it, a terminal or a Telegram, Discord or Slack chat. With the storeseo entry in ~/.hermes/config.yaml and the sign-in cached, you write something like preview new meta for the collection Summer Dresses and keep its URL handle. Hermes calls storeseo_start, then storeseo_get_resource_seo for the collection, then storeseo_preview_resource_optimization. StoreSEO generates the text server-side and projects the score without saving, and the preview spends AI Content Optimizer credit whether or not you apply it. The draft appears in the conversation with the projected score. Reply in the same place: apply the meta and skip the FAQ. Hermes passes only those axes to storeseo_apply_resource_optimization and the skipped parts stay available from the draft for an hour. Because Hermes remembers earlier sessions, it can recall which products you already optimized and which you skipped, but treat that memory as a convenience rather than a record: storeseo_get_resource_seo shows what is actually saved in the store, including whether FAQs or an AI snippet already exist. To keep the agent to reading and previewing while you experiment, narrow its tools with the include list on the storeseo server so the apply and settings tools are not exposed, then widen it when you are ready. Hermes can also run storeseo_get_ai_optimization_settings to tell you what StoreSEO already generates for new resources. If the shop's AI model matters to you, that read also lists the model in use. Prompts: - Preview StoreSEO meta for my five lowest-scoring products, one at a time, and show each draft and score change. Do not apply anything. - Preview a new meta title and description for the Summer Dresses collection and keep the current URL handle. - Apply the meta from that preview and skip the FAQ, then tell me what still fails. In Hermes Agent: If you edit config.yaml to add or narrow the storeseo tools, start a new Hermes session or run /reload-mcp, otherwise the open session keeps the old tool list and may still expose, or still hide, the preview and apply tools. A hermes cron job has nobody to say yes, so give it reads only for this job: previews would spend credit unattended, and the apply is a write that needs your answer. Ask for those in a chat you can reply to. Q: Can I restrict Hermes Agent to previews so it cannot apply anything? A: Yes. Hermes lets you narrow a server's tools with include and exclude lists in config.yaml, so you can leave out storeseo_apply_resource_optimization. Previews still spend AI Content Optimizer credit, and you need to restore the tool before you can apply a draft. Q: Does Hermes remember which products it already optimized? A: Hermes keeps memory across sessions, but the saved state in your store is the reliable record. Ask it to read the resource with StoreSEO before deciding, because storeseo_get_resource_seo shows the current fields, score and any FAQs already saved. --- ## Optimize meta tags and content with OpenCode URL: https://storeseo.com/agent/opencode/optimize-meta (full page: https://storeseo.com/agent/opencode/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with OpenCode means asking the terminal agent to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for one Shopify resource, then applying the draft you approve from the prompt. You start OpenCode in your theme project, check that opencode mcp list shows storeseo as connected, and type a request such as preview new meta for my lowest-scoring product and wait for me. The model sees the tools as storeseo_ plus the tool name and calls storeseo_start, then storeseo_get_resource_seo, then storeseo_preview_resource_optimization. StoreSEO writes the draft on its own side and projects the new SEO score, plus the GEO score for products, without saving anything. The preview spends AI Content Optimizer credit even if you never apply it. The draft is printed in the terminal, which makes it easy to keep. Ask OpenCode to paste the title and description into a notes file in the project, or to compare the proposed URL handle with the product's current one before you decide. When the handle changes, you should see the old and new URL first, because applying creates a redirect. Say apply everything except the handle, and OpenCode calls storeseo_apply_resource_optimization with the skipped fields listed. The result returns the recomputed score and checklist, which OpenCode summarizes as what is fixed and what still needs you. OpenCode's permission rules can match tool names with a pattern such as storeseo_*. The docs describe deny rules but no ask-first mode, so StoreSEO's own request for your yes in the chat is the main check on writes. If you want a client-side stop, deny storeseo_apply_resource_optimization and storeseo_update_ai_optimization_settings by name while you review drafts, and remove the rule when you are ready to apply. Prompts: - Use StoreSEO to preview new meta for my five lowest-scoring products. Show each draft and the score change, and do not apply anything. - Preview a meta title, description and FAQ for the page About Us, keep the handle, and save the draft text to meta-draft.md. - Apply the previewed meta but skip the tags and the URL handle. In OpenCode: OpenCode V2 expects servers under mcp.servers while older versions put the entry directly under mcp, so a config copied from a tutorial may leave storeseo missing and the model with no preview tool at all. Because the documented permission rules are deny rules and not an ask-first mode, a deny pattern on storeseo_* also blocks the read and preview tools, so deny only the apply and settings tools by name if you want drafts but no writes. Q: Can I stop OpenCode from applying a draft while I review it? A: Yes. Add a permission rule that denies storeseo_apply_resource_optimization by name. Previews and reads still run, and you remove the rule when you are ready to apply. A broad pattern that matches every StoreSEO tool would block the preview tool too. Q: Does OpenCode need the StoreSEO tools added in every project? A: Only if you added the server without the global flag. opencode mcp add writes to the project configuration by default, so run it with the global flag to make StoreSEO available in every project. --- ## Optimize meta tags and content with Zed URL: https://storeseo.com/agent/zed/optimize-meta (full page: https://storeseo.com/agent/zed/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Zed means asking the Agent Panel to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for a Shopify resource, then approving the draft in the panel and at Zed's tool prompt. You open the Agent Panel in the window where your theme is loaded and, with a green dot beside storeseo under Settings, AI, MCP Servers, type a request such as preview the meta for my lowest-scoring product. Zed's default setting is confirm, so the panel asks before each tool action, including the reads. Expect a prompt for storeseo_start, another for storeseo_get_resource_seo, and another for storeseo_preview_resource_optimization. That last one generates the draft on StoreSEO's side, projects the new score without saving anything, and spends AI Content Optimizer credit when you allow it. The generated meta title, description, tags, FAQs and AI snippet are shown in the panel, and a proposed URL handle comes with the old and new address because applying creates a redirect. This is where Zed gives you two separate checks. First StoreSEO tells the agent to wait for your yes in the conversation. Then, when the agent calls storeseo_apply_resource_optimization, Zed's own prompt appears and you allow or reject that single call. If you reply apply the meta and skip the FAQ, the call carries only those axes and the skipped parts stay available from the draft for an hour. Several prompts in a row can feel slow when you work through a handful of products. Rather than setting the default to allow, give the read tools their own permission with keys of the form mcp:storeseo:tool_name, and leave the preview and apply tools on confirm. The preview prompt is a good moment to remember the credit cost, and the apply prompt is the last point where you can stop a handle change that would create a redirect. Prompts: - Use StoreSEO to preview new meta for my lowest-scoring product. Show me the draft and score change and apply nothing yet. - Preview a meta title, description and FAQ for the collection Summer Dresses and keep the current URL handle. - Apply the meta from that preview and skip the FAQ. In Zed: Zed asks before every tool action by default, reads included, and each preview prompt is a point where credit is about to be spent. Setting agent.tool_permissions.default to allow removes Zed's check, so the confirmation StoreSEO asks the agent for in the panel becomes the only one before storeseo_apply_resource_optimization writes. Allow individual read tools with per-tool keys and leave the preview, apply and settings tools on confirm. Q: Why does Zed prompt me for every StoreSEO call? A: Zed's agent.tool_permissions.default is confirm, which asks before a tool action runs. You can allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_get_resource_seo and keep the preview and apply tools on confirm. Q: Do I approve the apply twice in Zed? A: Effectively yes. StoreSEO tells the agent to ask for your yes in the panel before it applies, and Zed then shows its own tool prompt for that call. The Zed prompt is the hard stop, and the in-panel confirmation is an instruction the agent follows. --- ## Optimize meta tags and content with Kiro URL: https://storeseo.com/agent/kiro/optimize-meta (full page: https://storeseo.com/agent/kiro/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Kiro means asking the agent in its IDE chat panel to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for a Shopify resource, then applying the draft you approve. You work in Kiro with your theme or storefront project open, and the storeseo server defined in .kiro/settings/mcp.json or the user-level file. In the chat panel you type a request such as preview new meta for the product Linen Throw Blanket and show me the score change. Kiro calls storeseo_start, then storeseo_get_resource_seo, then storeseo_preview_resource_optimization. If you listed the read tools in autoApprove they run straight away; otherwise Kiro prompts for each. The preview is generated by StoreSEO on its own side, projects the new SEO score without saving, and spends AI Content Optimizer credit. Kiro works in a plan-and-act style, so it often wants to break a request into steps. For this job, keep the steps small: one resource, one preview, one decision. When the draft appears, with a meta title, description, tags, FAQs, an AI snippet and sometimes a new URL handle, read the old and new address before you agree, since applying creates a redirect. A reply such as apply all of it but not the tags makes the agent call storeseo_apply_resource_optimization with the axes you chose, and the result shows the recomputed score and what still fails. Kiro's per-server settings are useful here. The autoApprove list can cover the read tools, and disabledTools can hide the apply and settings tools from the agent while you are only experimenting with drafts. StoreSEO tells the agent to wait for your yes before applying, but that is an instruction the agent follows, so the autoApprove list is where you decide whether Kiro itself adds a second prompt for the write. Prompts: - Use StoreSEO to preview new meta for my lowest-scoring product. Show the draft and the projected score and wait for my answer before applying. - Preview a meta title, description and FAQ for my Shipping Policy page and keep the handle. - Apply that preview except the tags, then show me the new score and the checklist that remains. In Kiro: autoApprove with a wildcard would let Kiro run every StoreSEO tool without asking, which includes storeseo_apply_resource_optimization and the preview tool that spends credit. List individual read tools such as storeseo_get_resource_seo instead, so Kiro's prompt remains a second check on the write. A workspace file at .kiro/settings/mcp.json also overrides a user-level storeseo entry, so if the wrong store answers, check which file defines it. Q: Should I put the preview tool in Kiro's autoApprove list? A: Only if you are comfortable spending AI Content Optimizer credit without a prompt. A preview generates content and writes nothing to the store, but it costs credit each time. Keep the apply and settings tools out of autoApprove so Kiro still prompts for writes. Q: Can Kiro hide the apply tool while I only review drafts? A: Yes. The disabledTools list on the storeseo server hides tools from the agent, so adding storeseo_apply_resource_optimization there lets you generate and read previews without any way to apply them until you remove it. --- ## Optimize meta tags and content with Antigravity URL: https://storeseo.com/agent/antigravity/optimize-meta (full page: https://storeseo.com/agent/antigravity/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with Antigravity means asking the agent in Google's IDE side panel to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for a Shopify resource, then applying the draft you approve. You open the agent side panel in Antigravity with your storefront project loaded, and once storeseo shows as authenticated under Agent settings, Customizations you type a request such as preview new meta for my lowest-scoring product. Tools you have not configured run in Ask mode by default, so the agent needs your approval before each call. It runs storeseo_start, then storeseo_get_resource_seo, then storeseo_preview_resource_optimization. StoreSEO generates the draft on its own side and projects the score without saving. The preview spends AI Content Optimizer credit, whether or not you apply it. When the draft comes back, with a meta title, description, tags, FAQs and an AI snippet, read it against the page it will sit on. Antigravity can open the theme template in the same window, so check how the title is rendered and whether the theme adds a store-name suffix before the new meta title looks too long. If a new handle is proposed you should see the old and new URL, because applying creates a redirect. Reply apply the meta and skip the handle, and the agent calls storeseo_apply_resource_optimization with only those parts. There are two layers of confirmation. StoreSEO tells the agent to wait for your yes in the conversation, and Antigravity's Ask mode then prompts for the call itself. The Ask mode prompt is the hard stop. If you later add a policy of the form mcp(storeseo/tool), you decide per tool whether Antigravity stops asking, and the preview and apply tools are the ones worth leaving on Ask. Prompts: - Use StoreSEO to preview new meta for my lowest-scoring product and show me the draft and projected score. Apply nothing yet. - Preview a meta title, description and FAQ for the collection Summer Dresses and keep the current URL handle. - Apply the meta from that preview but skip the URL handle and the FAQ. In Antigravity: Antigravity reads remote servers from the serverUrl field and does not support url or httpUrl, so an entry copied from another client leaves storeseo without tools and the preview call never appears. The sign-in also ends with a code you paste back into Agent settings, so finish it in the same window. If you allow mcp(storeseo/*) in a policy, Antigravity stops asking before storeseo_apply_resource_optimization, and only the confirmation StoreSEO asks the agent for remains. Q: Will Antigravity ask before it applies a StoreSEO draft? A: By default, yes. MCP tools you have not configured run in Ask mode, which requires your approval before execution. A policy that allows every StoreSEO tool at once turns that prompt off, so allow only specific read tools if you want to keep it for writes. Q: Does previewing meta in Antigravity change my store? A: No. The preview generates a draft and a projected score without saving anything. It spends AI Content Optimizer credit, and the store changes only when the agent calls the apply tool after your yes. --- ## Optimize images and alt text with Claude URL: https://storeseo.com/agent/claude/image-seo (full page: https://storeseo.com/agent/claude/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing images with Claude means asking Anthropic's chat assistant to find the heaviest and unlabelled images in your Shopify store and having StoreSEO compress them and write alt text and file names, one image per call, with a restore for each change. You open a chat on claude.ai or in Claude Desktop with the StoreSEO connector switched on and type something like find my heaviest product images and tell me which ones have no alt text. Claude calls storeseo_start first and then storeseo_list_images, which is a read. The listing shows, for every image, whether compression, alt text and file name are done, and by default it puts the largest unoptimized files at the top. Claude turns that into a short table in the chat, with the product each image belongs to, and you can ask it to narrow the list to one product or to images with no alt text at all. Fixing is a separate step. Claude tells you how many images it plans to touch and what kind of spend that means: compression counts against the Image Optimizer quota, while alt text and file names use AI credit. After your yes it calls storeseo_optimize_image once per image, with the axes you picked, and the connector shows each call in the conversation. The response reports compression, file name and alt text separately, so Claude can say that one image was compressed, got new alt text and was refused on a rename without bundling it all into one vague success. If you dislike a result, you say so in the same chat and Claude calls storeseo_restore_image for that image. Because the work is one image per call, a bulk run in a chat is a long string of tool calls. That suits Claude well when you keep the batch to a product or a handful of images and read the summary after each. Claude also keeps the earlier listing in context, so you can say do the next five without describing the filter again. Renaming a file changes its storefront URL, and Claude can remind you of that before it starts. Prompts: - Using StoreSEO, list my ten heaviest unoptimized product images and tell me how much Image Optimizer quota compressing them will use. - Which of my product images have no alt text? Show them grouped by product, then write alt text for the ones on my best seller. - Restore the original alt text on the Canvas Tote image you just changed, and leave its compression as it is. In Claude: Claude only calls StoreSEO in a chat where the connector is switched on in the Search and Tools menu, so a missing optimize tool usually means it is toggled off for that conversation. Anthropic advises Allow always only for servers you trust, and for this job it is safer to leave storeseo_optimize_image and storeseo_restore_image on ask: with one call per image, a long run means many prompts, which is the price of an approval you control. A run that renames collection or article images can sit for a couple of minutes per image while StoreSEO re-uploads them, so keep those batches small. Q: Can Claude compress my Shopify images without the Image Optimizer add-on? A: No. Compression needs the Image Optimizer add-on and counts against its monthly quota. Claude can still ask StoreSEO to write alt text and file names, which spend AI Content Optimizer credit instead, and the report for each image shows which part applied and which was refused. Q: Can I ask Claude to undo an image change? A: Yes. Tell Claude which image and which part to revert, and it calls storeseo_restore_image for compression, alt text, file name or any combination. Restoring compression brings back the original file, and restoring a file name points the storefront URL back at the original. --- ## Optimize images and alt text with ChatGPT URL: https://storeseo.com/agent/chatgpt/image-seo (full page: https://storeseo.com/agent/chatgpt/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing Shopify images with ChatGPT means using a developer mode app connected to StoreSEO so the assistant can list your heaviest or unlabelled images, then compress them and write alt text and file names one image per call, with a restore if you dislike a result. You work in a ChatGPT conversation with the StoreSEO app selected in the composer, and you ask for something like show me product images that have no alt text. ChatGPT calls storeseo_start and then storeseo_list_images. Both are read-only, and ChatGPT uses the read-only annotation to tell them apart from writes. What comes back is a listing of each image with its compression, alt text and file name status, which ChatGPT summarizes in the chat as a table grouped by product, with the largest unoptimized files first unless you ask for another order. The fixing step is where ChatGPT differs from a terminal. When it reaches storeseo_optimize_image or storeseo_restore_image, OpenAI's developer mode treats them as write actions and shows a confirmation with the payload, so you can read which image and which axes it is about to run. That is on top of the yes StoreSEO tells the agent to wait for, and the confirmation is the hard stop. You can choose to remember the approval for the rest of the conversation, which saves clicks over a run of many images but also removes the pause on each one, so use it only for a batch whose scope you already agreed. Each call handles one image with the axes you chose, and the response reports compression, file name and alt text separately. ChatGPT can then tell you that an image was compressed but its rename was refused, or that alt text was regenerated, without hiding it behind a single success message. Ask for the work in small batches, such as one product at a time, so the summaries stay readable and each payload you approve is short. Prompts: - Using the StoreSEO app, list my product images with no alt text, grouped by product, and do not change anything yet. - Compress the three heaviest product images and tell me the Image Optimizer quota before you start. - Write alt text only, not file names, for every image on the product Linen Shirt, one image at a time. In ChatGPT: Write actions in ChatGPT depend on your plan. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education on the web, but as we read OpenAI's help center, full write support is in beta on Business, Enterprise and Edu workspaces and Pro accounts get read access only. On a plan without write actions ChatGPT can still list images and tell you what is wrong, but storeseo_optimize_image and storeseo_restore_image will not run, so apply the fixes from the StoreSEO app or another agent. Q: Will ChatGPT ask me before it compresses or renames an image? A: StoreSEO tells the agent to say what a run will cost and wait for your yes, and ChatGPT's developer mode shows a confirmation for write actions such as storeseo_optimize_image. The remember option stops that pause for the rest of the conversation, so keep it off unless the batch is agreed. Q: Why does ChatGPT list my images but not change them? A: Listing is a read, so it runs on any plan that can connect. Optimizing and restoring are writes, and ChatGPT only runs them where your plan or workspace allows write actions in developer mode, so check that setting or apply fixes in the StoreSEO app. --- ## Optimize images and alt text with Claude Code URL: https://storeseo.com/agent/claude-code/image-seo (full page: https://storeseo.com/agent/claude-code/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing image SEO with Claude Code means asking Anthropic's terminal agent to list your heaviest and unlabelled Shopify images through StoreSEO, then compress them and write alt text and file names one image per call, while it checks your theme code for the URLs a rename changes. You run Claude Code in the theme repository and type a request such as list my twenty heaviest product images and flag any with no alt text. It calls storeseo_start, then storeseo_list_images, and the calls scroll past in the terminal as reads. Each row carries the status of compression, alt text and file name, so Claude Code can sort the worklist by size or by missing alt text and write it to a file in the folder if you want to work through it over a few sessions. The terminal changes what happens around the fix. Before it optimizes anything, Claude Code can search the repository for the file names it is about to change, because a new file name gives the image a new storefront URL and any Liquid section or JSON template that hard-coded the old URL keeps serving the old file. You see that grep result in the same session, decide which images are safe to rename, and only then approve storeseo_optimize_image. Each call handles one image and returns a per-axis report, which Claude Code can append to a changelog file so you have a record of what was compressed, renamed or refused. If a result is wrong, you ask for a restore and Claude Code calls storeseo_restore_image for the axes you name. Because the session stays open, you can alternate between listing a batch, optimizing it, and re-listing to confirm the status columns changed, which is also how you recover when a response is lost: Claude Code re-lists the image instead of retrying, since a retry would regenerate and spend again. Prompts: - List my twenty heaviest unoptimized product images with StoreSEO, write them to image-worklist.md, and change nothing yet. - Before renaming any image files, grep this theme for the current file names and tell me which ones are hard-coded. - Compress and add alt text to the images on my Canvas Tote product, one call per image, and show me each report. In Claude Code: Claude Code warns when one tool result passes 10,000 tokens, and a listing of images with all three status blocks can reach that on a large catalog. Ask for pages of twenty or filter to one product or to images with no alt text rather than the whole library. If you added an allow rule for the StoreSEO tools to stop prompts, it also covers storeseo_optimize_image and storeseo_restore_image, so a loose rule removes the pause on every image; scope it to the read tools if you only meant to speed up listing. Q: Can Claude Code tell me which theme files use an image before I rename it? A: Yes, because it works in the folder you started it in. Ask it to search the theme for the current file name before you approve a rename, since a new file name changes the image's storefront URL and anything that hard-coded the old one keeps the old file. Q: Can Claude Code run image optimization on many images in one command? A: Not in a single StoreSEO call. The tool takes one image per call, so Claude Code works through a batch as a series of calls and reports each result, and you should keep the batch to a size you are happy to review. --- ## Optimize images and alt text with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/image-seo (full page: https://storeseo.com/agent/claude-cowork/image-seo/index.md). Job: Image SEO and alt text, described once above. Doing image SEO with Claude Cowork means delegating the image review to Anthropic's task agent, which reads your Shopify image statuses through StoreSEO and hands back a worklist, while the compression, alt text and file name changes wait for a task where you are present to say yes. In Cowork you do not chat turn by turn; you describe a task and let Claude work through it. For images that might be review every product image in my store, group the ones with no alt text by product and rank the heaviest unoptimized files. Cowork uses the StoreSEO custom connector you added under Customize, Connectors, calls storeseo_start and then storeseo_list_images, which are reads, and returns a document or summary you review afterwards. The listing already tells it which of compression, alt text and file name are done, so the report can say exactly what is outstanding on each product. Fixing the images is a different kind of task. storeseo_optimize_image compresses against the Image Optimizer quota and writes alt text and file names against AI credit, so StoreSEO tells the agent to name the spend and wait for your yes before it runs one. Cowork's value is that you can hand off long work, but a yes needs you in the conversation, so run the fixing task while you are watching, in small batches, and read each per-image report as it comes back. If something is wrong, ask for storeseo_restore_image on that image and the axes you want reverted. Cowork can also run scheduled tasks, which suits the read half of this job well. A weekly task can list unoptimized images, count the ones with no alt text and write you a short summary, so you know when it is worth starting a fixing session. It does not need to change anything to do that. Prompts: - Review all my product images with StoreSEO, group those with no alt text by product, and write a ranked list of the heaviest unoptimized files. Do not change anything. - Every Monday, use StoreSEO to count images with no alt text and list the five heaviest unoptimized images. Report only: do not optimize, rename or restore anything. - I am here now: compress and write alt text for the images on my Linen Shirt product, one at a time, and show me each report. In Claude Cowork: A scheduled Cowork task runs without you in the conversation, so nobody is there to give the yes that storeseo_optimize_image and storeseo_restore_image wait for. Keep scheduled image tasks to reading statuses and writing a report, and say in plain words that it must not optimize, rename or restore anything. Run the actual fixing as a task you start and watch, because compression spends Image Optimizer quota and alt text and file names spend AI credit, and a rename changes the image's storefront URL. Q: Can a scheduled Claude Cowork task optimize my images on its own? A: It should not. A scheduled task runs without you present, so there is no one to say the yes StoreSEO tells the agent to wait for. Use scheduled tasks to list images and report what is unoptimized, and start fixing yourself when you are there. Q: Where do I add StoreSEO for Cowork? A: Under Customize, then Connectors, inside Cowork. If StoreSEO does not appear there, add it in that list rather than assuming a connector added elsewhere carried over, then sign in and pick your Shopify store. --- ## Optimize images and alt text with Cursor URL: https://storeseo.com/agent/cursor/image-seo (full page: https://storeseo.com/agent/cursor/image-seo/index.md). Job: Image SEO and alt text, described once above. Running image SEO with Cursor means using the editor's agent, in the theme project you already have open, to list your heaviest and unlabelled Shopify images through StoreSEO and then compress them and write alt text and file names one image per call. You open the theme in Cursor, start an agent chat and ask something like show me product images with no alt text and the five heaviest unoptimized files. The agent calls storeseo_start and storeseo_list_images, and Cursor shows each tool call in the chat with its arguments. The result is a read of image statuses, so the agent can name each image's product and say whether compression, alt text and file name are done. Since the theme files are open beside it, you can also ask it to search the project for an image's file name before you decide to rename anything. Cursor asks for approval before an MCP tool runs unless your Run Modes allow it. That makes it the hard stop in front of storeseo_optimize_image, which compresses against the Image Optimizer quota and writes alt text and file names against AI credit. StoreSEO also tells the agent to name the spend and wait for your yes, so a run involves both your answer in the chat and Cursor's own approval of each call. Each call handles one image with the axes you chose and returns a report for compression, file name and alt text separately, which the agent summarizes after every image. If you do not like an outcome, ask the agent to restore it: storeseo_restore_image reverts compression, alt text or file name for that image. Because file name changes alter the storefront URL, the editor context helps here, since the agent can point to the Liquid or JSON files that reference the old name. Prompts: - Using StoreSEO, list the twenty heaviest unoptimized product images and search this theme for hard-coded references to their file names. - Write alt text for every image on the product Ceramic Mug, one storeseo_optimize_image call per image, and wait for my yes before the first one. - Restore the file name on the last image you renamed and tell me which theme file still refers to the old name. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you had to switch some StoreSEO tools off. If storeseo_list_images, storeseo_optimize_image or storeseo_restore_image is one of them, the agent will say it cannot do the job or will fall back to describing what to do. Open the storeseo entry in Cursor's MCP settings and make sure these three are on, switching off tools you do not need for images, such as the rank and competitor lookups, instead. Q: Why can Cursor not find the image tools for StoreSEO? A: Cursor caps active tools at about 40 across all servers and StoreSEO has 44, so at least four are switched off. Check the storeseo entry in MCP settings and turn on storeseo_list_images, storeseo_optimize_image and storeseo_restore_image, then switch off tools this job does not need. Q: Does Cursor approve each image optimization separately? A: By default Cursor asks before an MCP tool runs, and storeseo_optimize_image handles one image per call, so a batch means one approval per image unless a Run Mode allows it. Keep the optimize and restore tools on ask, because that setting is the hard stop. --- ## Optimize images and alt text with Codex URL: https://storeseo.com/agent/codex/image-seo (full page: https://storeseo.com/agent/codex/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing Shopify images with Codex means asking OpenAI's coding agent, from your terminal or editor, to read your image statuses through StoreSEO and then compress them and write alt text and file names one image per call, with a restore for any change you reject. You start Codex in the theme repository and ask it to list the product images that have no alt text and the heaviest ones that are not compressed. Codex calls storeseo_start and storeseo_list_images, which are reads, and under a writes approval mode they run without a prompt because they are marked read-only. The listing gives each image's compression, alt text and file name status, and Codex can turn it into a worklist file in the repository or search the theme for the file names before you rename anything. The server's approval setting in config.toml decides what happens next. With default_tools_approval_mode set to writes or prompt, Codex stops before storeseo_optimize_image and storeseo_restore_image, which is the hard stop, while StoreSEO also tells the agent to name the Image Optimizer quota or AI credit it will spend and wait for your yes. Each call handles one image with the axes you picked and returns a report for compression, file name and alt text separately, so Codex can tell you which part applied and which was refused. Codex shares config.toml between the CLI, the IDE extension and the ChatGPT desktop app, so the same approval settings apply wherever you run the job. Timing matters in this job. A file name change on a collection or article image re-uploads it, and Codex waits 60 seconds for a tool by default, so those calls can run out of time before StoreSEO finishes. If that happens, re-list the image to see its state rather than repeating the call, because a retry regenerates and spends again. Prompts: - List my images with no alt text via StoreSEO, group them by product and write the result to images-todo.md. Make no changes. - Compress the ten heaviest unoptimized product images, one call per image, and stop for my yes before the first call. - Rename and add alt text to the images on my Collection Summer banner, and tell me about the URL change before you start. In Codex: Codex waits 60 seconds for a tool call by default, and a file name change on a collection or article image re-uploads it, which StoreSEO says can take up to a couple of minutes. Raise tool_timeout_sec for the storeseo server before renaming those images. If a call times out, re-list the image with storeseo_list_images to see what applied rather than retrying, because a retry regenerates and spends again. Do not set the server's approval mode to approve, since that removes the prompt in front of every image write. Q: Why did my Codex image rename time out? A: Renaming a collection or article image re-uploads it, which can take up to a couple of minutes, and Codex waits 60 seconds by default. Raise tool_timeout_sec for the storeseo server in config.toml, and re-list the image before retrying so you do not spend credit twice. Q: Which Codex approval mode suits image optimization? A: Use writes or prompt for the storeseo server. Reads such as listing images then run freely while storeseo_optimize_image and storeseo_restore_image stop for approval, and that setting is the hard stop in addition to the yes StoreSEO tells the agent to wait for. --- ## Optimize images and alt text with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/image-seo (full page: https://storeseo.com/agent/gemini-cli/image-seo/index.md). Job: Image SEO and alt text, described once above. Doing image SEO with Gemini CLI means asking Google's terminal agent to list your heaviest and unlabelled Shopify images through StoreSEO and then compress them and write alt text and file names one image per call, confirming each tool call as it goes. You run gemini in your theme folder and ask which product images have no alt text and which unoptimized files are heaviest. The model sees the StoreSEO tools as mcp_storeseo_ followed by the tool name, calls storeseo_start and storeseo_list_images, and shows each call in the terminal. Both are reads. The listing carries each image's compression, alt text and file name status, so Gemini CLI can print a ranked list or write it to a file in the folder for the next session. Gemini CLI asks you to confirm an MCP tool call unless the server is trusted, and for this job that confirmation matters because storeseo_optimize_image compresses against the Image Optimizer quota and writes alt text and file names against AI credit. StoreSEO also tells the agent to name the spend and wait for your yes in the chat, so you answer once for the plan and once for each call the CLI prompts. Each call handles one image and returns a report for each axis, which Gemini CLI summarizes. If you want a tighter surface, includeTools on the server can limit what the model sees to the three image tools plus storeseo_start. A rename changes the image's storefront URL, so ask Gemini CLI to search the theme for the old file name first, the way you would in any repository session. If you reject a result, storeseo_restore_image reverts compression, alt text or file name for that image, and the agent can re-list it afterwards to confirm the status changed. Prompts: - Use StoreSEO to list my product images with no alt text, grouped by product. Do not change anything. - Compress the five heaviest unoptimized product images, one call each, and ask me before the first call. - Search this theme for the file names of the images you are about to rename, then write alt text and file names for the Linen Shirt images. In Gemini CLI: Do not set trust on the storeseo server for this job. Trust skips every tool confirmation, and with one image per call that would let storeseo_optimize_image and storeseo_restore_image run without any prompt, leaving only the confirmation StoreSEO asks for in chat. Google's site also says Gemini CLI was replaced by Antigravity CLI on June 18th, 2026 for the unpaid tier and Google One users, so if that is you, check which command-line agent your account can still use before relying on this setup. Q: Should I trust the StoreSEO server in Gemini CLI to speed up image runs? A: No. Trust skips every tool confirmation, so storeseo_optimize_image would run on each image with no prompt from the CLI. Keep it off and let Gemini CLI ask, since its confirmation is the hard stop alongside the yes StoreSEO tells the agent to wait for. Q: Can I limit Gemini CLI to just the image tools? A: Yes. The includeTools list on the server controls which tools the model sees, so you can include storeseo_start, storeseo_list_images, storeseo_optimize_image and storeseo_restore_image and leave the rest out for an image-only session. --- ## Optimize images and alt text with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/image-seo (full page: https://storeseo.com/agent/github-copilot/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing Shopify images with GitHub Copilot means using Agent mode in VS Code to list your heaviest and unlabelled images through StoreSEO and then compress them and write alt text and file names one image per call, with the theme workspace open alongside. You open the theme workspace in VS Code, switch Copilot Chat to Agent mode and ask it to list product images with no alt text and the heaviest unoptimized files. Copilot calls storeseo_start and storeseo_list_images, which are reads, and the tool calls show in the chat. The listing gives each image's compression, alt text and file name status, and since the workspace is open Copilot can search the theme for any file name before you decide to rename it, which is useful because a rename changes the image's storefront URL. VS Code puts a confirmation in front of tool calls, and you choose how broadly to allow them. For this job keep storeseo_optimize_image and storeseo_restore_image on the confirmation prompt: it is the hard stop, and StoreSEO also tells the agent to name the Image Optimizer quota or AI credit a run will spend and wait for your yes. Each call handles one image and returns a report for compression, file name and alt text separately, so Copilot can say which part applied and which was refused rather than reporting one result for the batch. If a result is wrong, ask Copilot to restore it with storeseo_restore_image for the axes you name, then re-list the image to confirm. Because the server entry lives in .vscode/mcp.json or your user configuration, check MCP: List Servers if the image tools do not appear in the chat's tools list. Prompts: - In Agent mode, use StoreSEO to list my images with no alt text and the five heaviest unoptimized files. Change nothing. - Search this workspace for the file names of the Canvas Tote images, then compress them and write alt text, keeping the file names as they are. - Restore the last image you optimized, all three parts, and re-list it so I can see the status. In GitHub Copilot: VS Code reads servers from the servers key in .vscode/mcp.json, so a StoreSEO entry copied from a Cursor or Claude snippet that uses mcpServers will not load and the image tools never show up in Agent mode. With one image per call, a long run also means many confirmation prompts. Resist allowing the optimize and restore tools for the whole session to save clicks, since that prompt is the hard stop in front of compression quota, AI credit and URL-changing renames. Q: Why are the StoreSEO image tools missing in Copilot Agent mode? A: Check that the server entry uses the servers key in .vscode/mcp.json with type http, that MCP: List Servers shows storeseo running, and that the tools icon in Copilot Chat lists it. On Copilot Business and Enterprise an admin must also enable the MCP servers in Copilot policy. Q: Can Copilot check my theme before it renames an image? A: Yes. In Agent mode it can search the open workspace for the current file name, so you learn whether a Liquid file or JSON template hard-codes it before you approve a storeseo_optimize_image call that includes the file name axis. --- ## Optimize images and alt text with Windsurf URL: https://storeseo.com/agent/windsurf/image-seo (full page: https://storeseo.com/agent/windsurf/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your store's images with Windsurf means asking the Cascade agent in your editor to find heavy or unlabelled images through StoreSEO and then compress, caption and rename them one image at a time, with a restore for each change. You are in the Windsurf window you already use for the theme, with the Cascade panel open beside a Liquid file. You type something like find my ten heaviest unoptimized product images and tell me what compressing them will use. Cascade calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, and then storeseo_list_images, whose default sort puts the largest uncompressed files on top. Each row shows three axes, compression, alt text and file name, with an optimized count out of three, so the answer reads as a short worklist rather than a wall of data. The editor context is the useful part. A product image with no alt text often sits in a theme snippet that hard-codes an image URL or an empty alt attribute, and Cascade can open that file while StoreSEO shows which images lack alt text on the Shopify side. Before it changes anything, ask Cascade to state the image count and the kind of spend: compression uses the Image Optimizer add-on quota, while alt text and file names use AI Content Optimizer credit. Then storeseo_optimize_image runs once per image with the axes you chose, and the reply reports each axis as applied, already applied, skipped or refused. Cascade can reach all 44 StoreSEO tools within its 100-tool limit, so the image tools are available without trimming. If the store has no Image Optimizer add-on, compression is refused while alt text and file names still apply, and Cascade should say so instead of retrying. A file name change moves the image to a new storefront URL, which matters when a theme file hard-codes the old one; with the file open next to the chat, you can search for it right away. Prompts: - Use StoreSEO to list my ten heaviest unoptimized product images. Tell me the quota compression will use before you start. - Which product images have no alt text? Check the theme files in this workspace for hard-coded image URLs before we rename anything. - Optimize every image on the product Canvas Tote: compression, alt text and file name. Wait for my yes first. In Windsurf: Windsurf's MCP page describes the legacy Cascade agent, and the Devin Local agent that opens by default in new tabs reads different configuration files, so check which agent you are in if the image tools do not appear. The page also says nothing about how Cascade asks before running a tool, so treat StoreSEO's confirmation as an instruction the agent follows and read each optimize call before you accept it. Because storeseo_optimize_image handles one image per call, a bulk run is many calls in the panel. Q: Can Windsurf find the theme code that uses an image I renamed? A: Yes. A new file name gives the image a new storefront URL, and Cascade can search the workspace for the old file name in your theme files. StoreSEO cannot see your theme code, so that search is a separate step. Q: Does Windsurf compress images without the Image Optimizer add-on? A: No. StoreSEO refuses compression without the add-on, while alt text and file names still apply on the same call. Cascade should report the refused axis and carry on with the others. --- ## Optimize images and alt text with Grok URL: https://storeseo.com/agent/grok/image-seo (full page: https://storeseo.com/agent/grok/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images with Grok means chatting on grok.com with StoreSEO added as a custom connector, so Grok can find heavy or unlabelled product images and compress, caption and rename them one at a time without opening Shopify admin. You are in a Grok conversation on grok.com, with StoreSEO already added from the connectors page as a custom connector. There is no editor and no terminal, so everything happens in the chat. Ask which of your product images have no alt text, or which are the heaviest, and Grok opens a StoreSEO session with storeseo_start, then calls storeseo_list_images. The listing shows compression, alt text and file name for each image, and the default order puts the largest uncompressed files first, so the reply starts with the images that cost your pages the most. What you get back is a plain-language list: which images are done, which are partly done and which are untouched, with the product each one belongs to. Search by product title scopes it to one product, and Grok should check that every row really belongs to the product you meant, because a title search can also match file names from other resources. Before it starts optimizing, ask it to name the number of images and the kind of spend, since compression draws on the Image Optimizer add-on quota and alt text and file names draw on AI Content Optimizer credit. Then storeseo_optimize_image runs once per image with the axes you agreed to, and Grok reports applied, already applied, skipped or refused for each. If you dislike a result, ask Grok to restore that image and it calls storeseo_restore_image for the axis you name, bringing back the original file, the earlier alt text or the old file name. Renaming changes the image's storefront URL, which Grok should mention, and a collection or article image can take a couple of minutes because the rename re-uploads it. Prompts: - Use StoreSEO to show my ten heaviest unoptimized product images and tell me what compressing them will cost in quota. - Which product images have no alt text at all? Write alt text for the ones on my best seller after I say yes. - Put back the original alt text on the last image you changed. In Grok: xAI's connector documentation does not describe a per-tool approval setting, so on grok.com the main check on image changes is StoreSEO's instruction to name the cost and wait for your yes, and that works only if you read the plan before replying. On Grok Business and Enterprise, an admin has to provision the connector in the cloud console first. Image work is one call per image, so a store with hundreds of images means a long chat; scope each request to one product or one batch. Q: Can Grok write alt text for all my product images at once? A: It works through them one image per call, so a large catalog takes many calls in one conversation. Ask for one product or a batch of ten at a time, and expect it to name the AI credit use before it starts. Q: Can I tell Grok exactly what file name or alt text to use? A: No. StoreSEO writes both from the product's own context, and you can only steer the language. If you want specific wording, edit the alt text yourself in Shopify admin. --- ## Optimize images and alt text with OpenClaw URL: https://storeseo.com/agent/openclaw/image-seo (full page: https://storeseo.com/agent/openclaw/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your store's images with OpenClaw means messaging the agent runtime you host yourself from WhatsApp, Slack or its Control UI, and letting it find and fix heavy or unlabelled images through StoreSEO one image at a time. OpenClaw runs on your own machine or server, and you talk to it from a chat app on your phone. You send find the five heaviest product images that are not compressed yet, and the agent calls storeseo_start and then storeseo_list_images through the storeseo server you added with openclaw mcp add. It replies in the same thread with the product, the file size and which of the three axes, compression, alt text and file name, are still open. You can run openclaw mcp doctor with its probe option first to confirm the StoreSEO tools are listed. Changes are a conversation. Tell the agent which images and which axes you want, and ask it to state the count and the spend before it starts: compression uses Image Optimizer quota, alt text and file names use AI Content Optimizer credit. Then it calls storeseo_optimize_image once per image and reports each axis. If a result is wrong, a message such as restore the file name on that image makes it call storeseo_restore_image. Because a rename changes the image's storefront URL, ask the agent to list the new URLs in its reply. The scheduler suits the audit half of this job. An automation created with openclaw automations create can run on a cron expression in an isolated session and announce the result in a Telegram chat, for example a Monday count of images with no alt text and the five heaviest uncompressed files. That run should be report-only, because nobody is there to say yes to a write. When the report arrives and you want the fixes, reply in the chat, where you can answer the agent's question about scope and cost. Prompts: - Use StoreSEO to list my ten heaviest uncompressed product images. Read only, and tell me the quota compression would use. - Every Monday at 8am, tell me how many product images have no alt text and which five are heaviest. Do not optimize or restore anything. - Optimize the images on the product Market Basket: compression, alt text and file name. Ask me before you start. In OpenClaw: A scheduled OpenClaw run cannot wait for your yes, so an automation for this job should call only storeseo_list_images and report; leave storeseo_optimize_image and storeseo_restore_image to a chat you can answer. OpenClaw also has a per-server approval setting, set with openclaw mcp configure storeseo --approval, and a mode that skips the check would leave StoreSEO's own confirmation as the only stop before an image is rewritten. Run the mcp commands on the installation that owns the connector. Q: Can OpenClaw optimize my images on a schedule? A: It can run a scheduled check, but keep that run to reads. An automation has nobody to say yes, so have it report heavy and unlabelled images, then ask for the optimization in a chat message you can answer. Q: How do I check that OpenClaw can see the StoreSEO image tools? A: Run openclaw mcp doctor storeseo with the probe option. It opens a live connection and lists the tools, and you should see storeseo_list_images, storeseo_optimize_image and storeseo_restore_image in the list. --- ## Optimize images and alt text with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/image-seo (full page: https://storeseo.com/agent/hermes-agent/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images with Hermes Agent means asking Nous Research's agent, from a terminal or a chat app, to find heavy or unlabelled images through StoreSEO and optimize them one at a time, remembering your scope between sessions. You reach Hermes from a terminal or from Telegram, Discord or Slack, and the storeseo entry in ~/.hermes/config.yaml gives it the StoreSEO tools. Ask for the product images with no alt text, heaviest first, and Hermes calls storeseo_start and then storeseo_list_images. The listing carries a compression, alt text and file name block for each image plus how many of the three axes are done, so Hermes can group the answer by product and tell you which images are fully done and which are not. Hermes keeps memory across sessions, which suits a job that takes several sittings. Tell it you are working through the Linen Apron range this week, and in the next session you can say carry on with the next product without restating the scope. StoreSEO still opens its own session each time for the store snapshot. Before any run, ask Hermes to state the number of images and the spend: compression uses the Image Optimizer add-on quota, while alt text and file names use AI Content Optimizer credit, and asking for both text axes together is billed as one generation. storeseo_optimize_image takes one image per call with the axes you list, and Hermes reports applied, already applied, skipped or refused for each. If you dislike the wording or the new file name, ask it to restore that axis and it calls storeseo_restore_image. A rename changes the image's storefront URL, so ask Hermes to include the new address in its reply, and note that collection and article images take a couple of minutes because the rename re-uploads them. Prompts: - Use StoreSEO to list my product images with no alt text, heaviest first, and show the list before writing anything. - We are doing the Linen Apron range this week. Optimize its images one at a time and remember where we stopped. - Every Friday, count my product images with no alt text and report it to Telegram. Read only: do not optimize or restore anything. In Hermes Agent: After you edit config.yaml, start a new session or run /reload-mcp, or the open session will not see the storeseo server and its image tools. A hermes cron job for this work has nobody to say yes, so keep it to storeseo_list_images and a report, and put optimizing and restoring in a chat you can answer. If you want an agent that can only look, use the tools include list to expose just the StoreSEO list tools, since Hermes' own approval controls are not described in the page this record was checked against. Q: Does Hermes remember which images it already optimized? A: It remembers what you told it in earlier sessions, but the real state is in StoreSEO. Ask it to list the images again and it reads the three axes on each one, so a half-finished range is easy to pick up. Q: Can I limit Hermes to reading my images and never optimizing them? A: Yes. Set a tools include list for the storeseo server so it exposes only the start and list tools. Then Hermes can report on images but has no optimize or restore tool to call. --- ## Optimize images and alt text with OpenCode URL: https://storeseo.com/agent/opencode/image-seo (full page: https://storeseo.com/agent/opencode/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images with OpenCode means asking the open-source terminal agent to find heavy or unlabelled images through the StoreSEO MCP server and then compress, caption and rename them one image at a time. You run OpenCode in the theme repository or any folder, with the storeseo server added through opencode mcp add and signed in through the browser. You type find the ten heaviest unoptimized product images, and the model calls storeseo_start, then storeseo_list_images. The tools reach the model as storeseo_ followed by the tool name, and each call shows in the session. The rows carry the three image axes, compression, alt text and file name, so the reply is a ranked list you can scan before deciding what to spend. The terminal makes the follow-up work concrete. Ask OpenCode to write the list to a file, or to search the theme for the product's old image file name before you rename it, since a file name change moves the image to a new storefront URL. Ask it to state the count and the kind of spend before it begins: compression uses the Image Optimizer add-on quota, while alt text and file names use AI Content Optimizer credit. Then storeseo_optimize_image runs once per image with the axes you chose, and the model reports applied, already applied, skipped or refused for each. If a result is wrong, ask for a restore and the model calls storeseo_restore_image for the axis you name. Restoring compression brings back the original file, and you do it before re-compressing with different settings. OpenCode's permission rules can match tool names with a pattern, so you can hide the optimize and restore tools for a session where you only want to audit, then lift the rule when you are ready for changes. Prompts: - Use StoreSEO to list my ten heaviest unoptimized product images and tell me the quota compression would use. Do not change anything. - Find product images with no alt text, write the list to images-todo.md, then wait for my yes before optimizing any of them. - Restore the file name on the last image you renamed, and grep this repository for the old file name. In OpenCode: OpenCode's documentation describes deny rules for MCP tools but no ask-first mode, so a client-side stop on image changes means denying the storeseo optimize and restore tools by name; otherwise StoreSEO's confirmation in the chat is the main check. Also, opencode mcp add writes to the project config unless you pass --global, so an image session started in a different folder will not see the storeseo server. There is no remove command, so to take it out you delete the config entry. Q: Can I stop OpenCode from compressing images while I audit? A: Yes. Add a permission rule that denies the optimize and restore tools by name, then the model can still call storeseo_list_images but cannot change any image. Remove the rule when you want changes. Q: Why does OpenCode say it cannot see the image tools in another project? A: opencode mcp add saves the server to the current project unless you pass --global. Run it again with --global, or add the entry to the project you are in. --- ## Optimize images and alt text with Zed URL: https://storeseo.com/agent/zed/image-seo (full page: https://storeseo.com/agent/zed/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images in Zed means asking the Agent Panel to find heavy or unlabelled images through StoreSEO and optimize them one image at a time, with Zed's own tool prompt in front of every call. You have Zed open on your theme project with the Agent Panel beside the code, and storeseo listed under Settings, AI, MCP Servers with a green dot. You ask which product images have no alt text, heaviest first. The agent calls storeseo_start and then storeseo_list_images, and because Zed asks before running a tool action by default, each of those reads shows an approval prompt first. Once you accept, you get a list with the three axes for every image, compression, alt text and file name, plus how many are done. That default prompt shapes the job. A listing is only two or three calls, but an image run is one storeseo_optimize_image call per image, so a batch of ten means ten prompts in Zed on top of the confirmation StoreSEO tells the agent to ask for. You can cut the noise for audits by allowing the read tools with per-tool keys such as mcp:storeseo:storeseo_list_images, and leave storeseo_optimize_image and storeseo_restore_image on confirm. Ask the agent to state the number of images and the spend first: compression uses Image Optimizer quota, while alt text and file names use AI Content Optimizer credit. Because the panel sits in your editor, a rename has a visible follow-through. A new file name changes the image's storefront URL, and the agent can search the open theme for the old file name and show you each match. If a result is wrong, a request to restore the alt text, the file name or the compression calls storeseo_restore_image, and Zed asks you again before it runs. Prompts: - Use StoreSEO to list my ten heaviest unoptimized product images and tell me what compressing them will use. - Which product images have no alt text? List them, then search this theme for any that are hard-coded by file name. - Compress and rename the images on the product Canvas Tote, and tell me the new URLs before and after. In Zed: Zed's default is to ask before every tool action, reads included, so an image audit and a batch of single-image optimize calls each bring a prompt. Allow only the read tool with a key like mcp:storeseo:storeseo_list_images and keep storeseo_optimize_image and storeseo_restore_image on confirm. If you set agent.tool_permissions.default to allow, Zed stops asking and the confirmation StoreSEO tells the agent to ask for is the only check before an image is compressed or renamed. Q: Why does Zed ask me before every image call? A: agent.tool_permissions.default is confirm unless you change it, so the Agent Panel asks before each tool action. Allow the read tools by name to reduce prompts and keep the optimize and restore tools on confirm. Q: Can Zed show me where my theme uses a renamed image? A: Yes. A new file name gives the image a new storefront URL, and the Agent Panel can search the open project for the old file name. StoreSEO only reports the new address, so the search happens in your theme. --- ## Optimize images and alt text with Kiro URL: https://storeseo.com/agent/kiro/image-seo (full page: https://storeseo.com/agent/kiro/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images in Kiro means asking the agent in AWS's IDE to find heavy or unlabelled images through StoreSEO and optimize them one image at a time, with your autoApprove list deciding what runs without a prompt. You are in Kiro with your Shopify theme open and the storeseo server defined in .kiro/settings/mcp.json or in your user file. In the chat you ask for the ten heaviest product images that are not compressed. Kiro calls storeseo_start, then storeseo_list_images, and the answer lists each image with its compression, alt text and file name state and how many of the three axes are done. Since Kiro handles the sign-in in the browser, there is no token to paste unless you chose the header route. The autoApprove list is the lever for this job. If you list storeseo_list_images, the audit runs without a prompt, while storeseo_optimize_image and storeseo_restore_image stay outside the list, so Kiro prompts you for each image it is about to change. That is on top of the confirmation StoreSEO tells the agent to ask for, which should name the image count and the spend: compression uses the Image Optimizer add-on quota, and alt text and file names use AI Content Optimizer credit, with both text axes billed as one generation. Kiro's planning helps with the follow-through. After a rename, the new storefront URL is in the optimize report, and the agent can search the theme for the old file name and propose the edits as a change in the same project. If a result is wrong, ask for a restore of one axis and Kiro calls storeseo_restore_image. Collection and article images take a couple of minutes to rename because the image is re-uploaded, so expect a slower call there. Prompts: - Use StoreSEO to list my ten heaviest unoptimized product images and tell me what compressing them will use. - Which product images have no alt text? Show the list first, then wait for my yes before writing any. - Rename the images on the product Linen Apron, then search the theme for the old file names and list what would need updating. In Kiro: Putting storeseo_optimize_image in autoApprove, or a wildcard, makes Kiro run it without asking, and then the confirmation StoreSEO tells the agent to ask for is the only check before each image is compressed, captioned or renamed. List only the read tools there. A workspace .kiro/settings/mcp.json also overrides a user-level storeseo entry, so if the images belong to the wrong store, check which file defines the server, and Kiro's default approval behavior with an empty list is not described in the documentation checked. Q: Which StoreSEO image tools should I put in Kiro's autoApprove list? A: Only the read tools, storeseo_start and storeseo_list_images. Leave storeseo_optimize_image and storeseo_restore_image out, so Kiro prompts you before it changes or restores an image. Q: Can Kiro update my theme after an image is renamed? A: It can search the project for the old file name and propose edits as a change you review. The rename itself happens in Shopify through StoreSEO, and the new URL is in the optimize report. --- ## Optimize images and alt text with Antigravity URL: https://storeseo.com/agent/antigravity/image-seo (full page: https://storeseo.com/agent/antigravity/image-seo/index.md). Job: Image SEO and alt text, described once above. Optimizing your Shopify images in Antigravity means asking Google's agentic IDE to find heavy or unlabelled images through StoreSEO and optimize them one image at a time, with Ask mode prompting before each tool runs. You have Antigravity open on your theme with the agent side panel showing the storeseo server, which you added with the serverUrl field in mcp_config.json and signed in by pasting back an authorization code. You ask the agent for the ten heaviest product images that are not compressed. It calls storeseo_start, then storeseo_list_images, and the reply lists each image's compression, alt text and file name status and the count of axes done. Tools you have not configured run in Ask mode, so expect an approval prompt before each of those calls. That prompt matters most when you move from listing to changing. The agent should name the number of images and the spend before it starts: compression uses the Image Optimizer add-on quota, while alt text and file names use AI Content Optimizer credit. Each storeseo_optimize_image call handles one image with the axes you chose, so a batch is a series of prompts, and the report for each call says applied, already applied, skipped or refused. If you have no add-on, compression is refused while alt text and file names still apply, and the agent should say so. Because the agent also edits your theme, a rename can be followed through in the same window. The new file name gives the image a new storefront URL, which is in the report, and the agent can search the project for the old name and propose edits. If a result is wrong, ask for a restore of the axis and the agent calls storeseo_restore_image. For audit-only sessions, an mcp(storeseo/storeseo_list_images) policy pattern can allow the read while the optimize tool stays in Ask mode. Prompts: - Use StoreSEO to list my ten heaviest unoptimized product images and tell me the quota compression would use. - Which product images have no alt text? Show me the list before you write anything. - Optimize the images on the product Canvas Tote, then search this project for the old image file names. In Antigravity: If you add a policy that allows mcp(storeseo/*), Antigravity stops asking and the confirmation StoreSEO tells the agent to ask for is the only check before an image is compressed, captioned or renamed, so allow the listing tool by name instead. The server entry must use serverUrl, because Antigravity's documentation says url and httpUrl are not supported, and the sign-in ends with a code you paste into Agent settings, under Customizations. Q: Will Antigravity ask before it compresses an image? A: By default, MCP tools you have not configured run in Ask mode, so Antigravity asks before each call. If you allow a tool with an mcp policy pattern, that tool stops prompting, so allow only the read tool. Q: Why can Antigravity not find the StoreSEO image tools? A: Check that the entry in mcp_config.json uses serverUrl and not url, and that you finished the sign-in by pasting the authorization code in Agent settings. Then list the tools again. --- ## Get products ready for AI search with Claude URL: https://storeseo.com/agent/claude/ai-search-readiness (full page: https://storeseo.com/agent/claude/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Claude means asking Anthropic's chat assistant to read a product's GEO score in StoreSEO, explain which AI-readiness criteria fail, and preview the FAQs and AI snippet that would fix them before anything is saved. You are in a Claude chat tab on the web or in Claude Desktop, with the StoreSEO custom connector switched on for the conversation. You type something like: which of my products would an AI shopping agent struggle to understand, and fix the Merino Beanie first. Claude calls storeseo_start, which returns the conversation id every other tool needs and a GEO overview of your catalog, then storeseo_get_resource_seo for the product. The answer carries a geo block with the score and each AI-readiness criterion that passes or fails, and Claude reads those out by their visible titles instead of internal identifiers. From there the job is a short conversation. Claude tells you which failing criteria a tool can fix, namely FAQs, the AI snippet, image alt text and product schema, and which only you can change in the app or in Shopify admin, such as the product title, category or a missing barcode. It then calls storeseo_preview_resource_optimization, which generates FAQs and an AI snippet with a summary, benefits and best-for line, and projects the new SEO and GEO scores without saving. The preview spends AI Content Optimizer credit, so Claude should tell you that first. You read the draft in the chat, ask for changes by describing them, and say yes to the parts you want. Only then does Claude call storeseo_apply_resource_optimization. If product schema is off, Claude can also offer storeseo_update_schema_settings. That one is a shop-wide switch, so it lifts the schema criterion on every product at once, and Claude should say so before it asks. A thin description is a separate path: storeseo_scan_product_description returns questions that Claude puts to you in the chat, and storeseo_write_product_description writes from your answers. Because the whole exchange sits in one tab, the draft, your answers and the old description text that comes back from a rewrite all stay visible in the scroll. Prompts: - Which of my products have the lowest GEO score? Show what is missing for AI shopping agents, change nothing yet. - Make my Merino Beanie ready for AI search. Preview the FAQs and AI snippet first and tell me the credit cost. - Is product schema switched on in my store? If not, explain what turning it on changes across all products. In Claude: Claude asks before it runs a tool unless you chose Allow always for it. Anthropic advises Allow always only for servers you trust, and for this job it is safer to keep the apply, schema and write-description tools on ask, because a rewritten description is saved at once with no undo. On Team and Enterprise plans an owner adds the connector under Organization settings, Connectors, and on the Free plan StoreSEO uses the one custom connector slot you get. Check the Search and Tools menu in each new chat to confirm the connector is on. Q: Can Claude see my GEO score without changing my store? A: Yes. storeseo_start and storeseo_get_resource_seo are read-only, and the second one returns the GEO score with each AI-readiness criterion for a product. Nothing is saved until you approve an apply or a write. Q: Does a Claude chat show me the generated FAQs before they go live? A: Yes. The preview returns the FAQs and AI snippet as a draft that Claude prints in the chat, along with the projected scores. You can ask for changes or skip parts, and the draft expires after one hour if you do not apply it. --- ## Get products ready for AI search with ChatGPT URL: https://storeseo.com/agent/chatgpt/ai-search-readiness (full page: https://storeseo.com/agent/chatgpt/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with ChatGPT means connecting StoreSEO as a custom MCP app, then letting ChatGPT scan a product description for AI shopping agents, put StoreSEO's questions to you, and write a stronger description from your answers. You work in a ChatGPT conversation with the StoreSEO app selected, usually in developer mode, because that is where OpenAI lets a connected server do more than read. A request such as: scan the description of my Ceramic Pour Over for AI shopping agents, starts with storeseo_start and then storeseo_get_resource_seo for the product. That tells ChatGPT the GEO score and whether the product has a focus keyword, which this job needs and which only the StoreSEO app can set. Then it calls storeseo_scan_product_description, which spends a few AI credits and scores the description on seven pillars: identity, materials, dimensions, features, use cases, what is included and care. The scan returns questions for the facts StoreSEO cannot infer, and ChatGPT is supposed to put them to you word for word and in order, not answer them itself. This is where a chat window is a good fit: you type the real materials, the dimensions and what is in the box into the reply box, one message. ChatGPT then calls storeseo_write_product_description with your answers. StoreSEO writes the copy itself, so you cannot paste in a finished description for it to save. The write replaces the live description in Shopify immediately, so ChatGPT should ask first, and the response includes the previous description once, which you should ask ChatGPT to paste back into the chat so you keep a copy. FAQs and the AI snippet are the other half of the checklist. ChatGPT previews them with storeseo_preview_resource_optimization, shows the draft and the projected GEO score, and applies it after your yes. ChatGPT also shows each tool payload for you to read before you approve it, and that payload is the last chance to see exactly what is about to be sent to your store. Prompts: - Scan the description of my best-selling product for AI shopping agents and ask me every question StoreSEO returns, word for word. - Rewrite the description of Ceramic Pour Over from my answers. Paste the old description back to me before and after you write. - Which products have a GEO score under 50, and which of those are missing a focus keyword? In ChatGPT: ChatGPT connects from OpenAI's servers, and which plan can use write actions is not the same everywhere. OpenAI's developer guide lists developer mode for several paid plans, while its help center, as we read it, gives Pro accounts read access only and puts write actions in beta on Business, Enterprise and Edu workspaces. If writes are not available to you, use ChatGPT to scan and report and finish the rewrite in the StoreSEO app. Because a description write has no undo, read the payload before approving and avoid the option that remembers approval for the rest of the conversation. Q: Can ChatGPT rewrite a Shopify product description through StoreSEO? A: Yes, if your ChatGPT plan allows write actions for connected apps. It scans first, asks you StoreSEO's questions, then calls the write tool with your answers. StoreSEO writes the copy itself and saves it to Shopify immediately. Q: Why will StoreSEO not accept a description I wrote myself? A: The write tool takes your answers to the scan questions, not finished text. Facts you want in the copy belong in those answers, and StoreSEO writes the description around them and re-checks it. To keep your own wording, edit the description in Shopify admin. --- ## Get products ready for AI search with Claude Code URL: https://storeseo.com/agent/claude-code/ai-search-readiness (full page: https://storeseo.com/agent/claude-code/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Claude Code means asking Anthropic's terminal agent to read a product's GEO score in StoreSEO, check whether product JSON-LD is on, and apply the FAQs and AI snippet you approve without leaving your repository. You run Claude Code from the folder you already work in, often the theme repository, and type a request in plain words: check why the Merino Beanie scores low for AI search and fix what you can. It calls storeseo_start, then storeseo_get_resource_seo, and each call scrolls past in the terminal with its name. The geo block in the result lists the AI-readiness criteria. A typical gap is that product schema is off, which is a shop-wide setting rather than a product setting, so Claude Code should tell you that turning it on credits the schema criterion on every product before it calls storeseo_update_schema_settings. Developers get one thing here that chat users do not: the agent can see your code. If the theme already prints its own Product JSON-LD in a snippet, ask Claude Code to compare that markup with what StoreSEO would publish before you enable the schema, so you do not ship two competing Product blocks. That check is about files in your folder, and StoreSEO plays no part in it. After the schema decision, Claude Code calls storeseo_preview_resource_optimization, which generates FAQs and an AI snippet and projects the new SEO and GEO scores without saving. It spends AI credit, and the draft expires after an hour. You approve in the conversation, and Claude Code then calls storeseo_apply_resource_optimization. The apply returns the re-analysed checklist, so Claude Code can report which criteria now pass and which still need you, with one link to the product's fix page rather than one per item. If you stop before applying and the agent turned schema on only for this flow, it should switch that schema back off. Longer runs can be written to a Markdown file in the repository so the worklist survives the session. Prompts: - Use StoreSEO to read the GEO checklist for the Merino Beanie and list what is missing. Do not change anything. - Check whether product schema is enabled in StoreSEO, compare it with the JSON-LD in snippets/ in this theme, and tell me if they would clash. - Preview FAQs and an AI snippet for my three lowest GEO products, show the projected scores, and wait for my yes before applying. In Claude Code: Claude Code asks before a tool runs unless you added an allow rule. If you allow the StoreSEO tools to stop the prompts, that rule also covers storeseo_update_schema_settings, the apply tool and the description write, and the confirmation StoreSEO asks for in the chat becomes the only check, so keep those on ask. Without --scope user the server exists only in the project where you added it. Claude Code also warns when one tool result passes 10,000 tokens, so ask for a few products at a time, not the whole catalog. Q: Can Claude Code compare my theme's JSON-LD with StoreSEO's product schema? A: Claude Code can read the Liquid and JSON-LD files in your folder, and StoreSEO tells it whether product schema is enabled through its settings tools. Comparing the two is the agent's own work, so it is worth asking for before you switch the shop-wide schema on. Q: What happens if I stop after Claude Code enabled product schema? A: StoreSEO's workflow tells the agent to switch off only the schema it turned on for this flow if you decide not to apply the rest, and to leave a schema you already had on. Check the result with the schema settings tool if you are unsure. --- ## Get products ready for AI search with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/ai-search-readiness (full page: https://storeseo.com/agent/claude-cowork/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Claude Cowork means using Anthropic's desktop agent to track StoreSEO GEO scores on a schedule, then working through fixes with you in a live session, because a scheduled task has nobody there to say yes. Cowork runs on your desktop and can use the apps connected to it, so StoreSEO sits beside your files and other connectors. There are two different ways to use it for this job. In a live session you ask: which of my products are weakest for AI shopping agents, and prepare fixes for the worst three. Cowork calls storeseo_start, which returns a GEO overview, then storeseo_get_resource_seo on each candidate and reads out the failing criteria. It previews FAQs and an AI snippet with storeseo_preview_resource_optimization and waits for your yes before storeseo_apply_resource_optimization. The second way is a recurring task, and it needs different wording. A task scheduled in Cowork runs on its cadence without you in the conversation, so it cannot wait for an approval. For this job that points to a weekly report: read the GEO overview, list the lowest-scoring products and the criteria that fail, and write you a summary or a file. The prompt should say in plain words not to apply, generate, preview, publish, write descriptions or change schema. Previews spend AI credit even though they save nothing, so a report-only task should skip them too. Put the two together. Monday's task reports which products fell behind or never improved. You open Cowork when you have time, paste the list in and say: work through these with me. That session is where product schema, FAQs, the AI snippet and any description rewrite happen, with the agent asking and you answering. Cowork can keep the weekly summaries as files on your computer, which gives you a trail of how scores moved between runs. Prompts: - Every Monday, use StoreSEO to read my GEO overview and my five lowest GEO products with their failing criteria. Summarize only. Do not preview, apply, write or change anything. - Open a session: take the three weakest AI-readiness products from my last report and prepare FAQs and AI snippets for them. Preview and wait for me. - Compare this week's GEO report with last week's file and tell me which products improved. In Claude Cowork: A scheduled task runs without you in the conversation, so nobody is there to say yes. Keep any scheduled StoreSEO task for this job to reads and reports, and tell it in plain words not to preview, apply, write descriptions or change schema settings. Cowork also follows the custom connector rules of Claude: the Free plan allows one custom connector, and on Team and Enterprise an owner adds the connector under Organization settings, Connectors. If StoreSEO is missing in Cowork, check Customize, Connectors inside Cowork itself. Q: Can Claude Cowork improve my AI-readiness scores on a schedule? A: It can report on them on a schedule, which is the safe use. A scheduled task has nobody present to approve changes, so keep it to reading GEO scores and writing a summary, and make the fixes in a live Cowork session where you can say yes. Q: Why should a scheduled Cowork task skip previews? A: A preview writes nothing to your store but it spends AI Content Optimizer credit, and each re-run spends again. A weekly task that previews every low product could use your credit with no one reading the drafts, so read first and preview when you are in the session. --- ## Get products ready for AI search with Cursor URL: https://storeseo.com/agent/cursor/ai-search-readiness (full page: https://storeseo.com/agent/cursor/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Cursor means using the agent chat inside your theme project to read StoreSEO's GEO score for a product, preview FAQs and an AI snippet, and apply them once you approve, with the right tools left switched on. You sit in Cursor's editor with the theme project open and the agent chat beside your files. You type: the Merino Beanie looks thin to AI shopping agents, check its GEO score and fix what StoreSEO can. The agent calls storeseo_start, then storeseo_get_resource_seo, and Cursor shows each tool call in the chat with its arguments so you can read them. The result's geo block lists which AI-readiness criteria pass and which fail, and the agent splits them into things a tool can fix, such as FAQs, AI snippet, alt text and product schema, and things only you can change, such as the product title or category. The first thing that can go wrong in Cursor is not StoreSEO at all. Cursor allows about 40 active tools across all your servers and StoreSEO has 44, so you have already switched some off in the MCP settings. If the agent says it has no tool to preview or apply, open the settings and check that storeseo_preview_resource_optimization and storeseo_apply_resource_optimization are on. This job also uses storeseo_get_resource_seo, storeseo_update_schema_settings, storeseo_scan_product_description and storeseo_write_product_description, so keep those six plus storeseo_start enabled and switch off tools for jobs you are not doing today. Because the agent works in your project, you can ask it to cross-check the theme's own JSON-LD before enabling StoreSEO's product schema, and to keep the preview drafts in a scratch file while you decide. The preview generates the FAQs and snippet, projects the new SEO and GEO scores, and spends AI credit without saving. After your yes in the chat, the agent applies it and reports the re-analysed checklist, so you see which criteria now pass. Prompts: - Use StoreSEO to check the GEO score of the Merino Beanie and list what an AI shopping agent cannot find. Read only. - Preview FAQs and an AI snippet for the Merino Beanie, show the projected GEO score, and wait for me before applying. - Which StoreSEO tools do you have right now? List the ones for AI search readiness and tell me if any are missing. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you must turn some off in Cursor's MCP settings. For this job, make sure storeseo_get_resource_seo, storeseo_preview_resource_optimization and storeseo_apply_resource_optimization stay on, and if you want schema or a description rewrite, storeseo_update_schema_settings and the scan and write tools as well. If you set a StoreSEO tool to run without asking, the confirmation StoreSEO asks for in chat is the only check left, so keep the apply and write tools on ask. Q: Why can Cursor not find the StoreSEO tool for AI search readiness? A: Cursor allows about 40 active tools across all servers and StoreSEO has 44, so one you need may be switched off. Open Cursor's MCP settings, find the storeseo server and turn on the preview, apply and get SEO tools, switching off tools for other jobs. Q: Does a .cursor/mcp.json entry make StoreSEO available in every project? A: No. A server in .cursor/mcp.json belongs to that project only. Put it in ~/.cursor/mcp.json if you want StoreSEO's AI-readiness tools in every project you open. --- ## Get products ready for AI search with Codex URL: https://storeseo.com/agent/codex/ai-search-readiness (full page: https://storeseo.com/agent/codex/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Codex means running OpenAI's terminal agent against StoreSEO to read a product's GEO checklist, preview FAQs and an AI snippet, and apply only the parts you approve while Codex's own tool approval stays on. You start Codex in a terminal, normally in the project that holds your theme, with the storeseo server registered in config.toml or through codex mcp add. A request such as: read the GEO checklist for Ceramic Pour Over and tell me what is missing for AI shopping agents, makes Codex call storeseo_start and then storeseo_get_resource_seo. The geo block lists every AI-readiness criterion that passes or fails. Codex prints what it is about to call, and with the server's approval mode set to writes it runs reads straight away and stops to ask before a tool that changes something. The preview is where you shape the result. storeseo_preview_resource_optimization generates FAQs and an AI snippet and projects the new SEO and GEO scores without saving, and it spends AI credit, so Codex should name that before it runs one. Generation can take a while. Codex waits 60 seconds for a tool by default, so if a preview times out on a long product, raise tool_timeout_sec for the storeseo server and run it again; re-previewing spends credit again, which is a good reason to set the timeout before the first try. Applying is selective. storeseo_apply_resource_optimization accepts a subset, so you can approve the FAQs and AI snippet and leave out, say, a rewritten URL handle, and the skipped fields stay applicable from the same draft for one hour. When the apply returns, Codex has the re-analysed checklist and can report what now passes. For a thin description Codex uses storeseo_scan_product_description, relays the questions to you in the terminal, and calls storeseo_write_product_description with your answers after a yes. Prompts: - Using the storeseo server, read the GEO checklist for Ceramic Pour Over and list failing criteria. Do not call anything that writes. - Preview FAQs and an AI snippet for Ceramic Pour Over and show me the projected GEO score. Apply nothing until I answer. - Apply only the FAQs from the last preview and skip everything else. Then re-read the GEO checklist. In Codex: Set the storeseo server's default_tools_approval_mode to writes or prompt, not approve. Approve skips Codex's prompt, and then the confirmation StoreSEO asks for in the chat is the only check, which matters here because a description write has no undo. Codex waits 60 seconds for a tool by default and generation can take longer, so raise tool_timeout_sec for this server before you preview a long product, since each repeat of a preview spends AI credit again. A project-level .codex/config.toml only loads in trusted projects. Q: Can Codex apply only some of the generated AI-readiness content? A: Yes. The apply tool accepts a subset, so Codex can save the FAQs and skip the snippet or a URL handle. Skipped content stays applicable from the same draft until it expires one hour after the preview. Q: Why did my Codex preview time out for a product? A: Codex waits 60 seconds for a tool by default and generation can take longer. Raise tool_timeout_sec for the storeseo server in config.toml, then run the preview again. A second preview spends AI credit again. --- ## Get products ready for AI search with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/ai-search-readiness (full page: https://storeseo.com/agent/gemini-cli/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Gemini CLI means using Google's terminal agent to scan a Shopify product's description through StoreSEO, answer the questions it returns at the prompt, and have StoreSEO write an agent-ready description from your facts. You run gemini in a terminal and the storeseo server is already added with gemini mcp add, either for the project or for your user. You ask: scan the description of my Ceramic Pour Over for AI shopping agents. Gemini CLI calls storeseo_start, then storeseo_get_resource_seo, which returns the GEO block and shows whether the product has a focus keyword. If it does not, the scan will not run and the agent should tell you to set one in the StoreSEO app, because no tool here can set it. Then storeseo_scan_product_description scores seven pillars and returns the questions. A terminal is a plain place to answer questions. Gemini CLI should print each question as StoreSEO wrote it, in order, and wait. Where a field comes with options, such as a Shopify category attribute, pick one or give your own value; where it carries a suggested value, it comes from what you saved in Shopify or from the current description, so confirm or change it. You type the answers in one reply and the agent passes them to storeseo_write_product_description. Gemini CLI shows a confirmation for the tool call, which is separate from the yes you give in the conversation, and the second one is the hard stop. The write replaces the live description in Shopify immediately, with no undo and no revision history, and the response returns the old text once. Ask Gemini CLI to print it or save it to a file in your working folder straight after the write. If you only want the FAQs and AI snippet, use storeseo_preview_resource_optimization and storeseo_apply_resource_optimization instead, which show a draft and projected GEO score first. Prompts: - Scan the description of Ceramic Pour Over for AI shopping agents. Print each StoreSEO question exactly as returned and wait for my answers. - After you write the new description, save the previous description to old-description.html in this folder. - Which of my products lack a focus keyword, so that a description scan would not run? In Gemini CLI: Do not set --trust or trust: true on the storeseo server. Trust skips every tool confirmation in Gemini CLI, and then the confirmation StoreSEO asks for in the chat is the only check, which is a poor trade for a tool that saves a description with no undo. The browser sign-in needs a browser on the same machine, so over SSH or in a headless session use the access token header. gemini mcp add defaults to project scope, and Google's site says Antigravity CLI replaced Gemini CLI for some accounts, so check which agent your account can use. Q: Does Gemini CLI answer StoreSEO's description questions for me? A: It should not. StoreSEO tells the agent to put each question to you word for word and in order, and your answers become stated facts in the new description. If you decline to answer, the agent can skip the questions at your request. Q: How do I keep the old product description after a Gemini CLI rewrite? A: StoreSEO returns the previous description once in the write response and keeps no backup. Ask Gemini CLI to save it to a file in your working folder before you move on, since it is the only remaining copy. --- ## Get products ready for AI search with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/ai-search-readiness (full page: https://storeseo.com/agent/github-copilot/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with GitHub Copilot means using agent mode in VS Code to read a product's StoreSEO GEO score, decide on product schema, and apply the FAQs and AI snippet you approve, with the editor confirming each tool call. You open the Copilot Chat view in VS Code, switch to agent mode, and the storeseo server from .vscode/mcp.json is listed among its tools. You type: why is the Merino Beanie weak for AI search, and what can StoreSEO fix. Copilot calls storeseo_start and storeseo_get_resource_seo, and the geo block in the result tells it which AI-readiness criteria fail. Because you are already in the theme repository, you can also ask Copilot to look at how the theme prints Product JSON-LD, so you decide with both views whether to enable StoreSEO's product schema. VS Code puts a confirmation card in front of each tool call that is not a plain read, showing the tool name and its arguments, and that card is separate from the yes you give in chat. For this job the cards you will see are storeseo_update_schema_settings, storeseo_apply_resource_optimization, and, for a thin description, storeseo_write_product_description. Read the arguments on each. A schema change is shop-wide, so it affects every product, and a description write replaces the live text with no undo. The preview, storeseo_preview_resource_optimization, writes nothing but spends AI credit. The usual order is read, decide on schema, preview, apply. The apply returns the re-analysed checklist, so Copilot can list what now passes and what still needs you, for example a product category or barcode that only Shopify admin or the app can change. If the server was added through a workspace file committed to the repository, remember teammates share that file, so keep tokens out of it and let each person sign in with OAuth. Prompts: - @workspace Use StoreSEO to read the GEO checklist for the Merino Beanie and list what an AI shopping agent cannot find. Do not change anything. - Check whether product schema is enabled in StoreSEO and compare it with the JSON-LD in this theme before I turn it on. - Preview FAQs and an AI snippet for the Merino Beanie, show the projected GEO score, and wait for my confirmation before applying. In GitHub Copilot: VS Code uses the servers key in .vscode/mcp.json, while mcpServers belongs to the portable .mcp.json at the project root, so a Cursor or Claude snippet pasted unchanged will not work. On Copilot Business and Enterprise an organization admin must enable MCP servers in Copilot policy. Keep the apply, schema and description-write tools on VS Code's confirmation prompt, because StoreSEO asks for your yes in the chat and the editor's confirmation is the hard stop. A .vscode/mcp.json committed to the repository is shared with your team. Q: Does Copilot agent mode confirm StoreSEO writes in VS Code? A: VS Code shows a confirmation for tool calls that go beyond reads, with the tool name and arguments, unless you allowed that tool. StoreSEO also tells the agent to wait for your yes in chat, and the editor prompt is the hard stop. Q: Can Copilot check my theme before turning on StoreSEO product schema? A: Yes. Copilot can read your theme files and StoreSEO reports whether product schema is enabled, so you can compare them before the shop-wide switch. Spotting a duplicate Product block is the agent's own work, not a StoreSEO check. --- ## Get products ready for AI search with Windsurf URL: https://storeseo.com/agent/windsurf/ai-search-readiness (full page: https://storeseo.com/agent/windsurf/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Windsurf means asking the Cascade agent to read a product's GEO score in StoreSEO, fix the criteria a tool can fix, and rewrite a thin description from your own answers, all from the editor where your theme is already open. You sit in the Cascade panel with your Shopify theme repository open, and you type something like make Merino Beanie ready for AI shopping agents, preview first. Cascade calls storeseo_start, which returns the conversation id every other tool needs along with a GEO overview of your products, then storeseo_get_resource_seo for that product. The product's geo block lists the score and each AI readiness criterion that passes or fails: FAQs, the AI snippet, image alt text, product schema and the description itself. Those first calls are reads and show up in the panel as plain tool calls. From there Cascade works in the order the job needs. If product schema is off, it can read the setting with storeseo_get_schema_settings and ask before turning it on with storeseo_update_schema_settings, which is a store-wide switch. It then calls storeseo_preview_resource_optimization to generate FAQs and an AI snippet and to project the new SEO and GEO scores without saving anything; that spends AI credit, and StoreSEO tells the agent to say so first. Because you are in an editor, you can ask Cascade to paste the previewed FAQs next to the product template you are editing, or compare the projected checklist with a fix list kept in the repository, before you say apply. For a thin description, Cascade runs storeseo_scan_product_description and puts StoreSEO's questions to you in the chat, word for word and in order: materials, dimensions, what is in the box and so on. You type your answers right there, and Cascade passes them to storeseo_write_product_description. Criteria only you can fix, such as the product title, price or the focus keyword, come back as a single link to the product's fix page rather than a link per criterion. Prompts: - Use StoreSEO to find my five products with the lowest GEO score and list what each is missing for AI shopping agents. Read only. - Make the product Merino Beanie ready for AI search: check its GEO criteria, then preview FAQs and an AI snippet and tell me the credit cost first. - Scan the description of Ceramic Pour Over for AI agents, ask me StoreSEO's questions one by one, and keep the old description in a file called old-description.html here. In Windsurf: Windsurf's MCP documentation does not describe how Cascade approves tool calls, so you cannot count on the editor to stop the description write. StoreSEO's instruction to ask first is something the agent follows, and storeseo_write_product_description saves to Shopify at once with no undo and no revision history. Before you say yes, ask Cascade to save the previous description from the write response into a file in your repository. Also check which agent you are in: the MCP page applies to the legacy Cascade agent, while the Devin Local agent reads its own configuration files. Q: Can Windsurf rewrite a product description without me answering questions? A: Only if you decline to answer or authorize a bulk run without them. By default Cascade scans the description, puts StoreSEO's questions to you in the chat, and passes your answers to the write step so the new copy states your facts instead of guesses. Q: Does the product need anything before Cascade can scan its description? A: Yes. The product needs a focus keyword, which you set in the StoreSEO app, because no tool can set one. The scan and the rewrite both spend AI credit, and the product's fix page link covers anything only you can change. --- ## Get products ready for AI search with Grok URL: https://storeseo.com/agent/grok/ai-search-readiness (full page: https://storeseo.com/agent/grok/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Grok means chatting on grok.com with the StoreSEO connector switched on, so Grok can read each product's GEO score, preview the fixes StoreSEO can generate, and rewrite a thin description once you have answered its questions. You open a conversation on grok.com with the StoreSEO custom connector available and ask which products AI shopping agents will struggle with. Grok calls storeseo_start, then storeseo_list_resources sorted by score so the weakest products come first, and storeseo_get_resource_seo for the ones that matter. Each product carries a GEO score and a list of failing criteria, and Grok turns that into a short answer in the chat: this one has no FAQs, that one has no AI snippet, a third has a two-line description. All of that is read-only. Fixing is a conversation in the same thread. Grok can call storeseo_preview_resource_optimization to draft FAQs and an AI snippet for one product, which spends AI credit but saves nothing, and it shows you the draft and the projected score before anything is applied. If you say yes it calls storeseo_apply_resource_optimization with only the parts you accepted. When the description is the problem, Grok calls storeseo_scan_product_description and asks you StoreSEO's questions exactly as written, one after another, so you answer in plain sentences in the chat: what it is made of, how big it is, what comes in the box. The last step is the one to slow down on. storeseo_write_product_description replaces the live description at once, so Grok should say that, name the credit cost and wait for your yes. Because grok.com is a chat tab, you are reading proposals in a conversation rather than approving each call in a dialog, which suits a few products at a time better than a whole catalog. Prompts: - Use StoreSEO to find my lowest GEO-scoring products and tell me what each is missing for AI shopping agents. - Preview FAQs and an AI snippet for my best seller. Tell me what it will cost in AI credit before you run the preview. - Scan the description of Ceramic Pour Over for AI agents and ask me StoreSEO's questions one at a time. Do not write anything until I confirm. In Grok: xAI's connector documentation does not describe a per-tool approval setting for Grok, so the pause before a write comes from StoreSEO's instruction to the agent and from you reading what Grok proposes. That matters most for storeseo_write_product_description, which saves the new description at once with no undo. Ask Grok to paste the previous description from the write response into the chat before you move on, so you have a copy. Only the chat connector on grok.com is covered here, and on Grok Business and Enterprise a team admin has to provision it first. Q: Can I answer StoreSEO's description questions inside a Grok chat? A: Yes. Grok relays each question from the scan word for word and in order, and you reply in the same conversation. Your answers are passed to the write step so the description states your facts. Q: Does Grok ask before it applies FAQs or rewrites a description? A: StoreSEO tells Grok to preview, name the cost and wait for your yes. xAI's connector documentation does not describe a client-side approval setting, so read each proposal before you agree. --- ## Get products ready for AI search with OpenClaw URL: https://storeseo.com/agent/openclaw/ai-search-readiness (full page: https://storeseo.com/agent/openclaw/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with OpenClaw means messaging your own agent runtime from a chat app to fix GEO gaps in StoreSEO, and scheduling a report-only automation that tells you which products still fail AI readiness. OpenClaw runs on a machine you control, and you reach it through a chat app or the CLI. You send a message such as which products are weakest for AI shopping agents, and OpenClaw calls storeseo_start, storeseo_list_resources sorted by score and storeseo_get_resource_seo for the candidates. The reply in your chat app names the lowest GEO scores and the criteria each one fails: FAQs, AI snippet, alt text, product schema or a thin description. Those are reads and run straight away. Fixes happen in a message thread you can answer. You ask for a preview on one product, OpenClaw calls storeseo_preview_resource_optimization after saying it spends AI credit, and it sends back the drafted FAQs, the snippet and the projected score. You reply yes and it calls storeseo_apply_resource_optimization. For a description, OpenClaw calls storeseo_scan_product_description and sends you StoreSEO's questions in order; you answer in the chat, and the rewrite only runs after you confirm that it replaces the current text with no undo. OpenClaw has its own per-server approval setting, set with openclaw mcp configure storeseo --approval, which is the hard stop if you want the client to prompt before writes. The scheduler is where OpenClaw differs. An automation created with openclaw automations create can run every Monday in an isolated session and announce the result to your chat. For this job that run should be report-only: list the products with the lowest GEO score and the failing criteria, and nothing else. Nobody is there to answer the scan's questions or to say yes to a write, so you act on the report afterwards from a message you can reply to. Prompts: - Use StoreSEO to list my ten products with the lowest GEO score and the top failing criterion on each. Read only. - Preview FAQs and an AI snippet for Merino Beanie, tell me the credit cost first, and wait for my yes before applying anything. - Create a Monday automation that reports my five lowest GEO-scoring products and what each is missing. Reads only: no previews, scans or writes. In OpenClaw: A scheduled OpenClaw run has nobody to answer the questions storeseo_scan_product_description returns and nobody to say yes to storeseo_write_product_description, which replaces the live description with no undo. Keep the automation prompt to reads and say so in the prompt itself, as in the Monday report. Do the preview, the scan and the rewrite in a chat you can answer, and check that openclaw mcp configure storeseo --approval is not set to a mode that skips the client's check on writes. Q: Can an OpenClaw automation rewrite product descriptions on a schedule? A: It should not. The description scan returns questions for you to answer, and the rewrite replaces the live text with no undo. A scheduled automation can report GEO scores and failing criteria, and the rewrite belongs in a message you can answer. Q: How do I get the report into my chat app? A: Create the automation with the announce, channel and to options of openclaw automations create, and write the prompt as a read-only request. OpenClaw runs it in an isolated session and sends the result there. --- ## Get products ready for AI search with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/ai-search-readiness (full page: https://storeseo.com/agent/hermes-agent/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Hermes Agent means asking it, from a terminal or a chat app, to read your products' GEO scores in StoreSEO and fix the gaps you approve, with a hermes cron job for a weekly read-only report. You start a Hermes session in a terminal or message it from a chat app and ask which products are least ready for AI shopping agents. Hermes calls storeseo_start, then storeseo_list_resources sorted by score and storeseo_get_resource_seo for the weakest products. The GEO block on each product gives a score and the failing criteria, so the answer is concrete: this product has no FAQs, that one has no AI snippet, a third has alt text missing on two images and a description of a single sentence. Every one of those calls is a read. Hermes keeps memory across sessions, so you can come back to the same worklist the next day. When you pick a product, it calls storeseo_preview_resource_optimization to draft FAQs and an AI snippet and project the new GEO score, after saying that the preview spends AI credit. You review the draft, say yes, and it calls storeseo_apply_resource_optimization. For a thin description it runs storeseo_scan_product_description and puts StoreSEO's questions to you one by one; you answer, and only after you confirm that the new text replaces the live one does it call storeseo_write_product_description. The built-in scheduler suits the reporting half of this job. A hermes cron job with a prompt like list my five lowest GEO-scoring products and what each is missing can run every Monday and deliver the result to Telegram or another chat target. It must stay report-only, because a scheduled run has no one to answer the scan's questions or approve a write. Act on the report in a live session afterwards. Prompts: - Use StoreSEO to list my products with a GEO score under 50 and what each is missing. Read only, change nothing. - Make Merino Beanie ready for AI search: preview FAQs and an AI snippet, show me the projected score, and wait for my yes. - Scan the description of Ceramic Pour Over for AI agents and ask me every question StoreSEO returns, word for word. In Hermes Agent: A hermes cron job cannot answer the description scan's questions or say yes to a write, so keep scheduled prompts to reads and state that in the prompt. If you want to enforce it, use Hermes' tools include list to expose only the StoreSEO list and get tools to that job. Edits to config.yaml need a new session or /reload-mcp, and a token pasted into the file is a secret to keep out of version control. Q: Can hermes cron fix my GEO scores automatically? A: It can report them. A scheduled job has no one to answer the scan's questions or approve a write, so ask it for a read-only list of low GEO scores and apply fixes in a live session. Q: Does Hermes remember my answers to the description questions? A: Hermes keeps memory across sessions, but StoreSEO's scan is meant to put its questions to you each time, and the product must have a focus keyword first. Treat the questions as yours to answer. --- ## Get products ready for AI search with OpenCode URL: https://storeseo.com/agent/opencode/ai-search-readiness (full page: https://storeseo.com/agent/opencode/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with OpenCode means working in the terminal session you already have open, where the agent reads a product's GEO score in StoreSEO, previews the fixes, and rewrites a thin description once you have answered its questions. You run OpenCode in your theme project and ask for what you want in plain words: which products are weakest for AI shopping agents, and what would fix the worst one. OpenCode calls storeseo_start first, then storeseo_list_resources by score and storeseo_get_resource_seo for the product you name. The GEO block comes back with the score and every criterion that passes or fails, and OpenCode summarizes it in the terminal. These are reads, and OpenCode lists them as tool calls as they run. To fix a product, OpenCode checks storeseo_get_schema_settings and, if product schema is off, tells you that enabling it is a store-wide switch before it calls storeseo_update_schema_settings. It then calls storeseo_preview_resource_optimization for FAQs and an AI snippet, which spends AI credit and saves nothing, and shows the projected score. After your yes, storeseo_apply_resource_optimization applies what you accepted. If you decide not to apply the rest, the schema it turned on for this flow should be turned back off. Descriptions go through storeseo_scan_product_description, with StoreSEO's questions asked in order in the terminal and your answers typed back. Because OpenCode is a terminal tool you can keep the evidence in the repository: ask it to write the before and after GEO checklists, or the previous description returned by the write, to a Markdown file next to your theme. Tool names appear with the server name in front, so a deny rule for the storeseo server can hide a tool you do not want the agent to reach. Prompts: - Use StoreSEO to list my five lowest GEO-scoring products and what each is missing. Read only. - Make Merino Beanie ready for AI search. Check product schema first, then preview FAQs and an AI snippet and wait for my yes. - Scan the description of Ceramic Pour Over for AI agents, ask me the questions, and save the old description to old-description.html in this folder before writing. In OpenCode: OpenCode's documentation describes deny rules for MCP tools but no ask-before-running mode, so the pause before storeseo_write_product_description comes from StoreSEO's instruction to the agent and from you reading the terminal. That tool saves at once with no undo. If you want a client-side stop, deny the write tools by name until you are ready to use them, and remember that opencode mcp add only writes to the current project unless you pass --global. Q: Can I stop OpenCode from rewriting descriptions at all? A: Yes. OpenCode documents deny rules for MCP tools, so you can hide storeseo_write_product_description and run only previews and reads until you decide otherwise. Tool names carry the server name in front. Q: What if I enabled product schema and then change my mind? A: Tell OpenCode to turn it back off. Schema is a shop-wide setting, so the job's rule is that the agent only switches off the schema it enabled for this flow, never one you already had on. --- ## Get products ready for AI search with Zed URL: https://storeseo.com/agent/zed/ai-search-readiness (full page: https://storeseo.com/agent/zed/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Zed means asking the Agent Panel to read a product's GEO score in StoreSEO, preview FAQs and an AI snippet, and rewrite a thin description from your answers, with Zed's own prompts as a second check on every tool call. You open the Agent Panel next to your theme files and type something like make Merino Beanie ready for AI search, preview first. The agent calls storeseo_start, then storeseo_get_resource_seo for the product, and Zed shows each call in the thread. By default Zed asks for approval before running any tool action, reads included, so the first pass is a handful of confirmations: start, list, get. The GEO block that comes back gives the score and each AI readiness criterion, which the agent turns into a short list of what is missing. The fixing steps then pass through two gates. StoreSEO tells the agent to say what a preview costs in AI credit and to wait for your yes before storeseo_preview_resource_optimization, then to show the drafted FAQs, the AI snippet and the projected score before storeseo_apply_resource_optimization. Zed adds its own confirm prompt on each call, so you approve the same step twice in different places: once in the chat, once in the dialog. For a description, the agent runs storeseo_scan_product_description and asks you StoreSEO's questions in the thread, and you type the answers in the panel before storeseo_write_product_description runs. You can tune the friction. Zed's tool permissions accept per-tool keys in the form mcp:storeseo:storeseo_list_resources, so you can allow the read tools and leave the preview, apply, schema and write tools on confirm. That keeps a catalog scan quick without removing the check on the steps that cost credit or change the store. Prompts: - Use StoreSEO to list my five lowest GEO-scoring products and what each is missing for AI shopping agents. Read only. - Is product schema turned on for my store? If not, tell me what enabling it changes before you touch it. - Scan the description of Ceramic Pour Over for AI agents and ask me StoreSEO's questions one at a time. Do not write until I say so. In Zed: Zed asks before every tool action by default, so an AI readiness pass over several products means many prompts. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_get_resource_seo and keep storeseo_scan_product_description, storeseo_write_product_description and the apply and schema tools on confirm, because they spend credit or change the store. If you set the default to allow, Zed's check disappears and StoreSEO's own request to wait for your yes is the only pause before a description is replaced. Q: Why does Zed prompt me even for reading GEO scores? A: Zed's default tool permission is confirm, which prompts before any tool action. You can allow specific read tools with per-tool keys and leave writes on confirm. Q: Do I approve a description rewrite in Zed or in the chat? A: Both. StoreSEO tells the agent to ask in the chat and wait for your yes, and Zed's own confirm prompt appears on the call unless you allowed that tool. The write replaces the live description with no undo. --- ## Get products ready for AI search with Kiro URL: https://storeseo.com/agent/kiro/ai-search-readiness (full page: https://storeseo.com/agent/kiro/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Kiro means asking its agent, from the chat panel in your IDE, to read a product's GEO score in StoreSEO, preview the fixes and rewrite a thin description, with an autoApprove list you control deciding which calls run unprompted. You work in Kiro with your theme project open and ask in the chat panel which products AI shopping agents will find hardest to read. The agent calls storeseo_start, then storeseo_list_resources sorted by score and storeseo_get_resource_seo for the weakest ones. The GEO score and the failing criteria come back, and the agent reports them in the chat: missing FAQs, no AI snippet, image alt text gaps, a one-line description. A sensible autoApprove list covers exactly these read tools, so the audit runs without a prompt on each call. The write side stays outside that list. The agent calls storeseo_preview_resource_optimization to draft FAQs and an AI snippet and to project the new GEO score, after telling you it spends AI credit; Kiro should prompt for it, since you have not listed it, and that prompt is a useful moment to check the cost. You review the draft in the chat, say yes, and storeseo_apply_resource_optimization applies what you accepted. For a thin description the agent runs storeseo_scan_product_description, asks you StoreSEO's questions in order, and passes your answers to storeseo_write_product_description once you have agreed that it replaces the live text. Because Kiro supports both a workspace file at .kiro/settings/mcp.json and a user file at ~/.kiro/settings/mcp.json, you can keep a read-only autoApprove list in the shared workspace config and a different one personally. If the sign-in points at the wrong store, check which of the two files defines storeseo, since the workspace entry wins on a name clash. Prompts: - Use StoreSEO to list my five lowest GEO-scoring products and what each is missing for AI shopping agents. Read only. - Preview FAQs and an AI snippet for Merino Beanie, tell me the AI credit cost first, and apply nothing until I confirm. - Scan the description of Ceramic Pour Over for AI agents and ask me every question StoreSEO returns, in order. In Kiro: Keep storeseo_scan_product_description, storeseo_write_product_description, storeseo_update_schema_settings and storeseo_apply_resource_optimization out of Kiro's autoApprove list. The scan spends AI credit, the write replaces the live description with no undo, and the schema switch is store-wide. A wildcard in autoApprove would run every StoreSEO tool without Kiro asking, so list individual read tools such as storeseo_start and storeseo_get_resource_seo and leave the rest for Kiro's own prompt. Q: Which StoreSEO tools should I put in Kiro's autoApprove for this job? A: Only reads, such as storeseo_start, storeseo_list_resources and storeseo_get_resource_seo. Leave the scan, preview, apply, schema and description write tools out so Kiro still prompts for them. Q: Does Kiro need a focus keyword on the product before a description rewrite? A: Yes. StoreSEO's scan and write require a focus keyword on the product, which you set in the StoreSEO app. Kiro cannot set one with any tool. --- ## Get products ready for AI search with Antigravity URL: https://storeseo.com/agent/antigravity/ai-search-readiness (full page: https://storeseo.com/agent/antigravity/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Antigravity means asking Google's agentic IDE to read a product's GEO score in StoreSEO, preview the FAQs and AI snippet it can generate, and rewrite a thin description once you have answered StoreSEO's questions. You work in the Antigravity agent panel with your theme project open and ask which products AI shopping agents will struggle with. The agent calls storeseo_start, then storeseo_list_resources sorted by score and storeseo_get_resource_seo for the candidates. Each product's GEO block returns a score and the criteria that fail, which the agent summarizes: no FAQs, no AI snippet, thin description, alt text missing on a couple of images. Antigravity runs unconfigured MCP tools in Ask mode, so expect an approval request before each call in this first pass unless you have set a policy. For a fix, the agent calls storeseo_preview_resource_optimization after saying it spends AI credit, and shows the drafted FAQs, the AI snippet and the projected GEO score. StoreSEO tells it to wait for your yes before storeseo_apply_resource_optimization, and Ask mode adds Antigravity's own approval on the call, so the step is checked twice. If product schema is off, the agent explains that it is a store-wide setting before using storeseo_update_schema_settings, and it turns off only a schema it enabled for this flow if you stop short of applying. Descriptions go through storeseo_scan_product_description, with StoreSEO's questions put to you in the panel word for word and your answers typed back. Only then does the agent propose storeseo_write_product_description, with the warning that the new text replaces the current one at once. Since Antigravity works in your project, ask it to save the previous description from the write response to a file there. Prompts: - Use StoreSEO to list my five lowest GEO-scoring products and what each is missing for AI shopping agents. Read only. - Make Merino Beanie ready for AI search: preview FAQs and an AI snippet, show me the projected GEO score, and wait for my yes. - Scan the description of Ceramic Pour Over for AI agents, ask me StoreSEO's questions in order, and keep the old description in a file in this project. In Antigravity: Antigravity runs MCP tools you have not configured in Ask mode, which gives you a prompt on each call. If you add a policy that allows mcp(storeseo/*), that prompt disappears for every StoreSEO tool, including storeseo_write_product_description, which replaces the live description with no undo. StoreSEO's request to wait for your yes then becomes the only pause. Allow individual read tools instead, and keep the scan, preview, apply, schema and write tools in Ask mode. Q: Will Antigravity ask before it rewrites a product description? A: By default yes, because unconfigured MCP tools run in Ask mode, and StoreSEO also tells the agent to ask in the chat. A policy that allows the whole storeseo server removes the first of those checks. Q: Why does the agent ask me questions before writing a description? A: StoreSEO's scan returns questions about materials, dimensions, what is included and care. Your answers become stated facts in the new copy instead of guesses, and the agent should put each question to you word for word. --- ## Publish llms.txt and agents.md with Claude URL: https://storeseo.com/agent/claude/llms-txt-agents-md (full page: https://storeseo.com/agent/claude/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Claude means asking Anthropic's chat assistant, through the StoreSEO connector, to check what your Shopify store serves to AI crawlers and shopping agents, generate fresh files and switch back to an older version when one is wrong. You work in a Claude chat tab on the web or in Claude Desktop, with the StoreSEO custom connector switched on for the conversation. A good first message is simply whether your llms.txt and agents.md are up to date. Claude calls storeseo_start, then storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Those are reads, so they run straight away and come back as a short summary: your saved llms.txt settings, the latest version and whether a rebuild is due, the public URL, and whether /agents.md serves a custom file or Shopify's default. Generating is where the chat pauses. Generating llms.txt writes a template into your live theme and uploads a copy to Shopify Files in the same run, so StoreSEO tells Claude to warn you first, and Claude should ask before it calls storeseo_generate_llms_txt. It then polls the status until generation finishes, which takes a minute or two, and you can leave the tab open while it does. Generating agents.md is a separate request: it publishes live at once and spends AI Content Optimizer credit, so Claude should name both before it calls storeseo_generate_agents_md. When it finishes you get the published text and the URL back in the chat, which is a comfortable place to read and adjust your reply before asking for another pass. Versions are easy to manage in conversation. Ask to see them and Claude calls storeseo_list_llms_txt_versions, shows which one is active, and can make an older one live with storeseo_activate_llms_txt_version. Deleting a version and resetting agents.md are the two steps that cannot be undone, and Claude's own tool prompts give you a second chance to stop each one. Prompts: - Check my llms.txt and agents.md with StoreSEO. Which versions are live and is a rebuild due? - Generate a new llms.txt with products and collections only. Tell me what it writes to before you run it. - List my llms.txt versions and make last week's the live one. Say which version you will activate before you do. In Claude: On the Free plan Claude allows one custom connector, so StoreSEO may be taking the only slot you have. On Team and Enterprise an owner adds the connector under Organization settings, Connectors before you can sign in. Anthropic advises Allow always only for servers you trust; for this job keep the generate, activate, delete and reset tools on ask, because StoreSEO's warning in the chat is an instruction and Claude's own prompt is what actually stops a live theme write. Q: Can Claude publish llms.txt to my Shopify store? A: Yes. With the StoreSEO connector, Claude can call storeseo_generate_llms_txt, which writes the file into your live theme and uploads a copy to Shopify Files. StoreSEO tells Claude to warn you before it starts, and Claude's tool prompt is your hard stop. Q: Does Claude show me the agents.md it wrote? A: Yes. storeseo_generate_agents_md returns the published markdown, its public URL and the credit used, so Claude can paste the text into the chat. The file is already live by then, so ask Claude to preview your intent first and reset it if you do not like the result. --- ## Publish llms.txt and agents.md with ChatGPT URL: https://storeseo.com/agent/chatgpt/llms-txt-agents-md (full page: https://storeseo.com/agent/chatgpt/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with ChatGPT means using a developer-mode app connected to StoreSEO to read what your Shopify store serves to answer engines and shopping agents, then approving each generate, activate or reset action one at a time. You are in a ChatGPT conversation with the StoreSEO app enabled. Ask whether your llms.txt and agents.md are live and ChatGPT runs storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. These are marked read-only, so they run without a confirmation card. You get the public URLs, the latest llms.txt version, whether settings changed since the last generation and whether /agents.md is a custom file or Shopify's default. Writes behave differently here than in most clients. OpenAI's developer mode asks for approval on write actions by default and lets you open the payload before you accept. Take that option for storeseo_generate_llms_txt, because the arguments carry the resource types, the schedule and the content sections that get saved for scheduled runs, and the run writes a template into your live theme. For storeseo_generate_agents_md the card is your last look before a file replaces Shopify's default and spends AI Content Optimizer credit. ChatGPT then polls storeseo_get_llms_txt_status until generation finishes and reads you the result. To roll back, ask ChatGPT to list the llms.txt versions and activate an older one, or to reset agents.md. Both appear as approval cards, and the reset is the one to read twice because the custom text is discarded with no history. If your plan only has read access in developer mode, ChatGPT can still do the status and version checks, and you generate from another agent or the StoreSEO app, then come back to ChatGPT to confirm the result is live. Prompts: - Using StoreSEO, show my llms.txt status and agents.md status. Do not generate anything yet. - Generate agents.md for my store. Tell me what it will cost in AI credit and that it publishes live, then wait for my yes. - List my llms.txt versions and tell me which one is active before we change anything. In ChatGPT: Write actions depend on your ChatGPT plan. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web, while the help center, as we read it, gives Pro accounts read access only and puts write actions in beta on workspace plans. OpenAI also lets you remember an approval for the rest of a conversation. For this job, do not use it: each generate, activate, delete and reset deserves its own look at the payload. Q: Why does ChatGPT show an approval card when it generates llms.txt? A: StoreSEO marks generate, activate, delete and reset as write actions, and OpenAI's developer mode asks for approval on write actions by default. You can inspect the payload on the card, which shows the settings that will be saved and used for scheduled runs. Q: Can I use ChatGPT to check llms.txt if my plan has no write actions? A: Yes. storeseo_get_llms_txt_status, storeseo_get_agents_md_status and storeseo_list_llms_txt_versions are read-only tools. You can inspect the live state in ChatGPT and apply the generate and rollback steps from the StoreSEO app or another connected agent. --- ## Publish llms.txt and agents.md with Claude Code URL: https://storeseo.com/agent/claude-code/llms-txt-agents-md (full page: https://storeseo.com/agent/claude-code/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Claude Code means using Anthropic's terminal agent, connected to StoreSEO, to check what your Shopify store serves to AI crawlers, generate both files and manage llms.txt versions without leaving the folder you already work in. You run Claude Code from your theme repository or any folder, and type a request such as checking both discovery files and rebuilding llms.txt without blog posts. Claude Code opens with storeseo_start, then runs storeseo_get_llms_txt_status and storeseo_get_agents_md_status, and each call prints in the terminal as a read. The summary names the public URLs, the active version and whether a rebuild is due. Because you are in a theme repository, you can ask it to compare what the store serves with any llms.txt template you keep in the repo before deciding what to regenerate. Generation needs a decision. storeseo_generate_llms_txt writes templates/llms.txt.liquid into the live theme and uploads a copy to Shopify Files, so Claude Code should tell you that and wait for your yes in the chat, and its own permission prompt then appears as a second check. The run queues quickly and takes a minute or two; Claude Code polls storeseo_get_llms_txt_status until generation_ongoing is false, so leave the session running. storeseo_generate_agents_md is the credit-spending one: it publishes at once, replacing Shopify's default, and returns the text, so you can ask Claude Code to write the returned markdown to a file in the repo as a copy before you edit your brand details. Rolling back is a short exchange. storeseo_list_llms_txt_versions shows which version is active, storeseo_activate_llms_txt_version switches, and storeseo_delete_llms_txt_version and storeseo_reset_agents_md are the two that cannot be undone. Note that a file named AGENTS.md in your repository is a different thing from the agents.md StoreSEO publishes to your storefront, so say which one you mean. Prompts: - Use StoreSEO to check my llms.txt and agents.md. Tell me which version is live and whether a rebuild is due. - Regenerate llms.txt. Warn me that it writes to the live theme, wait for my yes, then poll until it finishes. - Generate agents.md for the store, save the returned text to storefront-agents.md in this folder, and tell me the AI credit it used. In Claude Code: If you allow the StoreSEO tools in your Claude Code permissions to skip prompts, the confirmation StoreSEO asks for in the chat becomes the only check before a live theme write, so scope the allow rule to the read tools and keep generate, activate, delete and reset on ask. Without --scope user the server is only registered for the current project, so a status check from another folder will find no StoreSEO tools. Q: Can Claude Code wait for llms.txt generation to finish? A: Yes. storeseo_generate_llms_txt queues the run and returns, and Claude Code polls storeseo_get_llms_txt_status until generation_ongoing is false and the latest generation time has advanced. That takes about one to two minutes, so keep the session open. Q: Is the agents.md StoreSEO publishes the same as an AGENTS.md in my repository? A: No. StoreSEO's agents.md is served from your storefront at /agents.md for AI shopping agents. A file with a similar name in your code repository is instruction text for coding agents. Claude Code does not move one into the other unless you ask it to copy text. --- ## Publish llms.txt and agents.md with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/llms-txt-agents-md (full page: https://storeseo.com/agent/claude-cowork/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Claude Cowork means delegating the StoreSEO status checks, generation and version switching to Anthropic's agent for delegated work, and keeping any scheduled run report-only so it never needs a yes you cannot give. You hand Cowork a task in plain words, such as getting llms.txt and agents.md ready for the store, and it works through the connected StoreSEO tools while you do something else. It starts with storeseo_start, reads storeseo_get_llms_txt_status and storeseo_get_agents_md_status, and brings back a short report on the live version, the public URLs and whether a rebuild is due. Reading the state is the delegable part. Generating is not, because storeseo_generate_llms_txt writes into your live theme and storeseo_generate_agents_md publishes live and spends AI Content Optimizer credit, and StoreSEO tells the agent to wait for you on both. So the useful pattern is two steps. First ask Cowork to check and report, then, with the task open and you present, tell it to run the generation you approved. It will poll storeseo_get_llms_txt_status through the minute or two the run takes and report when the versions list shows the new file active. For agents.md, review the returned text in Cowork before you decide whether to keep it or reset it. Scheduling is where Cowork is different from a chat tab. Anthropic documents that scheduled tasks run remotely on their cadence, even when the computer is asleep, and that they can use connected apps. A weekly task can read both statuses and tell you when a rebuild is due, which is a good fit for llms.txt because it changes when your catalog does. It is the wrong place for generate, activate, delete or reset, since nobody is present to say yes. Prompts: - Use StoreSEO to check my llms.txt and agents.md and give me a short report. Do not generate, activate, delete or reset anything. - Every Monday, read my llms.txt status and agents.md status and tell me if a rebuild is due. Report only. - I am here now. Regenerate llms.txt, wait for it to finish and confirm which version is active. In Claude Cowork: A scheduled Cowork task runs without you in the conversation, so keep any StoreSEO task here to reads and reports and tell it in plain words not to generate, activate, delete or reset. Cowork shares Claude's custom connector rules, so the Free plan allows one custom connector, and on Team and Enterprise an admin toggle and an owner-added connector are both prerequisites. If StoreSEO is missing in Cowork, check Customize, Connectors there. Q: Can a scheduled Cowork task regenerate my llms.txt every week? A: It can run, but it should not. Generating llms.txt writes to your live theme, and StoreSEO tells the agent to wait for your yes, which nobody can give to a scheduled run. Use a scheduled task to read the status and report a due rebuild, then approve the generation yourself. StoreSEO can also regenerate llms.txt on its own saved schedule. Q: What should I put in a Cowork task so it does not publish agents.md by accident? A: State the limit in the prompt: read the statuses, report, and do not generate, activate, delete or reset anything. Also keep your client approval mode strict for the task, since the wording in the task is an instruction and the approval mode is the actual stop. --- ## Publish llms.txt and agents.md with Cursor URL: https://storeseo.com/agent/cursor/llms-txt-agents-md (full page: https://storeseo.com/agent/cursor/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md in Cursor means asking the editor's agent, through the StoreSEO entry in mcp.json, to check what your Shopify store serves to AI crawlers and shopping agents, generate both files and switch llms.txt versions beside your theme code. You are in Cursor's agent chat with the theme project open. Type whether the store has a current llms.txt and agents.md and the agent calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Cursor shows each tool call in the chat and, by default, asks for approval before it uses an MCP tool, so you may click through the reads as well. The answer lists the public URLs, the active version and whether a rebuild is due, and because the theme is open in the editor you can ask the agent to look at your theme's templates folder alongside the status. The generation tools are the ones to pay attention to. storeseo_generate_llms_txt queues a run that writes a template into the live theme and a copy into Shopify Files, and storeseo_generate_agents_md publishes live and spends AI Content Optimizer credit. StoreSEO tells the agent to name those effects and wait for your yes in the chat, and Cursor's approval prompt is the second gate. After a generation the agent polls storeseo_get_llms_txt_status until it finishes, and you can ask it to open the returned agents.md text in a new editor tab to read before deciding whether to keep it. Version work is a short exchange: storeseo_list_llms_txt_versions, then storeseo_activate_llms_txt_version to make an older version live. Deleting a version and resetting agents.md cannot be undone, so confirm each by name. Do not confuse the storefront agents.md with an AGENTS.md rules file in your project: one is served to shopping agents at your store's /agents.md, the other is instruction text for coding agents and StoreSEO never touches it. Prompts: - Use StoreSEO to check my llms.txt and agents.md status and show the active llms.txt version. - Generate llms.txt without blog posts. Say it writes to the live theme, wait for me, then poll until it finishes. - List llms.txt versions and activate the previous one. Tell me which version number you are switching to first. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you have to switch off at least four. If you turned off storeseo_activate_llms_txt_version, storeseo_delete_llms_txt_version or storeseo_reset_agents_md to make room, Cursor will report that the tool is missing when you ask for a rollback, so turn it back on for that session instead of assuming the version cannot be changed. Q: Which StoreSEO tools should I switch off in Cursor to stay under the limit? A: Pick tools for jobs you do not run in Cursor, such as image or Google report tools. For this job you need the llms.txt and agents.md status and generate tools, and the version and reset tools if you plan to roll back. Cursor's own limit is about 40 active tools across servers. Q: Does Cursor ask before generating llms.txt? A: By default Cursor asks for approval before it uses MCP tools, and StoreSEO tells the agent to warn you that generation writes to your live theme. If you set the tool to run without asking, only the confirmation in the chat is left, so keep generate tools on ask. --- ## Publish llms.txt and agents.md with Codex URL: https://storeseo.com/agent/codex/llms-txt-agents-md (full page: https://storeseo.com/agent/codex/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Codex means asking OpenAI's coding agent, connected to StoreSEO, to check what your Shopify store serves to AI crawlers, generate llms.txt and agents.md and switch versions from the CLI, the desktop app or the IDE extension. You are in a Codex session, in the terminal or the IDE extension, with the storeseo server configured in config.toml and signed in through codex mcp login. Ask whether the store's llms.txt and agents.md are current and Codex calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. With the server's default_tools_approval_mode set to writes, those reads run without a prompt, since StoreSEO marks them read-only, and the answer names the active version, the public URLs and whether a rebuild is due. Writes are where the approval mode earns its place. storeseo_generate_llms_txt writes into the live theme and Shopify Files, storeseo_generate_agents_md publishes live and spends AI Content Optimizer credit, and with writes mode Codex prompts for each of those, on top of StoreSEO telling the agent to name the effect and wait for your yes. After llms.txt generation Codex polls storeseo_get_llms_txt_status, which is a series of quick reads, so the 60 second default timeout is not a problem there. storeseo_generate_agents_md is an AI generation inside one call, so it is the one that can outlast the default. Rollback uses storeseo_list_llms_txt_versions and storeseo_activate_llms_txt_version, and you can lock the destructive pair down further with per-tool approval so storeseo_delete_llms_txt_version and storeseo_reset_agents_md always prompt. A file called AGENTS.md in your repository is a different thing from the storefront agents.md StoreSEO publishes at your store's /agents.md, so name the one you mean when you prompt. The first is instruction text for coding agents, and StoreSEO never reads or writes it, which keeps the two jobs cleanly apart. Prompts: - Check llms.txt and agents.md with StoreSEO and tell me the live version and public URLs. - Generate agents.md. State the AI credit use and that it publishes live, then wait for my approval before calling the tool. - Show my llms.txt versions, then switch to version 3 after you confirm it with me. In Codex: Codex waits 60 seconds for a tool to finish by default, and storeseo_generate_agents_md runs an AI generation inside the call, so raise tool_timeout_sec for the storeseo server if it times out. Set default_tools_approval_mode to writes or prompt, not approve: approve skips the prompt, and then the confirmation StoreSEO asks for in chat is the only check before a live theme or storefront write. Q: Why did my agents.md generation time out in Codex? A: Codex waits 60 seconds for a tool by default, and storeseo_generate_agents_md composes the file with AI in a single call. Raise tool_timeout_sec for the storeseo server in config.toml, then check storeseo_get_agents_md_status before generating again, since the file may already have been published. Q: Which Codex approval mode suits llms.txt and agents.md? A: Use writes or prompt for the storeseo server. Writes lets the read-only status tools run and prompts for generate, activate, delete and reset. Avoid approve, which skips the prompt for every tool on that server. --- ## Publish llms.txt and agents.md with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/llms-txt-agents-md (full page: https://storeseo.com/agent/gemini-cli/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Gemini CLI means asking Google's terminal agent, connected to StoreSEO, to check what your Shopify store serves to AI crawlers, generate both files and roll llms.txt back to an earlier version. You run gemini in a terminal after adding the StoreSEO server and signing in through the browser. Ask whether the store's llms.txt and agents.md are live and Gemini CLI calls the StoreSEO tools, which it names with an mcp_ prefix and the server name, so storeseo_get_llms_txt_status appears as mcp_storeseo_storeseo_get_llms_txt_status in the tool list. After storeseo_start it reads both statuses and reports the active version, the public URLs and whether a rebuild is due. Gemini CLI asks you to confirm tool calls unless a server is trusted, which is useful here. When the agent reaches storeseo_generate_llms_txt you see a confirmation prompt, and the agent should already have told you in the conversation that the run writes a template into your live theme and a copy into Shopify Files. The same applies to storeseo_generate_agents_md, which publishes live and spends AI Content Optimizer credit. After generating, it polls storeseo_get_llms_txt_status through the minute or two the run takes, and you can leave the terminal open. For rollback, ask for the versions and name the one you want. Gemini CLI calls storeseo_list_llms_txt_versions and then storeseo_activate_llms_txt_version, and for deleting a version or resetting agents.md you get both the agent's question and the CLI's confirmation. If you run the CLI over SSH with no browser, the sign-in cannot complete, so use the access token header from Settings, AI Toolkit. Either way, the status checks are a cheap first command in a fresh session, because they confirm the connection and show the live state in one go before you decide whether to generate anything. Prompts: - Use StoreSEO to check my llms.txt and agents.md. Report the live version and public URLs only. - Generate a new llms.txt. Tell me it writes to the live theme, wait for my yes, then poll the status until it is done. - List my llms.txt versions and reset nothing. I only want to see which one is active. In Gemini CLI: Do not set --trust or trust: true on the storeseo server for this job. Trust skips every tool confirmation, and then the confirmation StoreSEO asks for in chat is the only check before a live theme write or a reset of agents.md. Google's site also says Gemini CLI was replaced by Antigravity CLI on June 18th, 2026 for unpaid tier and Google One users, so confirm which terminal agent your account can use first. Q: Will Gemini CLI ask before it generates llms.txt? A: Yes, if you have not trusted the server. Gemini CLI asks you to confirm tool calls by default, and StoreSEO tells the agent to say that generation writes to your live theme and wait for your yes. Trust mode removes the CLI prompt, so leave it off. Q: How does Gemini CLI sign in to StoreSEO on a remote machine? A: The browser OAuth sign-in needs a browser on the same machine and does not work headless. On a server or over SSH, create a personal access token in Settings, AI Toolkit and send it as an Authorization Bearer header in the server entry. --- ## Publish llms.txt and agents.md with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/llms-txt-agents-md (full page: https://storeseo.com/agent/github-copilot/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with GitHub Copilot means using agent mode in VS Code, connected to StoreSEO through .vscode/mcp.json, to check what your Shopify store serves to AI crawlers, generate both files and roll llms.txt back, beside your theme code. You open Copilot Chat in agent mode in VS Code, with the theme folder open and the StoreSEO server listed in the tools picker. Ask whether llms.txt and agents.md are current and Copilot calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Each call appears in the chat as a collapsible tool step with its inputs and result, and the answer names the active version, the public URLs and whether a rebuild is due. Generation is where the editor's own confirmation matters. storeseo_generate_llms_txt writes a template into the live theme and uploads a copy to Shopify Files in one run, and storeseo_generate_agents_md publishes live and spends AI Content Optimizer credit. StoreSEO tells Copilot to say so and wait for your yes in the chat, and VS Code's tool confirmation is the hard stop, so read the arguments on the card before you continue. Copilot then polls storeseo_get_llms_txt_status until generation finishes, and you can ask it to open the returned agents.md text next to your theme files for a read. Versions work as in other clients: storeseo_list_llms_txt_versions, then storeseo_activate_llms_txt_version. Deleting a version and resetting agents.md are permanent. If your repository also has an AGENTS.md or Copilot instructions file, those are instructions for coding agents, a separate thing from the /agents.md your storefront serves, so keep the two names straight when you prompt. StoreSEO only works on the files your storefront serves and never edits anything in your repository, which means your commits stay yours. Run the status tools again after any change to confirm what the storefront now serves. Prompts: - Using StoreSEO, check my llms.txt and agents.md status and tell me which llms.txt version is live. - Generate llms.txt for products and collections. Warn me about the live theme write and wait for my confirmation. - Reset my storefront agents.md to Shopify's default. Confirm with me first, because the custom text cannot be restored. In GitHub Copilot: VS Code reads the servers key in .vscode/mcp.json, while mcpServers belongs to the portable .mcp.json, so a Cursor or Claude snippet pasted unchanged will not load. On Copilot Business and Enterprise an organization admin must enable the MCP servers in Copilot policy. Keep the generate, activate, delete and reset tools on VS Code's confirmation prompt, since that prompt is the hard stop before a live theme write. Q: Does Copilot agent mode confirm before it publishes llms.txt? A: StoreSEO tells the agent to warn you that generation writes to your live theme and to wait for your yes, and VS Code shows a tool confirmation you can read before it runs. The editor prompt is the hard stop, so leave it on for generate tools. Q: Can I commit the StoreSEO entry in .vscode/mcp.json to my theme repository? A: You can commit the URL entry, but it is shared with your team, so keep any access token out of it and use input variables if you ever add one. OAuth sign-in avoids storing a token in the file at all. --- ## Publish llms.txt and agents.md with Windsurf URL: https://storeseo.com/agent/windsurf/llms-txt-agents-md (full page: https://storeseo.com/agent/windsurf/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Windsurf means asking the Cascade agent in your editor to check which AI discovery files your Shopify store serves and to have StoreSEO generate them, with the live-theme write and the credit cost spelled out before anything runs. You are in Windsurf with the Cascade panel open, usually in the theme repository, and you type something like: is my llms.txt current, and does the store serve a custom agents.md. Cascade calls storeseo_start first, then storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Those are reads, so they run straight away and come back with the saved generation settings, the latest llms.txt version, whether a rebuild is due, and whether /agents.md is a StoreSEO file or Shopify's default. The status calls also return the public URLs, so Cascade can fetch the live text and show you what an answer engine would read. Generation is where the conversation slows down on purpose. storeseo_generate_llms_txt writes a template into your live theme and uploads a copy to Shopify Files in the same run, so StoreSEO tells the agent to say that and wait for your yes in the Cascade chat. If you want a different shape, say it in the same message: products and collections but not blog posts, or a weekly schedule. The settings you pass are saved and reused by scheduled runs. After the yes, Cascade polls storeseo_get_llms_txt_status until generation_ongoing is false, which takes a minute or two. storeseo_generate_agents_md is the other write: it publishes at once, replaces Shopify's default and spends AI credit. Because Windsurf is an editor, the follow-up work stays next to your code. When the file is live you can ask Cascade to open the theme's llms.txt template and compare it with what StoreSEO wrote, or to list the stored versions with storeseo_list_llms_txt_versions and switch back to an older one. Deleting a version and resetting agents.md are permanent, so ask for those one at a time and read the confirmation before you answer. Prompts: - Use StoreSEO to check my llms.txt and agents.md status. Which version is live and is a rebuild due? Change nothing. - Generate a new llms.txt with products and collections but no blog posts. Tell me what it writes to my theme before you run it. - List my llms.txt versions, then open the live template in the theme folder and tell me whether it matches the active version. In Windsurf: Windsurf's MCP page says its mcp_config.json setup applies to the legacy Cascade agent, while the Devin Local agent, the default for new tabs, reads Devin CLI configuration files. If storeseo tools are missing when you ask for the llms.txt status, check which agent the tab is using before you debug the file. The MCP page also does not describe Cascade's tool-approval behavior, so for a job that writes to your live theme and publishes agents.md, treat the confirmation StoreSEO asks the agent for as an instruction Cascade follows and read each proposed write in the chat before you accept it. Q: Does Windsurf write the llms.txt file into my theme itself? A: No. The write is done by StoreSEO when Cascade calls storeseo_generate_llms_txt, which puts a template in your live theme and a copy in Shopify Files. Cascade can open the theme folder to compare, but the generation runs on the StoreSEO side. Q: Can Cascade check agents.md without generating anything? A: Yes. storeseo_get_agents_md_status is a read. It tells Cascade whether a custom StoreSEO file is live or Shopify's default is serving, and gives the public URL to read the current text. --- ## Publish llms.txt and agents.md with Grok URL: https://storeseo.com/agent/grok/llms-txt-agents-md (full page: https://storeseo.com/agent/grok/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Grok means chatting on grok.com with the StoreSEO connector attached, so you can ask which AI discovery files your Shopify store serves and have them generated, with each live publish named before you agree to it. You are in a Grok chat with the StoreSEO custom connector available, and you have no terminal or config file in sight. Ask whether your store is ready for AI answer engines and Grok opens a StoreSEO session with storeseo_start, then reads storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Both are reads. You get back whether an llms.txt exists, which version is active, whether settings changed since the last build, and whether /agents.md is a custom file or Shopify's default. That is a useful first answer on its own, since many stores have neither file. When you ask Grok to create the missing file, the conversation is the control. StoreSEO tells the agent that llms.txt generation writes to your live theme and Shopify Files, and that agents.md generation publishes immediately, replaces Shopify's default and spends AI Content Optimizer credit. Expect Grok to say so and ask. Your yes in the chat sends storeseo_generate_llms_txt or storeseo_generate_agents_md. The llms.txt run takes a minute or two and Grok polls the status until it finishes; agents.md returns the published text and its URL straight away, so you can read what went live in the same reply. Grok is a good place to review wording, because the whole result sits in one conversation. Ask it to summarize what the agents.md says about your brand and shipping, or to compare the active llms.txt version with the previous one from storeseo_list_llms_txt_versions. Rolling back is a request too: activating an older version republishes it, while deleting a version or resetting agents.md is permanent, so Grok should confirm each before it runs. Prompts: - Use StoreSEO to check my llms.txt and agents.md. Tell me what is live and what is missing. - Write an agents.md for my store and publish it. Tell me what it costs and that it goes live before you do it. - Show me my llms.txt versions and tell me which one is live right now. In Grok: xAI's connector documentation describes no per-tool approval setting for custom connectors, so for this job the confirmation StoreSEO asks the agent for is the main check in the Grok chat. That matters because generating agents.md publishes live and spends credit, and resetting it discards the custom text with no history. Read what Grok proposes, including which tool it is about to call, before you say yes. On Grok Business and Enterprise, a team admin also has to provision the connector in the cloud console before you will see any of these tools. Q: Can Grok publish llms.txt to my Shopify store? A: Yes, through the StoreSEO connector. After you agree, Grok calls storeseo_generate_llms_txt, which writes a template into your live theme and a copy into Shopify Files, then Grok checks the status until the run finishes. Q: How do I get my old agents.md back in Grok? A: You can ask Grok to reset it, which makes Shopify's default serve again within seconds. The custom text is discarded with no version history, so copy it out of the chat first if you may want it again. --- ## Publish llms.txt and agents.md with OpenClaw URL: https://storeseo.com/agent/openclaw/llms-txt-agents-md (full page: https://storeseo.com/agent/openclaw/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with OpenClaw means using your own always-on agent runtime, reached from a chat app or the CLI, to check which AI discovery files your store serves and to have StoreSEO regenerate them when you say so. OpenClaw runs on your machine or server and you talk to it from WhatsApp, Telegram, Slack, Discord, the Control UI or the CLI. Once storeseo is added with openclaw mcp add and signed in with openclaw mcp login, you message it the way you would a colleague: is my llms.txt current. OpenClaw opens a StoreSEO session with storeseo_start and reads storeseo_get_llms_txt_status and storeseo_get_agents_md_status. The reply lands in the same chat, naming the active version, whether a rebuild is due and whether /agents.md is custom or Shopify's default. The status checks suit OpenClaw's scheduler, which is the real difference here. An automation created with openclaw automations create can run every Monday in an isolated session, read both statuses and announce the result in a channel. Keep that job to reads. A scheduled run has nobody to say yes, and storeseo_generate_llms_txt writes into your live theme while storeseo_generate_agents_md publishes at once and spends credit. If the Monday report says a rebuild is due, reply in the chat. OpenClaw then asks you to confirm, runs the generation after your yes and polls until it finishes. OpenClaw also has its own per-server approval setting, configured with openclaw mcp configure storeseo --approval and one of approve, prompt or auto. For this job, prompt is the sensible choice for the write tools, because it adds a client-side stop on top of the confirmation StoreSEO asks the agent for. You can narrow the 44 tools with openclaw mcp tools include and exclude options, for example to expose only the status and version tools to a reporting agent and keep generation, activation, deletion and reset for a separate one. Prompts: - Use StoreSEO to check my llms.txt and agents.md status and tell me if a rebuild is due. Read only. - Every Monday at 8am, send me which llms.txt version is live and whether agents.md is custom or default. Do not generate anything. - A rebuild is due. Tell me what generating llms.txt writes to my theme and then wait for my yes. In OpenClaw: A scheduled OpenClaw automation cannot wait for a yes, so keep the weekly llms.txt and agents.md job report-only and tell it in the prompt not to generate, activate, delete or reset anything. Do the writes in a message you can answer. If you set openclaw mcp configure storeseo --approval to auto, nothing in OpenClaw stops a generation, so StoreSEO's own request to confirm is then the only check. Also run the mcp commands on the installation that owns the connector, or the report will fail to find storeseo. Q: Can OpenClaw regenerate my llms.txt on a schedule? A: StoreSEO can regenerate llms.txt on a schedule saved in its own settings, which an agent can set through storeseo_generate_llms_txt. An OpenClaw automation is better used for a read-only weekly status report, because a scheduled run cannot give the yes that a live theme write needs. Q: Where does the OpenClaw report about llms.txt appear? A: In the channel you give the automation with the announce, channel and to options, such as a Telegram chat. Past runs are listed with openclaw automations runs followed by the job ID. --- ## Publish llms.txt and agents.md with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/llms-txt-agents-md (full page: https://storeseo.com/agent/hermes-agent/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Managing llms.txt and agents.md with Hermes Agent means using a memory-keeping agent in your terminal or chat app to see which AI discovery files your store serves and, when a version is wrong, to have StoreSEO switch back or reset it. Hermes Agent runs from a terminal or from Telegram, Discord and Slack, and once the storeseo entry is in ~/.hermes/config.yaml and you have run hermes mcp login storeseo, you can ask it about the files AI crawlers read. A typical first message is which llms.txt version is live. Hermes calls storeseo_start, then storeseo_get_llms_txt_status and storeseo_list_llms_txt_versions. Both are reads. The status says which version the public /llms.txt serves, and the list shows up to ten stored versions with their dates, whether they were manual or scheduled, and delivery statuses. This job suits Hermes because its memory carries across sessions. Tell it once that you prefer the version without blog posts, and a later session can compare against that preference. The point to understand is that the live file is the active version, which is not always the newest. A fresh generation activates itself, but storeseo_activate_llms_txt_version can put an older one back, republishing its stored content to your theme. StoreSEO tells the agent to name the version and wait for your yes first. storeseo_delete_llms_txt_version is permanent, and deleting the active one promotes the next newest, so Hermes should say which version will go live. agents.md has no versions. storeseo_get_agents_md_status tells you whether a custom file or Shopify's default serves, and storeseo_reset_agents_md removes the custom text for good. If you want a copy before a reset, ask Hermes to fetch the public URL and save the text into your Hermes folder or a file in the working directory first. Hermes' tools include list can restrict the agent to the status and list tools if you want a session that can look but not change anything. Prompts: - Use StoreSEO to list my llms.txt versions and tell me which one is live and which is newest. - Switch my live llms.txt back to the version from before the last rebuild. Name the version and wait for my yes. - Fetch my current agents.md and save it to agents-backup.md here, then tell me what resetting it would do. In Hermes Agent: Start a new Hermes session or run /reload-mcp after editing config.yaml, or the open session will not see the storeseo tools. Hermes' own tool-approval controls are not described in its MCP documentation, so for deleting versions and resetting agents.md rely on the confirmation StoreSEO asks the agent for and read each request before you answer. A hermes cron job for this work has nobody to say yes, so keep it to status and version listings and do rollbacks in a chat you can answer. Q: Can Hermes Agent switch my store back to an older llms.txt? A: Yes. It lists the stored versions with storeseo_list_llms_txt_versions and, after your yes, calls storeseo_activate_llms_txt_version, which republishes that version to your storefront. StoreSEO keeps the latest ten. Q: What does Hermes do when I ask it to reset agents.md? A: After you confirm, it calls storeseo_reset_agents_md, which makes Shopify's default serve again within seconds. The custom text is discarded with no history, so ask Hermes to save a copy from the public URL first if you want one. --- ## Publish llms.txt and agents.md with OpenCode URL: https://storeseo.com/agent/opencode/llms-txt-agents-md (full page: https://storeseo.com/agent/opencode/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with OpenCode means asking the open-source terminal agent to read your store's AI discovery file status through StoreSEO, then generate, switch or reset a file once you confirm. You run OpenCode from your project folder, and after opencode mcp add and opencode mcp auth storeseo, the StoreSEO tools are available as storeseo_ prefixed tools in the session. Ask whether your agents.md is live and OpenCode calls storeseo_start, then storeseo_get_agents_md_status and storeseo_get_llms_txt_status. They are reads, so they run straight away. You see the tool names scroll by, and the answer says whether a custom file is live, which llms.txt version is active and whether a rebuild is due. Writes go through the conversation. For storeseo_generate_agents_md StoreSEO tells the agent to say that the file publishes live at once, replaces Shopify's default and spends AI Content Optimizer credit, and to wait for your yes. For storeseo_generate_llms_txt it tells the agent to say that a template is written into your live theme and a copy uploaded to Shopify Files in the same run. OpenCode then polls storeseo_get_llms_txt_status until generation finishes. Since you are already in a project, you can ask it to save the returned agents.md text into a file, so you hold a copy before any reset. OpenCode's documentation describes deny rules for MCP tools but no ask-first mode, so the permissions entry is your client-side lever. A permissions entry with the action storeseo_storeseo_generate_llms_txt and a deny effect hides that tool entirely, which suits a session meant only to check statuses. Because tool names are the server name plus the tool name, check the exact names with opencode mcp list or the /mcps view before you write a rule, and note that opencode mcp add writes to the project unless you pass --global. Prompts: - Use StoreSEO to check my llms.txt and agents.md status and summarize it in five lines. Do not generate anything. - Generate agents.md for my store and save the returned text to agents-md-copy.md here. Tell me the cost and that it publishes live before you run it. - List my llms.txt versions and tell me whether the active one is the newest. In OpenCode: OpenCode documents deny rules for MCP tools but no ask-first mode, so for this job the confirmation StoreSEO asks the agent for is the main check. Writes here include publishing agents.md live and writing to your theme, so read each proposal before you agree. If you want a client-side stop, deny the generate, activate, delete and reset tools by name in a permissions entry. Also, OpenCode V2 expects servers under mcp.servers while older versions put the entry directly under mcp, so check the layout your version documents if the tools do not appear. Q: How do I stop OpenCode from generating files by accident? A: Add permissions entries with a deny effect for the write tools you do not want, such as the generate and reset tools, using the server name plus the tool name. The status and list tools stay available for read-only checks. Q: Does OpenCode keep a copy of my agents.md before I replace it? A: Not by itself, and StoreSEO keeps no history for agents.md. Ask OpenCode to fetch the public URL and write the text to a file in your project before generating or resetting, so you have your own copy. --- ## Publish llms.txt and agents.md with Zed URL: https://storeseo.com/agent/zed/llms-txt-agents-md (full page: https://storeseo.com/agent/zed/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Zed means using the Agent Panel in the editor to have StoreSEO check, generate and reset the AI discovery files your Shopify store serves, with Zed's tool prompts as a second check on each call. You are in Zed with the Agent Panel open next to your theme files, and you ask whether the store has a current llms.txt and a custom agents.md. The agent calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. With Zed's default of confirm for tool actions, each of those calls shows an approval prompt before it runs, even though they only read. That is the unusual part of this client: an audit that touches three tools means three clicks before you see an answer. The prompts become useful once you reach the writes. StoreSEO tells the agent to say what storeseo_generate_llms_txt will do, which is to write a template into your live theme and upload a copy to Shopify Files, and what storeseo_generate_agents_md will do, which is publish live and spend credit. You answer in the panel, and then Zed prompts again when the tool call is about to run. Two checks stack: the one StoreSEO asks the agent for, and the one Zed enforces. The Zed one is the hard stop, because it blocks the call until you click. You can tune that without losing the stop on writes. Per-tool keys in agent.tool_permissions use the format mcp:storeseo:tool_name, so you can allow the read tools such as storeseo_get_llms_txt_status and storeseo_list_llms_txt_versions and leave generate, activate, delete and reset on confirm. Since Zed is an editor, you can also open the theme's llms.txt template in the next tab and compare it with the active version after a generation, or paste the agents.md text into a project file for a diff. Prompts: - Use StoreSEO to check my llms.txt and agents.md status. Which version is live and is a rebuild due? - Generate a new llms.txt and tell me what it writes to my theme first. I will confirm in the panel. - Reset agents.md to Shopify's default, but fetch the current text and save it into this project first. In Zed: Zed's default is to prompt before every tool action, reads included, so checking both statuses and listing versions takes several approvals. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_get_llms_txt_status and keep the write tools on confirm. Setting agent.tool_permissions.default to allow removes Zed's check, and then the confirmation StoreSEO asks the agent for is the only stop before a live theme write or an irreversible reset, so use allow sparingly for this job. Q: Why does Zed ask me to approve a read of my llms.txt status? A: Zed's default tool permission is confirm, which prompts before any tool action, including reads. You can allow the read tools by name with keys in the mcp:storeseo:tool_name format and keep the generate and reset tools on confirm. Q: Can Zed show me what changed after llms.txt was regenerated? A: Yes, in a practical way. After the generation finishes you can ask the agent to open the theme's llms.txt template or fetch the public URL and compare it with a copy you saved earlier, using the editor's own diff tools. --- ## Publish llms.txt and agents.md with Kiro URL: https://storeseo.com/agent/kiro/llms-txt-agents-md (full page: https://storeseo.com/agent/kiro/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Kiro means asking AWS's agentic IDE to read your store's AI discovery file status through StoreSEO and to generate or switch files, while you decide which tools Kiro may run without asking. You are working in Kiro, often on a theme or storefront project, and you ask whether the store serves a current llms.txt and a custom agents.md. Kiro calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. The answer names the saved generation settings, the active version and its age, whether a rebuild is due, and whether /agents.md is custom or Shopify's default. If you listed those read tools in autoApprove in your mcp.json, they run without a prompt, which makes a status check quick. Kiro's configuration is where this job gets safer or looser. autoApprove takes tool names, and disabledTools hides tools from the agent. A sensible setup for AI discovery files lists the read tools in autoApprove, namely storeseo_get_llms_txt_status, storeseo_get_agents_md_status and storeseo_list_llms_txt_versions, and leaves storeseo_generate_llms_txt, storeseo_generate_agents_md, storeseo_activate_llms_txt_version, storeseo_delete_llms_txt_version and storeseo_reset_agents_md out. StoreSEO tells the agent to name the live theme write, the publish and the credit cost before any of them and to wait for your yes. Kiro's own prompt then sits on top of that. After a generation you can keep working in the same project. Ask Kiro to compare the llms.txt template in the theme with the active version, to write the returned agents.md text into a spec or docs file so the team can review what AI shopping agents are being told, or to list versions and pick an older one to activate. A rollback of llms.txt republishes a stored version, but deleting a version and resetting agents.md cannot be undone, so confirm those individually. Prompts: - Use StoreSEO to check my llms.txt and agents.md status and list any llms.txt versions. Read only. - Generate a new agents.md and save the returned text into docs/agents-md.md. Tell me the credit cost and that it goes live first. - Activate the previous llms.txt version. Name the version and wait for my yes. In Kiro: A wildcard in autoApprove would let Kiro run every StoreSEO tool without asking, including the ones that write to your live theme, publish agents.md and delete llms.txt versions. List individual read tools instead. Kiro's default behavior when autoApprove is empty is not described on its MCP configuration page, so check what your Kiro does before you rely on it. Also remember that a workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, which can point the job at a different store than you expect. Q: Which StoreSEO tools should I put in Kiro's autoApprove for llms.txt? A: The read tools: storeseo_start, storeseo_get_llms_txt_status, storeseo_get_agents_md_status and storeseo_list_llms_txt_versions. Leave generate, activate, delete and reset out so Kiro prompts before they run. Q: Does Kiro wait for llms.txt generation to finish? A: The agent polls storeseo_get_llms_txt_status after storeseo_generate_llms_txt until generation_ongoing is false, which takes about a minute or two. You can keep working in the project while it does. --- ## Publish llms.txt and agents.md with Antigravity URL: https://storeseo.com/agent/antigravity/llms-txt-agents-md (full page: https://storeseo.com/agent/antigravity/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Antigravity means using Google's agent-first IDE to have StoreSEO check and generate the AI discovery files your Shopify store serves, with Antigravity's Ask mode approval in front of each tool call. You are in the Antigravity IDE with the agent panel open, probably beside the theme folder, and you ask which AI discovery files your store serves. The agent calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status. Antigravity runs MCP tools in Ask mode unless you configured a policy for them, so each call waits for your approval in the panel before it executes, reads included. The answers come back with the active llms.txt version, whether a rebuild is due and whether /agents.md is a custom file or Shopify's default. When you ask for a generation, two things happen in order. First the agent, following StoreSEO's instructions, says what the call does: storeseo_generate_llms_txt writes a template into your live theme and a copy into Shopify Files, and storeseo_generate_agents_md publishes live and spends AI Content Optimizer credit. Then, after your yes in the conversation, Antigravity's Ask mode puts the tool call itself in front of you for approval. The second is the hard stop, because the call does not run until you accept it. The agent then polls the llms.txt status for the minute or two the build takes. Antigravity's workspace gives you a good place to review the result. Ask the agent to open the llms.txt template in the theme and the text returned by storeseo_generate_agents_md, and to check them against your catalog and brand voice. If a version is wrong, storeseo_list_llms_txt_versions shows the stored ones and storeseo_activate_llms_txt_version switches the live one. The same sign-in and serverUrl setup you used for other StoreSEO jobs applies here, so nothing extra is needed for these tools. Prompts: - Use StoreSEO to check my llms.txt and agents.md status. Which version is active and is a rebuild due? - Generate a fresh llms.txt with products, collections and pages. Tell me what gets written to my live theme before I approve. - Compare my live agents.md with my store's products and tell me whether anything it says is out of date. In Antigravity: Antigravity's MCP tools run in Ask mode unless a policy says otherwise, so expect an approval prompt on every StoreSEO call here, including the status reads. If you add a policy that allows mcp(storeseo/*), the prompts stop and the confirmation StoreSEO asks the agent for is the only stop before a live theme write, an agents.md publish or an irreversible reset. Allow individual read tools instead. The entry also has to use serverUrl, since Antigravity does not support url or httpUrl for remote servers. Q: Will Antigravity ask me before publishing agents.md? A: Yes, in two places. StoreSEO tells the agent to say that agents.md publishes live and spends credit and to wait for your yes, and Antigravity's Ask mode then prompts for approval of the tool call unless you set a policy that allows it. Q: Can I allow only the llms.txt status checks in Antigravity? A: Yes. Use an mcp policy pattern that names the specific read tools, such as the llms.txt status and version list tools, and leave the generate, activate, delete and reset tools on Ask mode. --- ## Set up structured data with Claude URL: https://storeseo.com/agent/claude/structured-data (full page: https://storeseo.com/agent/claude/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Claude means asking the chat assistant to check which JSON-LD schema types StoreSEO has switched on for your Shopify store, collect your business details in the conversation and send back a complete settings object once you say yes. You are in a claude.ai tab or the Claude Desktop app with the StoreSEO connector switched on for the chat, and you type something like which schema types are enabled on my store. Claude calls storeseo_start, which hands it the conversation id and a snapshot of the store, then storeseo_list_schema_types, and answers with a short table: each type, whether it is on, and whether it takes settings. Product, review, sale dates, product merchant, organization and local business are the configurable ones; the rest, including the four FAQ types, are on or off only. Filling in organization or local business schema is a conversation rather than a form. Claude calls storeseo_get_schema_settings for the type, reads the field template it returns, and then asks you for whatever is blank, such as the legal name, logo URL, contact email, street address and opening hours. You answer in the chat, in any order, and can paste a block copied from your contact page. Claude shows the finished values back to you before it calls storeseo_update_schema_settings, so you can correct a typo while it is still only text on your screen. Because the connection comes from Anthropic's servers and the tools show up in the chat as expandable tool calls, you can open any call and read exactly what was sent. That is useful here, since an update replaces the whole settings object and you will want to see that nothing you set earlier was dropped. Once it is done, ask Claude to list the types again to confirm the status on the live store. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on and which need settings. - Set up organization schema. Read the current settings first, then ask me for anything that is missing before you save. - Turn on product schema, then review schema, and tell me if either is refused. In Claude: Claude's own prompts for tool calls are the hard stop here, and storeseo_update_schema_settings changes the markup on your live storefront. If you chose Allow always for the StoreSEO connector to avoid prompts, that choice covers this write tool as well, so keep the update tool on ask and let the read tools run freely. Also check the connector is on for the chat in the Search and Tools menu, because a new conversation does not always have it enabled. Q: Can Claude fill in my organization schema from a message I paste into the chat? A: Yes. Paste your business name, address and contact details, and Claude maps them onto the field template that storeseo_get_schema_settings returns. It should show you the finished values and wait for your yes before it saves them. Q: Why does Claude read the schema settings before it changes them? A: A settings update in StoreSEO replaces the stored object instead of merging into it. Claude reads the current values first so the object it sends back keeps every field you already filled in. --- ## Set up structured data with ChatGPT URL: https://storeseo.com/agent/chatgpt/structured-data (full page: https://storeseo.com/agent/chatgpt/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Configuring structured data with ChatGPT means using the StoreSEO app you added in developer mode to list your store's JSON-LD schema types, read a type's settings and send them back complete, with ChatGPT's own confirmation step before the write. You open a chat, pick the StoreSEO app from the developer-mode tool menu in the composer, and ask which schema types are enabled. ChatGPT calls storeseo_start and storeseo_list_schema_types and shows each call in the conversation. Reads such as these are what ChatGPT's developer mode treats as read-only, so they typically run without a prompt, and you get a plain list of types with their on or off state and which ones take settings. To set up local business schema, ChatGPT reads the type with storeseo_get_schema_settings, takes the field template from the result and asks you for the missing pieces: business name, address, phone number, opening hours. Organization and local business share the same basic information, so ChatGPT should tell you that filling in one fills in both, which is easy to miss if you only asked for one. When you agree, it calls storeseo_update_schema_settings with the whole settings object and the status you want. That update is a write action, so this is where ChatGPT shows its own confirmation. OpenAI's documentation says write actions need approval by default and that you can inspect the payload before they run. Do that: the payload is the complete settings object, and checking it is how you catch a blank field that would overwrite something you set before. Treat the option to remember approval for the rest of the conversation with care, because later calls in the same chat, such as switching a type off, would then go through without a prompt. Prompts: - In StoreSEO, which JSON-LD schema types are enabled on my store, and which ones can be configured? - Read my local business schema settings, tell me which fields are empty and ask me for them before changing anything. - Enable product schema in StoreSEO. If review schema is refused, tell me why instead of retrying. In ChatGPT: Whether ChatGPT can run storeseo_update_schema_settings depends on your plan. OpenAI's help center, as we read it, gives Pro accounts read access only and puts write actions in beta on Business, Enterprise and Edu workspaces, so on a plan without write actions you can still list types and read settings but must save them from the StoreSEO app or another agent. Where writes are available, read the payload in the confirmation before you approve it. Q: Can ChatGPT read my StoreSEO schema settings if it cannot write them? A: Yes. storeseo_list_schema_types and storeseo_get_schema_settings are read-only tools, so a ChatGPT plan with read access can still show you which types are on and what each one currently holds. Q: What should I check in ChatGPT's confirmation before a schema update? A: Check that the settings object is complete, with your earlier values still present and the new business details filled in, and that the status matches what you asked for. The update replaces the stored object wholesale. --- ## Set up structured data with Claude Code URL: https://storeseo.com/agent/claude-code/structured-data (full page: https://storeseo.com/agent/claude-code/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Claude Code means letting Anthropic's terminal agent read your Shopify store's JSON-LD schema configuration through StoreSEO and write complete settings back, while it can also see the schema your theme outputs in the same folder. You launch Claude Code in the theme repository and ask which schema types are on for the store. It calls storeseo_start, then storeseo_list_schema_types, and prints the result in the terminal as each call completes. The terminal is where this job differs most from a chat: Claude Code can also grep your theme for hand-written application/ld+json blocks, so it can warn you when the theme already prints Organization or Product markup that StoreSEO would then duplicate. For a type you want to configure it calls storeseo_get_schema_settings, takes the field template and fills what it can from files you already have. A store with a contact page, a footer template or a brand JSON file in the repository gives it the name, address and logo path without you typing them, but it should still show you what it found and let you correct it, because StoreSEO's rule is that business details come from you and are never invented. When you agree it sends the whole object through storeseo_update_schema_settings, and you see the call and its arguments in the terminal. Dependencies are worth saying out loud in a terminal session. Review, sale dates and product merchant schema need product schema on, and disabling product also disables product merchant. Ask Claude Code to list the types before and after the change and to paste both lists into a notes file in the repository, which gives you a record of what the markup looked like before you touched it. Prompts: - Use StoreSEO to list my schema types, then grep this theme for any ld+json blocks that could duplicate them. - Read my organization schema settings and fill the empty fields from the footer and contact templates in this repo. Show me the values before you save. - Enable product schema through StoreSEO, then list the types again and save the before and after to schema-notes.md. In Claude Code: Claude Code asks before a tool runs unless you have allowed it, and an allow rule written for the whole StoreSEO server also covers storeseo_update_schema_settings. Scope any rule to the list and get tools if you want the prompts gone for audits, and leave the update tool on ask, since it changes live storefront markup. If you added the server without --scope user it exists only in the folder where you ran the command, so a different project will not see it. Q: Can Claude Code compare StoreSEO's schema with the JSON-LD in my theme? A: Yes. Claude Code can search the theme files in your folder for ld+json blocks and set them beside what storeseo_list_schema_types reports, so you can spot types that would be printed twice. Q: Will Claude Code edit my theme files when I ask for structured data? A: Only if you ask it to. The StoreSEO tools change the schema settings held by StoreSEO, while theme files are separate edits that Claude Code makes through its own file tools, which you approve in the terminal. --- ## Set up structured data with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/structured-data (full page: https://storeseo.com/agent/claude-cowork/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Claude Cowork means delegating the schema work as a task: Cowork uses the StoreSEO connector to list your Shopify store's JSON-LD types, read their settings and hand back a report you review before any change is saved. In Cowork you describe the outcome rather than steer each call, for example check my structured data and tell me what is missing before I launch the sale. Cowork opens its working view, calls storeseo_start and storeseo_list_schema_types, then storeseo_get_schema_settings for each configurable type, and returns a single summary: which types are on, which configurable ones have empty settings, and which dependencies are unmet, such as review schema requested while product schema is off. You get a reviewable document instead of a scroll of tool calls. Doing the writing in Cowork works best as a second step. Read the report, supply the business details it asks for in the task thread, and tell it to go ahead. Cowork then calls storeseo_update_schema_settings with the complete object for the type, and it should stop and ask first, because StoreSEO tells the agent to wait for your yes before a write. Organization and local business share basic information, so one set of details settles both. The scheduling feature is the other thing that makes Cowork different, and it suits the read half of this job well. A weekly task can list the types and read the configurable ones, then tell you if product schema was switched off or an FAQ type started being refused. Keep that scheduled task report-only, and do the changes in a task you are present for, so the yes always comes from a person who has just read the report and knows which type is about to change. Prompts: - Use StoreSEO to check my structured data: list every JSON-LD type, read the configurable ones and write me a report of what is missing. Change nothing. - Prepare local business schema for my store. List the details you still need from me, and wait for my answers before you save anything. - Every Monday, list my StoreSEO schema types and tell me if any type that was on is now off. Do not update, enable or disable anything. In Claude Cowork: A scheduled Cowork task runs while you are away, so nobody can say yes to storeseo_update_schema_settings. Write the schedule as read-only in plain words, naming the list and get tools and forbidding updates, and make changes in a task you are watching. If StoreSEO is missing inside Cowork, check Customize, Connectors there, and on Team and Enterprise ask an admin whether Cowork is switched on. Q: Can Claude Cowork check my schema on a schedule? A: Yes, as a report. A scheduled Cowork task can list the schema types and read the configurable ones each week and tell you what changed, as long as you instruct it not to update, enable or disable anything. Q: Does Cowork use the same StoreSEO connector as Claude? A: It uses the same custom connector model. If StoreSEO is not showing in Cowork, open Customize, then Connectors inside Cowork and add https://mcp.storeseo.com/mcp there. --- ## Set up structured data with Cursor URL: https://storeseo.com/agent/cursor/structured-data (full page: https://storeseo.com/agent/cursor/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data in Cursor means asking the editor's agent to list your Shopify store's JSON-LD types through StoreSEO, read a type's settings and send them back, while Cursor's tool cap and run modes decide whether the schema tools are available at all. You have the theme project open in Cursor and ask the agent in the chat panel which schema types StoreSEO has enabled. It calls storeseo_start and storeseo_list_schema_types, and each call appears inline in the chat. By default Cursor asks before it runs an MCP tool, so on the first use you approve the call in the panel, and the result comes back as a plain list of types with their state. The step that catches people here comes before any of that. StoreSEO exposes 44 tools and Cursor allows about 40 active tools across all your servers, so you have switched at least four off in MCP settings. If storeseo_get_schema_settings or storeseo_update_schema_settings was one of the ones you disabled, the agent will tell you it has no such tool, or quietly fall back to describing what it would do. Open the StoreSEO entry in Cursor's MCP settings and check that the three schema tools are enabled before you start; they are few enough to keep on while you drop tools for jobs you do not use. Being in an editor gives you an advantage. The agent can open your theme files beside the StoreSEO results, find ld+json blocks that already exist and point at lines that would duplicate what StoreSEO outputs. When you ask it to set up organization schema, it reads the current settings and field template first, asks you for anything missing, and only after your yes calls the update with the complete object. You can then review the theme diff and the schema change in one sitting. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which ones are on. If a schema tool is unavailable, say which one. - Read my product merchant schema settings, show me the field template, and tell me what I need to fill in before it can be enabled. - Search this theme for ld+json blocks, then use StoreSEO to check whether organization schema is already on so we do not publish it twice. In Cursor: With about 40 active tools allowed across every server, the schema tools may be among the ones you switched off to fit StoreSEO's 44. Before you start, open the StoreSEO entry in Cursor's MCP settings and confirm storeseo_list_schema_types, storeseo_get_schema_settings and storeseo_update_schema_settings are enabled, and keep the update on ask rather than in a run mode that lets it go through unprompted. Q: Why does Cursor say it cannot find the StoreSEO schema update tool? A: Cursor allows about 40 active tools across all servers and StoreSEO has 44, so you must switch some off. If storeseo_update_schema_settings was among them, turn it back on in the StoreSEO entry in MCP settings and switch off tools you use less. Q: Can Cursor change my theme and my StoreSEO schema in one go? A: Yes, as two separate kinds of edit. The agent changes theme files through Cursor's editing tools, which you review as a diff, and changes StoreSEO schema settings through storeseo_update_schema_settings after you say yes. --- ## Set up structured data with Codex URL: https://storeseo.com/agent/codex/structured-data (full page: https://storeseo.com/agent/codex/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Codex means asking OpenAI's coding agent to list your Shopify store's JSON-LD schema types through the StoreSEO server in config.toml, read settings and write the complete object back after your yes. You run codex in the theme repository, or use the same server from the IDE extension or the ChatGPT desktop app, all of which read the same config.toml. You ask which schema types are on, and Codex calls storeseo_start and storeseo_list_schema_types, printing each call in its transcript. If you signed in with codex mcp login storeseo earlier, the read calls go straight through, and you get a short list of types with their state and whether each takes settings. The approval mode you set on the server decides what happens at the write. With default_tools_approval_mode set to writes, Codex runs the list and get tools that are marked read-only without asking and stops to prompt you before storeseo_update_schema_settings, which is exactly the split this job wants. With prompt, every call asks, which is slower but fine for a one-off setup. With approve, nothing asks, and the only check left is the confirmation StoreSEO tells the agent to request in the conversation, so avoid it for this server. A typical run goes: Codex reads organization settings with storeseo_get_schema_settings, takes the field template, fills what you have given it and anything it can read from files in the repository, and asks about the rest. It then sends the full object once you approve. Because Codex works on the repository you started it in, you can ask it to write the before and after type lists to a file there, which is a convenient record of what changed on the storefront and something you can hand to a client or a colleague afterwards. Prompts: - Use the storeseo MCP server to list my JSON-LD schema types and say which are enabled. Do not call any update tool. - Read my local business schema settings, list the empty fields and ask me for them. After my yes, save the full object and enable the type. - Enable product schema through StoreSEO, then list the types again and write both lists to schema-before-after.md. In Codex: Codex waits 60 seconds for a tool by default, which is plenty for the schema tools, so the usual timeout advice for generation does not apply here. The setting that matters is the server's approval mode: use writes or prompt so Codex stops before storeseo_update_schema_settings, and avoid approve, because then no client prompt stands between the agent and a change to your live storefront markup. Q: Which Codex approval mode suits StoreSEO schema changes? A: Writes mode suits it well, because Codex runs the read-only list and get tools without asking and prompts before storeseo_update_schema_settings. Prompt mode asks for every call, and approve mode skips the prompts entirely. Q: Does the Codex IDE extension see the same StoreSEO server as the CLI? A: Yes. The CLI, the IDE extension and the ChatGPT desktop app share config.toml, so a storeseo server added once is available in each of them, including its approval settings. --- ## Set up structured data with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/structured-data (full page: https://storeseo.com/agent/gemini-cli/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Gemini CLI means asking Google's terminal agent to call the StoreSEO tools that list your Shopify store's JSON-LD types, read a type's settings and write the full settings object back after you confirm. You start gemini in the theme folder and ask which schema types are enabled. Gemini CLI exposes the server's tools under names such as mcp_storeseo_storeseo_list_schema_types, and the model calls storeseo_start first and then the list tool. With trust off, which is the default, Gemini CLI shows a confirmation for each MCP tool call, and you choose to allow it once or for the session. Allowing the read tools for the session is reasonable here and saves repeated prompts during an audit. For a configurable type the model calls storeseo_get_schema_settings, reads the field template and asks you for the missing business details. Since Gemini CLI reads the folder you started it in, it can pull a company name or address from a README, a brand file or the theme's footer, then show you what it found before you accept it. When you confirm, it calls storeseo_update_schema_settings with the full object. Organization and local business share their basic information, so it should tell you that one save covers both. The control worth using here is the includeTools list on the server entry in settings.json. If you only want Gemini CLI to audit schema, list the three schema tools and nothing else, and the model cannot reach the apply or delete tools in that session at all. That is a stricter guard than a prompt, and it keeps tool lists short, which helps the model pick the right call. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on and which accept settings. - Read my organization schema settings, then ask me for any empty field. Do not call an update tool until I say yes. - Turn on product schema in StoreSEO, then check whether review schema can be enabled and tell me if it is refused. In Gemini CLI: Do not set trust on the storeseo server, because trust skips every tool confirmation and then storeseo_update_schema_settings can change live markup with no prompt from the client. The browser sign-in also needs a browser on the same machine, so over SSH or in a headless session use the access token header instead. Check too that your account can still use Gemini CLI, since Google says it was replaced by Antigravity CLI for some unpaid and Google One users on June 18th, 2026. Q: Can I limit Gemini CLI to only the StoreSEO schema tools? A: Yes. Add an includeTools list to the storeseo entry in settings.json naming storeseo_start, storeseo_list_schema_types, storeseo_get_schema_settings and storeseo_update_schema_settings, and the model sees only those from that server. Q: Why does Gemini CLI prompt me for each StoreSEO call? A: Trust is off by default for MCP servers, so Gemini CLI asks before each tool call. You can allow a tool for the session, which is sensible for reads, while keeping the update tool on a prompt. --- ## Set up structured data with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/structured-data (full page: https://storeseo.com/agent/github-copilot/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with GitHub Copilot means asking Agent mode in VS Code to call the StoreSEO tools that list your Shopify store's JSON-LD types, read a type's settings and write them back, with VS Code's confirmation dialog in front of the write. You have the theme workspace open in VS Code with Copilot Chat in Agent mode, and the storeseo server started from .vscode/mcp.json. You ask which schema types are on for the store. Agent mode calls storeseo_start and storeseo_list_schema_types, and VS Code shows a confirmation card for each tool call, with the arguments visible and a choice to allow it once, for the session or for the workspace. For the read tools, allowing them for the session keeps an audit quick. To set up local business schema, the agent calls storeseo_get_schema_settings, takes the field template and asks you for what it lacks. Because it sits in your workspace it can read the theme's footer and contact templates to propose values, and it should present them for you to correct. Organization and local business share basic information, so one set of answers fills both. You then confirm storeseo_update_schema_settings, and the card shows the full settings object that will be sent, which is the thing to read before you click continue. A workspace-level server file brings one practical point. A .vscode/mcp.json committed to the repository is shared with your team, so a colleague who opens the project gets the same StoreSEO server, and each person signs in to the store themselves. Decide among you who is allowed to approve schema updates, since the change goes to the live storefront rather than to the repository, and a pull request review will not catch it. Agreeing that in advance saves an awkward surprise later. Prompts: - Use StoreSEO to list which JSON-LD schema types are enabled and which ones take settings. - Get my organization schema settings from StoreSEO and list every empty field. Ask me for the values before you update anything. - Enable product schema through StoreSEO. If review schema depends on it, tell me the order you will do it in and wait for my yes. In GitHub Copilot: A schema update goes to your live storefront, not to the repository, so no pull request or diff review will show it. VS Code's tool confirmation is the hard stop, so choose Allow once rather than for the workspace on storeseo_update_schema_settings. On Copilot Business and Enterprise plans an organization admin must also enable MCP servers in Copilot policy, otherwise the server will not start. Q: Will a schema change from Copilot appear in my git diff? A: No. storeseo_update_schema_settings changes the schema configuration held by StoreSEO for your live store, so nothing in the workspace changes. Only edits to theme files show up in git. Q: Do my teammates need to sign in separately if .vscode/mcp.json is committed? A: Yes. The file shares the server definition, not your credentials, so each person opens MCP: List Servers, starts storeseo and signs in with their own Shopify access. --- ## Set up structured data with Windsurf URL: https://storeseo.com/agent/windsurf/structured-data (full page: https://storeseo.com/agent/windsurf/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Windsurf means asking the Cascade agent in your editor to check which StoreSEO JSON-LD schema types are on, fill in your organization or product settings from the field template, and switch types on once you agree. You are in Windsurf, usually with a theme or app repository open, and you type the request into the Cascade panel: which schema types are enabled on my store, and is organization schema set up. Cascade calls storeseo_start first to get the conversation id, then storeseo_list_schema_types, which returns every type with an on or off flag and whether it is configurable. Those two calls are reads, so you see them appear in the panel as tool calls and the answer is a short table of what is on, what is off and which types take settings. To configure a type, Cascade calls storeseo_get_schema_settings for it. For product, review, product merchant, organization and local business that response includes a field template, the full shape of the settings object. Cascade fills the template with details you give it, such as business name, logo URL and contact email, and shows you the finished object before it sends anything. Because storeseo_update_schema_settings replaces the stored settings instead of merging into them, Cascade has to read first and send the whole object back. Organization and local business share their basic business information, so a change in one shows up in the other, which is worth telling Cascade if your two entries differ. The editor is useful here in one specific way: if your theme already prints its own JSON-LD, you can ask Cascade to search the repository for existing script tags of type application/ld+json before you switch StoreSEO's types on, so you do not publish the same entity twice. That search is a Windsurf file operation and has nothing to do with StoreSEO. The update call itself changes your live storefront markup, so StoreSEO tells the agent to describe the change and wait for your yes in the chat. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on, which are off and which take settings. - Read my organization schema settings and field template, then draft the full settings object from this business info before sending anything. - Before we enable StoreSEO schema, search this repo for application/ld+json in the theme so we do not duplicate it. In Windsurf: Windsurf's MCP page says mcp_config.json applies to the legacy Cascade agent, while the default Devin Local agent reads Devin CLI configuration, so confirm which agent your tab uses before assuming storeseo is loaded. Cascade also caps MCP tools at 100 across all servers, and StoreSEO takes 44 of them, so with several servers enabled the three schema tools may be among those you have to leave on. The docs do not describe Cascade's tool-approval behavior, which means the confirmation StoreSEO asks the agent for is your main check on a schema update. Read the settings object Cascade proposes before you accept the write. Q: Does Windsurf need to read my theme files to set up StoreSEO schema? A: No. The schema tools work on your StoreSEO settings through the MCP server. Reading the theme is optional and only useful to spot JSON-LD your theme already prints. Q: Why does Cascade read the schema settings before every update? A: Because storeseo_update_schema_settings replaces the stored object instead of merging. Cascade reads the current settings and field template, edits them and sends the complete object, so no existing field is dropped. --- ## Set up structured data with Grok URL: https://storeseo.com/agent/grok/structured-data (full page: https://storeseo.com/agent/grok/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Grok means chatting with xAI's assistant at grok.com, with the StoreSEO connector added, to review your JSON-LD schema types and have it send complete organization, product or review settings once you confirm. You are in a Grok conversation with the StoreSEO custom connector enabled, and you ask in plain words: which structured data is switched on in my store. Grok opens a StoreSEO session with storeseo_start, then calls storeseo_list_schema_types and reports back in the chat. You get each schema type, whether it is on and whether it is one of the configurable ones: product, review, product merchant, organization and local business. Every one of those calls is a read, so the answer arrives without you doing anything but asking. Business details are where a chat is convenient, because you can simply type them. Tell Grok your business name, address, phone number and logo URL, and it calls storeseo_get_schema_settings for organization to fetch the current settings and the field template, then lays out the complete settings object in the conversation for you to check. Nothing is invented: if you leave out the opening hours for a local business, the agent should ask rather than fill a guess. When you say yes, it calls storeseo_update_schema_settings with the full object and the on status, and reports whether the call was accepted or refused. Refusals are useful information in a chat. If the FAQ schema types are not in your plan, or you try to enable review schema before product schema is on, StoreSEO says which requirement failed and Grok relays it in plain words instead of retrying. You can then decide whether to turn product schema on first or leave FAQ schema alone. Because Grok has no editor or file system in this setup, any copy of the settings you want to keep is whatever you paste out of the chat. Prompts: - With StoreSEO, which JSON-LD schema types are on in my store, and which of them can I configure? - Show me my current organization schema settings and what fields are still empty. Do not change anything yet. - Here are my business name, logo URL and support email. Prepare the complete organization settings and ask me before you send them. In Grok: xAI's connector documentation does not describe a per-tool approval setting, so in a Grok chat the main check on a schema change is the confirmation StoreSEO instructs the agent to ask for. Read the settings object Grok proposes before you say yes, because the update replaces the stored settings and changes the markup on your live storefront. On Grok Business and Enterprise a team admin has to provision the connector in the cloud console before you can use it, and the connector must reach the StoreSEO server over the public internet. This page covers the chat connector at grok.com, not xAI's terminal agent, which is configured separately. Q: Can Grok turn on organization schema from a chat? A: Yes, through the StoreSEO connector. It reads the current settings and field template, shows you the complete object and sends it with storeseo_update_schema_settings after you agree. Q: What happens if Grok asks to enable a schema type my plan does not include? A: StoreSEO rejects the call and names the missing feature or dependency, such as the FAQ Schema feature or product schema being off. Grok should report that to you instead of trying again. --- ## Set up structured data with OpenClaw URL: https://storeseo.com/agent/openclaw/structured-data (full page: https://storeseo.com/agent/openclaw/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with OpenClaw means messaging your self-hosted agent runtime to review StoreSEO's JSON-LD schema types and configure organization or product settings, with scheduled checks kept to reports. You reach OpenClaw through a chat app such as Telegram or Slack, or through its Control UI, and ask which schema types are on for your store. The agent calls storeseo_start, then storeseo_list_schema_types, and replies in the same thread with each type, its on or off state and whether it is configurable. The storeseo server was added once with openclaw mcp add and signed in with openclaw mcp login, and that sign-in lives on the installation that owns the connector, so the same credentials serve both a chat message and a scheduled run. A change is a conversation in that thread. You give the business details, OpenClaw calls storeseo_get_schema_settings to fetch the stored settings and the field template, and it posts the complete settings object back to you for review. Only after you answer yes does it call storeseo_update_schema_settings. The update replaces the stored object, which is why the read comes first, and enabling organization, local business or product merchant needs a complete object in the same call or already stored. If StoreSEO refuses because a dependency is missing, the reply in your chat says which one. The scheduler is what makes OpenClaw different for this job. An automation created with openclaw automations create can wake the agent every week in an isolated session, list the schema types and announce the result in your chat, so a type that was switched off or a business detail that went missing shows up without you asking. That run has nobody to say yes, so the prompt should ask for reads only: list the types, read the settings, report the differences. Edits stay in a message you can answer. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on. Read only. - Read my local business schema settings and tell me which fields are empty. Do not send an update. - Draft the complete organization settings from my business details, post them here and wait for my yes before you update. In OpenClaw: A scheduled OpenClaw automation cannot wait for a yes, so a recurring schema check should ask only for storeseo_list_schema_types and storeseo_get_schema_settings and say plainly that it must not call the update tool. Run the openclaw mcp commands on the installation that owns the connector, since a Gateway or agent on another host needs its own reload before it sees storeseo. OpenClaw also has a per-server approval setting through openclaw mcp configure, and a mode that skips the check would leave StoreSEO's own confirmation as the only stop on a schema update. If you want the scheduled agent unable to write at all, narrow its tools with openclaw mcp tools and exclude the update tool. Q: Can an OpenClaw automation switch schema types on by itself? A: It should not. A scheduled run has nobody to answer, so keep it to reading and reporting. Ask for changes in a chat message where you can say yes, or exclude the update tool from the agent. Q: Where does OpenClaw keep the StoreSEO sign-in for scheduled runs? A: On the installation that owns the connector. openclaw mcp login saves the credentials there, so scheduled automations reuse the same sign-in as your chat messages. --- ## Set up structured data with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/structured-data (full page: https://storeseo.com/agent/hermes-agent/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Hermes Agent means asking Nous Research's open-source agent from a terminal or chat app to review your StoreSEO JSON-LD schema types and send complete settings once you agree. You talk to Hermes from a terminal or from Telegram, Discord or Slack, and ask which structured data your Shopify store publishes. Hermes calls storeseo_start, then storeseo_list_schema_types, and answers with each schema type, its on or off state and whether it takes settings. The storeseo entry sits in the mcp_servers block of config.yaml with auth: oauth, so the sign-in was done once with hermes mcp login and Hermes refreshes the tokens itself. If you edited that file during a session, start a new one or run /reload-mcp first. Configuring a type follows the same read-then-write order every time. Hermes calls storeseo_get_schema_settings, which returns the current settings and the field template, edits the object with the details you provide and shows you the complete result. Only after your yes does it call storeseo_update_schema_settings, which replaces the stored settings instead of merging them. Hermes keeps memory across sessions, so a later conversation can start from what it learned: which types you switched on last week, or that you want the organization logo to stay as it is. It still reads the current settings again before writing, because the store may have changed. Hermes also has hermes cron, which runs a prompt on a schedule and delivers the answer to a chat target. A weekly job that lists the schema types and reports any that are off is a sensible use. The run has nobody to confirm a write, so the prompt should be limited to the list and get tools. Changes wait for a chat you can answer. Prompts: - Use StoreSEO to list my JSON-LD schema types, then tell me which ones are off and which take settings. - Read my product schema and review schema settings. Remember that review needs product schema on, and do not change anything yet. - Set up organization schema from my business details. Show me the full settings object and wait for my yes. In Hermes Agent: Hermes only sees the storeseo server in a session started after config.yaml changed, so start a new session or run /reload-mcp before you blame the connection. A token kept in config.yaml for a machine with no browser is a secret, so keep the file out of version control and revoke the token in StoreSEO under Settings, AI Toolkit if it leaks. A hermes cron job cannot wait for your yes, so give a scheduled schema check only reads and say it must not call the update tool. The record for Hermes does not document its own tool-approval controls, so treat StoreSEO's confirmation as an instruction the agent follows, and use the include and exclude lists to expose only the list and get tools to a read-only agent. Q: Will Hermes Agent remember my business details for schema settings? A: Hermes keeps memory across sessions, so it may recall details you gave it before. It should still read the current StoreSEO settings before an update, because the update replaces the stored object. Q: Can a hermes cron job change schema settings? A: Keep it from doing so. A scheduled run has nobody to say yes, so ask it to list types and read settings only, and make changes in a chat you can answer. --- ## Set up structured data with OpenCode URL: https://storeseo.com/agent/opencode/structured-data (full page: https://storeseo.com/agent/opencode/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with OpenCode means asking the open-source terminal agent to check which StoreSEO JSON-LD schema types are on and to send complete organization, product or review settings once you approve. You run OpenCode in a project folder and ask which schema types your store has enabled. The model sees StoreSEO tools named with the server prefix, such as storeseo_storeseo_list_schema_types, and it calls storeseo_start first, then the list tool. The terminal shows each call and the answer is a plain summary: which types are on, which are off and which accept settings. OAuth is on by default, so the sign-in you did with opencode mcp auth storeseo is stored by OpenCode, not in your project files. For a configurable type the agent calls storeseo_get_schema_settings and receives the current settings plus the field template. This is where a terminal helps: you can keep your business details in a file in the repository, such as a short JSON or Markdown note, and tell OpenCode to build the settings object from it. It shows you the complete object, and after your yes calls storeseo_update_schema_settings. Remember that the update replaces the stored settings, so an object missing the logo drops the logo. Organization and local business share their business information, so keep the two consistent. If you want a hard stop on the write, OpenCode lets you add a permission rule with a pattern that matches the tool name and denies it. That turns the session into a safe audit of your schema while you still use the same connection. Lift the rule when you are ready to apply, then ask again. Prompts: - Use StoreSEO to list the JSON-LD schema types for my store and show which are enabled. Read only. - Read the organization schema settings, then fill the field template from business.md in this folder and show me the object before sending. - Check whether review schema can be enabled. Tell me what it depends on, then ask before changing anything. In OpenCode: OpenCode V2 expects servers under mcp.servers, while older versions put the server directly under mcp, so check the layout your installed version documents before editing the config. OpenCode's docs describe deny rules for MCP tools but no ask-first mode, so the confirmation StoreSEO asks the agent for in the terminal is your main check on a schema update unless you deny the update tool by name. Tool names reach the model as server_tool, which is why a pattern such as storeseo_* can hide or deny tools without disconnecting the server. There is also no remove command, so delete the entry from the config and run opencode mcp logout storeseo to clear the sign-in. Q: How do I stop OpenCode from changing my schema settings? A: Add a permission rule that denies the storeseo_update_schema_settings tool by name. The list and get tools keep working, so you can audit your schema types without the agent being able to write. Q: Can OpenCode build schema settings from a file in my project? A: Yes. Point it at a note with your business details and ask it to fill the field template from storeseo_get_schema_settings. Check the full object it shows before you approve the update. --- ## Set up structured data with Zed URL: https://storeseo.com/agent/zed/structured-data (full page: https://storeseo.com/agent/zed/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Zed means asking the Agent Panel in your editor to read your StoreSEO JSON-LD schema types and settings, then send complete settings once you approve both Zed's prompt and the agent's question. You open Zed's Agent Panel and ask which schema types your Shopify store has switched on. The agent calls storeseo_start, then storeseo_list_schema_types, and Zed shows each call in the thread. Zed's default for agent.tool_permissions is confirm, so it asks before it runs a tool action, reads included. Expect a prompt for each of the first three calls in a schema review, and approve them to see the table of types, their on or off state and which of them accept settings. Setting up organization or local business schema continues in the same thread. The agent calls storeseo_get_schema_settings to get the current settings and the field template, fills it from details you type or from a file open in your project, and shows you the complete object. Then there are two separate checks on the write. StoreSEO tells the agent to describe the change and wait for your yes in the conversation, and Zed's own prompt for storeseo_update_schema_settings appears when the agent tries to run it. The update replaces the stored settings, so read the object in the panel before you approve either one. You can make the prompts less repetitive without losing the stop on writes. Zed allows per-tool keys of the form mcp:storeseo:storeseo_list_schema_types, so you can set the list and get tools to allow and leave the update tool on confirm. A schema audit then runs without interruption and only the change asks you. Prompts: - Use StoreSEO to list my JSON-LD schema types and show which are on. Read only. - Read the product schema settings and field template, then tell me which fields are empty. - Prepare the full organization settings from this business info and wait for my yes before calling the update tool. In Zed: Zed asks before every tool action by default, so a schema review means several prompts even though the first three calls only read. Allow the read tools with per-tool keys such as mcp:storeseo:storeseo_get_schema_settings and leave mcp:storeseo:storeseo_update_schema_settings on confirm, so Zed's own prompt stays in front of the write. If you set agent.tool_permissions.default to allow, Zed stops asking and the confirmation StoreSEO tells the agent to request becomes the only check on a change to your live storefront markup. A token in settings.json, if you use one instead of the OAuth prompt, is a secret to keep out of version control. Q: Why does Zed prompt me for read-only StoreSEO schema calls? A: Zed's default is confirm for tool actions, reads included. Set per-tool keys for the StoreSEO list and get tools to allow and leave the update tool on confirm. Q: Who confirms a schema change in Zed, StoreSEO or Zed? A: Both can. StoreSEO tells the agent to wait for your yes in the thread, and Zed's tool permission setting is the hard stop when it prompts before storeseo_update_schema_settings runs. --- ## Set up structured data with Kiro URL: https://storeseo.com/agent/kiro/structured-data (full page: https://storeseo.com/agent/kiro/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Kiro means asking the AWS agentic IDE's chat to read your StoreSEO JSON-LD schema types and settings and then send complete organization, product or review settings after you approve. You work in Kiro with your project open and ask the agent in the chat panel which structured data your Shopify store has on. It calls storeseo_start, then storeseo_list_schema_types, and replies with each type, its on or off state and whether it takes settings. The server is defined in .kiro/settings/mcp.json for the project or in ~/.kiro/settings/mcp.json for every project, and Kiro applies a saved change straight away. If the sign-in points at the wrong store, check which of the two files defines storeseo, because the workspace file wins. To configure a type, the agent calls storeseo_get_schema_settings, takes the field template and fills it from what you tell it or from a file in the project, then shows you the complete settings. The update replaces the stored object instead of merging, and organization and local business share their business information, so the agent should check both before it writes. After your yes it calls storeseo_update_schema_settings and reports the outcome. If StoreSEO rejects the call because the plan lacks the FAQ Schema feature or product schema is off, Kiro relays that reason. Kiro gives each server an autoApprove list and a disabledTools list, and they suit this job. List the read tools, storeseo_start, storeseo_list_schema_types and storeseo_get_schema_settings, under autoApprove so the audit runs without prompts, and leave the update tool off it so Kiro still prompts you before a write. If you only want an audit, add the update tool to disabledTools and the agent cannot call it at all. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on, off and configurable. - Read my local business schema settings and field template, then list what is missing before changing anything. - Draft complete organization settings from the details in this spec file and wait for my yes before updating. In Kiro: An autoApprove entry with a wildcard would let Kiro run every StoreSEO tool without prompting, and that includes storeseo_update_schema_settings, which replaces your schema settings and changes live storefront markup. Name the individual read tools in autoApprove instead, so Kiro's own prompt stays in front of the write, and use disabledTools if you want an audit-only setup. A workspace .kiro/settings/mcp.json overrides a user-level entry with the same server name, which matters if two projects point at different stores. A token under headers is a secret, so do not commit a workspace file that holds one, and revoke it in StoreSEO under Settings, AI Toolkit if it leaks. Q: Which StoreSEO schema tools should I put in Kiro's autoApprove list? A: The read tools only: storeseo_start, storeseo_list_schema_types and storeseo_get_schema_settings. Leave storeseo_update_schema_settings off the list so Kiro prompts before a write. Q: Can I make Kiro unable to change schema at all? A: Yes. Add storeseo_update_schema_settings to the server's disabledTools list in mcp.json. The agent can still list and read schema types but cannot call the update tool. --- ## Set up structured data with Antigravity URL: https://storeseo.com/agent/antigravity/structured-data (full page: https://storeseo.com/agent/antigravity/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Antigravity means asking Google's agentic coding product to read your StoreSEO JSON-LD schema types and settings and send complete settings once you approve the call in Ask mode. You open Antigravity's agent side panel in your project and ask which schema types your Shopify store has enabled. The agent calls storeseo_start and then storeseo_list_schema_types. Tools you have not configured run in Ask mode by default, so Antigravity asks for your approval before each call, and you approve the reads to see the list of types, their on or off state and which accept settings. The server entry in mcp_config.json uses the serverUrl field, and the sign-in ended with a code you pasted back under Customizations in Agent settings. To configure organization, product or review schema, the agent calls storeseo_get_schema_settings, which returns the stored settings and a field template. It fills the template from your details, which can come from a file in the workspace if you point it at one, and shows you the complete object. The update replaces the stored settings, so the agent has to send every field, not only the one you changed. When you agree in the thread it calls storeseo_update_schema_settings, and Antigravity's Ask mode shows its own approval for that call too. That gives you two separate checks on the change. StoreSEO tells the agent to describe it and wait for your yes, and Antigravity's approval prompt is the client-side stop. If you want fewer prompts for the audit, a policy of the form mcp(storeseo/storeseo_list_schema_types) allows a single read tool and leaves the update tool on Ask. Prompts: - Use StoreSEO to list my JSON-LD schema types and tell me which are on, off and configurable. - Read my product merchant schema settings and tell me what it depends on before I enable it. - Build complete organization settings from the details in this workspace file and wait for my yes before updating. In Antigravity: Antigravity's documentation says the remote field is serverUrl and that legacy fields such as url or httpUrl are not supported, so an entry copied from another client's config will not connect until you rename it. Unconfigured MCP tools run in Ask mode, which suits a schema update because it replaces your stored settings and changes live storefront markup. A policy that allows mcp(storeseo/*) would stop Antigravity asking about every StoreSEO tool, and then the confirmation StoreSEO tells the agent to request is the only check. Allow individual read tools such as storeseo_list_schema_types instead, and keep storeseo_update_schema_settings on Ask. Q: Why does Antigravity ask before listing my schema types? A: Unconfigured MCP tools run in Ask mode, which requires approval before each call. Allow the individual read tools with a mcp(server/tool) policy if you want the audit to run without prompts. Q: What does Antigravity send when it updates StoreSEO schema settings? A: The complete settings object, based on the field template from storeseo_get_schema_settings, plus the on or off status. The update replaces the stored settings, so it has to include every field. --- ## Manage sitemaps and indexing with Claude URL: https://storeseo.com/agent/claude/sitemaps (full page: https://storeseo.com/agent/claude/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Claude means asking Anthropic's chat assistant which of your Shopify products, collections, pages and posts are left out of the XML sitemap or set to noindex or nofollow in StoreSEO, changing those flags one resource at a time, and setting up an HTML sitemap page from the same conversation. You work in a Claude chat on the web or in Claude Desktop, with the StoreSEO custom connector switched on for that conversation. A typical first message is which products are set to noindex right now. Claude calls storeseo_start, which returns the conversation id every other tool needs, and then storeseo_list_resource_sitemaps for products with the noindex filter on. The reply is a short table in the chat with each resource's sitemap, noindex and nofollow state, and because the tool also takes a search term and a resource type, you can narrow it to a collection, a page or your blog posts by saying so in plain words. Changing a flag is a separate step. When you say to exclude a draft product from the sitemap and set it to noindex, Claude names that one resource and the two flags it is about to set, and waits for your yes, because noindex removes a page from search results. After you agree, it calls storeseo_update_resource_sitemap, which writes the matching Shopify metafields, and then lists the resource again so the answer you get reflects the store rather than the call it just made. Flags can also be changed in the StoreSEO app between messages, which is why that second read matters. The HTML sitemap runs the same way. Claude reads storeseo_get_html_sitemap_settings first, then generates and enables the page with storeseo_update_html_sitemap, and calls storeseo_list_sitemap_menus so it can show your navigation menus with their edit links. Adding the sitemap link to a menu is something you do in Shopify admin; Claude tells you which menu suits and then marks setup complete once you say it is done. Prompts: - Use StoreSEO to list my products that are set to noindex or excluded from the sitemap, and tell me which ones look like a mistake. - Exclude the product Gift Card Test from the sitemap and set it to noindex. Show me the flags afterwards. - Set up an HTML sitemap for my store, then show me my menus so I can add the link in the footer menu. In Claude: Claude's own prompts for write actions are the hard stop, and the connector menu lets you act on that: open Search and Tools and switch off storeseo_update_resource_sitemap and storeseo_update_html_sitemap for a conversation that should only inspect flags. If you pick Allow always on those two, a noindex change is no longer held by Claude's prompt, and the only check left is the yes StoreSEO tells Claude to wait for. Also remember the Free plan allows one custom connector, so StoreSEO uses that slot. Q: Can Claude set a product to noindex in StoreSEO from a chat? A: Yes. Claude calls storeseo_update_resource_sitemap for one resource at a time, after you agree. It writes the noindex flag as a Shopify metafield and then lists the resource again so you see the current state. Q: Will Claude add the HTML sitemap link to my store's menu? A: No. Claude generates the HTML sitemap and lists your menus with their edit links, but you add the link to a menu in Shopify admin. Then you tell Claude, and it marks setup complete. --- ## Manage sitemaps and indexing with ChatGPT URL: https://storeseo.com/agent/chatgpt/sitemaps (full page: https://storeseo.com/agent/chatgpt/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with ChatGPT means using OpenAI's chat assistant, connected to StoreSEO through developer mode, to read which Shopify resources are excluded from the sitemap or marked noindex or nofollow, change those flags one resource at a time, and generate an HTML sitemap page. You sit in a ChatGPT conversation with the StoreSEO app you created in developer mode available to that chat. Ask which collections are noindex and ChatGPT runs storeseo_start for the conversation id, then storeseo_list_resource_sitemaps with the resource type set to collection and the noindex filter. Reads like these run straight away and ChatGPT shows the call in the conversation, so you can open it and check the arguments it sent. What returns is a list of collections with their sitemap, noindex and nofollow flags, which ChatGPT turns into a short summary and a question about what you want to do with each. A flag change goes through two gates. StoreSEO tells ChatGPT to name the resource and wait for your yes before it calls storeseo_update_resource_sitemap, and ChatGPT's own write confirmation then shows the payload, the resource id and the booleans, for you to approve. Read that payload, because a mistyped resource id is easier to catch there than after the metafield is written. After the write, ChatGPT lists the flags again in the same turn so the answer reflects the store, not just the call it made. For the HTML sitemap, ChatGPT reads storeseo_get_html_sitemap_settings, calls storeseo_update_html_sitemap to generate and enable the page, and then calls storeseo_list_sitemap_menus to hand you the menu edit links. Putting the link in a menu is your step in Shopify admin; ChatGPT cannot do it. Once you confirm it is done, one more update marks setup complete. If the HTML sitemap already has custom settings, ask ChatGPT to read them first and show you the full object before it sends anything, because an update replaces the settings wholesale. Prompts: - Using StoreSEO, which of my collections and pages are set to noindex? List them with their sitemap flags. - Show me which blog posts are excluded from the sitemap, then ask me before changing any of them. - Generate an HTML sitemap for my store with StoreSEO and tell me which of my menus to add the link to. In ChatGPT: Write actions in ChatGPT depend on your plan and on developer mode. OpenAI's developer guide lists developer mode for several plans, while its help center, as we read it, gives Pro accounts read access only and puts write actions in beta on Business, Enterprise and Edu workspaces. On a plan without write actions, ChatGPT can still list every sitemap flag and tell you what to change, and you make the noindex or sitemap edit in the StoreSEO app. Where writes are available, avoid the option that remembers approval for the rest of the conversation, since a noindex flag hides a page from search results. Q: Why can ChatGPT read my sitemap flags but not change them? A: Write actions through developer mode depend on your ChatGPT plan and workspace settings. If your plan only has read access, use ChatGPT to list the flags and apply the change in the StoreSEO app. Q: What does ChatGPT show me before it sets a page to noindex? A: ChatGPT asks you to approve the write action and lets you inspect the payload, which names the resource and the flags. StoreSEO also tells ChatGPT to name the resource and wait for your yes first. --- ## Manage sitemaps and indexing with Claude Code URL: https://storeseo.com/agent/claude-code/sitemaps (full page: https://storeseo.com/agent/claude-code/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Claude Code means asking Anthropic's terminal agent to read the sitemap, noindex and nofollow flags StoreSEO holds for your Shopify resources, change them one resource at a time after your yes, and walk you through the HTML sitemap setup without leaving the folder you are working in. You run Claude Code from your theme repository or any folder and type a request such as show me every product with noindex set, and tell me if any of them are live in the storefront. Claude Code calls storeseo_start, then storeseo_list_resource_sitemaps with filter_on set to noindex and filter_value set to 1, and prints each call in the terminal as it runs. Because it works in your repository, it can also read your theme files, so a robots meta tag in a Liquid template or a hard-coded noindex in the layout shows up next to the StoreSEO flags rather than as a mystery later. To change a flag, tell it which resource and which flag. Claude Code names the resource, says that noindex takes the page out of search results, and waits for your yes in the terminal before calling storeseo_update_resource_sitemap. The tool takes any subset of sitemap_disabled, no_index and no_follow and writes the matching Shopify metafields. Claude Code then lists the resource again before it tells you the state, which matters when someone is also editing flags in the StoreSEO app. If you want a record, ask it to write the before and after flags to a Markdown file in the folder. The HTML sitemap is a short sequence. Claude Code reads storeseo_get_html_sitemap_settings, calls storeseo_update_html_sitemap with setup_step 1 to generate and enable the page, then storeseo_list_sitemap_menus for the menu edit links and the sitemap URL. You add the link in Shopify admin and tell it so, and a last update marks setup complete. When you customize the settings, ask it to show the full object it will send, because settings are replaced wholesale rather than merged. Prompts: - Use StoreSEO to list every product with noindex set. Then check my theme for any hard-coded noindex tags and tell me if they conflict. - Set the page Old Shipping Policy to noindex and exclude it from the sitemap. Ask me first, then show the flags after. - Set up the HTML sitemap with StoreSEO, show me my menus and edit links, and write the steps I still have to do by hand to sitemap-setup.md. In Claude Code: If you add an allow rule for the StoreSEO tools in Claude Code so the read calls stop prompting, that rule also covers storeseo_update_resource_sitemap and storeseo_update_html_sitemap, and then the confirmation StoreSEO asks for in the chat is the only check on a noindex change. Scope the allow rule to storeseo_start, storeseo_list_resource_sitemaps, storeseo_get_html_sitemap_settings and storeseo_list_sitemap_menus. Also add the server with --scope user, otherwise it exists only in the project folder where you first ran the command. Q: Can Claude Code check my theme for noindex tags as well as the StoreSEO flags? A: Yes, because Claude Code reads the folder you start it in. The StoreSEO flags come from storeseo_list_resource_sitemaps, and a robots meta tag in a template is something it finds by reading your theme files. Q: Does Claude Code change the HTML sitemap settings without reading them first? A: It should not. StoreSEO's tool description tells the agent to read storeseo_get_html_sitemap_settings first, because an update replaces the whole settings object. Ask to see the full object before it sends the update. --- ## Manage sitemaps and indexing with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/sitemaps (full page: https://storeseo.com/agent/claude-cowork/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Claude Cowork means handing Anthropic's delegated-work agent a review of which Shopify resources StoreSEO has excluded from the sitemap or set to noindex or nofollow, getting one summary back, and approving any flag change or HTML sitemap setup in the conversation. In Cowork you describe the task rather than steer it call by call. A good one reads go through my products, collections, pages and blog posts, list everything that is noindex, nofollow or excluded from the sitemap, and tell me which look unintended. Cowork opens a StoreSEO session with storeseo_start and then calls storeseo_list_resource_sitemaps once per resource type, paging as needed, because the tool lists one type at a time. You come back to a single report grouped by resource type, with the flags per resource and a short note on each flag that looks out of place, such as a live collection set to noindex. Fixes come after that report. You pick the resources and Cowork asks for your yes before each storeseo_update_resource_sitemap call, naming the resource and the flag, since noindex takes a page out of search results. The tool changes one resource per call, so a long list becomes a series of calls, each followed by a re-list so the final numbers match the store. For the HTML sitemap, Cowork reads the current settings, generates the page, lists your menus with their edit links and leaves the menu step to you in Shopify admin. Because Cowork can run scheduled tasks, the useful recurring version of this job is a check, not a fix. A weekly task can list noindex and excluded resources and tell you what changed since last time, which catches a flag set by accident in the StoreSEO app or by a teammate. Nobody is in the conversation when a scheduled run executes, so the schedule has to be report-only. Prompts: - Use StoreSEO to list every product, collection, page and article that is noindex, nofollow or excluded from the sitemap, and group them by type with a note on any that look unintended. - Every Monday, use StoreSEO to list resources with noindex set and tell me which are new since last week. Do not change, generate or delete anything. - Work through the noindex pages I approve below, one at a time, asking me before each change, and give me a final list of the current flags. In Claude Cowork: A scheduled Cowork task runs without you in the conversation, so nobody can say yes to a noindex change. Keep the scheduled version to storeseo_list_resource_sitemaps and the other reads, and write in the task prompt that it must not call storeseo_update_resource_sitemap or storeseo_update_html_sitemap. Apply fixes from an interactive Cowork task instead, where you can approve each one. StoreSEO also needs to be added under Customize, Connectors inside Cowork, and on Team and Enterprise an owner adds it first. Q: Can a scheduled Cowork task fix noindex flags by itself? A: It should not be set up that way. A scheduled run has nobody to give the yes StoreSEO tells the agent to wait for, so use a scheduled task to list and report flags and make the changes in a task you are present for. Q: Why does Cowork list resources more than once in a sitemap review? A: storeseo_list_resource_sitemaps lists one resource type at a time, covering products, collections, pages or articles. A store-wide review therefore needs one call per type, plus more calls when a type has several pages of results. --- ## Manage sitemaps and indexing with Cursor URL: https://storeseo.com/agent/cursor/sitemaps (full page: https://storeseo.com/agent/cursor/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Cursor means using the editor's agent chat to read StoreSEO's sitemap, noindex and nofollow flags for your Shopify resources, change them one resource at a time after your yes, and set up an HTML sitemap while the theme's Liquid files are open beside it. You open the agent chat in the project that holds your Shopify theme and ask which of your pages and collections are noindex. Cursor calls storeseo_start and then storeseo_list_resource_sitemaps, and each call appears in the chat as a tool block you can expand to see its arguments and the raw response. By default Cursor asks for approval before an MCP tool runs, so even a read may wait for you to press run until you allow it. The answer is a plain list of resources with their sitemap, noindex and nofollow flags, and because the theme is open, you can ask the same chat to search the templates for robots meta tags that would override what StoreSEO sets. To change a flag, name the resource. Cursor says which flags it will set and waits for your yes in the chat, then calls storeseo_update_resource_sitemap, which writes the matching Shopify metafields for that one resource. It lists the resource again afterwards so you see the stored state. The two checks sit at different layers: the yes is an instruction StoreSEO gives the agent, and Cursor's run prompt on the tool call is the hard stop, so a noindex change has to pass both unless you have set the tool to run without asking. For the HTML sitemap, Cursor reads storeseo_get_html_sitemap_settings, calls storeseo_update_html_sitemap to generate and enable the page, then storeseo_list_sitemap_menus for the menu edit links. Adding the link to a menu is your step in Shopify admin. If you later want the page styled to match the theme, that is a normal edit in the open project, and you can ask Cursor to make it in the same chat. Prompts: - Use StoreSEO to list my pages and collections that are noindex or excluded from the sitemap, then grep this theme for robots meta tags that could conflict. - Exclude the collection Test Bundles from the sitemap and set it to noindex. Ask me before you run the update. - Set up the HTML sitemap with StoreSEO, list my menus with their edit links and tell me which one to add the link to. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you have to switch some off in Cursor's MCP settings. The sitemap job needs five of them: storeseo_list_resource_sitemaps, storeseo_update_resource_sitemap, storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap and storeseo_list_sitemap_menus. If the agent says it has no tool for sitemap flags, one of these is probably the one you turned off. Keep the two update tools on Cursor's ask setting, since a noindex write is hard to notice later. Q: Why does Cursor say it cannot change sitemap flags in StoreSEO? A: Cursor allows about 40 active tools and StoreSEO has 44, so you turn some off. Check that storeseo_update_resource_sitemap is still enabled in Cursor's MCP settings, along with the list tool for sitemap flags. Q: Can Cursor find noindex tags in my theme as well? A: Yes. Cursor reads the project you have open, so it can search Liquid templates for robots meta tags while StoreSEO reports the flags it has set. StoreSEO does not edit theme files itself. --- ## Manage sitemaps and indexing with Codex URL: https://storeseo.com/agent/codex/sitemaps (full page: https://storeseo.com/agent/codex/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Codex means asking OpenAI's coding agent, from a terminal or the IDE extension, to read the sitemap, noindex and nofollow flags StoreSEO holds for your Shopify resources, change them one resource at a time after your yes, and set up an HTML sitemap. You start Codex in the repository that holds your theme, or in any folder, and ask for the flags: which products and pages are noindex, and are any of them linked from my navigation. Codex calls storeseo_start for the conversation id and then storeseo_list_resource_sitemaps, once per resource type you ask about, and prints each tool call in the session. Because it also reads the repository, it can compare the flags with the menu files in the theme. The same setup works in the IDE extension and the ChatGPT desktop app, since they read the same config.toml. A flag change is a named request. You say set Old Returns Policy to noindex and exclude it from the sitemap, Codex states the resource and the two flags and waits for your yes, and then calls storeseo_update_resource_sitemap. Where the server's approval mode is set to writes, Codex also prompts on that call because the tool is not marked read-only, which gives you a second stop. After the write it re-lists the resource, because flags can be edited in the StoreSEO app at the same time and you want the current value, not the call's own response. The HTML sitemap goes in order: storeseo_get_html_sitemap_settings, then storeseo_update_html_sitemap with setup_step 1 to generate and enable the page, then storeseo_list_sitemap_menus for the menu edit links and the sitemap URL. Adding the link to a menu is a manual step in Shopify admin, and a last update marks setup complete once you confirm. If you customize the settings, ask Codex to show the full object first, since the update replaces the stored settings instead of merging them. Prompts: - Use StoreSEO to list all pages with noindex or excluded from the sitemap, then compare with the menus in this theme and flag any linked page that is noindex. - Set the product Sample Pack to noindex and exclude it from the sitemap. Wait for my yes, then list its flags again. - Set up the HTML sitemap with StoreSEO. Read the current settings first, generate it, and list the menus I can add it to. In Codex: Set the storeseo server's default_tools_approval_mode in config.toml to writes or prompt. Writes lets the list and get tools run and stops on storeseo_update_resource_sitemap and storeseo_update_html_sitemap, which is the pairing this job wants. Approve skips the prompt, and then the yes StoreSEO tells Codex to wait for is the only check on a noindex change. A project-level .codex/config.toml only loads in a trusted project, so use ~/.codex/config.toml if the server seems to be missing. Q: Which Codex approval mode suits StoreSEO sitemap changes? A: The writes mode fits well. It lets the read tools run and prompts for any tool not marked read-only, so storeseo_update_resource_sitemap and storeseo_update_html_sitemap stop for you, while listing flags does not. Q: Does the Codex setup carry over to the IDE extension? A: Yes. The CLI, the IDE extension and the ChatGPT desktop app read the same config.toml, so the StoreSEO server and its approval settings apply in all of them. --- ## Manage sitemaps and indexing with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/sitemaps (full page: https://storeseo.com/agent/gemini-cli/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Gemini CLI means asking Google's command-line agent to read the sitemap, noindex and nofollow flags StoreSEO holds for your Shopify resources, change them one resource at a time after your yes, and run the HTML sitemap setup from the terminal. You run Gemini CLI in a terminal, usually in the theme folder, and ask which collections and pages are set to noindex. The model sees StoreSEO's tools under names like mcp_storeseo_storeseo_list_resource_sitemaps, and it calls storeseo_start first and then the listing tool with the resource type and the noindex filter. Gemini CLI shows each call and, unless you trust the server, asks you to confirm it before it runs, so even a read can wait on a prompt in your first session. What comes back is a list of resources with their sitemap, noindex and nofollow flags in the terminal. For a change, you name the resource and the flag. Gemini CLI states what it will set and waits for your yes, which StoreSEO tells the agent to do before any write, and then calls storeseo_update_resource_sitemap. Gemini CLI's own confirmation prompt appears on that tool call too, so a noindex change passes two checks. The agent then lists the flags again before it reports the result, since someone may have changed them in the StoreSEO app in the meantime. The HTML sitemap follows the same order as in every client: read storeseo_get_html_sitemap_settings, generate and enable the page with storeseo_update_html_sitemap, then call storeseo_list_sitemap_menus for the menu edit links. You add the link to a menu in Shopify admin, and the agent marks setup complete after you say it is done. If a settings object is involved, ask Gemini CLI to print the full object it plans to send, because an update replaces the stored settings. If the model has too many StoreSEO tools to choose from, includeTools in settings.json can narrow the list to these five. Prompts: - Use StoreSEO to list my products that are noindex or excluded from the sitemap and tell me which look like mistakes. - Set the article Draft Announcement to nofollow and noindex. Ask me first, then show the flags afterwards. - Set up an HTML sitemap with StoreSEO and show me the menus and edit links so I can add it to the footer menu. In Gemini CLI: Do not set trust: true or pass --trust on the storeseo server in Gemini CLI. Trust skips every tool confirmation, and for this job that means a noindex change is held only by the yes StoreSEO tells the agent to wait for. Also add the server with --scope user if you want it outside one project, and remember the browser sign-in needs a browser on the same machine; over SSH or headless, use the access token header instead. Q: How do I stop Gemini CLI from changing sitemap flags without a prompt? A: Leave trust off for the storeseo server, which is the default. Then Gemini CLI asks you to confirm each MCP tool call, including storeseo_update_resource_sitemap, on top of the yes StoreSEO tells the agent to wait for. Q: Can I show Gemini CLI only the sitemap tools? A: Yes. The includeTools list on the server entry in settings.json limits which StoreSEO tools the model sees, so you can keep the five sitemap tools and storeseo_start and hide the rest. --- ## Manage sitemaps and indexing with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/sitemaps (full page: https://storeseo.com/agent/github-copilot/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with GitHub Copilot means using Agent mode in Visual Studio Code to read StoreSEO's sitemap, noindex and nofollow flags for your Shopify resources, change them one resource at a time after your yes, and set up an HTML sitemap while your theme files stay open next to the chat. You open Copilot Chat in Agent mode in the VS Code window that holds your theme and ask which products and pages are noindex. Copilot calls storeseo_start, which returns the conversation id, and then storeseo_list_resource_sitemaps, and VS Code shows each tool call in the chat with its input and output. You pick the StoreSEO tools you want from the tools picker, which is also how you keep the list short. The answer is a list of resources with their sitemap, noindex and nofollow flags, and with the theme open you can ask Copilot to look for robots meta tags in templates that might conflict. Changing a flag is one resource at a time. When you ask for a page to be set to noindex and excluded from the sitemap, Copilot names the page and both flags and waits for your yes in the chat. VS Code then shows its own confirmation for the tool call, so the write passes through StoreSEO's instruction and the editor's prompt. Once you allow it, storeseo_update_resource_sitemap writes the matching Shopify metafields, and Copilot lists the resource again so the flags you see are the stored ones. For the HTML sitemap, Copilot reads storeseo_get_html_sitemap_settings, calls storeseo_update_html_sitemap to generate and enable the page, then storeseo_list_sitemap_menus for the menu edit links and the sitemap URL. Adding the link to a menu happens in Shopify admin, not in a tool, and you tell Copilot when it is done so it can mark setup complete. If your team shares a .vscode/mcp.json, each person signs in to their own StoreSEO account, so the flags each of you sees are for the store you authorized. Prompts: - Use StoreSEO to list my pages and collections that are noindex or excluded from the sitemap, then search this workspace for robots meta tags in the theme. - Set the page Old Terms to noindex and exclude it from the sitemap. Wait for my confirmation, then list its flags again. - Set up an HTML sitemap with StoreSEO, then list my menus and edit links so I can add the sitemap URL. In GitHub Copilot: Keep storeseo_update_resource_sitemap and storeseo_update_html_sitemap on VS Code's confirmation prompt, because that prompt is the hard stop on a noindex change. If you let VS Code remember an approval for the server, the yes StoreSEO tells Copilot to wait for is the only check left. Also note that .vscode/mcp.json uses the servers key rather than mcpServers, and that on Copilot Business and Enterprise an organization admin has to enable MCP servers in Copilot policy before StoreSEO will run. Q: Does Copilot ask before setting a page to noindex? A: StoreSEO tells Copilot to name the page and wait for your yes, and VS Code shows a confirmation on the storeseo_update_resource_sitemap call unless you have chosen to remember approvals. Keep that prompt on for this tool. Q: Why does my Copilot setup not see the StoreSEO sitemap tools? A: Check that the server is started in MCP: List Servers, that the storeseo tools are ticked in the tools picker, and, on Business or Enterprise plans, that an admin has enabled MCP servers in Copilot policy. --- ## Manage sitemaps and indexing with Windsurf URL: https://storeseo.com/agent/windsurf/sitemaps (full page: https://storeseo.com/agent/windsurf/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Windsurf means asking Cascade, the agent panel in Cognition's editor, to read which Shopify resources StoreSEO leaves out of the sitemap or marks noindex, change those flags one resource at a time and set up the HTML sitemap page. You work in the Cascade panel, usually with the theme repository open beside it, and type a request such as which pages are set to noindex. Cascade calls storeseo_start first, because every StoreSEO tool needs the conversation id it returns, and then storeseo_list_resource_sitemaps for the resource type you named: products, collections, pages or articles. It can filter the listing to rows where the sitemap flag, noindex or nofollow is on, so a question about a stray noindex on a collection comes back as a short list rather than the whole catalog. Each call shows in the panel, and all of them are reads. Changing a flag is where the order matters. A noindex flag removes a page from search results, so Cascade should name the resource, say which of the three flags it will set and wait for your yes before it calls storeseo_update_resource_sitemap. Windsurf's MCP page does not describe a tool-approval prompt, so read each proposal in the chat before you accept it. After the write, ask Cascade to list that resource again, because the flags can also be changed in the StoreSEO app and an earlier listing may be stale. The same session handles the HTML sitemap: storeseo_get_html_sitemap_settings first, then storeseo_update_html_sitemap, then storeseo_list_sitemap_menus for the menu edit links. The editor setting helps here. Cascade can open the theme's footer or header section in the same window while you decide where the sitemap link should live, and you add the link in Shopify admin yourself, since no tool does it. The sitemap tools work on every active plan, and the whole server's 44 tools fit under Cascade's cap of 100, so you do not need to switch anything off to run this job. Prompts: - Use StoreSEO to list my collections that are set to noindex or excluded from the sitemap. Read only. - Exclude the product Gift Card Test from the sitemap and set it to noindex. Tell me what you will change and wait for my yes. - Set up the HTML sitemap with StoreSEO, then list my menus and show me which edit link to open in Shopify admin. In Windsurf: Windsurf's MCP page says its configuration applies to the legacy Cascade agent, and the Devin Local agent, the default for new tabs, reads Devin CLI configuration files instead. If the StoreSEO sitemap tools do not appear, check which agent your tab is using before you edit mcp_config.json again. The page also does not describe Cascade's approval behavior for MCP tools, so for a noindex change treat StoreSEO's confirmation as an instruction the agent follows and read the proposed resource and flags before you accept the call. Q: Can Windsurf's Cascade change noindex on several products at once? A: The StoreSEO tool sets flags on one resource per call. Cascade can run it for each product you name, but it should list the products and wait for your yes first, because a noindex flag removes a page from search results. Q: Does Cascade add the HTML sitemap link to my menu? A: No. Cascade generates the HTML sitemap and lists your menus with their edit links, and you add the link to a menu in Shopify admin. After that you can ask Cascade to mark setup complete. --- ## Manage sitemaps and indexing with Grok URL: https://storeseo.com/agent/grok/sitemaps (full page: https://storeseo.com/agent/grok/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Grok means asking xAI's chat assistant, through the StoreSEO connector on grok.com, which of your Shopify products, collections, pages and articles are noindex, nofollow or left out of the sitemap, and having it fix them one resource at a time. You sit in an ordinary Grok chat with the StoreSEO connector available and ask in plain words, for example whether any of your blog posts are set to nofollow. Grok opens a StoreSEO session with storeseo_start, then calls storeseo_list_resource_sitemaps with the resource type article and a filter on the nofollow flag. What comes back is a list of titles with their three flags, which Grok turns into a short answer. Nothing here needs a terminal or a config file, and the questions work on a phone as well as on a desktop, so checking a page before you send a link is quick. For a change, the conversation is the control point. A noindex flag removes the page from search results, so StoreSEO tells the agent to name the resource and the flag and wait for your yes before it calls storeseo_update_resource_sitemap. xAI's connector documentation describes no separate per-tool approval setting, so your own reading of the proposal is the check: confirm the product title and which of the three flags it will set. Once the write returns, ask Grok to re-list that resource, because the flags can also change in the StoreSEO app and the first listing may be out of date. The HTML sitemap works the same way. Grok reads storeseo_get_html_sitemap_settings, generates the page with storeseo_update_html_sitemap after your yes, and then lists your navigation menus with storeseo_list_sitemap_menus so you know which one to edit. You add the sitemap link in Shopify admin yourself, and then Grok can mark setup complete. Every sitemap tool works on any active plan, including the free one. Prompts: - Which products in my store are excluded from the sitemap or set to noindex? List them with their flags. - Set the page Old Shipping Policy to noindex and leave it in the sitemap. Show me what you will change first. - Do I have an HTML sitemap yet? If not, set one up and tell me which menu to add it to. In Grok: xAI's connector documentation does not describe a per-tool approval setting, so the only stop in a Grok chat is your reading of what it proposes. For this job that means checking the exact title and which flag before you say yes, because noindex takes a page out of search results and a similar product name is easy to confuse. On Grok Business and Enterprise a team admin has to provision the StoreSEO connector in the cloud console before members can use it, so the sitemap tools will not appear until that is done. Q: Can Grok set noindex on a Shopify page through StoreSEO? A: Yes, one resource per call. StoreSEO tells Grok to name the page and the flag and wait for your yes, and xAI's documentation describes no separate approval setting, so check the title before you agree. Q: Does a Grok chat need my computer to be online for the sitemap tools? A: No. Grok reaches the StoreSEO server over the public internet, so nothing on your machine needs to be reachable. The sign-in only has to finish once in a browser. --- ## Manage sitemaps and indexing with OpenClaw URL: https://storeseo.com/agent/openclaw/sitemaps (full page: https://storeseo.com/agent/openclaw/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with OpenClaw means asking the self-hosted agent runtime, from a chat app or the terminal, to read StoreSEO's sitemap, noindex and nofollow flags for your Shopify resources, change them on request and run a read-only check of those flags on a schedule. OpenClaw is the one client here where the same job can start in two places. In a chat or the CLI you ask which products are set to noindex, and the agent calls storeseo_start, then storeseo_list_resource_sitemaps for the resource type, filtered to rows with noindex on. The answer comes back in the channel you asked in. Because the connector lives on the installation that owns it, a message from your phone reaches the same StoreSEO session as one typed at the machine, and openclaw mcp doctor storeseo --probe tells you the connection works before you rely on it. Changes stay conversational. A noindex flag removes a page from search results, so StoreSEO tells the agent to name the resource and the flag and wait for your yes before storeseo_update_resource_sitemap. OpenClaw also has its own approval setting for the server, set with openclaw mcp configure storeseo --approval, and that is the hard stop if you want the client to refuse a write rather than rely on the instruction. After a change, ask for the listing again; the flags can be edited in the StoreSEO app too. The HTML sitemap follows the usual order of storeseo_get_html_sitemap_settings, then storeseo_update_html_sitemap, then storeseo_list_sitemap_menus. Scheduling is what OpenClaw adds. An automation made with openclaw automations create can run weekly in an isolated session and announce the result in a channel, which suits a check that nothing important slipped into noindex after a bulk edit. Write the prompt as a report: list the resources with noindex or the sitemap flag on and say nothing else. A scheduled run has nobody to answer a question, so it must not ask for changes. Prompts: - Use StoreSEO to list every product and collection that is noindex or excluded from the sitemap. Read only. - Exclude the product Gift Card Test from the sitemap and set it to noindex. Name the resource and ask me before you write. - Create a weekly automation that lists noindex and nofollow flags on my pages and articles and announces the list in this chat. Reads only. In OpenClaw: A scheduled OpenClaw automation runs in an isolated session with nobody to say yes, so a sitemap automation must be report-only: ask it to list flags, never to change them, and do not word the prompt as a fix. A noindex change should happen in a message you can answer. If you want the client itself to refuse writes from unattended runs, set the server's approval mode with openclaw mcp configure storeseo --approval to something other than the mode that skips the check, since StoreSEO's confirmation is an instruction the agent follows rather than a lock. Q: Can an OpenClaw automation switch pages to noindex on its own? A: It should not. A scheduled run has nobody to answer StoreSEO's confirmation, so keep the prompt to listing flags and set OpenClaw's approval mode for the server so a write cannot slip through unattended. Q: How do I get a weekly list of noindex pages in my chat channel? A: Create an automation with openclaw automations create, a cron expression and a prompt that asks StoreSEO for the resources with noindex on, then add the announce option and your channel. It runs in an isolated session and posts the result. --- ## Manage sitemaps and indexing with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/sitemaps (full page: https://storeseo.com/agent/hermes-agent/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Hermes Agent means asking Nous Research's agent, which remembers earlier sessions, to read StoreSEO's sitemap, noindex and nofollow flags on your Shopify resources, change them one at a time on request and report on them from a schedule. You start a Hermes session, or message the agent from the chat target it delivers to, after adding the storeseo entry to mcp_servers in config.yaml and running hermes mcp login storeseo. Ask which collections are excluded from the sitemap and Hermes calls storeseo_start, then storeseo_list_resource_sitemaps with the resource type collection and a filter on the sitemap flag. The reply is a list of titles with their sitemap, noindex and nofollow values. Hermes keeps what it learned, so a follow-up such as now check the pages the same way does not need you to restate the store. A change goes through the confirmation StoreSEO asks the agent for. A noindex flag removes a page from search results, so Hermes should name the resource and the flags and wait for your yes before storeseo_update_resource_sitemap, and it should list the resource again afterwards because the flags can also change in the StoreSEO app. Hermes' own approval controls are not confirmed in its documentation, so use its tool filters when you need a hard stop: an include list that exposes only the list and get tools gives you a read-only agent, and a second profile can keep the update tools. The HTML sitemap runs through storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap and storeseo_list_sitemap_menus, and you add the sitemap link to a menu in Shopify admin yourself. Hermes also has hermes cron create, which takes a schedule and a prompt and delivers the result with --deliver. For this job, that means a Monday message listing everything that is noindex or excluded from the sitemap, which catches an accidental flag after a bulk edit. The prompt must be report-only, because nobody is present at the run to say yes. Prompts: - Use StoreSEO to list the products with noindex or excluded from the sitemap. Read only, change nothing. - Set the page Staff Login to noindex and exclude it from the sitemap. Tell me what you will change and wait for my yes. - Generate my HTML sitemap with StoreSEO, then list my menus so I know where to add the link. In Hermes Agent: A hermes cron job runs with nobody present to answer StoreSEO's confirmation, so a sitemap job must be a report: list the flags, never change them. The agent record does not confirm Hermes' own approval controls, so for a hard stop use Hermes tool filters to expose only the StoreSEO list and get tools to the scheduled agent. Also start a new session or run /reload-mcp after editing config.yaml, otherwise the open session does not see the server, and keep a token out of version control if you used the header form. Q: Can I give Hermes Agent read-only access to my sitemap flags? A: Yes. Use Hermes' tool include list to expose only the StoreSEO list and get tools, such as storeseo_list_resource_sitemaps and storeseo_get_html_sitemap_settings, and leave the update tools out for that agent. Q: What should a hermes cron prompt for noindex checks say? A: Ask it to use StoreSEO to list the resources with noindex or excluded from the sitemap and to deliver the list, and tell it not to change anything. A scheduled run cannot answer a confirmation, so it should only read. --- ## Manage sitemaps and indexing with OpenCode URL: https://storeseo.com/agent/opencode/sitemaps (full page: https://storeseo.com/agent/opencode/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with OpenCode means asking the open-source terminal agent to read StoreSEO's sitemap, noindex and nofollow flags on your Shopify products, collections, pages and articles, change them one resource at a time and set up the HTML sitemap. You open OpenCode in the theme project, with the storeseo server added by opencode mcp add and signed in with opencode mcp auth storeseo, and type a request such as which pages are excluded from the sitemap. Tool names reach the model as server_tool, so the agent calls storeseo_start and then storeseo_list_resource_sitemaps, asking for the page resource type and filtering on the sitemap flag. Each call prints in the session as it runs and every one is a read. The same session can hold the theme code, so a question about why a template carries a robots meta tag sits next to the StoreSEO flag for that page. Writing a flag is a separate step. A noindex flag removes a page from search results, so StoreSEO tells the agent to name the resource and the flags and wait for your yes before storeseo_update_resource_sitemap. OpenCode's MCP documentation describes deny rules but no ask-before-running mode, so StoreSEO's confirmation in the chat is the main check unless you add a client-side stop. A permission rule matching storeseo_update_resource_sitemap with the deny effect hides that tool entirely, which is a clean way to run a read-only audit session. Then ask the agent to list the resource again, since the flags can change in the StoreSEO app. The HTML sitemap uses storeseo_get_html_sitemap_settings first, because an update replaces the settings wholesale, then storeseo_update_html_sitemap to generate and enable the page, then storeseo_list_sitemap_menus for the edit links. You add the sitemap link to a menu in Shopify admin; no tool does it. Everything in this job runs on any active plan, so it works on the free plan too. Prompts: - Use StoreSEO to list my products that are noindex or excluded from the sitemap. Read only, no writes. - Exclude the product Gift Card Test from the sitemap and set it to noindex. Show me the resource and flags first and wait for my yes. - Check my HTML sitemap settings with StoreSEO, set it up if it is missing and list the menus I can add it to. In OpenCode: OpenCode's documentation describes deny rules for MCP tools but no setting that asks before running one, so for a noindex change StoreSEO's confirmation in the chat is the main check. If you want a client-side stop, add a permission rule with the deny effect for storeseo_update_resource_sitemap and storeseo_update_html_sitemap, and lift it only for the session where you mean to write. Also note that opencode mcp add writes to the project configuration unless you pass --global, so a sitemap session in another folder will not see the server. Q: How do I stop OpenCode from changing sitemap flags by accident? A: Add a permission rule with the deny effect for the StoreSEO update tools, for example storeseo_update_resource_sitemap. OpenCode's docs describe deny rules for MCP tools, so the agent can still list flags but cannot write them. Q: Does OpenCode need the config layout for the HTML sitemap tools? A: No extra setup is needed beyond the storeseo server entry. In OpenCode V2 it sits under mcp.servers, and the sitemap tools then work on any active StoreSEO plan. --- ## Manage sitemaps and indexing with Zed URL: https://storeseo.com/agent/zed/sitemaps (full page: https://storeseo.com/agent/zed/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Zed means asking the Agent Panel to read StoreSEO's sitemap, noindex and nofollow flags on your Shopify resources, approve each change in Zed's own prompt and set up the HTML sitemap, next to your theme code. You open the Agent Panel in the editor with the theme folder loaded and type a request such as which collections are noindex. The agent calls storeseo_start, then storeseo_list_resource_sitemaps for the collection resource type. Zed's default tool-permission value is confirm, so each of those calls stops for your approval in the panel before it runs, reads included. For an audit that touches only the listing tools, that is a handful of clicks, and you can reduce them by allowing individual read tools with keys such as mcp:storeseo:storeseo_list_resource_sitemaps. This is the client where the approval prompt does the heavy lifting. When the agent proposes storeseo_update_resource_sitemap, StoreSEO has already told it to name the resource and the flags and wait for your yes, and Zed adds its own confirmation on the call itself, so you see the tool call before it runs. Keep the write tools on confirm: a noindex flag removes a page from search results. After the write, ask the agent to list the resource again, because the flags can be changed in the StoreSEO app too. The HTML sitemap goes through storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap and storeseo_list_sitemap_menus, and the menu edit links it returns are opened in Shopify admin. The editor pays off when the answer touches code. If the flags look right but a collection page still carries a stray robots meta tag, the agent can open the template in the same window and show you the line, while StoreSEO shows the flag it holds. The sitemap tools work on every active plan. Prompts: - Use StoreSEO to list my pages that are excluded from the sitemap or set to noindex. Read only. - Set the collection Holiday Bundle back to indexable and into the sitemap. Show me the call before you make it. - Set up my HTML sitemap with StoreSEO and list the menus so I can add the link in Shopify admin. In Zed: Zed asks before every tool action by default, reads included, so a sitemap audit of several resource types means several prompts. Allow only the listing tools with per-tool keys like mcp:storeseo:storeseo_list_resource_sitemaps and leave storeseo_update_resource_sitemap and storeseo_update_html_sitemap on confirm. Setting agent.tool_permissions.default to allow removes Zed's check, and then the confirmation StoreSEO asks the agent for is the only one before a noindex flag is written, so use allow sparingly. A token in settings.json is a secret, so keep that file out of a public repository. Q: Will Zed ask before StoreSEO changes a noindex flag? A: By default yes. Zed's tool permission is confirm, so the Agent Panel prompts before it runs a tool call, and StoreSEO also tells the agent to name the resource and wait for your yes. Setting the default to allow removes Zed's prompt. Q: How do I stop Zed prompting for every sitemap listing? A: Add per-tool permission keys of the form mcp:storeseo:storeseo_list_resource_sitemaps set to allow for the read tools, and leave the update tools on confirm so a write still stops for your approval. --- ## Manage sitemaps and indexing with Kiro URL: https://storeseo.com/agent/kiro/sitemaps (full page: https://storeseo.com/agent/kiro/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Kiro means asking the AWS agentic IDE to read StoreSEO's sitemap, noindex and nofollow flags on your Shopify resources, change them one at a time and set up the HTML sitemap, with autoApprove deciding which calls run without a prompt. You work in Kiro's chat panel with the theme project open and the storeseo entry saved in .kiro/settings/mcp.json. Ask which articles are set to nofollow and the agent calls storeseo_start, then storeseo_list_resource_sitemaps for the article resource type, filtered to nofollow on. A good first configuration puts the reads in autoApprove: storeseo_start and storeseo_list_resource_sitemaps, plus storeseo_get_html_sitemap_settings and storeseo_list_sitemap_menus, so an audit runs without a prompt on each call. Writes stay out of that list. A noindex flag removes a page from search results, so StoreSEO tells the agent to name the resource and the flags and wait for your yes before storeseo_update_resource_sitemap, and with the tool left out of autoApprove Kiro's own prompt is a second check on the same call. The Kiro docs do not state what the prompt does when autoApprove is empty, so confirm the behavior on your first write rather than assuming it. After a change, ask the agent to list the resource again, since the flags can also be edited in the StoreSEO app. The HTML sitemap then follows storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap and storeseo_list_sitemap_menus. The workspace file is useful for a team: committing a .kiro/settings/mcp.json with the read tools in autoApprove gives everyone the same safe defaults, as long as no token sits in it. The menu edit links come back from the list call and you add the sitemap link in Shopify admin yourself. Everything in this job works on any active StoreSEO plan. Prompts: - Use StoreSEO to list my articles that are nofollow or excluded from the sitemap. Read only. - Exclude the page Staff Login from the sitemap and set it to noindex. Name the resource and wait for my yes. - Check my HTML sitemap settings with StoreSEO, generate the page if it is missing and list the menus for the link. In Kiro: Kiro's autoApprove list is the setting to get right for this job. List only the read tools, such as storeseo_list_resource_sitemaps and storeseo_get_html_sitemap_settings, and keep storeseo_update_resource_sitemap and storeseo_update_html_sitemap out of it, because anything in the list runs without Kiro asking and a wildcard would include the noindex write. A workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, so if the agent seems to read the wrong store, check which file defines storeseo. Q: Which StoreSEO sitemap tools are safe to put in Kiro's autoApprove? A: The read tools: storeseo_start, storeseo_list_resource_sitemaps, storeseo_get_html_sitemap_settings and storeseo_list_sitemap_menus. Leave the two update tools out so Kiro's own prompt stays a second check on a noindex or sitemap change. Q: Can Kiro hide the sitemap write tools completely? A: Yes. Kiro's disabledTools list hides tools from the agent. Add storeseo_update_resource_sitemap and storeseo_update_html_sitemap for a read-only audit session and remove them when you want to write. --- ## Manage sitemaps and indexing with Antigravity URL: https://storeseo.com/agent/antigravity/sitemaps (full page: https://storeseo.com/agent/antigravity/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Antigravity means asking Google's agentic IDE to read StoreSEO's sitemap, noindex and nofollow flags on your Shopify resources, change them one at a time with Ask mode in front of each call and set up the HTML sitemap. You work in the agent panel with the theme project open and the storeseo server added to mcp_config.json with the serverUrl field. Ask which products are set to noindex and the agent calls storeseo_start, then storeseo_list_resource_sitemaps for the product resource type, filtered to noindex on. Antigravity runs MCP tools you have not configured in Ask mode, which requires your approval before execution, so each call arrives as a prompt in the panel, reads included. The answer is a list of titles with their three flags. Because Ask mode is the default, a write passes two checks. StoreSEO tells the agent to name the resource and the flags and wait for your yes before storeseo_update_resource_sitemap, since a noindex flag removes a page from search results, and Antigravity then asks again before it runs the call. If the prompts for reads become tedious on a large store, a policy pattern of the form mcp(storeseo/storeseo_list_resource_sitemaps) lets you allow just that tool and leave the update tools in Ask mode. Afterwards ask the agent to list the resource again, because the flags can also be edited in the StoreSEO app. The HTML sitemap runs through storeseo_get_html_sitemap_settings, storeseo_update_html_sitemap and storeseo_list_sitemap_menus. The agent hands you the menu edit links and you add the sitemap link in Shopify admin; no tool does it, and the agent can then mark setup complete. If the connection was set up with the sign-in code flow, remember that the authorization code is pasted back in Agent settings, under Customizations. These tools work on any active StoreSEO plan. Prompts: - Use StoreSEO to list my collections and pages that are excluded from the sitemap or set to noindex. Read only. - Set the product Gift Card Test to noindex and exclude it from the sitemap. State the resource and flags, then wait for my yes. - Check my HTML sitemap with StoreSEO and set it up if it is missing. Then list my menus and show me the edit link. In Antigravity: Antigravity's remote field is serverUrl, and its documentation says legacy fields such as url or httpUrl are not supported, so an entry copied from another client will not connect until you rename it. Unconfigured MCP tools run in Ask mode, so a sitemap audit means a prompt per call until you add a policy pattern such as mcp(storeseo/storeseo_list_resource_sitemaps) for the reads. Do not allow a wildcard for the whole server: it would let the noindex write run without Antigravity asking, and StoreSEO's confirmation would be the only check left. Q: Will Antigravity ask before StoreSEO sets noindex on a page? A: By default yes. MCP tools you have not configured run in Ask mode, which requires approval before execution, and StoreSEO also tells the agent to name the resource and wait for your yes. Q: Why does my StoreSEO entry not load in Antigravity? A: Check the field name. Antigravity uses serverUrl for remote servers and its documentation says url and httpUrl are not supported, so an entry copied from another client needs that one field renamed. --- ## Track keyword rankings with Claude URL: https://storeseo.com/agent/claude/keyword-tracking (full page: https://storeseo.com/agent/claude/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Claude means telling Anthropic's chat assistant which searches matter to your Shopify store and letting it use StoreSEO's rank tools to add them, read where you appear on Google and compare you with competitors. You work in a Claude chat tab, on the web or in Claude Desktop, with the StoreSEO custom connector switched on for that conversation. A message such as track linen shirt for men in the United States on mobile makes Claude call storeseo_start, then storeseo_list_rank_locations and storeseo_list_rank_languages, because the tracker only accepts the exact country and language names it lists. Those are lookups, so they run straight away. Claude then tells you the keyword, country, language and device it intends to add and waits for your reply before it calls storeseo_add_keywords. Positions do not come back in the same message. StoreSEO queues the first check in the background, so a keyword you added a minute ago shows as pending in storeseo_list_keywords. The better chat questions are about keywords you already track: how have they moved this month, which dropped out of the top ten, where does a named competitor sit on each. Claude reads storeseo_list_keywords for the latest position, change and SERP features, then storeseo_get_keyword_history for the dated checks behind one keyword, and writes the answer as a short table in the chat. Keyword clusters fit a chat well because you can read the result as it lands. Name a product with a focus keyword and Claude calls storeseo_generate_keyword_cluster only after it has told you the estimated AI credit and you have agreed. It then polls storeseo_get_keyword_cluster until the background job finishes, and presents the related keywords in the order StoreSEO returns them, highest search volume first, with relevance and difficulty beside each. Rank and competitor tools are Pro plan tools, so on another plan Claude should explain the upgrade instead of calling them. Prompts: - Use StoreSEO to track organic cotton tote bag in Canada, English, on mobile. Tell me what you will add before you add it. - How have my tracked keywords moved this month? Show the biggest drops first and where my competitors rank. - Build a keyword cluster for my best-selling candle. Tell me the estimated credit first and wait for my yes. In Claude: Claude's own prompts for write actions are the hard stop here, and Anthropic advises using Allow always only for servers you trust. For rank tracking that matters most for storeseo_delete_keyword and storeseo_delete_competitor: deleting a keyword erases its entire rank history and cannot be undone, so keep those tools on ask even if you let the read tools run freely. Also check that the StoreSEO connector is switched on in the Search and Tools menu for the conversation, or Claude will not see the rank tools at all. Q: Why does Claude say my new keyword has no position yet? A: StoreSEO checks ranks in the background. When Claude adds a keyword the first check is only queued, so the position appears in the keyword list after that check completes. Ask Claude to list your keywords again a little later. Q: Does Claude need a Pro StoreSEO plan to track keywords? A: The rank and competitor tools are Pro plan tools, so Claude can only add keywords and read rank history on Pro. On another plan Claude should explain the upgrade. Keyword clusters are separate and use AI Content Optimizer credit. --- ## Track keyword rankings with ChatGPT URL: https://storeseo.com/agent/chatgpt/keyword-tracking (full page: https://storeseo.com/agent/chatgpt/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with ChatGPT means using the StoreSEO app you created in developer mode to ask where your Shopify store ranks on Google, add new keywords to track and compare against competitor domains from a ChatGPT conversation. You are in a ChatGPT conversation with the StoreSEO app attached, and the first thing ChatGPT does with a request like how are my tracked keywords doing is call storeseo_start. Reading is the comfortable part of this job in ChatGPT: storeseo_list_keywords returns each keyword's latest Google position, change, SERP features and competitor positions, and storeseo_get_keyword_history returns the dated checks behind one of them. Ask for the keywords that fell out of the top ten, or for a comparison by country, and ChatGPT turns the rows into a table. Adding keywords is where ChatGPT's developer mode shows. Before storeseo_add_keywords, ChatGPT first calls storeseo_list_rank_locations and storeseo_list_rank_languages so the country and language match the tracker's list exactly, and StoreSEO tells it to confirm the keyword, market and device with you. Then ChatGPT's own write-action approval appears, and you can inspect the payload before you allow it. The approval can be remembered for the rest of the conversation, which is convenient for a batch of keywords but deserves care, because the same remembered approval would cover a delete later in the chat. Competitor tracking and clusters follow the same pattern. storeseo_add_competitors takes up to five domains in total, and there is no edit: to swap one, the agent deletes it with storeseo_delete_competitor and adds the new domain, after your yes on each. A cluster for a product starts with the credit estimate stated to you, then ChatGPT polls storeseo_get_keyword_cluster until the background job ends and lists related keywords by search volume. Rank and competitor tools need the Pro plan. Prompts: - Using the StoreSEO app, show my tracked keywords with their latest Google position and what changed since the last check. - Add example-competitor.com as a competitor. Tell me how many of the five slots that uses before you add it. - Show the rank history for organic cotton tote since July and say whether the drop came from a competitor moving up. In ChatGPT: Write actions depend on your ChatGPT plan. OpenAI's developer guide lists developer mode for several plans, while OpenAI's help center, as we read it, gives Pro accounts read access only and puts write actions in beta on Business, Enterprise and Edu workspaces. On a read-only setup ChatGPT can still read ranks, history and competitor positions, but it cannot add or delete keywords, so do those from the StoreSEO app or another agent. Be careful with the option that remembers approval for the conversation before any delete. Q: Can ChatGPT delete a tracked keyword in StoreSEO? A: Yes, with the storeseo_delete_keyword tool, when your plan allows write actions. It removes the keyword and its whole rank history for good, so read the approval payload before you allow it and avoid remembering approval for the conversation. Q: Why does ChatGPT list countries before adding a keyword? A: The tracker accepts only exact country and language names from its own lists, so ChatGPT looks them up with the rank location and language tools instead of typing them from memory. That avoids a keyword being rejected. --- ## Track keyword rankings with Claude Code URL: https://storeseo.com/agent/claude-code/keyword-tracking (full page: https://storeseo.com/agent/claude-code/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Claude Code means asking Anthropic's terminal agent to manage your StoreSEO keyword list, read Google positions and history, compare competitors and write the result into a file beside your theme code. You start Claude Code in your theme folder or any working directory and type the request in the terminal, for example add linen shirt and linen shirt for men for the United States on mobile. Claude Code calls storeseo_start, looks up the market with storeseo_list_rank_locations and storeseo_list_rank_languages, and prints the exact keywords, country, language and device it plans to add. Each tool call scrolls past as it runs, and Claude Code's own permission prompt appears before storeseo_add_keywords unless you have allowed that tool. StoreSEO also tells the agent to wait for your yes in the conversation. The terminal is a good place to turn rank data into something you keep. After the first checks finish, ask Claude Code to read storeseo_list_keywords and write rank-report.md in the repository: keyword, market, latest position, change, SERP features and where each competitor sits. Later sessions can compare against that file, and storeseo_get_keyword_history supplies the dated checks when one keyword needs a closer look. Because positions are checked in the background, a keyword added this morning may still be pending when you ask, and the report should say so rather than guess. Keyword clusters work in the same flow. Claude Code names the product, states the estimated AI credit from storeseo_generate_keyword_cluster, waits for your yes and then polls storeseo_get_keyword_cluster until the background job completes, which takes a couple of minutes. The cluster comes back sorted by search volume with relevance and difficulty, so you can paste the best terms into a content brief in the same folder. Rank and competitor tools need the Pro plan. Prompts: - Use StoreSEO to list my tracked keywords and write rank-report.md here with position, change and competitor positions. Do not add or delete anything. - Track linen shirt for men in the United States on mobile. Show me the exact country and language names you will use first. - Which tracked keywords dropped out of the top ten? Pull the history for each and summarize the dates it fell. In Claude Code: Claude Code warns when a single tool result passes 10,000 tokens, and a keyword listing that carries competitor positions for every check can get there on a store tracking many keywords. Ask for a page at a time, filter by country or device, or search for one keyword instead of listing everything. If you added an allow rule for the StoreSEO tools to cut prompts, it also covers storeseo_delete_keyword, which erases a keyword's rank history permanently, so allow the read tools by name and keep the delete tools on ask. Q: Can Claude Code write my keyword rankings to a file? A: Yes. Claude Code works in the folder you started it in, so it can read your tracked keywords and history through StoreSEO and write a Markdown or CSV report there. Ask for a fresh report in a later session to compare against it. Q: Why is my Claude Code keyword listing so large? A: Each keyword row includes the latest position, change, SERP features and competitor positions. On a big list the result can pass Claude Code's 10,000 token warning, so ask for smaller pages or filter by country, device or search term. --- ## Track keyword rankings with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/keyword-tracking (full page: https://storeseo.com/agent/claude-cowork/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Claude Cowork means connecting StoreSEO as a custom connector so Anthropic's desktop work agent can read your Google positions and competitor comparisons, and run a report-only rank check on a schedule. You give Cowork a task in plain words, for instance review how my tracked keywords moved this week and flag anything that left the top ten. Cowork calls storeseo_start, reads storeseo_list_keywords for the latest position, change, SERP features and competitor positions, and calls storeseo_get_keyword_history for any keyword that moved. The result is a written summary you can read, edit and reuse. This is the part of keyword tracking that suits Cowork best, because it is repeatable, read-only and needs no decision from you in the middle. Changing the keyword list is different. When you ask Cowork to track a new term, it looks up the exact market with storeseo_list_rank_locations and storeseo_list_rank_languages, says what it will add and waits for your yes before storeseo_add_keywords, which counts against your plan's keyword limit. Deleting a keyword or competitor erases its history for good, so do that only while you are in the conversation. Rank and competitor tools need the Pro plan, and a keyword added now shows its first position only after the background check completes. Cowork documents scheduled tasks that run on a cadence and can use connected apps, which makes a Monday rank summary easy to set up. A scheduled run has nobody in the conversation to say yes, so write the task as read-only and tell it not to add, delete or generate anything. A cluster, which spends AI credit and waits for your yes on the estimate, belongs in a task you start yourself rather than in a schedule. Prompts: - Use StoreSEO to summarize how my tracked keywords moved this week, with competitor positions. Do not change anything. - Every Monday, read my tracked keywords and write a short rank summary. Do not add, delete or generate anything. - Track organic cotton tote bag in the United Kingdom, English, on desktop. Show me what you will add and wait for my yes. In Claude Cowork: A scheduled Cowork task runs without you in the conversation, so nobody is there to approve a write. Keep a scheduled StoreSEO task to reads such as storeseo_list_keywords and storeseo_get_keyword_history, and say in the prompt that it must not add keywords, delete anything or build a cluster. Scheduled tasks cannot be tied to a local folder either, so the summary lands in Cowork rather than a file on your machine. If StoreSEO is missing in Cowork, look under Customize, Connectors there. Q: Can Claude Cowork check my keyword rankings on a schedule? A: Anthropic documents scheduled tasks in Cowork that use connected apps. Make the StoreSEO task read-only, so it lists keywords and history and writes a summary, and tell it not to add or delete keywords because nobody is there to say yes. Q: Will a scheduled Cowork task add keywords for me? A: It should not. Adding keywords counts against your keyword limit and StoreSEO tells the agent to wait for your yes, which a scheduled run cannot give. Add keywords in a conversation you are present in. --- ## Track keyword rankings with Cursor URL: https://storeseo.com/agent/cursor/keyword-tracking (full page: https://storeseo.com/agent/cursor/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Cursor means using the editor's agent panel with the StoreSEO MCP server to add Google keywords, read your positions and competitors, and keep the findings next to your storefront code. You are in Cursor with a theme or storefront project open, and you ask the agent in plain words to track a keyword or report on the ones you follow. The agent calls storeseo_start, then storeseo_list_rank_locations and storeseo_list_rank_languages so the country and language are the exact names the tracker accepts, and it shows you what it intends to add. Cursor asks for approval before it uses an MCP tool by default, and StoreSEO also tells the agent to wait for your yes before storeseo_add_keywords, which counts against your plan's keyword limit. Reading ranks is where the editor helps. storeseo_list_keywords returns each keyword's latest Google position, change, SERP features and competitor positions, and storeseo_get_keyword_history gives the dated checks. Ask Cursor to put the result in a notes file in the project, or to compare tracked keywords with the titles and meta in your theme files and point at pages that target a term you rank poorly for. Positions are checked in the background, so a keyword added a moment ago shows as pending until its first check completes. Competitors and clusters follow the same loop. storeseo_add_competitors accepts up to five domains in total, and changing one means deleting it and adding the new domain, each after your yes. For a cluster, the agent states the estimated AI credit, waits, calls storeseo_generate_keyword_cluster and polls storeseo_get_keyword_cluster for a couple of minutes, then lists related keywords by search volume. Rank and competitor tools need the Pro plan, and in Cursor there is one more thing to check first: whether those tools are switched on. Prompts: - Use StoreSEO to list my tracked keywords and write rank-notes.md in this project with position, change and competitors. - Track linen shirt for men in the United States on mobile. Confirm the exact country and language names first. - Build a keyword cluster for my top product and tell me the estimated credit before you run it. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you were told to switch off at least four. If you switched off the rank and competitor lookups, storeseo_add_keywords or storeseo_list_rank_locations may be among them, and the agent will say it cannot track keywords even though you are on Pro. Re-enable the rank tools in Cursor's MCP settings and switch off tools for jobs you do not use, such as image or sitemap tools, instead. Q: Why can Cursor not add keywords even though StoreSEO is connected? A: Cursor allows about 40 active tools, so you may have switched off some of StoreSEO's 44. Check that the rank location, rank language and add keywords tools are enabled in Cursor's MCP settings, and that your StoreSEO plan is Pro. Q: Can Cursor compare my tracked keywords with my theme files? A: Yes. Cursor reads your project, and StoreSEO supplies positions through storeseo_list_keywords, so the agent can point at templates or content where a term you rank poorly for is missing. It cannot edit keywords in StoreSEO without your yes. --- ## Track keyword rankings with Codex URL: https://storeseo.com/agent/codex/keyword-tracking (full page: https://storeseo.com/agent/codex/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Codex means asking OpenAI's coding agent to use StoreSEO's rank tools from the CLI, the IDE extension or the ChatGPT desktop app to add keywords, read Google positions and compare competitors. You run Codex in a terminal or editor, with StoreSEO configured as an mcp_servers table in config.toml, and ask for what you want in a sentence: track linen shirt for men in the United States on mobile. Codex calls storeseo_start, then storeseo_list_rank_locations and storeseo_list_rank_languages, and reports the exact names it will use. Whether Codex pauses before storeseo_add_keywords depends on your approval mode, and StoreSEO separately tells the agent to wait for your yes. The keyword counts against your plan's keyword limit. Codex lets you set approval per tool, which suits this job. Reads such as storeseo_list_keywords and storeseo_get_keyword_history can run without a prompt while storeseo_add_keywords, storeseo_add_competitors and the two delete tools stay on prompt. Deleting a keyword removes its entire rank history and cannot be undone, and deleting a competitor removes its comparison results, so those two deserve the stricter setting. Ask Codex to write the report into the repository, for example ranks.md with position, change, SERP features and competitor positions, and diff it against last week's file to see what moved and which competitor gained ground. A cluster is the one call where timing matters. The agent states the estimated AI credit and waits for your yes, calls storeseo_generate_keyword_cluster, and polls storeseo_get_keyword_cluster until the background job finishes, which takes a couple of minutes. Rank and competitor tools need the Pro plan, and positions for newly added keywords appear only after the background check completes, so a report written straight after adding them will show them as pending. Ask for a later refresh once the checks complete. Prompts: - Use StoreSEO to list my tracked keywords and write ranks.md here. Read only, add nothing. - Add linen shirt for men for the United States on mobile. Show me the exact location and language names before you add it. - Show the rank history for my worst-moving keyword and say which competitor moved up on it. In Codex: Set the storeseo server's default_tools_approval_mode to writes or prompt, not approve, and give storeseo_delete_keyword and storeseo_delete_competitor their own tools..approval_mode of prompt. With approve, Codex skips its prompt and the confirmation StoreSEO asks for in chat is the only check left, while a deleted keyword loses its whole rank history. Codex also waits 60 seconds for a tool by default, so raise tool_timeout_sec if a call times out. Q: Which Codex approval setting suits StoreSEO keyword tracking? A: Use writes or prompt as the server default and keep the delete tools on prompt. Codex then asks before adding or deleting keywords and competitors. Approve skips the prompt, which leaves only the confirmation StoreSEO asks for in chat. Q: Does Codex work in the IDE extension for rank tracking too? A: Yes. The Codex CLI, the IDE extension and the ChatGPT desktop app share the same configuration, so a StoreSEO server in config.toml is available in each. Rank and competitor tools still need the Pro plan. --- ## Track keyword rankings with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/keyword-tracking (full page: https://storeseo.com/agent/gemini-cli/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Gemini CLI means asking Google's terminal agent to call StoreSEO's rank tools so it can add keywords, read your Google positions and history and compare competitor domains for your Shopify store. You run gemini in a terminal and describe the job in a sentence: track organic cotton tote bag in Canada, English, on desktop. Gemini CLI calls storeseo_start, then storeseo_list_rank_locations and storeseo_list_rank_languages, because the tracker takes only the exact names in its lists. Gemini CLI asks for confirmation before it runs an MCP tool unless the server is trusted, and StoreSEO tells the agent to state the keyword, market and device and wait for your yes before storeseo_add_keywords. The keyword then counts against your plan's keyword limit. Reading is the everyday use. Ask which tracked keywords moved and Gemini CLI reads storeseo_list_keywords for the latest position, change, SERP features and competitor positions, then storeseo_get_keyword_history for the dated checks behind any keyword you care about. Because the sign-in needs a browser on the same machine, a headless box or an SSH session needs the access token header instead. Ask for the summary to be saved in the folder you launched from, so a later session can compare against it. Competitors and clusters work as in any client. storeseo_add_competitors takes up to five domains in total, with no edit: the old domain is deleted and the new one added, each after your yes, and each delete erases that competitor's comparison results. For a cluster, the agent states the estimated AI credit, waits, calls storeseo_generate_keyword_cluster and then polls storeseo_get_keyword_cluster until the background job completes. Rank and competitor tools need the Pro plan, and new keywords show a position only after their first background check. Prompts: - Use StoreSEO to list my tracked keywords with position and change, and save a summary to ranks.md here. Add nothing. - Track linen shirt for men in the United States on mobile. Show me the exact location and language you will use first. - Delete nothing, but tell me which tracked keywords have not left position 20 in their whole history. In Gemini CLI: Do not set --trust or trust: true on the storeseo server. Trust skips every tool confirmation in Gemini CLI, so the confirmation StoreSEO asks for in chat would be the only check before storeseo_delete_keyword, which erases a keyword's whole rank history for good. If you want fewer prompts, trim the server with includeTools or excludeTools instead. Google's site also says Gemini CLI was replaced by Antigravity CLI on June 18th, 2026 for unpaid tier and Google One users, so check which agent your account can use. Q: Why does Gemini CLI keep asking before it uses a StoreSEO rank tool? A: Gemini CLI confirms MCP tool calls unless the server is marked trusted. Leave trust off for StoreSEO, so the prompt stands between the agent and any keyword add or delete, and use includeTools if you want a smaller tool list. Q: Can Gemini CLI track keywords over SSH? A: Gemini CLI's browser sign-in needs a browser on the same machine, so over SSH without one use a personal access token from StoreSEO under Settings, AI Toolkit, sent with the --header option. Rank tools still need the Pro plan. --- ## Track keyword rankings with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/keyword-tracking (full page: https://storeseo.com/agent/github-copilot/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with GitHub Copilot means asking Copilot Chat in agent mode, inside VS Code, to call StoreSEO's rank tools to add keywords, read your Google positions and compare competitors while your theme code stays open. You are in VS Code with the storefront or theme project open, Copilot Chat in Agent mode, and StoreSEO started from MCP: List Servers. You ask for a keyword to be tracked or a report on the ones you follow, and Copilot calls storeseo_start, then storeseo_list_rank_locations and storeseo_list_rank_languages so the country and language match the tracker's lists exactly. VS Code shows a tool confirmation before an MCP tool runs, and StoreSEO tells the agent to state the keyword, market and device and wait for your yes before storeseo_add_keywords. The editor makes the reading side practical. Copilot calls storeseo_list_keywords for the latest position, change, SERP features and competitor positions, and storeseo_get_keyword_history for the dated checks behind one keyword. Ask it to write the result to a notes file in the workspace, or to check your collection and product templates for the terms you rank poorly for. A keyword added a moment ago shows as pending, because StoreSEO queues the first rank check in the background and the position appears in the list later. Competitors and clusters follow the same loop. storeseo_add_competitors accepts up to five domains in total, and a change means deleting the old domain and adding the new one, each after your yes. A cluster starts with the estimated AI credit being stated to you, then storeseo_generate_keyword_cluster, then polling storeseo_get_keyword_cluster until the background job finishes, and the related keywords come back sorted by search volume. Rank and competitor tools need the Pro plan, and on Copilot Business and Enterprise an administrator must enable the MCP servers in Copilot policy first. Prompts: - Use StoreSEO to list my tracked keywords and write rank-notes.md in this workspace. Do not add or delete anything. - Track linen shirt for men in the United States on mobile. Confirm the exact location and language names first. - Which of my tracked keywords dropped out of the top ten? Show the history for each. In GitHub Copilot: Keep storeseo_add_keywords, storeseo_add_competitors and the two delete tools on VS Code's tool confirmation prompt. StoreSEO asks for your yes in the chat, and the editor's confirmation is the hard stop, which matters because deleting a keyword removes its whole rank history and cannot be undone. On Copilot Business and Enterprise plans the MCP servers in Copilot policy must be enabled by an organization admin, or the rank tools will not appear in agent mode. Q: Why do I not see StoreSEO rank tools in Copilot agent mode? A: Check that the server is started in MCP: List Servers and that you are in Agent mode. On Copilot Business and Enterprise an organization admin must also enable the MCP servers in Copilot policy. Rank tools also need a Pro StoreSEO plan. Q: Can Copilot track keywords and edit my theme in the same session? A: Yes. Copilot reads your workspace and calls StoreSEO for positions, so it can suggest where a term belongs in your templates. Adding or deleting keywords still waits for your yes, and VS Code's confirmation is the hard stop. --- ## Track keyword rankings with Windsurf URL: https://storeseo.com/agent/windsurf/keyword-tracking (full page: https://storeseo.com/agent/windsurf/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Windsurf means asking the Cascade agent in your editor to add keywords to StoreSEO's Google rank tracker, read their latest positions and history, and compare them with up to five competitor domains while your Shopify theme stays open beside it. You sit in the Cascade panel with a theme repository open, and you type something like track linen shirts and linen shirt for men in the United States on mobile. Cascade calls storeseo_start first, then storeseo_list_rank_locations and storeseo_list_rank_languages, because the tracker only accepts the exact country and language names from those lists. It shows you the proposed keyword rows, with country, language and device spelled out, and waits for your yes before it calls storeseo_add_keywords. The editor shows each tool call in the chat, so you can see which lookup produced which name. Adding a keyword only queues the first Google check. Positions arrive in the background, so a keyword added a minute ago may show no rank yet, and Cascade should say so instead of guessing. Later you ask how the tracked keywords moved, and Cascade reads storeseo_list_keywords for the latest position, the change, SERP features and competitor positions, then storeseo_get_keyword_history for a dated trail on one keyword. Because the work happens in the editor, you can ask Cascade to put a keyword's history next to the product copy it is editing, or to write the movers into a notes file in the project. Competitors work the same way. storeseo_list_competitors shows the domains you track, and storeseo_add_competitors takes new ones after your yes, up to five in total. Rank and competitor tracking are Pro plan tools, so on another plan Cascade should explain the upgrade instead of calling them. Everything here checks Google only; AI answer engines are not tracked by this job. Prompts: - Use StoreSEO to look up the exact country and language names for rank tracking in Germany and German, then show me the keyword rows you would add before you add anything. - List my tracked keywords sorted by overall position, and flag any that dropped since the last check. Read only. - Read the rank history for organic cotton tote and write a short note about its trend into rank-notes.md in this project. In Windsurf: Cascade can use at most 100 MCP tools at once across every server, and StoreSEO uses 44 of them, so a heavy MCP setup can leave no room for the rank tools. Check the MCPs section and disable what you do not need. Windsurf's MCP page also says its mcp_config.json applies to the legacy Cascade agent, while the Devin Local agent reads other files, so confirm which agent you are in. The page does not describe Cascade's tool approval, so treat a keyword delete, which erases that keyword's whole rank history, as something to read carefully before you say yes. Q: Why does Cascade say my new keyword has no position yet? A: Adding a keyword with storeseo_add_keywords only queues the first Google check. The position appears in the keyword list when that background check completes, so ask Cascade to read the list again a little later. Q: Can Windsurf track keywords on a free StoreSEO plan? A: The MCP server connects on any active plan, but the rank and competitor tools are Pro plan tools. On other plans Cascade should explain that instead of calling them. --- ## Track keyword rankings with Grok URL: https://storeseo.com/agent/grok/keyword-tracking (full page: https://storeseo.com/agent/grok/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Grok means chatting with xAI's assistant on grok.com, through a custom StoreSEO connector, to add keywords to the Google rank tracker, read how they move and compare them with competitor domains in plain language. You open a chat with the StoreSEO connector available and ask where your store ranks for the keywords you care about. Grok opens a StoreSEO session with storeseo_start and lists your tracked keywords with storeseo_list_keywords, which returns the latest Google position, the change since the previous check, SERP features and the positions of your competitors. Everything comes back as chat text, so a good prompt asks for a short table or a ranked list of the biggest drops, which is easier to read in a conversation than a raw dump. To track something new you describe the market the way a merchant would, for example shopify linen shirts for the United States on mobile. Grok calls storeseo_list_rank_locations and storeseo_list_rank_languages to find the exact names the tracker accepts, repeats the keyword rows back to you and, after your yes, calls storeseo_add_keywords. The first check runs in the background, so a new keyword shows no position until it finishes. For one keyword's trail over time you ask for its history, and Grok reads storeseo_get_keyword_history, which returns dated positions with the competitor positions at each check. Competitor comparison is a conversation too: storeseo_list_competitors shows the domains you track, and a new domain is added with storeseo_add_competitors after you agree, up to five in total. There is no edit, so replacing a domain means deleting it and adding the new one. Rank and competitor tracking need the Pro plan, and the tracker checks Google only, not AI search engines. Grok does not need anything open on your side because the connector reaches StoreSEO over the public internet. Prompts: - Using StoreSEO, which of my tracked keywords moved the most this month, and where do my competitors rank for them? - I sell linen shirts. Find the exact country and language names for the United States and English, then propose three keywords to track on mobile and wait for my yes. - Show the rank history for organic cotton tote and tell me if the trend is up or down. In Grok: xAI's connector documentation does not describe a per-tool approval setting, so the only check on a change is the yes Grok asks you for in the conversation. That matters most for storeseo_delete_keyword and storeseo_delete_competitor, because deleting a keyword permanently erases its entire rank history. Read what Grok proposes before you agree. On Grok Business and Enterprise a team admin has to provision the connector in the cloud console first, so if the tools do not appear, ask your admin. This page covers the Grok chat connector, not xAI's terminal agent. Q: Can Grok check my Google rankings through StoreSEO? A: Yes, once the StoreSEO connector is added and you are on the Pro plan. Grok reads your tracked keywords with their latest Google position and change, and can read the dated history of any one keyword. Q: Does Grok track how my store appears in AI search answers? A: No. StoreSEO's rank tracker checks Google results by country, language and device. AI visibility is handled by the GEO score, llms.txt and agents.md jobs instead. --- ## Track keyword rankings with OpenClaw URL: https://storeseo.com/agent/openclaw/keyword-tracking (full page: https://storeseo.com/agent/openclaw/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with OpenClaw means asking your self-hosted agent, from WhatsApp, Telegram, Slack or its Control UI, to manage StoreSEO's Google rank tracker, and optionally letting an automation read your rankings on a schedule and post the movers to a chat. OpenClaw runs on your own machine or server and you talk to it in a chat app, so keyword tracking becomes a message. You write track shopify linen shirts in the United States on mobile, and the agent opens a session with storeseo_start, looks up the accepted names with storeseo_list_rank_locations and storeseo_list_rank_languages and replies with the exact rows it would add. After you answer yes in that same chat it calls storeseo_add_keywords. Checks run in the background, so the first positions appear a little later in storeseo_list_keywords. The part that sets OpenClaw apart is its scheduler. An automation, which openclaw cron also reaches, can wake the agent every Monday in an isolated session, read storeseo_list_keywords for the latest positions and changes, read storeseo_get_keyword_history for the keywords that moved, and announce the summary in a Telegram or Slack channel. That is a ranking report that arrives without you opening anything. Competitor positions come along in the same list, and you can list the domains you track with storeseo_list_competitors. Keep the scheduled prompt to reads. Nobody is there to say yes at eight on a Monday, so the job should never add, delete or generate anything; changes such as a new keyword or a new competitor domain belong in a message you can answer. Rank and competitor tracking are Pro plan tools and the tracker checks Google only. Prompts: - Use StoreSEO to list my tracked keywords and tell me which three dropped the most since the last check. Read only. - Create a Monday 8am automation that reports my keyword position changes and competitor positions to this Telegram chat. Reads only: do not add, delete or generate anything. - Track shopify linen shirts for the United States on mobile. Look up the exact names first and ask me before you add it. In OpenClaw: Run the openclaw mcp commands on the installation that owns the connector; a Gateway or agent on another host needs its own reload or restart before it sees the storeseo server. OpenClaw has a per-server approval setting, set with openclaw mcp configure storeseo --approval, and the add and delete tools are writes, so do not leave it on a mode that skips the check if you want the client to stop before them. A scheduled automation cannot wait for a yes, so write it as report-only and never let it call storeseo_delete_keyword, which erases a keyword's rank history for good. Q: Can OpenClaw send me a weekly keyword ranking report? A: Yes. An OpenClaw automation can run on a schedule in an isolated session, read your tracked keywords and their changes through StoreSEO and announce the result in a chat channel such as Telegram. Keep the prompt to reads. Q: Can a scheduled OpenClaw job add or delete keywords? A: It should not. A scheduled run has nobody to say yes, and deleting a keyword removes its whole rank history permanently. Ask for adds and deletes in a chat message you can answer. --- ## Track keyword rankings with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/keyword-tracking (full page: https://storeseo.com/agent/hermes-agent/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Hermes Agent means using Nous Research's open-source agent, from a terminal or a chat app, to manage StoreSEO's Google rank tracker, with its memory carrying your keyword plan across sessions and hermes cron delivering a ranking report on a schedule. You run Hermes in a terminal or reach it from Telegram, Discord or Slack, and you describe the keywords you want to follow. Hermes opens a StoreSEO session with storeseo_start, calls storeseo_list_rank_locations and storeseo_list_rank_languages to get the exact country and language names, and tells you the rows it would add, for example linen shirt for United States, English, desktop. When you answer yes it calls storeseo_add_keywords. The first Google check is queued in the background, so a fresh keyword has no position until storeseo_list_keywords shows one. Hermes keeps memory across sessions, which suits a job that unfolds over weeks. You can tell it which keywords you care about and why, and in a later session ask how they moved. It reads storeseo_list_keywords for the latest position, change, SERP features and competitor positions, and storeseo_get_keyword_history for the dated trail of one keyword. storeseo_list_competitors shows the domains you track, up to five, and a new one goes in through storeseo_add_competitors after your yes. For a standing report, hermes cron create takes a schedule and a prompt and delivers the answer to a chat target such as Telegram. Write that prompt as reads only, because a scheduled run has nobody to approve a write. Rank and competitor tracking are Pro plan tools, and the tracker covers Google, not AI search engines. Prompts: - Use StoreSEO to list my tracked keywords and summarize which ones are climbing and which are falling. Remember my top three keywords for next time. - Look up the exact names for Canada and French in the rank tracker, then ask me before adding linen shirt as a mobile keyword. - Schedule a Monday 9am hermes cron job that reports my keyword position changes to Telegram. Reads only, no adds or deletes. In Hermes Agent: Start a new Hermes session or run /reload-mcp after editing config.yaml, otherwise the open session does not see the storeseo server. If you want a read-only rank agent, use Hermes' tools include list to expose only the StoreSEO list and get tools, which keeps storeseo_delete_keyword out of reach. A scheduled hermes cron run cannot wait for a yes, so keep it to reads and ask for adds and deletes in a chat you can answer. If you connect with a personal access token, it sits in config.yaml, so keep that file out of version control. Q: Does Hermes Agent remember my tracked keywords between sessions? A: Hermes keeps memory across sessions, so you can tell it which keywords matter and ask later how they moved. The positions themselves always come from StoreSEO's tracker, read with storeseo_list_keywords. Q: How do I get a weekly ranking report from Hermes? A: Create a hermes cron job with a schedule, a read-only prompt and a delivery target such as Telegram. The job reads your keyword positions and history through StoreSEO and sends the summary to the chat. --- ## Track keyword rankings with OpenCode URL: https://storeseo.com/agent/opencode/keyword-tracking (full page: https://storeseo.com/agent/opencode/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with OpenCode means running the open-source terminal agent in your theme project and asking it to add keywords to StoreSEO's Google rank tracker, read their positions and history, and compare them with competitor domains. You start OpenCode in the project folder and type a request such as show my tracked keywords and which ones fell. The agent calls storeseo_start, then storeseo_list_keywords, which returns each keyword's latest Google position, the change, SERP features and competitor positions. For one keyword it reads storeseo_get_keyword_history, a dated trail with the competitor positions at every check. Because the agent also works on your files, you can ask it to write the movers into a Markdown file or a CSV in the repository and keep a running log next to your theme code. Adding keywords is a short exchange. OpenCode calls storeseo_list_rank_locations and storeseo_list_rank_languages, because the tracker only accepts the exact names those lookups return, then shows you the rows it would add with country, language and device. After your yes in the terminal it calls storeseo_add_keywords. Checks run in the background, so a new keyword shows no position until a later storeseo_list_keywords. Competitors follow the same pattern with storeseo_list_competitors and storeseo_add_competitors, up to five domains. OpenCode hands tool names to the model as the server name plus the tool name, so StoreSEO tools look like storeseo_ followed by the tool, and permission patterns can match them. Rank and competitor tracking are Pro plan tools, and the tracker covers Google only. If your plan lacks them, the agent should say so rather than call the rank tools. Prompts: - Use StoreSEO to list my tracked keywords sorted by overall position and write the five biggest movers to rank-log.md. Read only. - Look up the exact country and language names for the United Kingdom and English, then show me the keyword rows you would add and wait for my yes. - Add example-competitor.com as a competitor, but first list my current competitors and tell me how many of the five slots are used. In OpenCode: OpenCode's documentation describes deny rules for MCP tools but no ask-first mode, so the yes StoreSEO asks the agent for in the terminal is the main check on writes. If you want a client-side stop, deny the delete tools by name with a pattern, because storeseo_delete_keyword permanently removes a keyword and its entire rank history. Also remember that opencode mcp add writes to the project configuration unless you pass --global, so a server added without it only exists in that project, and OpenCode V2 expects servers under mcp.servers. Q: Can I stop OpenCode from deleting tracked keywords? A: Yes. OpenCode permission rules can deny tools by name or pattern, so deny the StoreSEO delete tools if you want a client-side stop. Deleting a keyword erases its whole rank history and cannot be undone. Q: Can OpenCode save my keyword rankings to a file? A: Yes. OpenCode works in the folder you started it in, so you can ask it to write the positions or the biggest movers into a Markdown or CSV file there and update it in a later session. --- ## Track keyword rankings with Zed URL: https://storeseo.com/agent/zed/keyword-tracking (full page: https://storeseo.com/agent/zed/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Zed means asking the Agent Panel in the editor to add keywords to StoreSEO's Google rank tracker, read their positions and history and compare them with competitor domains, with Zed's own tool prompts running alongside. You open the Agent Panel in a Zed window that holds your theme project and ask which tracked keywords moved this month. The agent calls storeseo_start and then storeseo_list_keywords, and by default Zed asks before it runs each tool action, so you click allow for each call. That default is set by agent.tool_permissions.default, which is confirm unless you change it. The result is the latest Google position for each keyword, the change, SERP features and the positions of the competitors you track. To follow a new keyword you describe the market, and the agent calls storeseo_list_rank_locations and storeseo_list_rank_languages to get the names the tracker accepts. It then states the rows it will add, with country, language and device, and asks for your yes before storeseo_add_keywords. Zed shows its own confirmation for that call too, so a write gets two prompts: the agent's question in the panel and Zed's tool approval. The check runs in the background, so a new keyword has no position until storeseo_list_keywords shows one. For history, storeseo_get_keyword_history returns a dated trail for one keyword with competitor positions at each check, and storeseo_list_competitors shows the domains you track, up to five. Rank and competitor tracking are Pro plan tools and the tracker covers Google only. If your plan lacks them, the agent should explain the upgrade in the panel instead of calling the rank tools, and you can keep working on the theme in the same window while the first checks run. Because the Agent Panel sits next to your files, you can ask for the result as a note in the project. A request such as write the keywords that lost ground into rank-notes.md gives you a dated record you can compare against the next review, while the positions themselves always come from StoreSEO rather than from the agent's memory of an earlier chat. Prompts: - Use StoreSEO to list my tracked keywords and tell me which ones lost ground this month. Read only. - Read the rank history for linen shirt and summarize the trend in three sentences. - Find the exact rank tracker names for Australia and English, then ask me before adding organic cotton tote as a desktop keyword. In Zed: Zed asks before every tool action by default, reads included, so a ranking review with several lookups means several prompts. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_list_keywords and mcp:storeseo:storeseo_get_keyword_history, and leave storeseo_add_keywords and storeseo_delete_keyword on confirm. Setting agent.tool_permissions.default to allow removes Zed's check entirely, which leaves only the yes StoreSEO asks the agent for, and deleting a keyword erases its whole rank history, so use allow sparingly. Q: Why does Zed ask me to approve every StoreSEO keyword lookup? A: Zed's agent.tool_permissions.default is confirm by default, so the Agent Panel asks before each tool action, reads included. You can allow specific read tools with per-tool keys and keep the write tools on confirm. Q: Which StoreSEO tools should I leave on confirm in Zed for keyword tracking? A: Leave storeseo_add_keywords, storeseo_add_competitors, storeseo_delete_keyword and storeseo_delete_competitor on confirm. Deleting removes rank history for good, and adding counts against your plan's limits. --- ## Track keyword rankings with Kiro URL: https://storeseo.com/agent/kiro/keyword-tracking (full page: https://storeseo.com/agent/kiro/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Kiro means asking AWS's agentic IDE to manage StoreSEO's Google rank tracker from the chat panel: add keywords, read their positions and history and compare them with competitor domains, with an autoApprove list deciding which calls skip a prompt. You open the Kiro chat panel in the workspace that holds your theme and ask for a ranking review of your tracked keywords. Kiro calls storeseo_start and storeseo_list_keywords, which return each keyword's latest Google position, the change, SERP features and competitor positions, and it can read storeseo_get_keyword_history for the dated trail of any one keyword. Tools not in your autoApprove list make Kiro ask before they run, so a first review may show a prompt per call until you decide which reads to approve. Adding keywords starts with the lookups. Kiro calls storeseo_list_rank_locations and storeseo_list_rank_languages for the exact names the tracker accepts, then lays out the rows, country, language and device, and waits for your yes before storeseo_add_keywords. The first Google check is queued in the background, so a new keyword shows no position until a later list call. Competitors work the same way: storeseo_list_competitors shows what you track, and storeseo_add_competitors adds a domain after your yes, up to five in total. Because Kiro plans multi-step work, you can ask for a full pass in one request, such as look up my market, propose keywords, wait for my yes and then show competitor positions. Rank and competitor tracking are Pro plan tools, and the tracker checks Google only, not AI search engines. If your plan lacks the rank tools, Kiro should explain the upgrade instead of calling them. Keep the market lookups in the loop rather than letting the agent guess a country or language, because a keyword added with a name the tracker does not recognize will not be accepted. Prompts: - Use StoreSEO to list my tracked keywords and group them by whether they went up, down or stayed flat. Read only. - Propose five keywords to track for my linen shirt collection in the United States on mobile, look up the exact names first and wait for my yes before adding any. - Show the rank history for organic cotton tote and the positions of my competitors for it. In Kiro: A workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, so if the tools behave as if they point at the wrong store, check which file defines storeseo. In autoApprove, list individual read tools such as storeseo_list_keywords, storeseo_get_keyword_history, storeseo_list_competitors and the two lookups, and keep storeseo_add_keywords and the delete tools out of it. A wildcard would run storeseo_delete_keyword without a prompt, and that tool erases a keyword's entire rank history for good. Q: Which StoreSEO keyword tools are safe to put in Kiro's autoApprove list? A: The read tools: storeseo_list_keywords, storeseo_get_keyword_history, storeseo_list_competitors, storeseo_list_rank_locations and storeseo_list_rank_languages. Keep the add and delete tools out so Kiro's prompt stays a second check. Q: Can Kiro plan a whole keyword tracking setup in one request? A: Yes. Kiro can look up your market, propose keywords and stop for your yes before adding them. The add call still waits for your answer, and it needs the Pro plan. --- ## Track keyword rankings with Antigravity URL: https://storeseo.com/agent/antigravity/keyword-tracking (full page: https://storeseo.com/agent/antigravity/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Antigravity means asking Google's agentic coding product, from the agent side panel, to add keywords to StoreSEO's Google rank tracker, read their positions and history and compare them with competitor domains while Ask mode gates each tool call. You open the agent side panel in the Antigravity IDE with your theme workspace loaded and ask for a ranking review. The agent calls storeseo_start and storeseo_list_keywords and returns each tracked keyword's latest Google position, its change, SERP features and competitor positions. Tools you have not configured run in Ask mode, which requires your approval before execution, so each call arrives with an approval prompt until you allow the read tools you trust. To track new keywords you describe the market. The agent calls storeseo_list_rank_locations and storeseo_list_rank_languages because the tracker takes only the exact names they return, then states the rows with country, language and device. StoreSEO tells it to wait for your yes, and Ask mode adds its own approval on top, before storeseo_add_keywords runs. Checks are queued and run in the background, so a new keyword has no position until storeseo_list_keywords shows one. For one keyword's trail, storeseo_get_keyword_history returns dated positions with competitor positions. Competitors are listed with storeseo_list_competitors and added with storeseo_add_competitors after your yes, up to five domains. Since Antigravity plans multi-step work, you can ask it to write the ranking movers into a file in the workspace for the next review. Rank and competitor tracking need the Pro plan, and the tracker covers Google only. If your plan lacks them, the agent should explain the upgrade instead of calling the rank tools. Take care with country and language: they must come from the lookup tools, so ask the agent to show you the names it found before it adds anything for a new market. Prompts: - Use StoreSEO to list my tracked keywords and summarize which ones dropped out of the top ten. Read only. - Find the exact rank tracker names for Japan and Japanese, then propose two keywords for my store and ask before adding them. - Read the history of linen shirt and write a short trend note into seo/rank-notes.md in this workspace. In Antigravity: Antigravity's remote server field is serverUrl, and its documentation says legacy fields such as url or httpUrl are not supported, so an entry copied from another client will not connect until renamed. Ask mode is the default for tools you have not configured, and a policy that allows every StoreSEO tool would stop those prompts, leaving only the yes StoreSEO asks the agent for before storeseo_delete_keyword wipes a keyword's rank history. Allow individual read tools such as storeseo_list_keywords instead. Q: Does Antigravity ask before running StoreSEO keyword tools? A: By default, yes. MCP tools you have not configured run in Ask mode, which requires your approval before execution. You can allow specific read tools with a policy and leave the add and delete tools in Ask mode. Q: Why does my StoreSEO server not connect in Antigravity? A: Check that the entry uses serverUrl. Antigravity does not support url or httpUrl for remote servers, so an entry copied from another client will not connect until you rename the field. --- ## Read Search Console and Analytics with Claude URL: https://storeseo.com/agent/claude/google-reports (full page: https://storeseo.com/agent/claude/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading your Google reports with Claude means asking Anthropic's chat assistant a question about clicks, queries or traffic sources and letting it pull the answer from your Search Console and GA4 data through the StoreSEO connector, with no Google tab open. You are in a Claude conversation on the web or in Claude Desktop with the StoreSEO custom connector switched on, and you type a question the way you would ask an analyst: which queries sent the most clicks last month, and how does that compare with the month before. Claude calls storeseo_start first to open the session, then storeseo_get_search_console_insights with the query dimension, and the tool call appears as a collapsible row in the chat. The reply comes back as totals for clicks, impressions, CTR and average position, the top rows ranked by clicks, and the change against the previous period of the same length. Traffic questions go to the second tool. Ask what visitors did and Claude calls storeseo_get_analytics_report with a report type that fits, such as search_traffic for organic Google clicks over time, traffic_source for channels, page_performance for pages or key_events for conversions. Because both tools return structured rows, Claude is good at the next step: it can lay the results out as a table, pick out the queries that moved most, or set a Search Console ranking against what GA4 recorded for the same pages. Both tools are reads, so you can keep asking follow-ups in the same chat without anything being applied to your store. Chat is the right place for the exploratory part of this job. You can ask for a different dimension, a narrower date range or a single country and device without rebuilding a report, and Claude keeps the earlier numbers in the thread so it can refer back to them. When a finding points at a fix, such as a page with plenty of impressions and a weak CTR, Claude can carry on into the meta job in the same conversation and wait for your yes before it applies anything. Prompts: - Which Google queries brought the most clicks in the last 28 days, and how did they change compared with the 28 days before? - Show my top ten pages in Search Console by impressions, with CTR and average position, and flag any with a CTR under two percent. - Pull GA4 traffic by source for last month and tell me which channel grew the most. In Claude: A Claude chat only reaches StoreSEO when the connector is switched on for that conversation, so if the Google tools are missing, check the Search and Tools menu before you assume Google is not connected. On the Free Claude plan StoreSEO uses the one custom connector slot you get, and on Team and Enterprise an owner adds the connector under Organization settings before members can sign in to it. If a report comes back with a message about Google, the fix is in the StoreSEO app, where you connect Search Console, Analytics and the GA4 property on a paid plan, not in Claude. Q: Can Claude read my Search Console and GA4 data through StoreSEO? A: Yes, once Google is connected in the StoreSEO app on a paid plan and a GA4 property is selected. Claude then calls the Search Console and Analytics report tools in the chat and reads the numbers back, comparing each with the previous period. Q: Does asking Claude for a report use AI credit? A: The two Google report tools are read-only reads, so they do not generate content. Credit is only spent when you later ask Claude to generate something, and StoreSEO tells the agent to say what it will cost first. --- ## Read Search Console and Analytics with ChatGPT URL: https://storeseo.com/agent/chatgpt/google-reports (full page: https://storeseo.com/agent/chatgpt/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Pulling Google reports with ChatGPT means asking OpenAI's assistant about your clicks, queries and visitors and letting it read Search Console and GA4 through the StoreSEO app you added in developer mode, then explain the changes in a chat. You open a ChatGPT conversation where the StoreSEO app is enabled and ask for the report in plain words, for example which pages gained impressions this month. ChatGPT calls storeseo_start to open the session, then storeseo_get_search_console_insights with the page dimension and the date range you named, and shows the tool call in the thread so you can open the request and response. The tool returns summary totals, the top rows by clicks and the per-metric change against the previous period, and ChatGPT turns that into a short read of what moved. For visitor behavior it calls storeseo_get_analytics_report. The report type decides the bundle: search_traffic and user_activity return rows by date, while demographics, page_performance, traffic_source and key_events return paginated rows with a totals summary. ChatGPT is comfortable with the follow-up that makes the two sources useful together, such as taking the pages with the most Search Console impressions and checking how GA4 says those pages performed, and it can present the result as a table or a short ranked list. Both Google tools are read-only and carry the read-only annotation, which OpenAI's documentation says ChatGPT uses to tell reads from writes, so a report session should not raise the write confirmation you see for applying content. That makes ChatGPT a fine place to review Google performance even on a plan where write actions are not available to you, and you can take any fix it finds to the StoreSEO app or to an agent that can apply it. Prompts: - Using StoreSEO, show my Search Console clicks and impressions by country for the last 28 days and compare with the previous period. - Which of my product pages have more than a thousand impressions and a CTR below two percent? List them with average position. - Give me the GA4 key events report for last month and tell me which traffic source led to the most of them. In ChatGPT: StoreSEO reaches ChatGPT through developer mode, which OpenAI lists under Settings, Security and login, and which you may not see on some plans or in a workspace where an admin has not enabled it. OpenAI describes the exact plan rules for write actions differently in different places, but reports are reads, so a report-only use of StoreSEO does not depend on write access. The Google tools still need a paid StoreSEO plan with Google connected and a GA4 property chosen in the app, and ChatGPT cannot connect Google for you. Q: Do I need ChatGPT write actions to read Search Console and GA4 reports? A: No. Both Google report tools are reads, so they work for reporting even if your ChatGPT plan only gives read access to connected apps. Applying changes is a separate step and is where write actions matter. Q: Why does ChatGPT say Google is not connected? A: Because the StoreSEO app has no Search Console or GA4 connection for your store yet. Connect Google and pick the GA4 property in the StoreSEO app on a paid plan, then ask ChatGPT again. --- ## Read Search Console and Analytics with Claude Code URL: https://storeseo.com/agent/claude-code/google-reports (full page: https://storeseo.com/agent/claude-code/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with Claude Code means asking Anthropic's terminal agent for Search Console and GA4 numbers through StoreSEO while you work in your theme repository, so the data lands beside the code you are about to change. You are in a terminal with Claude Code started in a theme folder, and you type a request such as which queries lost clicks this month and which pages gained impressions. Claude Code calls storeseo_start, then storeseo_get_search_console_insights once per dimension it needs, usually query and then page, and each call prints as it runs. The results include range totals for clicks, impressions, CTR and average position and the change against the previous equal-length period, so you read a comparison rather than a raw dump. Because Claude Code can read and write files in the working folder, the report does not have to stay on screen. Ask it to write the top queries and their position changes to a Markdown or CSV file, or to keep a dated reports folder, and a later session can compare a new pull with the saved one. For visitor data it calls storeseo_get_analytics_report with a report type like page_performance or traffic_source, and it can cross-reference those pages with the Liquid templates and content in the repository that produce them. That cross-referencing is the reason to do this job in a terminal. A page with many impressions and a weak CTR is a candidate for new meta, and Claude Code can tell you which template or product handle it belongs to before you move on to the optimization job, where it asks for your yes before applying anything. The two Google tools themselves are reads and run straight away. Prompts: - Use StoreSEO to pull Search Console by query for the last 28 days, compare with the previous period, and save the top 25 to reports/queries.md. - Which product pages have high impressions and a low CTR? Pull the page dimension, then tell me which templates in this repo render them. - Get the GA4 traffic_source report for last month and summarize which channels grew. Read only, change nothing. In Claude Code: Claude Code warns when a single tool result is larger than 10,000 tokens, and a Search Console pull by query or page with a large top_n can reach that on a busy store. Ask for the top twenty or fifty rows, or one narrow dimension at a time, instead of the longest list the tool will return. Also check which scope you added the server with: without the user scope StoreSEO is available only in the project folder where you added it, so a report started from another directory will not find the Google tools. Q: Can Claude Code save my Search Console data to a file? A: Yes. Claude Code works in the folder you started it in, so it can write the report it reads from StoreSEO to a Markdown or CSV file there. StoreSEO does not store the file for you, and a later session can pull fresh numbers and compare them with the saved ones. Q: Will Claude Code ask before running the Google report tools? A: That depends on your Claude Code permission settings. The tools are reads, but Claude Code can still prompt before an MCP tool runs unless you have allowed it. If you allow the StoreSEO tools, scope the rule to the read tools. --- ## Read Search Console and Analytics with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/google-reports (full page: https://storeseo.com/agent/claude-cowork/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Getting Google reports from Claude Cowork means delegating the Search Console and GA4 read to Anthropic's task agent, which uses the StoreSEO connector to pull the numbers and hand you a written summary, once or on a schedule. In Cowork you do not run the report step by step. You describe the result you want, for example a summary of last week's clicks, impressions and organic visitors with the biggest movers named, and Cowork works through it using the StoreSEO connector you added under Customize, Connectors. It opens the session with storeseo_start, reads storeseo_get_search_console_insights for the query and page dimensions, reads storeseo_get_analytics_report for the GA4 side, and returns a finished write-up for you to review instead of a conversation. This job suits Cowork better than most because both Google tools are read-only. A weekly task has to run without you in the conversation, and a task that only reads can do that safely: it pulls the previous equal-length period for the comparison, notes which queries and pages changed most and lists the traffic sources that grew or shrank. You choose the cadence, and Anthropic's documentation says scheduled tasks run remotely even when your computer is asleep, on plans that include them. The summary is where the follow-up work starts. If it flags a page with a falling CTR, you can open a normal Cowork conversation and ask for a preview of new meta, where you are present to give your yes. Keeping the scheduled task to reads and the fixes to a conversation you attend is the clean split between the two. Prompts: - Use StoreSEO to write me a one-page summary of last week's Search Console clicks, impressions, CTR and position, and the five queries that moved most. - Every Monday, read Search Console by page and GA4 traffic sources for the last 28 days and summarize what changed. Do not apply, generate, publish or delete anything. - Compare this month's GA4 organic search traffic with last month's and list the pages that drove the difference. In Claude Cowork: A scheduled Cowork task runs while you are not in the conversation, so it cannot wait for a yes. Word the task as report-only and say in plain words that it must not apply, generate, publish or delete anything, because the Google tools are reads but the task could otherwise reach other StoreSEO tools. Scheduled tasks and Cowork itself depend on your plan and, on Team and Enterprise, on an admin toggle. If StoreSEO is missing inside Cowork, add it under Customize, Connectors there rather than assuming another surface carried it over. Q: Can Claude Cowork send me a weekly Search Console report? A: Yes, as a scheduled task, on the plans where Anthropic offers them. Describe the report, tell it to use StoreSEO for Search Console and GA4, and keep it read-only. Cowork produces the summary for you to review. Q: Is it safe to schedule a StoreSEO task in Cowork? A: A report-only task is a reasonable use because the two Google tools only read. The task cannot ask you a question mid-run, so say explicitly that it must not apply, generate, publish or delete anything, and do those jobs in a conversation you are present for. --- ## Read Search Console and Analytics with Cursor URL: https://storeseo.com/agent/cursor/google-reports (full page: https://storeseo.com/agent/cursor/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports in Cursor means asking the editor's agent for Search Console and GA4 numbers through StoreSEO while your theme project is open, so the queries and pages it finds sit next to the files that render them. You have a Shopify theme open in Cursor and you ask the agent chat, in plain words, which pages get impressions but few clicks. The agent calls storeseo_start and then storeseo_get_search_console_insights with the page dimension, and Cursor shows each MCP call in the chat with its arguments, so you can read the date range and top_n the agent chose. The response carries range totals, the top rows ranked by clicks and the change against the previous equal-length period. The editor is the point. Once the agent has a page URL or a product handle from the report, it can search the project for the template, section or snippet that renders it and tell you what that page contains: its title tag logic, its schema blocks, its image tags. For visitor behavior it calls storeseo_get_analytics_report with page_performance or traffic_source, and it can put the two reports side by side in a file in the repository if you want to keep them with the project. Both Google tools are read-only. Cursor by default asks before an MCP tool runs, so expect a confirmation on the first calls unless you have changed the run mode, and approve the read tools to let a report flow. Keep that approval to reads, and leave the apply and generate tools on their prompt for the optimization jobs that come after the report. Prompts: - Use StoreSEO to pull Search Console by page for the last 28 days and tell me which templates in this theme render the top five. - Which queries have lost clicks compared with the previous period? Read only, and write the top 20 to notes/queries.md. - Get GA4 page_performance for last month and list the pages with the most views but the lowest engagement. In Cursor: StoreSEO exposes 44 tools and Cursor allows about 40 active tools across all your servers, so you have to switch some off, and a Google report fails quietly if the tool you switched off was one of the two it needs. If the agent says it cannot read Search Console, open Cursor's MCP settings and check that storeseo_get_search_console_insights and storeseo_get_analytics_report are still enabled, and disable tools for jobs you do not use instead. A server in .cursor/mcp.json belongs to that project only, so use ~/.cursor/mcp.json to keep it everywhere. Q: Why can Cursor not find the Search Console tool? A: Most often because the tool was switched off to stay under Cursor's cap of about 40 active tools. StoreSEO has 44 tools, so open the MCP settings, turn the Search Console and Analytics report tools back on and turn off tools for jobs you do not run. Q: Does Cursor need Google connected to read these reports? A: Google has to be connected in the StoreSEO app, on a paid plan, with a GA4 property selected for Analytics. Cursor cannot do that step. If it is missing, the tool returns a message pointing you to the app. --- ## Read Search Console and Analytics with Codex URL: https://storeseo.com/agent/codex/google-reports (full page: https://storeseo.com/agent/codex/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with Codex means asking OpenAI's coding agent, in the terminal or the IDE extension, for Search Console and GA4 numbers through StoreSEO and getting a comparison with the previous period back in the session. You start Codex in a theme repository, with StoreSEO already added to config.toml and signed in with codex mcp login, and ask for a report such as which countries and devices send the most clicks. Codex calls storeseo_start to open the session, then storeseo_get_search_console_insights once for the country dimension and again for device, and you see each call in the transcript. Each response has range totals, the top rows by clicks and the per-metric change against the previous equal-length period. For visitor data Codex calls storeseo_get_analytics_report, choosing demographics for country, traffic_source for channels or key_events for conversions. Because Codex works in your repository, it can write the results to a file, diff them against an earlier pull, or tie a page from the report to the template in the repo that produces it. The same config.toml is shared by the CLI, the IDE extension and the desktop app, so a report started in one place uses the same server settings as another. The approval mode on the storeseo server decides what you see while this runs. The Google tools are read-only, so with the writes mode Codex should run them without a prompt and still stop for anything that is not marked read-only, which is the setting to keep for the jobs that follow a report, such as applying new meta, where you want the prompt. If a pull is large, ask Codex to split it by dimension and keep the result in a file. Prompts: - Use StoreSEO to get Search Console clicks by country and by device for the last 28 days and compare with the previous period. - Pull my top 25 queries from Search Console and write them with position changes to reports/queries.csv. - Get the GA4 search_traffic report for the last three months and tell me which month was strongest. In Codex: Codex waits 60 seconds for a tool call by default, and a Search Console pull with a wide date range and a large top_n on a busy store can run long. If a report times out, raise tool_timeout_sec for the storeseo server in config.toml or ask for a narrower range or one dimension at a time. Keep the server's default_tools_approval_mode on writes or prompt, not approve, so the read tools run and the tools that change your store still stop for a decision. Q: Why does a Codex Search Console report time out? A: Codex waits 60 seconds for a tool to finish by default. A large pull can take longer, so raise tool_timeout_sec for the storeseo server in config.toml, or ask for a smaller top_n, a shorter range or a single dimension. Q: Does the Codex CLI use the same StoreSEO setup as the IDE extension? A: Yes. The CLI, the IDE extension and the desktop app share config.toml, so the storeseo server entry and its approval settings apply in all of them. A project-level file only loads in trusted projects, so the user file is simpler. --- ## Read Search Console and Analytics with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/google-reports (full page: https://storeseo.com/agent/gemini-cli/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with Gemini CLI means asking Google's command-line agent for Search Console and GA4 numbers through the StoreSEO MCP server and reading the comparison with the previous period in your terminal. You run Gemini CLI in a folder, with StoreSEO added under mcpServers or through gemini mcp add, and ask for the report in a sentence, such as which search appearances and devices drive my clicks. The model sees the tools as mcp_storeseo_ names and calls storeseo_start, then storeseo_get_search_console_insights with the dimension you asked for. Gemini CLI prints the call and asks for confirmation before running it unless the server is trusted, which is the default behavior you should leave in place. What returns is the range summary for clicks, impressions, CTR and average position, the top rows ranked by clicks, and the change against the previous equal-length period. For visitors Gemini CLI calls storeseo_get_analytics_report, for example with the traffic_source or user_activity report type, and you can follow up with a narrower range or a different dimension in the same session. Because the CLI reads the folder it was started in, it can also save a report to a file there or compare a new pull with an earlier one. Two details are particular to this client. Each confirmation you give covers a read, so a report session is a short run of approvals rather than a risk, and you can allow the Google tools for the session if you do not want to repeat them. And if you want the model to see fewer StoreSEO tools, an includeTools list on the server can limit it to the two Google tools plus storeseo_start. Prompts: - Use StoreSEO to show Search Console clicks by search appearance and by device for the last 28 days, compared with the previous period. - Get the top 20 queries from Search Console and save them with position changes to reports/queries.md. - Pull the GA4 traffic_source report for last month and tell me which channel grew the most. In Gemini CLI: Do not set trust: true or --trust on the storeseo server to avoid the confirmations during a report session. Trust skips every tool confirmation for that server, including the tools that apply content or change settings, and then the confirmation StoreSEO asks for in the chat is the only check. Allow the two Google read tools for the session instead, or use an includeTools list. If the browser sign-in cannot open because you work over SSH or in a headless session, use a personal access token from Settings, AI Toolkit as a header. Q: Should I trust the StoreSEO server in Gemini CLI to speed up reports? A: No. Trust skips confirmations for every tool on that server, not only the two Google report tools. Approve the read tools for the session, or limit the server with includeTools, and keep confirmations on for tools that write. Q: Why does Gemini CLI not show a Google report tool? A: Check that the server connected and that an includeTools or excludeTools list did not remove the Search Console or Analytics tool. Also confirm that Google is connected in the StoreSEO app on a paid plan. --- ## Read Search Console and Analytics with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/google-reports (full page: https://storeseo.com/agent/github-copilot/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with GitHub Copilot means asking Copilot Chat in Agent mode, inside Visual Studio Code, for Search Console and GA4 numbers through StoreSEO while your theme workspace stays open. You open Copilot Chat in VS Code, switch to Agent mode and ask for a report, for example which queries sent the most clicks last month. Copilot calls storeseo_start, then storeseo_get_search_console_insights with the query dimension, and VS Code shows the MCP tool call in the chat with a confirmation to continue, where you can inspect the arguments. Once it runs, the response carries range totals for clicks, impressions, CTR and average position, the top rows by clicks and the change against the previous equal-length period. Questions about visitors go to storeseo_get_analytics_report. Copilot picks a report type such as search_traffic, page_performance or traffic_source and returns totals and rows, and it can put the Search Console and GA4 views into a Markdown file in the workspace so the numbers sit in the repository you are working on. Since Agent mode can read the files that render a page, it can also point from a page in the report to the template or section behind it. VS Code gives you a granular choice on the confirmation prompt, such as allowing a tool for the session or the workspace. For this job it is reasonable to allow the two Google report tools, because they are reads. Keep the prompt on for the apply, generate and delete tools that follow a report, since StoreSEO asks for your yes in the chat but the editor's confirmation is the hard stop. Tool output also stays in the chat, so you can ask Copilot to turn a long report into a short list of pages worth checking. Prompts: - Use StoreSEO to get Search Console clicks and impressions by query for the last 28 days and compare with the previous period. - Pull my top pages from Search Console and write the top 20 with CTR and position to docs/search-pages.md in this workspace. - Get GA4 key_events for last month and show which traffic source led to the most events. In GitHub Copilot: On Copilot Business and Enterprise an organization admin has to enable MCP servers in Copilot policy, or the StoreSEO tools will not appear in Agent mode, so a missing Google tool can be a policy issue rather than a StoreSEO one. The server also has to be under the servers key in .vscode/mcp.json, not mcpServers, which belongs to the portable .mcp.json file. If you commit .vscode/mcp.json, remember it is shared with your team, and keep any token out of it. Q: Why are the StoreSEO Google report tools missing in Copilot Agent mode? A: Check that the server is listed under servers in .vscode/mcp.json and started, that you use VS Code 1.99 or later, and that on Business or Enterprise your admin enabled MCP servers in Copilot policy. Then confirm Google is connected in the StoreSEO app. Q: Can I allow the Google report tools without allowing the rest of StoreSEO? A: Yes. VS Code confirms tools one at a time, so you can allow the Search Console and Analytics report tools, which are reads, and keep the confirmation on for the tools that apply content or change settings. --- ## Read Search Console and Analytics with Windsurf URL: https://storeseo.com/agent/windsurf/google-reports (full page: https://storeseo.com/agent/windsurf/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Windsurf means asking the Cascade agent in your editor for clicks, impressions, queries, pages and traffic sources, which StoreSEO pulls from your connected Google Search Console and GA4 property while your Shopify theme stays open beside the answer. You are in Windsurf with the theme repository open and the Cascade panel docked on the side. Once StoreSEO is listed under MCPs, you type a question the way you would ask a colleague, for example which search queries brought the most clicks in the last 28 days. Cascade calls storeseo_start first, because every StoreSEO tool needs the conversation id it returns, then storeseo_get_search_console_insights with the query dimension. The result is the top rows ranked by clicks, the range totals and the previous period of the same length with the change per metric, all printed in the chat as a table. Analytics questions use the second tool. Asking where purchases came from makes Cascade call storeseo_get_analytics_report with the traffic_source report type, and a question about organic growth uses search_traffic, which returns rows by date. Both calls are reads, so there is nothing to accept before they run, but your editor still shows each call in the panel and you can expand it to see the arguments Cascade chose. Because the answer lands next to your code, it is easy to follow a finding into a file: pages with many impressions and few clicks can lead straight to the product template or the meta fields you are about to edit. Windsurf keeps the whole session in context, so you can narrow in place. Ask for the same report by device, then by country, then by page, and Cascade makes one more call each time rather than rebuilding the picture from scratch. If Google is not connected in StoreSEO, the tool says so and gives a link to finish setup in the app, since the agent cannot connect Google for you. You can then paste a summary into a pull request description or a comment in the theme file. Prompts: - Use StoreSEO to show my top 15 Google queries by clicks for the last 28 days, with the change against the previous 28 days. - Which of my pages have lots of impressions but few clicks? List them with CTR and average position so I can open the matching templates. - Pull the GA4 traffic_source report for last month and tell me which channel brought the most users. In Windsurf: Windsurf's MCP page says the mcp_config.json setup applies to the legacy Cascade agent, while the Devin Local agent, the default for new tabs, reads Devin CLI configuration files instead. If the report tools are missing, check which agent the tab is using before you debug the server. Cascade also allows 100 tools in total across every MCP server, and StoreSEO takes 44 of them, so a crowded setup may leave the two Google tools switched off. Open the MCPs section and confirm both are enabled. Q: Why does Windsurf say Google is not connected? A: The two report tools need a paid StoreSEO plan with Google connected, and Analytics reports also need a GA4 property selected. Connecting Google is done in the StoreSEO app, and the tool returns a link to it, because Cascade cannot connect Google for you. Q: Does Cascade change anything when it reads these reports? A: No. storeseo_get_search_console_insights and storeseo_get_analytics_report are read-only, so they only fetch figures. Windsurf's own approval behavior for MCP tools is not described on its MCP page, so watch the panel to see each call as it runs. --- ## Read Search Console and Analytics with Grok URL: https://storeseo.com/agent/grok/google-reports (full page: https://storeseo.com/agent/grok/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Grok means asking xAI's assistant in a normal chat which queries, pages and channels bring your Shopify store its visitors, while the StoreSEO connector fetches the figures from your Google accounts and Grok explains them in plain words. You open a chat on grok.com with the StoreSEO connector available and type the question you would otherwise open two Google dashboards for. Grok calls storeseo_start to open a session, then storeseo_get_search_console_insights for the clicks, impressions, click-through rate and average position of your top queries. The answer comes back in the conversation, with the last 28 days and the previous 28 days side by side, and you can read it on a laptop or in the xAI app on your phone without installing anything. The chat format suits follow-up questions. After the query list you can say now by country, and Grok makes another call with the country dimension, or ask for the same numbers by device, by page or by search appearance. For visitor behavior it switches to storeseo_get_analytics_report and picks a report type such as search_traffic, traffic_source or page_performance. Both tools only read, so the connector can run them straight away, and you can keep asking for different slices without anything being applied to your store. Grok is also good at the explaining part. Ask why a query with many impressions earns so few clicks and it will describe what a low click-through rate usually points to, using the figures StoreSEO returned. Keep in mind that the numbers are what Google reports for the range you named, and Grok does not forecast outcomes from them. If a Search Console figure points at a page worth fixing, say so and Grok can carry on into the meta or alt text jobs in the same chat, where StoreSEO asks for your yes before anything is written. Prompts: - Which Google queries sent my store the most clicks in the last 28 days, and how does that compare with the 28 days before? - Break my Search Console clicks down by device and by country, and tell me which segment grew the most. - Show my GA4 traffic sources for last month and explain in simple words which channel I should look at first. In Grok: xAI's connector documentation does not describe a per-tool approval setting, but these two reports are reads, so the question of confirmation only matters once you move on to a write. On Grok Business and Enterprise a team admin must provision the connector in the cloud console before members can use it, so if the StoreSEO connector is missing from your chat, ask your admin first. Google must also be connected inside StoreSEO, and Grok cannot do that step from the chat. Q: Can Grok connect my Google account to StoreSEO? A: No. Connecting Search Console and Google Analytics, and choosing the GA4 property, happens in the StoreSEO app. When Google is not connected, the report tool returns a link telling you what to set up, and Grok can pass that on. Q: Do I need a paid StoreSEO plan to ask Grok for Google reports? A: Yes. The reports need a paid StoreSEO plan with Google connected, and Analytics reports also need a GA4 property selected in the app. The pricing table shows the current plans. --- ## Read Search Console and Analytics with OpenClaw URL: https://storeseo.com/agent/openclaw/google-reports (full page: https://storeseo.com/agent/openclaw/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with OpenClaw means your own agent runtime asks StoreSEO for clicks, queries, pages and traffic sources, either when you message it in a chat app or on a schedule that announces the report in a channel without you opening anything. OpenClaw runs on your machine or server and you talk to it through a chat app such as Telegram, Slack or WhatsApp, or through its Control UI and CLI. Once the storeseo server is added and the probe passes, you message it a question like which queries brought the most Google clicks this month. OpenClaw calls storeseo_start, then storeseo_get_search_console_insights, and replies in the same thread with the ranked rows, the totals and the change against the previous period of equal length. A follow-up for GA4 triggers storeseo_get_analytics_report with the report type that fits, for example traffic_source. The reason to pick OpenClaw for this job is the scheduler. Automations, also reachable as openclaw cron, wake the agent in an isolated session at set times and can announce the result in a channel. A Monday morning job can read the Search Console summary, read the GA4 traffic source report and post a short digest to your team chat, so the figures reach people who never log in to either Google product. Because both tools are reads, this is a good fit for unattended runs: nobody has to be there to say yes. Keep the scheduled prompt report-only and say so in it. The prompt can ask for the top queries, the pages with the most impressions and the traffic sources, each compared with the previous period, and tell the agent not to preview, apply or generate anything. If a finding deserves action, such as a page with plenty of impressions and a weak click-through rate, reply in the chat that received the digest and ask for the change there, where you can answer StoreSEO's request for a yes. You can list past runs with openclaw automations runs followed by the job ID. Prompts: - Use StoreSEO to give me my top 10 Google queries by clicks this month and the change from last month. - Which pages get many impressions but a low click-through rate? Reply with page, impressions, CTR and average position. - Read the GA4 traffic_source report for the last 28 days and tell me which channel grew the most. In OpenClaw: A scheduled automation has nobody to say yes, so word the prompt as reads only. Both Google tools are reads, which makes them suitable, but a loose prompt such as fix the weak pages could lead the agent toward a write. OpenClaw also has its own per-server approval setting, openclaw mcp configure storeseo --approval with approve, prompt or auto, so check it before you leave an automation running. Run the mcp commands on the installation that owns the connector, because another host needs its own reload before it sees storeseo. Q: Can OpenClaw post my Search Console numbers to a chat on a schedule? A: Yes. An OpenClaw automation can run at a set time in an isolated session, call the two StoreSEO report tools and announce the result in a channel such as Telegram. Keep the prompt to reads, since a scheduled run cannot wait for a yes. Q: Why does the scheduled report say Google is not connected? A: The report tools need a paid StoreSEO plan with Google connected, and Analytics also needs a GA4 property selected. Both are set up in the StoreSEO app. The tool returns a link to the missing step, and the automation will repeat that message until you fix it there. --- ## Read Search Console and Analytics with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/google-reports (full page: https://storeseo.com/agent/hermes-agent/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Hermes Agent means Nous Research's open-source agent asks StoreSEO for your clicks, queries, pages and traffic sources, in a terminal session or a chat app, and can repeat the same check on a hermes cron schedule. Hermes Agent keeps memory across sessions and runs from a terminal or from chat apps such as Telegram, Discord and Slack, so you can ask about your store wherever you already are. After storeseo is added under mcp_servers in config.yaml and you have signed in with hermes mcp login storeseo, a question such as how did my Google clicks change this month starts with storeseo_start. Hermes then calls storeseo_get_search_console_insights, usually with the date dimension for a trend or the query dimension for a ranked list, and answers with the totals and the previous period beside them. Memory changes how the follow-ups feel. Because Hermes remembers earlier sessions, you can say compare this with what we found last week and it can refer to the figures it already reported, though it will read the report again for the current range rather than guess. For GA4 questions it calls storeseo_get_analytics_report with a report type such as user_activity, page_performance or key_events. Both tools are reads, so the answers arrive without a pause, and you can ask for the same data by device, country or search appearance with one more call. hermes cron create takes a schedule and a prompt and delivers the result to a chat target, so a weekly Google digest can land in Telegram on its own. Write that prompt as a report: ask for the top queries, the pages with high impressions and low click-through rate and the main traffic sources, each against the previous period, and say plainly not to preview, apply or generate anything. The rest of the job, such as rewriting meta for a weak page, belongs in a live session where StoreSEO can ask for your yes and you can give it. Prompts: - Use StoreSEO to list my top queries by clicks for the last 28 days and compare with the 28 days before. - Which of my pages earn many impressions but a CTR under two percent? Sort by impressions. - Read the GA4 user_activity report for the last three months and tell me whether engagement is rising or falling. In Hermes Agent: Start a new session or run /reload-mcp after you edit config.yaml, otherwise the session you already have open will not see the storeseo server and the two report tools will seem to be missing. For a cron job, remember that no one is there to say yes, so the prompt should ask for reads only. If you use a personal access token instead of OAuth, the token sits in config.yaml, so keep that file out of version control and revoke the token in Settings, AI Toolkit if it leaks. Q: Can hermes cron send me a Search Console report every week? A: Yes. hermes cron create takes a schedule and a prompt and can deliver the result to a chat target such as Telegram. Ask for reads only, because a scheduled run cannot wait for a yes, and both StoreSEO Google tools are read-only. Q: Does Hermes remember my earlier Google figures? A: Hermes keeps memory across sessions, so it can recall what it reported before. For a current number it still calls storeseo_get_search_console_insights or storeseo_get_analytics_report again, so you get the figures Google returns for the range you name. --- ## Read Search Console and Analytics with OpenCode URL: https://storeseo.com/agent/opencode/google-reports (full page: https://storeseo.com/agent/opencode/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with OpenCode means asking the open-source terminal agent for your store's clicks, queries, pages and traffic sources, which StoreSEO reads from your connected Google accounts and OpenCode prints next to the project you are working in. You run opencode in the repository for your theme or storefront and ask a question in the prompt box, for instance which pages earn impressions but few clicks. After opencode mcp auth storeseo has completed, OpenCode sees the tools under the storeseo prefix, so the model calls storeseo_start and then storeseo_get_search_console_insights with the page dimension. The reply is a ranked table of pages with clicks, impressions, CTR and average position, plus the range totals and the previous period of the same length, and it appears in the terminal alongside the files you have open. Because OpenCode is a coding agent, a finding can turn into a change in the same session without a context switch. Ask for the GA4 page_performance report with storeseo_get_analytics_report and the agent can match a weak landing page to its template, or write the figures into a Markdown file in the project so the next session can compare against them. You can also choose the dimension yourself: query, page, country, device, searchAppearance or date. Each choice is one more read call, and reads need no yes from you. OpenCode exposes tool names to the model as server_tool, and its permission rules can match a pattern such as storeseo_*. That is useful here: if you only want reporting, you can deny the write tools by name and leave the two Google reports, so the session cannot apply anything even if the conversation drifts. If you later decide to act on a report, such as new meta for a page with a weak click-through rate, lift that rule for the job and let StoreSEO ask for your yes before the write. Prompts: - Use StoreSEO to show Search Console clicks, impressions, CTR and position by page for the last 28 days, top 20 by clicks. - Compare my Search Console performance on mobile and desktop for the last 28 days and note the biggest change. - Fetch the GA4 key_events report for last month and save the figures to reports/ga4-key-events.md in this project. In OpenCode: OpenCode V2 expects servers under mcp.servers, while older versions put the server directly under mcp, so check the layout your installed version documents before you edit the file. opencode mcp add writes to the project configuration unless you pass --global, which means a server added without it will be missing in other projects and the report tools will not appear there. The docs describe deny rules but no ask-first mode, so deny the write tools by name if you want a client-side stop on a reporting session. Q: Can OpenCode save my Search Console report to a file? A: Yes. OpenCode works in the folder you started it in, so you can ask it to write the figures from storeseo_get_search_console_insights or storeseo_get_analytics_report to a Markdown or CSV file and compare them in a later session. Q: How do I stop OpenCode from using the StoreSEO write tools during a reporting session? A: Add a permission rule that denies the write tools by name, using the storeseo prefix, and leave the two Google report tools allowed. OpenCode documents deny rules for MCP tools, and the reports only need the read tools. --- ## Read Search Console and Analytics with Zed URL: https://storeseo.com/agent/zed/google-reports (full page: https://storeseo.com/agent/zed/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Zed means asking the Agent Panel in your editor for clicks, impressions, queries, pages and traffic sources, which StoreSEO fetches from your Google accounts, with Zed's own tool prompt deciding when each call may run. You are in Zed with your theme project open and the Agent Panel showing. Settings, AI, MCP Servers has a green dot next to storeseo, so you type a question such as which queries brought the most clicks from Google this month. The agent calls storeseo_start and then storeseo_get_search_console_insights, and the result appears in the panel as ranked rows with the totals and the previous period of equal length. A second question about visitors uses storeseo_get_analytics_report, with a report type like search_traffic or traffic_source. Zed behaves differently from most clients here because of agent.tool_permissions.default, which is confirm out of the box. That means the panel asks before it runs a tool action, and it does so for reads too. A report question can therefore show a prompt for storeseo_start, another for the Search Console call and another for the GA4 call. Nothing is wrong when that happens: it is Zed's own check, not StoreSEO, and you accept each one to let the report through. If the prompts get in the way of a longer reporting session, give the read tools their own permission using keys of the form mcp:storeseo:tool_name, for example mcp:storeseo:storeseo_get_search_console_insights, and set them to allow. Leave the apply, generate and update tools on confirm. That way the figures arrive without a click while a later write, such as new meta for a page with a weak click-through rate, still stops at Zed's prompt as well as at StoreSEO's own request for your yes. The answers stay in the panel, and you can copy the table into a note beside the code. Prompts: - Use StoreSEO to show my top 15 Google queries by clicks for the last 28 days, with the change against the previous period. - Which device type sends the most Search Console clicks, and which one grew fastest compared with the previous 28 days? - Read the GA4 search_traffic report for the last three months and describe the trend in two sentences. In Zed: Zed's default is to ask before every tool action, reads included, so one report question can mean three prompts. Allow only the two Google report tools and storeseo_start with per-tool keys such as mcp:storeseo:storeseo_get_search_console_insights, and leave the write tools on confirm. Setting agent.tool_permissions.default to allow removes Zed's check altogether, so StoreSEO's request for your yes in the chat would be the only one left. Use it sparingly, even for a session that only reads. Q: Why does Zed ask me to approve a report that only reads data? A: Zed's agent.tool_permissions.default is confirm by default, which covers every tool action, reads included. That prompt comes from Zed, not StoreSEO. You can allow individual read tools with keys of the form mcp:storeseo:tool_name and keep the write tools on confirm. Q: Does Zed need a Google sign-in for these reports? A: No. Google is connected inside the StoreSEO app, under a paid plan, and Analytics also needs a GA4 property selected there. Zed only signs in to StoreSEO, and the report tools return a link if a Google step is missing. --- ## Read Search Console and Analytics with Kiro URL: https://storeseo.com/agent/kiro/google-reports (full page: https://storeseo.com/agent/kiro/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Kiro means asking AWS's agentic IDE for your store's clicks, impressions, queries, pages and traffic sources, which StoreSEO reads from your connected Google accounts and Kiro lays out in the chat panel next to your project. You work in Kiro with the project open and the chat panel on the side. After storeseo is added to .kiro/settings/mcp.json or the user-level file and the browser sign-in is done, you ask a question like which queries earned the most clicks from Google last month. Kiro's agent calls storeseo_start and then storeseo_get_search_console_insights, and the reply lists the top rows ranked by clicks with the range totals and the previous period of equal length, including the change per metric. Analytics works the same way through storeseo_get_analytics_report. Ask which channels drove users and it picks the traffic_source report, ask about engagement over time and it uses user_activity, and ask about key events for a purchase-related view. Each report is one read, and you can narrow by dimension, so the same Search Console data can be split by page, country, device or search appearance. Kiro tends to plan multi-step work, so a request such as audit my Google performance and list the next three pages to improve can run several reads in a row before it answers. Kiro's autoApprove list is the useful lever for this job. Put storeseo_start, storeseo_get_search_console_insights and storeseo_get_analytics_report in it and a reporting session runs without a prompt for every call. Keep the apply, generate and update tools out of that list, so Kiro still stops before a write. A finding such as a page with plenty of impressions and a poor click-through rate can then lead into a preview in the same chat, and StoreSEO asks for your yes before anything is applied. The figures stay in the conversation, so you can ask Kiro to save them in a spec or notes file. Prompts: - Use StoreSEO to list my top queries by clicks for the last 28 days with the change from the previous 28 days. - Show the Search Console pages report and flag the pages with high impressions and a CTR under two percent. - Read GA4 traffic_source for last month and write a short summary of the channels into notes/traffic.md. In Kiro: A workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, so if the sign-in points at the wrong store or the report tools look unavailable, check which file defines storeseo. When you fill in autoApprove, list the individual read tools. A wildcard would also approve the write tools, and then Kiro would stop asking before StoreSEO applies content, which leaves only StoreSEO's own request for your yes. Keep the two Google tools in the list and nothing more. Q: Which StoreSEO tools should I put in Kiro's autoApprove for reports? A: storeseo_start, storeseo_get_search_console_insights and storeseo_get_analytics_report are all reads, so they are reasonable to list. Leave the apply, generate and update tools out so Kiro still prompts before a write. Q: Can Kiro compare this month's Search Console data with last month's? A: Yes. Both report tools include the previous period of the same length and the change per metric by default, so a 28 day range is compared with the 28 days before it. You can also name your own date range. --- ## Read Search Console and Analytics with Antigravity URL: https://storeseo.com/agent/antigravity/google-reports (full page: https://storeseo.com/agent/antigravity/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Antigravity means asking Google's agentic coding product, from its IDE side panel, for your store's clicks, queries, pages and traffic sources, which StoreSEO reads from your connected Google Search Console and GA4 property. You are in the Antigravity IDE with the agent side panel open and storeseo listed under MCP servers. Once Authenticate has finished and you have pasted the authorization code back into Agent settings, you ask for what you want to know, for example which pages earn the most Google clicks and how that changed since the previous period. The agent calls storeseo_start, then storeseo_get_search_console_insights with the page dimension, and returns the ranked rows, the totals and the per-metric change against the previous period of equal length. For visitor behavior the agent calls storeseo_get_analytics_report, choosing among search_traffic, user_activity, demographics, page_performance, traffic_source and key_events. Because Antigravity plans multi-step work, a broader request such as review my Google performance and propose where to look first can turn into several reads in a row: a Search Console query list, a page list and a GA4 channel report. The two StoreSEO tools do not write anything, so the plan is a sequence of lookups, and you read the result in the panel while your theme files sit in the editor. Antigravity runs MCP tools you have not configured in Ask mode by default, so expect an approval request before each call until you add a policy. Policy patterns of the form mcp(server/tool) let you allow specific tools, so mcp(storeseo/storeseo_get_search_console_insights) and mcp(storeseo/storeseo_get_analytics_report) are a sensible pair for a reporting habit. Leave the write tools in Ask mode, so a later step such as new meta for a page with a weak click-through rate still stops at Antigravity's prompt and at StoreSEO's own request for your yes. Prompts: - Use StoreSEO to show Search Console clicks, impressions, CTR and position by query for the last 28 days, top 20, with the previous-period change. - Which countries and devices send my Google clicks? Show both breakdowns and call out the biggest difference. - Pull GA4 page_performance for last month and list the five pages with the most active users. In Antigravity: The field is serverUrl. Antigravity's documentation says legacy fields such as url or httpUrl are not supported, so an entry copied from another client will not connect until you rename the field, and the two report tools will not show up. The sign-in also ends with a code you paste back into Agent settings, under Customizations, so finish it in the same window you started in. If you add a policy that allows mcp(storeseo/*), Antigravity stops asking for every StoreSEO tool, writes included. Q: Why does Antigravity ask me before each Search Console call? A: MCP tools you have not configured run in Ask mode by default, which requires your approval before execution. You can allow the two report tools with policy patterns of the form mcp(server/tool) and leave the write tools in Ask mode. Q: Does Antigravity need to be signed in to Google for these reports? A: No. StoreSEO reads the data using the Google connection you set up in the StoreSEO app, on a paid plan, with a GA4 property selected for Analytics. Antigravity only needs to authenticate to the StoreSEO server. --- ## Write SEO blog post drafts with Claude URL: https://storeseo.com/agent/claude/blog-posts (full page: https://storeseo.com/agent/claude/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Claude means describing the post you want in a chat, letting Claude hand the brief to StoreSEO, and finding a finished draft article in your Shopify blog a few minutes later. You sit in a Claude chat tab, on the web or in Claude Desktop, with the StoreSEO connector switched on in the Search and Tools menu. You type something like write a post about caring for linen shirts for the News blog. Claude calls storeseo_start first to open the session, then storeseo_list_blogs to find the numeric id of the blog the post should go into. Both are reads, so they run straight away and you see them listed as tool calls in the conversation. Before it writes anything Claude should come back with the brief in words: the topic, the keyword it plans to target, the length, whether you want a featured image, and what that will cost in AI credit. StoreSEO tells the agent to do this because the cost grows with word count and with the image. When you say yes, Claude calls storeseo_generate_blog_post, and Claude's own permission card for that write tool appears as well. Generation runs in the background for a few minutes, so Claude polls storeseo_get_blog_post_job and tells you how many steps are done out of the total, rather than calling the post finished early. When the job completes, Claude can read the draft with storeseo_get_resource_seo using the article id from the job result and summarize its SEO record in the chat. The post itself is a draft in Shopify, so you open the blog in Shopify admin, read it, edit what you want and press publish yourself. A chat is a good place to iterate on the brief: ask Claude for three angles first, pick one, and only then agree the cost. Prompts: - Use StoreSEO to list my blogs, then propose three post ideas for my store. Do not write anything yet. - Write a 1,200 word draft on how to care for linen shirts in my News blog, with a featured image. Tell me the credit cost and wait for my yes. - How far along is the post you started? Once it is done, check the SEO record of the draft and tell me what to improve. In Claude: Generation takes a few minutes and a chat only moves when you or Claude send a message, so the polling happens while Claude is working on your turn. If you close the tab or switch conversations before it reports back, the job still runs in StoreSEO. Open the conversation again, or look in the blog in Shopify admin, and ask Claude to check the job by its job id. Claude's prompt for the generate tool is the hard stop, so keep that tool on ask rather than Allow always. Q: Does Claude publish the blog post to my Shopify store? A: No. StoreSEO saves the post as a draft article in the blog you chose. You review it and publish it yourself from Shopify admin. Q: Why does Claude say I have no blogs to write into? A: Only blogs synced with StoreSEO are listed, and no tool in the chat can sync them. Sync your blogs in the StoreSEO app, then ask Claude to list them again. --- ## Write SEO blog post drafts with ChatGPT URL: https://storeseo.com/agent/chatgpt/blog-posts (full page: https://storeseo.com/agent/chatgpt/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with ChatGPT means briefing the chat, approving one StoreSEO write action, and picking up a finished draft article in your Shopify blog without leaving ChatGPT until it is time to publish. You work in a ChatGPT conversation with the StoreSEO app added through developer mode. You ask for a post in plain words, for example a gift guide for coffee lovers with a featured image. ChatGPT calls storeseo_start to open the session and storeseo_list_blogs to get the numeric blog id. Those are reads and ChatGPT tells reads from writes by the readOnlyHint annotation, so they go through without a confirmation. The write is storeseo_generate_blog_post. StoreSEO tells the agent to state the topic, keyword, length and expected AI credit before it calls, and to wait for your yes. When it does call, ChatGPT shows its own confirmation with the payload, which is where you can read the blog id, the brief and the image option before approving. OpenAI also offers to remember an approval for the rest of the conversation. For a tool that spends credit, approve each run instead. After that ChatGPT polls storeseo_get_blog_post_job and reports progress until the job is completed, failed or cancelled. When it completes, ask ChatGPT to run storeseo_get_resource_seo on the draft article to see its SEO record in the chat. The draft sits in your Shopify blog; you edit and publish it from Shopify admin. Which plans can approve write actions in ChatGPT varies, so check that before you plan to use this job. Prompts: - Using StoreSEO, list my Shopify blogs and tell me which one fits a gift guide. - Write a draft post targeting best gifts for coffee lovers, with a featured image. Tell me the AI credit cost and wait for my confirmation before you start. - Check the job for the blog post you started and tell me how many steps are done. In ChatGPT: Whether ChatGPT can run a write action such as storeseo_generate_blog_post depends on your plan and workspace settings. OpenAI's documentation and help center differ on which plans get write actions, and on a plan with read access only the generate call will not run. In that case use ChatGPT to list blogs and plan topics, and start the post from another connected agent. Where writes are available, read the payload in ChatGPT's confirmation before you approve it, and check that the brief, the blog and the featured image option match what you agreed with the agent a moment earlier. If you are unsure, ask for the plan again. Q: Can ChatGPT write a StoreSEO blog post on every plan? A: Not necessarily. Writing a post is a write action, and ChatGPT's support for write actions in developer mode depends on your plan and workspace. If it only has read access, ChatGPT can list your blogs but cannot start the post. Q: Will ChatGPT show the draft as finished while it is still writing? A: It should not. The draft article exists from the start of the run, but StoreSEO tells the agent to keep polling the job and to report progress until the status is completed, failed or cancelled. --- ## Write SEO blog post drafts with Claude Code URL: https://storeseo.com/agent/claude-code/blog-posts (full page: https://storeseo.com/agent/claude-code/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Claude Code means briefing Anthropic's terminal agent from the folder you already work in, and letting StoreSEO write the post into a draft article in your Shopify blog while you keep coding. You type the brief at the Claude Code prompt, for example a 1,500 word post on choosing a ring size for the Guides blog, targeting the keyword ring size guide. Claude Code calls storeseo_start, then storeseo_list_blogs to resolve the blog id, and you see each call scroll past in the terminal. Because Claude Code sits in your project folder, it can also read a content plan, a CSV of keywords or a brief written in Markdown there and turn each line into a request. The generate call is the one to pay attention to. StoreSEO tells Claude Code to state the credit cost and wait for your yes before storeseo_generate_blog_post, and Claude Code adds its own permission prompt for the tool unless you allowed it. Generation runs in the background for a few minutes. Claude Code polls storeseo_get_blog_post_job, reports steps completed out of total, and can carry on with other work in the same session between polls. When the job completes, ask it to read the draft with storeseo_get_resource_seo and write the SEO findings, or the article id, into a file in the repository. The post itself lands as a draft article in Shopify, never published live, so the last step is yours in Shopify admin. If you write several posts, do them one after another and agree the cost each time, so you can see the credit being spent before the next one starts. Prompts: - Use StoreSEO to list my blogs, then read content-plan.md in this folder and tell me which row to write first. - Write a 1,500 word draft for the Guides blog targeting ring size guide, no featured image. Tell me the credit cost and wait for my yes. - Check the blog post job you started, and when it completes read the draft's SEO record and save the article id to notes.md. In Claude Code: A permission rule that allows the StoreSEO tools to stop the prompts also covers storeseo_generate_blog_post, which spends credit, so scope an allow rule to the read tools and keep the generate tool on ask. Claude Code has a default tool timeout, but StoreSEO returns a job id straight away and the writing happens in the background, so the long wait is in polling, not in one blocked call. Keep polls modest in a long session instead of looping on the job as fast as possible. Q: Can Claude Code write several blog posts from a content plan file? A: Yes, one at a time. Claude Code can read the plan in your folder and start a post per row, but each generation spends AI credit, so agree the cost for each post before it starts. Q: Where does the finished post end up? A: As a draft article in the Shopify blog you picked. Claude Code can read its SEO record, but you review and publish it from Shopify admin. --- ## Write SEO blog post drafts with Claude Cowork URL: https://storeseo.com/agent/claude-cowork/blog-posts (full page: https://storeseo.com/agent/claude-cowork/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Claude Cowork means handing Anthropic's delegated-work agent a content brief, letting it drive StoreSEO to produce a draft article in your Shopify blog, and reviewing the result instead of supervising each step. In Cowork you describe the task rather than hold a conversation: write two drafts for my News blog from this list of topics, and tell me what each will cost first. Cowork opens a StoreSEO session with storeseo_start, calls storeseo_list_blogs for the blog id, and then comes back to you with the plan and the AI credit it expects to spend. This is the point where StoreSEO tells the agent to wait for your yes, and where Cowork's own permission step for the write tool applies. Once you agree, Cowork calls storeseo_generate_blog_post and then polls storeseo_get_blog_post_job. The writing takes a few minutes, which suits Cowork's style of working through a task on its own while you do something else. When a job completes, Cowork can read each draft's SEO record with storeseo_get_resource_seo and give you a short review: which post is ready, what the SEO checklist flags, and where the draft sits in the blog. You then open Shopify admin to edit and publish. Cowork also runs scheduled tasks, and this job is a place to be careful with them. Writing a post spends credit and needs your yes, and nobody is there to give it at three in the morning. A scheduled task can still do the safe half: list your blogs, read your existing articles with storeseo_list_resources, and send you a list of topic ideas for you to approve and start in a live session. Prompts: - Use StoreSEO to list my blogs and propose five post topics. Do not generate anything. - Write a draft on how to care for linen shirts for my News blog, about 1,200 words. Tell me the AI credit cost and wait for my yes before you start. - Check the state of the blog post job you started, and when it is done review the draft's SEO record and summarize it for me. In Claude Cowork: A scheduled Cowork task runs without you in the conversation, so it cannot wait for the yes that StoreSEO tells the agent to get before a credit-spending run. Keep scheduled tasks to reads such as listing blogs and suggesting topics, and tell the task in plain words not to generate, publish or delete anything. Start the actual post in a session where you are present to approve the cost. A connector added to Claude elsewhere may not appear in Cowork, so check Customize, Connectors inside Cowork first. Q: Can a Claude Cowork scheduled task write a blog post for me every week? A: It is not a good fit. Writing a post spends AI credit and StoreSEO tells the agent to wait for your yes, which a scheduled task cannot receive. Schedule a report-only task that suggests topics and start the post yourself. Q: Does Cowork publish the post when the job finishes? A: No. The result is a draft article in your Shopify blog. You review and publish it from Shopify admin. --- ## Write SEO blog post drafts with Cursor URL: https://storeseo.com/agent/cursor/blog-posts (full page: https://storeseo.com/agent/cursor/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Cursor means asking the editor's agent, from the project you have open, to have StoreSEO write a post into a draft article in your Shopify blog and to tell you when it is ready. You open the Cursor agent chat in your theme or content repository and ask for the post: a 1,000 word article on how often to wash denim, for the Guides blog. The agent calls storeseo_start and storeseo_list_blogs, and Cursor shows each MCP call in the chat so you can expand the arguments. If your repository holds a content calendar or a style guide, the agent can read it first and write the brief from it. Cursor asks for approval before an MCP tool runs unless you have changed the run mode, and StoreSEO adds its own instruction: state the credit cost, then wait for your yes. So before storeseo_generate_blog_post you get a plain message from the agent about the cost, and then Cursor's approval card, which shows the blog id and the brief. Approve the generate call each time. The agent then polls storeseo_get_blog_post_job until the job is completed, failed or cancelled, and reports the steps as it goes. The draft lands in the Shopify blog as a draft article. Ask the agent to read it with storeseo_get_resource_seo and write the article id and any SEO issues into a notes file in the repository, so the review list lives next to your other work. Publishing is done by you in Shopify admin. Prompts: - Use StoreSEO to list my blogs, then look at docs/content-calendar.md and suggest which post to write next. - Write a 1,000 word draft on how often to wash denim for the Guides blog. State the AI credit cost and wait for my approval before you start. - Check the blog post job and, when it completes, save the draft's SEO issues to docs/blog-notes.md. In Cursor: StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers, so you have to switch some off. This job needs storeseo_start, storeseo_list_blogs, storeseo_generate_blog_post, storeseo_get_blog_post_job and storeseo_get_resource_seo. If you trimmed the list to fit under the cap, check that the two blog tools are still on, otherwise the agent will say it cannot find them. Keep the generate tool on ask, because a tool set to run without asking leaves the chat confirmation as the only check. Reads such as listing blogs and polling the job are fine to leave on automatic if you want fewer prompts during a long generation. Q: Why can Cursor not find the blog post tools? A: Cursor allows about 40 active tools across all servers and StoreSEO has 44, so some are switched off. Open Cursor's MCP settings and make sure storeseo_list_blogs, storeseo_generate_blog_post and storeseo_get_blog_post_job are enabled. Q: Can Cursor save the blog post into my repository as well? A: StoreSEO saves the draft in your Shopify blog, not in your files. Cursor can read the draft's SEO record and write notes or the article id into your repository, but the post itself is edited and published in Shopify admin. --- ## Write SEO blog post drafts with Codex URL: https://storeseo.com/agent/codex/blog-posts (full page: https://storeseo.com/agent/codex/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Codex means giving OpenAI's coding agent a brief from the repository you run it in, and having StoreSEO write the post into a draft article in your Shopify blog while Codex tracks the job. You start Codex in a terminal, an IDE extension or the ChatGPT desktop app, all of which read the same config.toml, and ask for the post: an 1,100 word article on storing olive oil for the Recipes blog. Codex calls storeseo_start, then storeseo_list_blogs to get the blog id. These are reads, and with the server's default_tools_approval_mode set to writes, Codex lets read-only tools through and prompts for the rest. That setting is what puts the hard stop on storeseo_generate_blog_post. StoreSEO tells the agent to state the credit cost and wait for your yes, and Codex then shows its own approval prompt for the write tool, so you answer twice at the two layers. After approval the generation runs in the background for a few minutes, and Codex polls storeseo_get_blog_post_job, printing steps completed out of total until the status is terminal. Because the job is polled in short calls, the wait does not depend on one long tool call. When the job completes, ask Codex to read the draft's SEO record with storeseo_get_resource_seo and record the article id in a file in your working tree. You then review and publish the draft in Shopify admin. If your repository holds product copy or a tone guide, Codex can read it and fold it into the brief before you approve. Prompts: - Use StoreSEO to list my Shopify blogs and tell me their ids and how many articles each holds. - Write an 1,100 word draft on storing olive oil for the Recipes blog. Tell me the AI credit cost and wait for my yes before you start. - Poll the blog post job until it finishes, then read the draft's SEO record and write the article id to notes.md. In Codex: Codex waits 60 seconds for a tool to finish by default. StoreSEO answers storeseo_generate_blog_post with a job id and writes in the background, so the call itself should return quickly, but raise tool_timeout_sec for the storeseo server if you see timeouts on the start or on slow polls. Set the server's default_tools_approval_mode to writes or prompt, not approve, because approve skips the prompt for the generate tool and leaves the confirmation StoreSEO asks for in chat as the only check. Reads like listing blogs and polling the job can stay automatic, so only the credit-spending call interrupts you. Q: Does Codex stay blocked while the blog post is written? A: No single call blocks for the whole run. storeseo_generate_blog_post returns a job id and the writing happens in the background for a few minutes, while Codex polls storeseo_get_blog_post_job for progress. Q: Which Codex approval mode suits blog post drafts? A: Use writes or prompt for the storeseo server. Reads such as listing blogs and polling the job then run on their own, and storeseo_generate_blog_post, which spends credit, stops for your approval. --- ## Write SEO blog post drafts with Gemini CLI URL: https://storeseo.com/agent/gemini-cli/blog-posts (full page: https://storeseo.com/agent/gemini-cli/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Gemini CLI means asking Google's terminal agent for a post, confirming the cost, and having StoreSEO save the result as a draft article in your Shopify blog. You run Gemini CLI in a terminal in the folder you work from and type the request: a 900 word post on caring for leather boots for the Journal blog, with a featured image. The model sees StoreSEO tools under names like mcp_storeseo_storeseo_list_blogs, and it calls storeseo_start first and then storeseo_list_blogs to find the numeric blog id. The CLI prints each tool call as it runs. Gemini CLI asks you to confirm an MCP tool call before it runs unless the server has trust turned on. Leave trust off for StoreSEO. For storeseo_generate_blog_post, StoreSEO also tells the agent to state the AI credit cost, which grows with word count and the featured image, and to wait for your yes. You answer in the conversation and then at the CLI's confirmation prompt. The job then runs in the background for a few minutes while the agent polls storeseo_get_blog_post_job and reports progress. When the job completes, the agent can read the draft's SEO record with storeseo_get_resource_seo, using the article id from the job result, and summarize what to fix. The article is a draft in Shopify. You review it, edit it and publish it from Shopify admin. If you want to limit what the model can reach, an includeTools list on the server can expose only the five tools this job needs. Prompts: - Use StoreSEO to list my blogs and how many articles each one has. - Write a 900 word draft on caring for leather boots for the Journal blog with a featured image. Tell me the AI credit cost and wait for my yes. - Check the blog post job you started and report how many steps are done. When it finishes, read the draft's SEO record. In Gemini CLI: Do not set trust: true on the storeseo server. Trust skips every tool confirmation, so storeseo_generate_blog_post would run on the agent's say alone and the confirmation StoreSEO asks for in chat would be the only check. The browser sign-in also needs a browser on the same machine, so on a headless box or over SSH use the access token header, and then check that the token's connection still lists the blog tools. Google's site also says Gemini CLI was replaced by Antigravity CLI for unpaid tier and Google One users, so check which agent your account can use. Q: Should I turn on trust for StoreSEO in Gemini CLI to speed up blog drafts? A: No. Trust skips all tool confirmations, including the one for storeseo_generate_blog_post, which spends AI credit. Leave it off and approve the generate call when the CLI asks. Q: How do I limit Gemini CLI to just the blog tools? A: Add an includeTools list to the storeseo server in settings.json naming storeseo_start, storeseo_list_blogs, storeseo_generate_blog_post, storeseo_get_blog_post_job and storeseo_get_resource_seo. The model then sees only those tools from StoreSEO. --- ## Write SEO blog post drafts with GitHub Copilot URL: https://storeseo.com/agent/github-copilot/blog-posts (full page: https://storeseo.com/agent/github-copilot/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with GitHub Copilot means asking Agent mode in VS Code to have StoreSEO write a post into a draft article in your Shopify blog, then tracking the job from the same chat panel. You open Copilot Chat in VS Code, switch to Agent mode and ask for the post: a 1,300 word article on picking a yoga mat for the Guides blog. Agent mode calls storeseo_start and then storeseo_list_blogs, and VS Code lists the tools it used under the reply, so you can expand each call. The server comes from .vscode/mcp.json or your user configuration, and you can see it in MCP: List Servers. VS Code puts a confirmation in front of MCP tools, and StoreSEO adds its own instruction: the agent states the topic, keyword, length and credit cost, and waits for your yes. For storeseo_generate_blog_post you therefore get a message from Copilot and then the editor's confirmation card, where you can look at the arguments before you continue. The writing runs in the background for a few minutes. Copilot polls storeseo_get_blog_post_job and reports steps completed out of total, and the editor stays usable while it does. When the job completes, ask Copilot to read the draft with storeseo_get_resource_seo and, if you like, write the findings into a Markdown file in the workspace so they sit next to your code. The post itself is a draft article in Shopify. You edit it and publish it from Shopify admin. On Copilot Business and Enterprise an organization admin must have enabled MCP servers in Copilot policy before any of this works. Prompts: - Use the StoreSEO tools to list my Shopify blogs and their ids. - Write a 1,300 word draft on picking a yoga mat for the Guides blog. Tell me the AI credit cost and wait for my confirmation before you start. - Check the blog post job you started and, when it completes, save the draft's SEO issues to blog-notes.md in this workspace. In GitHub Copilot: Copy the StoreSEO config into .vscode/mcp.json with the servers key, not mcpServers, which belongs to the portable .mcp.json file. Keep storeseo_generate_blog_post on VS Code's confirmation prompt, because that prompt is the hard stop on a tool that spends credit. If you use an access token, keep it out of a committed .vscode/mcp.json, since the file is shared with your team, and use input variables instead. The same shared file means a teammate who opens the workspace gets the StoreSEO server too, so tell them it spends credit. Q: Why does Copilot Agent mode not show the StoreSEO blog tools? A: Check that your VS Code is version 1.99 or later, that the server is running in MCP: List Servers, and on Copilot Business or Enterprise that an admin has enabled MCP servers in Copilot policy. Then confirm the blog tools are selected in the tools picker. Q: Does Copilot publish the blog post on my store? A: No. StoreSEO saves it as a draft article in the Shopify blog you chose, and you publish it from Shopify admin. --- ## Write SEO blog post drafts with Windsurf URL: https://storeseo.com/agent/windsurf/blog-posts (full page: https://storeseo.com/agent/windsurf/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in Windsurf with StoreSEO means asking Cascade, from the editor you already work in, to choose a Shopify blog and have StoreSEO write a complete SEO post that lands as a draft article for you to review. You are in Windsurf with the Cascade panel open and StoreSEO listed in your MCP servers. You type the brief the way you would hand it to a colleague: the topic, the keyword, roughly how long, and whether you want a featured image. Cascade calls storeseo_start first, because every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs to find the blog the post will go into. That tool returns each blog's numeric id, title and article count, and only blogs already synced in the StoreSEO app appear, so an empty list sends you to the app rather than to another prompt. Before it writes anything, Cascade should tell you what the post will cost. Writing a post spends AI Content Optimizer credit, and the amount grows with the word count and with a featured image, so StoreSEO tells the agent to name the cost and wait for your yes. Once you agree, Cascade calls storeseo_generate_blog_post with the blog id and your brief. The writing runs in the background for a few minutes, so Cascade then calls storeseo_get_blog_post_job on the returned job id and reports how many steps are complete until the job is completed, failed or cancelled. While it runs, the draft article already exists in Shopify but is not finished, and Cascade should not describe it as done. The editor setting is a small advantage here. While the job runs you can keep working in the theme or the repository in another Cascade thread, and when the job completes you can ask Cascade to read the draft's SEO record with storeseo_get_resource_seo using resource type article. You can also ask it to save the brief and the keyword plan into a Markdown file in your project so the next post follows the same pattern. The article itself stays in Shopify as a draft; you open it in Shopify admin, edit it and publish it there. Prompts: - Use StoreSEO to list my blogs, then write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost before you start. - Write a StoreSEO draft post targeting best gifts for coffee lovers, with a featured image, and tell me how far along the job is. - Check the SEO record of the blog draft you just wrote and list what I should fix before publishing it. In Windsurf: Windsurf's MCP page says its configuration applies to the legacy Cascade agent, while the Devin Local agent, the default for new tabs, reads Devin CLI configuration files instead. If StoreSEO tools are missing when you ask for a blog draft, check which agent the tab is using. The page does not describe how Cascade approves MCP tool calls, so treat StoreSEO's cost confirmation as an instruction the agent follows and read the proposed call before you accept it. Cascade also caps tools at 100 in total across all enabled servers, and StoreSEO uses 44 of them. Q: Will Windsurf publish the blog post to my Shopify store? A: No. storeseo_generate_blog_post saves the post as a draft article in the blog you choose. You review it and publish it from Shopify admin. Q: Why does Cascade say I have no blogs to write into? A: storeseo_list_blogs only returns blogs synced with StoreSEO. Sync your blogs in the StoreSEO app, then ask Cascade to list them again. --- ## Write SEO blog post drafts with Grok URL: https://storeseo.com/agent/grok/blog-posts (full page: https://storeseo.com/agent/grok/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in Grok with StoreSEO means describing the post in a grok.com chat with the StoreSEO connector available, and getting a complete SEO article saved as a draft in the Shopify blog you pick. You are in a Grok chat on grok.com with the StoreSEO custom connector added and your Shopify store approved through the browser sign-in. You say what you want: a post on a topic, the keyword to aim at, a length, and whether it needs a featured image. Grok calls storeseo_start, which returns the conversation id every other StoreSEO tool needs, and then storeseo_list_blogs to show the blogs it can write into. Each blog comes back with its numeric id, title and article count, and only blogs synced in the StoreSEO app are listed, so a store that has never synced its blogs gets an empty answer and a pointer to the app. The cost conversation happens in the chat. Writing a post spends AI Content Optimizer credit, more for a longer post and more again for a featured image, and StoreSEO tells the agent to state the cost and wait for your yes. When you agree, Grok calls storeseo_generate_blog_post and then storeseo_get_blog_post_job with the job id. The job takes a few minutes, so you will usually see Grok report steps completed out of total rather than an article. Stay in the same chat while it runs, or come back to it and ask how far along the post is, and Grok will check the job again. When the status is completed you can have it read the draft's SEO record with storeseo_get_resource_seo using resource type article. What Grok gives you is a conversation, not a file. The article itself is not pasted into the chat as the deliverable; it exists as a draft in your Shopify blog, where you open it in Shopify admin, edit it and publish it. If you want a second post, ask for a different keyword in the same chat and Grok reuses the blog id it already found. Prompts: - Use StoreSEO to list my blogs and write a 1,000 word draft on how to choose running shoes. Tell me the credit cost before you start. - Write a StoreSEO blog draft targeting winter skincare routine for dry skin, with a featured image. - How far along is the blog post you started? Do not call it finished until the job says completed. In Grok: xAI's connector documentation does not describe a per-tool approval setting, so the check on spending credit is StoreSEO's instruction to name the cost and wait for your yes, which Grok follows in the conversation. Read the cost it quotes before you reply. Because the job runs for a few minutes, keep the chat open or return to it and ask for a status check, and do not treat a started job as a finished post. On Grok Business and Enterprise a team admin has to provision the connector first, so a missing StoreSEO tool may mean the admin step has not happened. Q: Can Grok publish the blog post on my store? A: No. StoreSEO saves what Grok generates as a draft article in your Shopify blog. You open it in Shopify admin, review it and publish it yourself. Q: What happens if I leave the Grok chat while the post is being written? A: The writing runs in the background in StoreSEO. Return to the chat and ask for a status check, and Grok calls storeseo_get_blog_post_job again until the job is completed, failed or cancelled. --- ## Write SEO blog post drafts with OpenClaw URL: https://storeseo.com/agent/openclaw/blog-posts (full page: https://storeseo.com/agent/openclaw/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with OpenClaw and StoreSEO means asking your OpenClaw agent, through whatever channel it answers on, to write a complete SEO post that StoreSEO saves as a draft in a Shopify blog you choose. OpenClaw is a runtime that stays on, so the way you reach it is a message rather than a window: a chat channel such as Telegram, or a session on the installation that owns the connector. You send the brief: topic, keyword, length and whether you want a featured image. The agent calls storeseo_start for the conversation id, then storeseo_list_blogs to find the blog and its numeric id. Only blogs synced in the StoreSEO app come back, so if the list is empty the next step is the app, not another message. The StoreSEO connection was set up with openclaw mcp add and openclaw mcp login, and the credentials live on that installation. Then comes the part that needs you. Writing a post spends AI Content Optimizer credit, and the cost grows with word count and a featured image, so StoreSEO tells the agent to say what it will cost and wait for your yes. A message in a chat is a natural place to give it. After the yes the agent calls storeseo_generate_blog_post and polls storeseo_get_blog_post_job, reporting how many steps are complete, until the job is completed, failed or cancelled. When it completes you can ask the agent to read the draft's SEO record with storeseo_get_resource_seo using resource type article, and then publish from Shopify admin. The runtime also invites an obvious idea that this job does not suit: putting the writing on an automation. A scheduled run has nobody to say yes, and generating a post spends credit and creates a draft every time it fires. Use openclaw automations for the reads around the job, for example a weekly report of how many articles each blog holds, and start the writing yourself from a message you can answer. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the cost and wait for my yes. - Write a StoreSEO draft on best gifts for coffee lovers with a featured image, and message me when the job finishes. - Check the blog post job you started and tell me how many steps are done. In OpenClaw: A scheduled OpenClaw automation cannot wait for a yes, and storeseo_generate_blog_post spends credit and creates a draft on every run, so do not put it in a cron-style prompt. Keep automations to reads, such as listing blogs or checking a job you started, and start each post from a message you can answer. OpenClaw has its own per-server approval setting, set with openclaw mcp configure storeseo --approval to approve, prompt or auto. If you want the client to stop before StoreSEO spends credit, do not leave the server on a mode that skips the check, because that setting, not StoreSEO, is the hard stop. Q: Can I schedule OpenClaw to write a blog post every week? A: It is not a good fit. A scheduled run has nobody to confirm the credit cost, and each run would create a new draft. Schedule read-only reports instead and start a post from a message you can answer. Q: Does OpenClaw publish the post to Shopify? A: No. StoreSEO saves the generated post as a draft article in the blog you choose. You review and publish it from Shopify admin. --- ## Write SEO blog post drafts with Hermes Agent URL: https://storeseo.com/agent/hermes-agent/blog-posts (full page: https://storeseo.com/agent/hermes-agent/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts with Hermes Agent and StoreSEO means telling Hermes in a chat session which post you want, and having it run StoreSEO's blog writer so the result is a draft article in your Shopify blog. Hermes Agent runs as a long-lived agent with a config file, a CLI and chat delivery targets, and you talk to it in a session. After adding StoreSEO under mcp_servers in config.yaml and signing in with hermes mcp login storeseo, you start a session, or run /reload-mcp in the one you have open, and describe the post: topic, keyword, length, featured image or not. Hermes calls storeseo_start first, since every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs, which gives each synced blog's numeric id, title and article count. A blog that is not synced in the StoreSEO app will not be in the list, and no tool here can sync it. Hermes then has to settle the cost with you. Writing a post spends AI Content Optimizer credit, which grows with the word count and with a featured image, and StoreSEO tells the agent to name the cost and wait for your yes before it calls storeseo_generate_blog_post. After your reply the writing runs for a few minutes in the background, and Hermes polls storeseo_get_blog_post_job with the job id, reporting progress and refusing to call the draft finished until the status is completed. A failed job returns an error message and the credit is settled automatically. On completion you can have it read the new draft with storeseo_get_resource_seo using resource type article. Hermes also has hermes cron, and it is worth being clear about what belongs there. Writing a post is not a good scheduled task: a cron run has nobody to say yes, and every run would spend credit and add a draft. Use cron for reports, for instance a weekly count of articles per blog, and run the blog writer from a session where you are present. You publish the finished draft from Shopify admin. Prompts: - Use StoreSEO to list my blogs, then write a 1,500 word draft on how to build a capsule wardrobe in the Journal blog. State the credit cost and wait for my yes. - Write a StoreSEO blog draft targeting best gifts for coffee lovers, without a featured image, and keep checking the job until it completes. - Check the SEO record of the blog draft from the last job and summarize what is still weak. In Hermes Agent: Start a new Hermes session or run /reload-mcp after editing config.yaml, otherwise the open session does not see StoreSEO and the blog tools appear to be missing. A hermes cron job cannot wait for a yes, and storeseo_generate_blog_post spends credit and creates a draft on every run, so keep scheduled jobs to reads and run the writer from a session where you can answer the cost question. If you want a read-only agent for audits, Hermes' tools include list can expose only the list and get tools, but then it cannot write posts either, so use a separate configuration when you want drafts. Q: Can hermes cron write a blog post on a schedule? A: It is not a good fit. A cron run cannot confirm the credit cost and would create a new draft each time. Use cron for read-only reports and run the blog writer in a session you are watching. Q: How do I know the Hermes blog draft is finished? A: Hermes polls storeseo_get_blog_post_job until the status is completed, failed or cancelled. Until then the draft article exists in Shopify but is not finished, and Hermes should not describe it as done. --- ## Write SEO blog post drafts with OpenCode URL: https://storeseo.com/agent/opencode/blog-posts (full page: https://storeseo.com/agent/opencode/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in OpenCode with StoreSEO means describing the post in your terminal session and letting StoreSEO's blog writer save a finished SEO article as a draft in the Shopify blog you pick. You are in OpenCode's terminal interface, usually inside a project, with StoreSEO added through opencode mcp add and signed in with opencode mcp auth storeseo. You type the brief: the topic, the keyword, a word count and whether a featured image is wanted. OpenCode calls storeseo_start first, since every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs. The result lists each synced blog with its numeric Shopify id, title and article count; if it is empty, the blogs have to be synced in the StoreSEO app, which no tool here can do. The next step is the one to read closely. Writing a post spends AI Content Optimizer credit, which grows with word count and with a featured image, so StoreSEO tells the agent to state the cost and wait for your yes. You answer in the session. OpenCode then calls storeseo_generate_blog_post with the blog id and your brief, and polls storeseo_get_blog_post_job with the returned job id. The writing takes a few minutes, and you see each poll as a tool call in the transcript with steps completed out of total, ending in completed, failed or cancelled. Once it completes, you can have OpenCode read the draft's record with storeseo_get_resource_seo using resource type article. Being in a terminal helps in small ways. You can keep the brief, the keyword list and the title options in a file in the project and tell OpenCode to use it, and you can keep working on code in another session while the job runs. What does not change is where the result lives: the post is a draft article inside Shopify, not a file in your repository, and you review it and publish it from Shopify admin. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Give me the credit cost and wait for my yes. - Read brief.md in this folder and write the StoreSEO blog draft it describes, with a featured image. Tell me how far along the job is. - Check the SEO record of the blog draft you just created and list the issues I should fix before publishing. In OpenCode: The OpenCode documentation describes deny rules for MCP tools but no ask-first mode, so a StoreSEO call that is not denied runs when the agent makes it. For this job that means the cost confirmation StoreSEO asks the agent for is the main check on spending credit, so read the cost it quotes before you answer. If you want a client-side stop, deny storeseo_generate_blog_post by name in the sessions where you are only auditing, and remove the rule for a session where you intend to write. Also remember that opencode mcp add writes to the project configuration unless you pass --global. Q: Does OpenCode ask me before it spends AI credit on a blog post? A: StoreSEO tells the agent to name the cost and wait for your yes, and OpenCode follows that in the session. OpenCode's own documentation describes deny rules but no ask-first mode, so deny the generate tool by name if you want a hard stop. Q: Where does the finished blog post end up? A: In the Shopify blog you chose, as a draft article. It is not written into your project folder. You review it and publish it from Shopify admin. --- ## Write SEO blog post drafts with Zed URL: https://storeseo.com/agent/zed/blog-posts (full page: https://storeseo.com/agent/zed/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in Zed with StoreSEO means asking the Agent Panel to choose a Shopify blog and run StoreSEO's blog writer, so a complete SEO post is saved as a draft for you to review and publish. You are in Zed with the Agent Panel open and StoreSEO showing a green dot under Settings, AI, MCP Servers. You type the brief: topic, keyword, length and whether to add a featured image. The agent calls storeseo_start, because every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs, which lists each synced blog with its numeric id, title and article count. A blog that is not synced in the StoreSEO app is not listed, and an empty list means you sync in the app first. Zed's own behavior shapes this job more than most. By default Zed asks before every tool action, reads included, so you will see a confirmation for storeseo_start, for storeseo_list_blogs and again for each poll of storeseo_get_blog_post_job. That is a lot of clicking for a job that runs for a few minutes, because the agent polls until the job is completed, failed or cancelled. Separately from Zed, StoreSEO tells the agent to name the credit cost and wait for your yes before storeseo_generate_blog_post, since writing a post spends AI Content Optimizer credit that grows with word count and a featured image. Those are two different questions, one from the client and one from the agent. A sensible setup is to allow the read tools you trust with per-tool keys, such as mcp:storeseo:storeseo_list_blogs and mcp:storeseo:storeseo_get_blog_post_job, and leave storeseo_generate_blog_post on confirm so Zed still stops before it spends credit. When the job completes, ask the agent to read the draft's SEO record with storeseo_get_resource_seo using resource type article. The post stays in Shopify as a draft, and you publish it from Shopify admin; nothing is written into your project. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost first. - Write a StoreSEO blog draft targeting best gifts for coffee lovers, with a featured image, and check the job until it is completed. - Read the SEO record of the blog draft the last job created and list what to fix before I publish it. In Zed: Zed's default is to ask before every tool action, reads included, and this job polls storeseo_get_blog_post_job for a few minutes, so you can face a string of prompts. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_get_blog_post_job and leave storeseo_generate_blog_post on confirm. If you set agent.tool_permissions.default to allow, Zed stops asking altogether, and then the cost confirmation StoreSEO asks the agent for in the panel is the only check before credit is spent, so use that setting sparingly. Q: Why does Zed keep asking me to confirm while the blog post is being written? A: Zed asks before tool actions by default, and the agent polls storeseo_get_blog_post_job until the job finishes. Allow that read tool with a per-tool key and keep the generate tool on confirm. Q: Will Zed publish the post to my Shopify blog? A: No. StoreSEO saves the post as a draft article in the blog you choose, and you publish it from Shopify admin. --- ## Write SEO blog post drafts with Kiro URL: https://storeseo.com/agent/kiro/blog-posts (full page: https://storeseo.com/agent/kiro/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in Kiro with StoreSEO means asking Kiro's agent to pick a Shopify blog and run StoreSEO's blog writer, so a finished SEO post is saved as a draft you review and publish yourself. You are in the Kiro IDE with StoreSEO defined in your MCP configuration, either in the workspace file .kiro/settings/mcp.json or in the user file. You describe the post in the agent chat: topic, keyword, length and whether it needs a featured image. Kiro calls storeseo_start first, since every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs, which returns each synced blog's numeric id, title and article count. Only blogs synced in the StoreSEO app appear, so an empty list sends you to the app. Two kinds of confirmation meet in this job. Kiro asks before it runs a tool unless the tool is in the autoApprove list for the server, and StoreSEO separately tells the agent to state the credit cost and wait for your yes before it writes. Writing a post spends AI Content Optimizer credit that grows with the word count and a featured image, so the second confirmation matters. After your yes, Kiro calls storeseo_generate_blog_post and then polls storeseo_get_blog_post_job with the job id. The run takes a few minutes, and the poll reports steps completed out of total until the status is completed, failed or cancelled. When it is done you can ask Kiro to read the draft with storeseo_get_resource_seo using resource type article. The practical configuration for this job is to put the quiet tools in autoApprove, for example storeseo_start, storeseo_list_blogs and storeseo_get_blog_post_job, so the polling does not interrupt you, and to leave storeseo_generate_blog_post out so Kiro still shows its own prompt before credit is spent. The draft stays in Shopify; you open it there, edit it and publish it. Nothing is added to your workspace files. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost and wait for my yes. - Write a StoreSEO blog draft targeting best gifts for coffee lovers with a featured image, and keep checking the job until it completes. - Check the SEO record of the draft you created and tell me which issues to fix before I publish it from Shopify admin. In Kiro: Keep storeseo_generate_blog_post out of autoApprove. That list tells Kiro to run a tool without asking, and a wildcard would run every StoreSEO tool, so the only prompt left before credit is spent would be the confirmation StoreSEO asks the agent for. List the individual read tools instead: storeseo_start, storeseo_list_blogs and storeseo_get_blog_post_job. Also check which file defines storeseo, because a workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, and a different sign-in could point the draft at another store. Q: Which StoreSEO tools should I put in Kiro's autoApprove for blog posts? A: The read tools: storeseo_start, storeseo_list_blogs and storeseo_get_blog_post_job. Leave storeseo_generate_blog_post out so Kiro still asks before credit is spent. Q: Does Kiro publish the post it writes? A: No. StoreSEO saves the generated post as a draft article in the Shopify blog you pick. You publish it from Shopify admin. --- ## Write SEO blog post drafts with Antigravity URL: https://storeseo.com/agent/antigravity/blog-posts (full page: https://storeseo.com/agent/antigravity/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog post drafts in Antigravity with StoreSEO means asking the agent in the IDE's side panel to choose a Shopify blog and run StoreSEO's blog writer, so a complete SEO post is saved as a draft for your review. You are in the Antigravity IDE with StoreSEO added in mcp_config.json using the serverUrl field and authenticated from Agent settings, under Customizations. In the agent panel you give the brief: topic, keyword, length and whether a featured image is wanted. The agent calls storeseo_start, because every StoreSEO tool needs the conversation id it returns, and then storeseo_list_blogs to list the blogs it can write into, each with a numeric id, title and article count. Only blogs synced in the StoreSEO app are returned, so an empty list means syncing in the app first. Before writing, the agent should tell you the cost. Writing a post spends AI Content Optimizer credit, which grows with the word count and a featured image, and StoreSEO tells the agent to name it and wait for your yes. Once you agree, the agent calls storeseo_generate_blog_post with the blog id and your brief, then polls storeseo_get_blog_post_job with the job id. The job runs for a few minutes in the background, and the poll reports steps completed out of total until the status is completed, failed or cancelled. A draft article exists in Shopify from the start of the run, so the agent should not call it finished early. When the job completes, you can have it read the draft with storeseo_get_resource_seo using resource type article. Working in an IDE means you can keep editing code while the job runs and keep the brief or keyword notes in a project file for the agent to read. The article itself does not come back into the project; it waits in your Shopify blog as a draft, and you review and publish it from Shopify admin. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost and wait for my yes. - Write a StoreSEO blog draft targeting best gifts for coffee lovers with a featured image, and report the job progress until it completes. - Read the SEO record of the blog draft you just wrote and list what to improve before I publish it. In Antigravity: If you add a policy that allows mcp(storeseo/*), Antigravity stops asking before StoreSEO tool calls, and then the confirmation StoreSEO asks the agent for is the only check before credit is spent on a post. Allow individual read tools such as storeseo_list_blogs and storeseo_get_blog_post_job instead, and leave storeseo_generate_blog_post to prompt. Also remember that the field in mcp_config.json is serverUrl, and an entry copied from another client with url will not connect, which shows up as missing blog tools rather than as an error about the post. Q: Should I allow every StoreSEO tool in Antigravity for blog posts? A: No. A blanket allow policy for the whole StoreSEO server removes Antigravity's prompt. Allow the read tools and leave storeseo_generate_blog_post to prompt, because it spends AI credit. Q: Where does the blog post written from Antigravity appear? A: As a draft article in the Shopify blog you chose, not in your project. You review it and publish it from Shopify admin. --- ## Audit your store's SEO with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/seo-audit (full page: https://storeseo.com/agent/chatgpt-dots/seo-audit/index.md). Job: Store SEO audit, described once above. Running a store SEO audit with a ChatGPT dot means giving an always-on OpenAI agent a standing StoreSEO check, so the lowest-scoring products, collections, pages and posts arrive as a recurring report instead of something you remember to open. You do not sit in front of this one. You tell your dot, once, to audit the store every Monday and say where the summary should land, and the task runs on its own cloud computer while you are doing something else. Each run opens with storeseo_start, which returns the conversation id and a snapshot of how many resources are optimized, need improvement or are not optimized. The dot then calls storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo on the handful at the bottom. Every one of those calls is a read, which is why an audit is the safest thing to put on a timer. What makes a dot different from a chat here is memory. OpenAI's help center describes dots as keeping context between conversations, so the Monday report can say which products climbed since last week, which are new at the bottom and which failing criterion keeps coming back, instead of restating a flat list. Ask for the report in the shape you will actually read: three bullet points, a table of the ten worst, or only changes since last run. If StoreSEO cannot be reached, tell the dot to say so rather than reuse last week's numbers. Keep the scheduled task to this reading. When the report shows a collection with a missing meta description, open a normal ChatGPT conversation with the same StoreSEO plugin and fix it there, where a preview can be shown and a yes can be given. The dot finds the problems; you decide which ones get changed, and the hard stop for changes is the approval setting on your plugin and your dot's custom rules. Prompts: - Every Monday at 8:00, use StoreSEO to list my ten lowest-scoring products and what changed since last week. Report only and change nothing. - Audit my collections and pages as well, worst first, and tell me which fixes I can ask for in a chat and which need me in the StoreSEO app. - If StoreSEO is unreachable on a run, tell me that instead of reusing the last report. In ChatGPT dots: A scheduled run has nobody watching it, so write the task as report only and check your dot's custom rules under Customize. StoreSEO's audit tools are all reads, but your plugin is shared with every other dot and ChatGPT chat you use, and a rule that lets actions run without asking would also cover the apply and generate tools if a later task wandered into them. Developer mode and the plan that allows it are set on the ChatGPT side, so a Business workspace may need an admin to publish the app before your dot can see it. Q: Does a scheduled ChatGPT dot audit change my Shopify store? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which are reads. Word the task as report only, and keep fixes for a conversation where you can see a preview first. Q: Can the dot tell me what improved since the last audit? A: It can compare against earlier reports when your dot keeps them in its memory, which OpenAI's help center describes for dots. Ask for changes since last run, and check the first few reports to see it is comparing correctly. --- ## Optimize meta tags and content with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/optimize-meta (full page: https://storeseo.com/agent/chatgpt-dots/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags and on-page content with a ChatGPT dot means splitting the work in two: the dot finds the resources whose meta fields are failing, and a supervised ChatGPT conversation previews and applies the StoreSEO rewrite. Meta rewriting is the job where a timer is least useful and a human is most useful, so the useful dot pattern is a hand-off. Your dot runs a scheduled check that calls storeseo_start and storeseo_get_resource_seo on the products, collections or pages you point it at, and reports which have a missing or too-short meta description, no focus keyword in the title or an empty AI snippet. That report costs nothing but reads. It arrives with a short list of the resources worth fixing first. The rewrite itself happens in a regular ChatGPT conversation using the same StoreSEO plugin. There you ask for storeseo_preview_resource_optimization on one resource, which makes StoreSEO write the meta title, description, tags, URL handle, FAQs and AI snippet server-side and projects the new score without saving anything. Previews spend AI Content Optimizer credit whether or not you apply them, so the agent should tell you the cost before it generates. When the handle would change, you see the old and new URL, because applying creates a redirect. Then storeseo_apply_resource_optimization writes only the parts you approved, and ChatGPT shows its own confirmation for that write. Do not put the preview on the schedule. A weekly task that previews ten products spends credit ten times a week on drafts nobody has read, and drafts expire after an hour anyway. storeseo_get_ai_optimization_settings is a fine read for the dot to include, because it lets the report say what StoreSEO already generates automatically for new products, but changing those settings with storeseo_update_ai_optimization_settings is a decision for a conversation. Prompts: - Every Friday, use StoreSEO to list products and pages whose meta description is missing or too short. Report only, no previews. - Preview a new meta title and description for the product Linen Shirt, keep the current URL handle, and tell me the credit cost first. - Apply the previewed meta description but skip the tags and the FAQ. In ChatGPT dots: A dot can run on a timer, and a preview spends credit each time it runs. Keep the scheduled task to storeseo_get_resource_seo reads and leave storeseo_preview_resource_optimization and storeseo_apply_resource_optimization to a supervised chat. StoreSEO tells the agent to name the cost and wait for your yes, but that is an instruction, and the hard stops are your plugin's write confirmation and your dot's custom rules. If your plan only gives ChatGPT read access to connected servers, previews may be all you can do from ChatGPT, and the apply step moves to the StoreSEO app. Q: Should I let my dot preview meta tags on a schedule? A: Not as a rule. Each preview spends AI Content Optimizer credit and expires after an hour, so a recurring preview pays for drafts nobody reviews. Schedule the check that finds weak meta, and preview from a chat when you are ready to read the result. Q: Does the dot write the meta copy itself? A: No. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet server-side. The dot asks for the preview, shows you what came back and applies the parts you approve. --- ## Optimize images and alt text with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/image-seo (full page: https://storeseo.com/agent/chatgpt-dots/image-seo/index.md). Job: Image SEO and alt text, described once above. Handling image SEO with a ChatGPT dot means letting an always-on agent watch which StoreSEO image checks (product, collection and article images lacking alt text, compression or a good file name) still fail, while the actual optimizing waits for a chat you are in. Image problems pile up quietly. A merchant adds twenty photos for a new range, nobody writes alt text, and the files stay at camera weight. That is exactly the kind of drift a dot is good at noticing. Schedule a task that calls storeseo_list_images with has_alt_text set to false, then again filtered to images that are not optimized, sorted so the heaviest files come first. The report names the resource each image belongs to and which of the three axes, compression, alt text and file name, is still open. Both calls are reads and cost nothing. Fixing is a different matter, and the tool shows why. storeseo_optimize_image works on one image per call, runs synchronously and takes the axes you ask for. Alt text and file names spend AI Content Optimizer credit, and compression needs the Image Optimizer add-on and counts against its monthly quota. An unattended dot looping over a long list would burn through both. So take the list the dot sent, open a ChatGPT conversation with the same StoreSEO plugin, and ask for the first five. Passing every axis you want in a single call per image avoids two calls racing on the same file. If you dislike a result, storeseo_restore_image reverts one or more axes, including bringing back the alt text you had before StoreSEO wrote one. Keep restore in the chat too. A dot that could undo an image on its own schedule could just as easily undo one you had kept on purpose. Prompts: - Every Wednesday, use StoreSEO to list images with no alt text and the five heaviest uncompressed images. Report only and change nothing. - Optimize alt text and file name for the images on the product Linen Shirt, one image per call, and tell me the credit cost first. - Restore the alt text on that last image to what I had before. In ChatGPT dots: Compression only works when the Image Optimizer add-on is on your store, and it counts against that add-on's monthly quota, so a scheduled dot should never loop storeseo_optimize_image across a catalog. Alt text and file names also spend AI credit on every image. Keep the timer to storeseo_list_images and run optimizing from a chat where you can see each call. A file name change swaps the image's underlying Shopify media, which is one more reason to keep it supervised. Q: Can my dot optimize images overnight on its own? A: It can if you instruct it to and your custom rules let tool calls run unattended, but we would not. Each image spends credit or add-on quota and a file name change swaps the media. Let the dot report and run optimization in a chat. Q: How does the dot know which images still need work? A: storeseo_list_images returns each image with the status of compression, alt text and file name, and it can filter to images with no alt text. The dot reads that list, so its report comes from your store's current state. --- ## Get products ready for AI search with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/ai-search-readiness (full page: https://storeseo.com/agent/chatgpt-dots/ai-search-readiness/index.md). Job: AI search readiness, described once above. Preparing products for AI search with a ChatGPT dot means pointing an always-on OpenAI agent at StoreSEO's GEO scores so it can report which listings AI shopping agents will struggle to read, while the rewriting stays in a conversation where you answer the questions. A ChatGPT dot is an interesting fit for this job because it is itself the kind of AI client that reads store data. The scheduled part is simple. Each week the dot calls storeseo_start, which includes a GEO overview for products, and then storeseo_get_resource_seo on the products you care about, reading the geo block for the score and the criteria that fail: a thin description, missing FAQs, no AI snippet, product schema switched off. The report ranks products by what is missing and notes which gaps a tool can close and which need you. The closing happens in chat. For FAQs and an AI snippet, storeseo_preview_resource_optimization drafts them without saving and spends AI credit, and storeseo_apply_resource_optimization writes what you approve. For a thin description the sequence is different and cannot be scheduled at all: storeseo_scan_product_description scores seven pillars, identity, materials, dimensions, features, use cases, what is included and care, and returns questions that only you can answer. The agent should put them to you word for word. Then storeseo_write_product_description saves a new description immediately, with no backup beyond the previous text it returns, so a yes is worth reading twice. Product schema is one more fix, via storeseo_update_schema_settings. Because the scan needs your answers and the rewrite is not reversible, a dot working alone would be guessing at facts about your products. Leave those to a supervised chat. The product also needs a focus keyword set in the StoreSEO app first, and no tool here can set one. Prompts: - Every Thursday, use StoreSEO to list my ten products with the lowest GEO score and the issues on each. Report only. - Scan the description of the product Linen Shirt and ask me the gap questions one by one. - Preview FAQs and an AI snippet for that product, and wait for my approval before applying. In ChatGPT dots: storeseo_write_product_description saves immediately and replaces the live description, and StoreSEO keeps no revision history beyond the previous text in the response. A dot cannot answer the scan's questions for you, and a scheduled task has nobody to say yes, so keep scanning and writing out of the timer. Check that your custom rules ask before a write, because StoreSEO's wait for your yes is an instruction, not a lock. A product without a focus keyword cannot be scanned until you set one in the StoreSEO app. Q: Can a dot rewrite my product descriptions on a schedule? A: It should not. The scan asks questions only you can answer, and the write replaces the live description with no history. A scheduled dot can report GEO scores and gaps, and the rewrite belongs in a chat. Q: What does the dot read to judge AI readiness? A: It reads storeseo_start for the GEO overview and storeseo_get_resource_seo for a product's GEO score and per-criterion analysis. Those are reads, so they can run on a schedule without credit. --- ## Publish llms.txt and agents.md with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/llms-txt-agents-md (full page: https://storeseo.com/agent/chatgpt-dots/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with a ChatGPT dot and StoreSEO means using an always-on agent to keep watch over the two discovery files AI crawlers and shopping agents read, while generating or replacing either file stays a deliberate step you take in a chat. These two files are published straight to your live theme, which makes the job a good example of what to automate and what to hold back. The watching is perfect for a dot. A weekly task calls storeseo_get_llms_txt_status and storeseo_get_agents_md_status, and the report says whether /llms.txt is serving, when it was last generated, whether settings changed since then so a rebuild is due, and whether a custom agents.md is live or Shopify's native default still serves. storeseo_list_llms_txt_versions adds which version is active. All are reads. The publishing is not a read. storeseo_generate_llms_txt writes templates/llms.txt.liquid into the live theme and uploads a copy to Shopify Files, and it runs for a minute or two, so the agent polls status until generation is done. storeseo_generate_agents_md spends AI credit and publishes live at once, replacing Shopify's default. Activating an older version with storeseo_activate_llms_txt_version changes what the public file says, and storeseo_delete_llms_txt_version and storeseo_reset_agents_md discard things permanently. Do those in a supervised ChatGPT conversation, where the agent has told you what goes live and you have said yes. A useful pairing: have the dot flag a stale file, then come to the chat with the report and say regenerate it. The dot saves you from remembering to check, and you keep the one-line approval on the part that touches your theme. Run the same check again after publishing, so the next report confirms the new version is the active one. Prompts: - Every Monday, use StoreSEO to check whether my llms.txt and agents.md are live and whether a rebuild is due. Report only. - Which llms.txt version is active, and how does it differ from the newest one? - Regenerate llms.txt now and tell me everything it will write to my theme before you start. In ChatGPT dots: Generation writes to your live theme, and agents.md generation spends AI credit and goes live immediately. A scheduled dot cannot wait for a yes, so keep the timer to the two status reads and the version list. StoreSEO's own settings can also schedule llms.txt regeneration, which is separate from your dot's task, so check what is already scheduled before you add a second thing that publishes. Deleting a version and resetting agents.md cannot be undone. Q: Can my dot regenerate llms.txt automatically each week? A: It could be told to, but generation writes to your live theme and uploads to Shopify Files. We suggest the dot reports whether a rebuild is due and you approve regeneration in a chat. Q: How do I undo agents.md if I do not like it? A: storeseo_reset_agents_md removes the custom file so Shopify's native default serves again. It discards the custom content with no version history, so it is a chat decision, not a scheduled one. --- ## Set up structured data with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/structured-data (full page: https://storeseo.com/agent/chatgpt-dots/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Managing structured data with a ChatGPT dot means asking an always-on OpenAI agent to read which JSON-LD schema types StoreSEO has switched on for your store, and keeping any change to those settings for a conversation where you can see the full object first. Schema settings rarely change, which is why drift is easy to miss: a theme update, a plan change or a teammate flipping a type off, and your products quietly stop emitting review markup. A dot is a decent watchman for that. Give it a monthly task that calls storeseo_list_schema_types and reports each type, whether it is enabled and whether it is configurable. For the configurable ones, product, review, product_merchant, organization and local_business, it can add storeseo_get_schema_settings and note what is filled in and what is empty. Every call is a read, and it costs no credit. Changing settings is trickier than it sounds, and the tool says why. storeseo_update_schema_settings replaces the stored settings object wholesale, with no merge, so the agent has to read the current settings first, edit the whole object using the field template and send it back. Some types depend on others: review, sale_dates and product_merchant need the product schema enabled first, and organization, local_business and product_merchant need a complete settings object in the same call. The four FAQ types need the FAQ Schema feature in your plan or the call is rejected. That level of care suits a conversation, where the dot can show you the object it is about to send and wait for your answer. On a timer it would send a replacement object nobody reviewed. So let the scheduled task say what is on and what is missing, and make the edit in chat. Prompts: - On the first of each month, use StoreSEO to list my schema types and tell me which are off or half-configured. Report only. - Read the product schema settings, show me the full object, and propose the change before you send anything. - Enable review schema after checking product schema is on, and tell me what else it depends on. In ChatGPT dots: Settings sent to storeseo_update_schema_settings replace what is stored, so an agent that skips the read can wipe fields you configured. A dot working unattended is the worst place for that. Keep the scheduled task to storeseo_list_schema_types and storeseo_get_schema_settings, and make edits in a chat where the object is shown first. The FAQ schema types are rejected on plans without the FAQ Schema feature, even for switching them off. Q: Can a ChatGPT dot fix my schema settings on its own? A: It can read them on a schedule, and that is what we would give it. Updates replace the whole settings object, so edits belong in a chat where the dot reads first and shows you the object. Q: Why was my schema change rejected? A: Usually a prerequisite is missing: review, sale_dates and product_merchant need product schema enabled, some types need complete settings, and the FAQ types need the FAQ Schema feature in your plan. --- ## Manage sitemaps and indexing with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/sitemaps (full page: https://storeseo.com/agent/chatgpt-dots/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with a ChatGPT dot means letting an always-on agent keep an eye on which products, collections, pages and articles are excluded or noindexed through StoreSEO, so a flag set by mistake does not sit unnoticed for months. Indexing flags are the opposite of image alt text: a single wrong one is costly and invisible. A product set to noindex stays out of search results, and nothing on your storefront looks different. That makes a read-only watcher valuable. Schedule a dot to call storeseo_list_resource_sitemaps for each resource type, products, collections, pages and articles, filtering on noindex, nofollow and sitemap exclusion, and to report anything on the list you did not intend. Keep a short note in the task of what is deliberately hidden, such as a thank-you page, so the report highlights only surprises. Changing a flag is one resource at a time through storeseo_update_resource_sitemap, which writes Shopify metafields for sitemap_disabled, no_index and no_follow. It spends no AI credit, but it changes whether search engines can find a page, so the dot should say which resource and which flag before it acts and wait for your yes in a conversation. The HTML sitemap is a longer thread: storeseo_get_html_sitemap_settings reads the state, storeseo_list_sitemap_menus gives you the menu edit links, and storeseo_update_html_sitemap with setup_step 1 generates the page. There is no step 2 in the agent flow, because you add the sitemap link to a menu yourself, and then step 3 marks setup complete. Because of that manual step and the replace-the-whole-config behavior of the settings call, set the HTML sitemap up in a chat with the dot walking you through it, and leave the dot's scheduled work to the flag audit. Prompts: - Every Tuesday, use StoreSEO to list products and pages that are noindexed or excluded from the sitemap, and flag any that are not on this list of intentional ones. Report only. - Set the collection Old Stock to noindex and tell me which flag you are writing before you do it. - Walk me through setting up the HTML sitemap and tell me when I need to add the menu link myself. In ChatGPT dots: A flag change affects whether search engines can find a page, and a scheduled dot has nobody to confirm it. Keep the timer to storeseo_list_resource_sitemaps reads and make flag changes in a chat. The HTML sitemap settings call replaces the whole config, so the dot must read it first. The menu link is added by you, because the chosen menu is not stored and no tool adds it. Q: Can my dot noindex pages on a schedule? A: It could if told to, but we would not. A wrong noindex hides a page from search and looks fine on the storefront. Let the dot report flags weekly and change them in a chat. Q: Does the dot set up the HTML sitemap for me? A: Mostly. It generates the page with storeseo_update_html_sitemap and hands you the menu links, but you add the link to a menu yourself before the final step marks setup complete. --- ## Track keyword rankings with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/keyword-tracking (full page: https://storeseo.com/agent/chatgpt-dots/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with a ChatGPT dot means letting an always-on agent read your StoreSEO rank data on a timer and bring back what moved, which is the one SEO job where a recurring report is the point. Rank tracking is a time series, and a dot is built for time series. Once keywords are being tracked, a Monday task calls storeseo_list_keywords for the latest Google position, rank delta, SERP features and competitor positions, and storeseo_get_keyword_history for any keyword that jumped or dropped. The report you want reads like a short briefing: the biggest climbers, the biggest losers, the keywords where a tracked competitor now sits above you. All of that is reading, but rank tracking is a Pro plan feature, so on other plans these calls will be refused and the dot should say so. Setting up the tracked list is a supervised job. storeseo_add_keywords needs the location and language to match the values from storeseo_list_rank_locations and storeseo_list_rank_languages exactly, uses a country as the location and counts against your plan's keyword limit. The response only confirms the keywords were queued, and positions show up later in storeseo_list_keywords, so the dot should not announce a rank straight after adding one. Competitors work the same way, up to five, through storeseo_add_competitors. Deleting a keyword with storeseo_delete_keyword removes its entire rank history, and there is no undo. Keyword clusters sit in between, and they are the one place a dot's patience helps. storeseo_generate_keyword_cluster spends AI credit, takes a couple of minutes and needs a focus keyword on the product, and storeseo_get_keyword_cluster is the poll. A dot is comfortable waiting on a poll like that, but the generation is a spend, so it belongs in a chat with your yes. Prompts: - Every Monday, use StoreSEO to list my tracked keywords and report the five biggest rank changes and any competitor that overtook me. Report only. - Add the keyword linen shirt for the United States in English on mobile, after checking the exact location and language values. - Build a keyword cluster for the product Linen Shirt and tell me the credit cost first. In ChatGPT dots: Rank tracking and competitors are Pro plan tools, so a scheduled dot on another plan will hit refusals every Monday and should tell you that instead of reporting an empty week. Ranks are also checked asynchronously: a keyword you add now has no position until a later check, so a report run straight afterwards will not show it. Keep adding and deleting keywords, and cluster generation, in a supervised chat. Q: Is keyword tracking a good job for a ChatGPT dot? A: Yes, for the reading. A weekly task over storeseo_list_keywords and storeseo_get_keyword_history reports movers and competitor changes. Adding keywords and building clusters stays in a chat, because they use plan limits and credit. Q: Why does a keyword I just added show no rank? A: StoreSEO queues the rank check when you add a keyword, and the position appears in storeseo_list_keywords after the check finishes. A dot reading straight away will find nothing yet. --- ## Read Search Console and Analytics with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/google-reports (full page: https://storeseo.com/agent/chatgpt-dots/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Getting Search Console and Analytics reports from a ChatGPT dot means giving an always-on agent a standing StoreSEO request for your store's Google numbers, so the monthly summary arrives written up instead of living in two dashboards. Of the ten StoreSEO jobs, this one fits a scheduled dot best, because nothing in it writes. A monthly task calls storeseo_get_search_console_insights with a dimension such as query or page, for the clicks, impressions, CTR and average position over a range, and storeseo_get_analytics_report with a report type such as search_traffic, traffic_source or page_performance. By default each looks back 28 days and compares against the previous equal-length period, so the deltas come with the data. The dot's job is to turn them into sentences: which queries are rising, which pages lost clicks, whether organic traffic moved as Search Console says it did. Give it a question rather than a dump. Ask where you rank between 8 and 20 for queries with real impressions, because those are the ones a better meta description or an FAQ might lift, then have the report end with two or three named pages to look at. Because OpenAI's help center describes dots as keeping context between conversations, you can ask for what changed since last month, and tell it to mention a missing connection plainly if Google is not linked. The reports need a paid StoreSEO plan with Google connected in the app, and GA4 needs a property selected there too. Nothing a dot reads here can change your store, so this is the one job where a rule that allows unattended calls costs you little, and you can say that explicitly in the task. Even so, word it as reads only, so a later edit to the task cannot widen it by accident. Prompts: - On the first of every month, use StoreSEO to pull Search Console clicks, impressions and average position for the last 28 days by query and page, and tell me the five biggest changes. - Pull the GA4 traffic source report for the same range and tell me whether organic search grew or shrank against the previous period. - Which queries do I rank between 8 and 20 for with high impressions? List the pages involved. In ChatGPT dots: These reports need a paid StoreSEO plan with Google connected, and GA4 needs a property selected in the app. If the connection lapses, a scheduled dot will get an error where it expected data, so tell it in the task to report that plainly rather than fall back on last month's numbers. The date range defaults to the last 28 days, so set an explicit range if you want a calendar month. Q: Can a ChatGPT dot send me a monthly Search Console summary? A: Yes, with a scheduled task that calls storeseo_get_search_console_insights and storeseo_get_analytics_report. Both are reads, and you need a paid StoreSEO plan with Google connected. Q: Why does my dot say Google reports are unavailable? A: Google reports need a paid StoreSEO plan with Google connected in the app, and GA4 needs a selected property. Check the connection in StoreSEO, then rerun the task. --- ## Write SEO blog post drafts with ChatGPT dots URL: https://storeseo.com/agent/chatgpt-dots/blog-posts (full page: https://storeseo.com/agent/chatgpt-dots/blog-posts/index.md). Job: Blog post drafts, described once above. Drafting blog posts with a ChatGPT dot means asking an always-on agent to start a StoreSEO generation, wait out the several minutes it takes and report on the saved draft, with the decision to spend the credit made by you in a chat. Blog generation is slow by design, and that suits an agent that does not need you to babysit a spinner. You pick a blog with storeseo_list_blogs, which returns the numeric id the generator needs and only lists blogs synced with StoreSEO. You ask for a post, the agent names the cost, and after your yes storeseo_generate_blog_post queues a job and returns a job_id. The post is written as a draft in your Shopify blog and nothing goes live. The dot then polls storeseo_get_blog_post_job, reporting steps completed out of total, until the job is completed, failed or cancelled. The dot adds something at the end. Once the job completes, it calls storeseo_get_resource_seo on the article, using the Shopify article id from the job payload, and tells you the SEO score and what still fails: a short meta description, no focus keyword in the title, images missing alt text. Then you open the draft in Shopify admin to read and publish it, because the draft only exists as a draft. Do not schedule the generation itself. Cost grows with word count and a featured image, so a weekly timer that writes a post spends credit every week on drafts nobody asked for. A better pattern is a scheduled read: have the dot list your articles with storeseo_list_resources, resource type article, sorted by score, and report the posts whose SEO is weakest. Write new posts, and rewrite old ones, in a conversation. Prompts: - List my blogs with StoreSEO, then tell me the credit cost of a 1,200 word post with a featured image about caring for linen before you start. - Generate that post as a draft in the News blog and poll the job until it finishes. - Once the draft is done, check its SEO score and tell me what still needs fixing. In ChatGPT dots: A generation spends AI Content Optimizer credit and takes a few minutes, and a dot that runs it from a timer would spend that credit with nobody reading the draft. Keep storeseo_generate_blog_post for a chat where you have said yes, and let a scheduled dot only read article scores. An empty storeseo_list_blogs result means your blogs have not been synced in the StoreSEO app, and no tool here can sync them. Drafts are never published by this tool; you publish them from Shopify admin. Q: Will my dot publish the blog post for me? A: No. storeseo_generate_blog_post saves a draft article in your Shopify blog, and you review and publish it from Shopify admin. The dot can poll the job and check the draft's SEO score. Q: Why does storeseo_list_blogs return nothing? A: Only blogs synced with StoreSEO are listed. Sync your blogs in the StoreSEO app first, then ask your dot again. --- ## Audit your store's SEO with Grok Bot URL: https://storeseo.com/agent/grok-bot/seo-audit (full page: https://storeseo.com/agent/grok-bot/seo-audit/index.md). Job: Store SEO audit, described once above. Running a store SEO audit with Grok Bot means giving one of xAI's named Bots the StoreSEO plugin and a standing question, so it reads your product, collection, page and post scores on its cloud computer and hands you a ranked worklist, even on days you never open the app. You work in a chat with one of your Bots, in the desktop or mobile app, and type @ to attach StoreSEO to the task. The first message can be as short as audit my store and tell me where to start. The Bot calls storeseo_start, which returns the conversation id every other StoreSEO tool needs plus a snapshot of how many resources are optimized, need improvement or are not optimized. It then calls storeseo_list_resources with the lowest scores first and storeseo_get_resource_seo on the worst candidates. All three are reads, although Grok Bot may still put its own Allow once, Always allow or Deny prompt in front of a tool the first time you use it. What makes this different from a chat assistant is that the audit can be a habit instead of a request. Because a Bot keeps its context and works on a persistent cloud computer, you can ask it to turn the same question into a routine: every Monday, list the weakest products, sort the fixes into what StoreSEO can generate and what only you can change in the app, and post the result in the conversation. The Bot shows you the next run time when it creates the routine, and the routine's recent run records show what it reported before, so you can tell whether the worklist is shrinking from week to week. Because nobody is watching a scheduled run, word the routine as audit only. The audit tools need no approval from StoreSEO's side, but a routine that is also allowed to preview or apply content would have no one to say yes. Keep the audit as the first half of a loop where you read the Monday report, pick two or three items, and then start a normal chat to preview and apply the fixes with the Bot waiting for you. Prompts: - Use @StoreSEO to rank my ten weakest products, name the biggest failing check on each and say which ones only I can fix in the app. - Create a routine for every Monday at 8:00 AM: use @StoreSEO to audit products, collections and pages, post the worst ten with their issues here, and change nothing. - Show me the last three reports from my SEO routine and tell me which products keep appearing in the worst ten. In Grok Bot: A routine only helps if its wording stays at report only. Grok Bot's documentation says a Test run performs real work and calls connected tools, so test the audit routine while it contains reads alone. If you later extend the same routine to preview or apply content, an Always allow rule on a StoreSEO write tool removes the second check in the client, and StoreSEO's instruction to wait for your yes becomes the only brake. Keep audit routines and fix conversations separate. Q: Can a Grok Bot run my StoreSEO audit while my laptop is closed? A: Yes. A routine runs on the Bot's cloud computer on its schedule, and background routines can run while your laptop is closed. Word it as audit only, because a scheduled run has nobody to answer an approval prompt. Q: Does the audit use any AI credit? A: The audit tools storeseo_start, storeseo_list_resources and storeseo_get_resource_seo only read your scores and checklists. Credit is spent later, when you ask for a preview or a generated file. --- ## Optimize meta tags and content with Grok Bot URL: https://storeseo.com/agent/grok-bot/optimize-meta (full page: https://storeseo.com/agent/grok-bot/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags with Grok Bot means asking a named xAI Bot to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for one Shopify resource, then applying only the parts of the draft you say yes to. You start in a chat with a Bot and attach the StoreSEO plugin with @. A good opening request names one resource, for example write better meta for my bestselling candle product, because each preview spends AI Content Optimizer credit. The Bot reads the current state with storeseo_get_resource_seo, then calls storeseo_preview_resource_optimization. StoreSEO writes the copy on its own servers, so the words come from StoreSEO rather than from Grok, and the response carries a draft id and the projected score change. The Bot lays the draft out in the chat so you can read the title, the description and the new URL handle next to the old one. The apply step is where Grok Bot's own controls come in. Calling storeseo_apply_resource_optimization is a write, so the Bot should ask you in plain words, and the client can add a prompt offering Allow once, Always allow or Deny. Choose Allow once while you are learning how the drafts look. If a handle changed, applying it creates a redirect from the old address, which is why the Bot should show both URLs before that call. You can approve only the meta and skip the handle, and the Bot passes just those fields along. The cloud computer helps with volume without changing the rules. You can ask the Bot to work through ten products in the same conversation, previewing one at a time and pausing for your yes before each apply. What it should not do is preview in a loop on a schedule: every preview spends credit whether or not you apply it, and a routine has no one to approve the result. Use a routine to find which products need meta, then do the previews in a chat. Prompts: - Use @StoreSEO to preview new meta for my Soy Candle product and show me the old and new title, description and handle before anything is saved. - Preview meta for my three lowest-scoring products, one at a time, and wait for my yes after each one before you apply it. - Apply only the meta title and description from that draft and skip the URL handle. In Grok Bot: Treat Always allow with care here. Grok Bot saves it as a rule for matching actions, so if you choose it on storeseo_apply_resource_optimization or storeseo_preview_resource_optimization, the client stops asking and only StoreSEO's own preview-first instruction remains. A Test run on a routine also performs real work, including connected tool calls, so never test a routine that contains a preview or apply. In Auto-review, an Ask first rule on the apply tool is the hard stop. Q: Does Grok write the meta text itself? A: No. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet on its own servers. The Bot fetches the draft, shows it to you and passes along only the parts you approve. Q: How do I keep a Grok Bot from applying meta without my yes? A: StoreSEO tells the agent to wait for your yes, but the hard stop is your client setting. Leave the apply tool on Allow once or add an Ask first rule in Auto-review, and avoid Always allow on it. --- ## Optimize images and alt text with Grok Bot URL: https://storeseo.com/agent/grok-bot/image-seo (full page: https://storeseo.com/agent/grok-bot/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing Shopify image SEO with Grok Bot means asking a persistent xAI Bot to list the heaviest or unlabelled images through StoreSEO and then optimize them one call at a time, so compression, alt text and file names get done in a long-running session. Image work is repetitive, which suits a Bot that keeps its place in a conversation. You attach StoreSEO with @ and ask which product images are heaviest or have no alt text. The Bot calls storeseo_start and then storeseo_list_images, which sorts by file size by default and reports compression, alt text and file name status for every image. That listing is a read, so the first answer is a shortlist with sizes and a count, and nothing has been spent yet. You can filter in plain words, for instance only images with no alt text on one collection. The write step is storeseo_optimize_image, and it takes one image per call with every wanted axis in the same call. A Bot does not run these in parallel against the same image, which matters because a file rename swaps the underlying Shopify media. Each call returns a report per axis, so the Bot can tell you that compression applied while alt text spent AI credit, or that compression was refused because the Image Optimizer add-on is missing. If you dislike one result, ask the Bot to call storeseo_restore_image for that axis. Because a Bot can keep going while you do something else, say the scope up front: five images, product images only, alt text and compression but no renaming. A renamed file gets a new storefront URL, so anything that hard-coded the old address keeps serving the old file, and the Bot should tell you which images it renamed. Do not put the optimize step in a routine. Use a routine at most to count unlabelled images each week, and do the optimizing in a conversation where you can answer. Prompts: - Use @StoreSEO to list my ten heaviest product images and tell me which have no alt text. Read only for now. - Optimize the alt text and compression of those first five images, one at a time, and tell me the result for each axis. - Restore the file name on the last image you optimized, I do not want the new URL. In Grok Bot: A Bot is happy to keep working, which makes Always allow on storeseo_optimize_image tempting for a batch of fifty images. Once saved, that rule means the client no longer stops between images, and each call spends Image Optimizer quota or AI credit. Use Allow once for a first image, state the number of images in your request, and use an Ask first rule in Auto-review if you want a pause before every call. If a response is lost, have the Bot re-list the image rather than retry. Q: Can Grok Bot optimize many images while I am away? A: It can keep working in a long conversation, but each storeseo_optimize_image call is a write that spends quota or credit. Set the number of images in your request and keep an Ask first rule on that tool unless you accept an unattended run. Q: What if I dislike how Grok Bot renamed or captioned an image? A: Ask it to call storeseo_restore_image for that image and name the axis, which can be compression, alt text or file name. Restoring a file name changes the image URL back as well. --- ## Get products ready for AI search with Grok Bot URL: https://storeseo.com/agent/grok-bot/ai-search-readiness (full page: https://storeseo.com/agent/grok-bot/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Grok Bot means asking a Bot to read each product's GEO score through StoreSEO, fix what a tool can fix, and rewrite a thin description after you answer StoreSEO's questions about the product. The GEO score measures how well AI shopping agents can understand a product, and it applies to products only. You attach StoreSEO with @ and ask the Bot which products are weakest for AI readiness. It calls storeseo_start, which includes a GEO overview, then storeseo_get_resource_seo on a product to read the failing criteria. Some of those are fixable by tools, such as FAQs, the AI snippet, image alt text and product schema, and others, such as the title, category and price, you change in Shopify or the StoreSEO app. A good Bot says which is which for each product. The description rewrite is the part where a Bot is more than a relay. storeseo_scan_product_description scores seven pillars and returns gap questions, and the Bot must put them to you word for word and in order, then pass your answers to storeseo_write_product_description. In a Grok Bot chat that is a short back and forth where you answer in your own words about materials, dimensions and care. The Bot has a filesystem on its cloud computer, so you can also ask it to save the old description to a file before the write, since the response returns the previous text once and StoreSEO keeps no history. Both the scan and the write are guarded in practice by your approval choices, and you should treat them differently. The scan and a preview spend credit but save nothing. The write replaces the live description at once with no undo. Do that one product at a time in a conversation, and keep it out of routines. A routine is fine for a weekly look at the GEO scores, worded as report only. Prompts: - Use @StoreSEO to list my five products with the lowest GEO score and say which failing criteria a tool can fix. - Scan the description of my Linen Throw product, ask me the gap questions word for word, and save the current description to a file before writing anything. - Preview FAQs and an AI snippet for my three weakest products and wait for my yes before applying each. In Grok Bot: storeseo_write_product_description saves immediately and cannot be undone, so it is the one tool where an Always allow rule in Grok Bot costs the most. With that rule the client no longer pauses, and StoreSEO's own request to wait for your yes is the only check left. Keep it on Allow once or an Ask first rule in Auto-review. The product also needs a focus keyword set in the StoreSEO app before a scan, so a Bot cannot fix that for you. Q: Does Grok Bot answer the description questions for me? A: It should not. StoreSEO tells the agent to put the scan's questions to you word for word and in order. If you decline to answer, or authorize a bulk run without questions, the write can proceed without them. Q: Can I get the old description back after a Grok Bot rewrite? A: Only from the copy returned by the write call, because StoreSEO keeps no history. Ask the Bot to save that text to a file on its cloud computer before it writes, and to show it to you afterwards. --- ## Publish llms.txt and agents.md with Grok Bot URL: https://storeseo.com/agent/grok-bot/llms-txt-agents-md (full page: https://storeseo.com/agent/grok-bot/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Grok Bot means asking a Bot to check which AI discovery files your Shopify store serves, then having StoreSEO generate and publish the missing ones to your live theme after you agree to each. Two files tell AI crawlers and AI shopping agents about your store: llms.txt, a curated map of your content, and agents.md, a description of who you are and how to buy. You attach StoreSEO to a Bot chat with @ and ask whether either file is live and current. The Bot calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status, all reads, and tells you what is served, when llms.txt last generated, and whether your settings changed since. If you want to see an older llms.txt version, storeseo_list_llms_txt_versions lists the latest ten. Generation is where the Bot has to be honest about cost and reach. storeseo_generate_llms_txt writes a template into your live theme and uploads a copy to Shopify Files in the same run, and it takes about one to two minutes. The Bot should say so, wait for your yes, then poll the status call until generation is no longer ongoing. A cloud computer is a good place to wait, because the Bot can carry on without your device. storeseo_generate_agents_md also publishes live and spends AI credit, and storeseo_reset_agents_md hands the file back to Shopify's default with no history. There is a natural routine here, and it is a quiet one. Ask the Bot every Monday to read both statuses and tell you if llms.txt is missing, if the active version is older than the newest, or if a rebuild is due. StoreSEO can regenerate llms.txt on its own saved schedule, so the routine does not need to generate anything. Leave generation, activation, deletion and reset to a conversation where you can answer the Bot's question. Prompts: - Use @StoreSEO to tell me whether my llms.txt and agents.md are live, when llms.txt last generated and which version is active. - Explain what generating llms.txt will write to my theme and Shopify Files, and wait for my yes before you start it. - Create a Monday routine that reads both file statuses with @StoreSEO and reports anything missing or out of date. Report only and generate nothing. In Grok Bot: A routine can easily drift into doing the work. If the instruction says keep llms.txt current, a Bot may decide to call storeseo_generate_llms_txt on every run, which publishes to the live theme with no one to say yes. Word the routine as read the statuses and report, and add an Ask first rule in Auto-review for the generate, activate, delete and reset tools. Remember that a Test run performs real work, so test only a routine that reads. Q: Will a Grok Bot routine regenerate llms.txt on its own? A: Only if you word the routine that way, and it is better not to. StoreSEO has its own saved schedule for llms.txt, so a routine can just read the status and report whether a rebuild is due. Q: How do I undo agents.md from Grok Bot? A: Ask the Bot to call storeseo_reset_agents_md, which makes Shopify's default file serve again within seconds. The custom text is gone for good, so the Bot should confirm with you first. --- ## Set up structured data with Grok Bot URL: https://storeseo.com/agent/grok-bot/structured-data (full page: https://storeseo.com/agent/grok-bot/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Grok Bot means asking a Bot to list your StoreSEO schema types, read the current settings of the ones you care about, and send back a complete, edited settings object once you confirm the change to your live markup. Structured data is a place where one careless write can overwrite good settings, so the sequence matters. You attach StoreSEO with @ and ask which schema types are on. The Bot calls storeseo_start and storeseo_list_schema_types, which shows every JSON-LD type, whether it is enabled and whether it can be configured. For one type, say organization, it calls storeseo_get_schema_settings, which returns the current values and a field template describing the full object to send back. These calls are reads, so you can have the Bot summarize the whole set in a single reply. Updates through storeseo_update_schema_settings replace the stored settings wholesale. That is a good fit for a Bot with a filesystem, because you can ask it to save the settings it just read to a file on its cloud computer before it edits anything. If the new version looks wrong, the saved copy is what you ask it to send back. You supply the business details, such as the legal name, address and opening hours, in the chat, and the Bot fills the template. It should not invent any of them. Some types depend on others. Review and product merchant schema need product schema on first, and the four FAQ types need the FAQ Schema feature in your plan, so the Bot may report a refusal rather than a success. Each change alters the markup your live storefront serves, so keep it in a conversation and out of routines. A routine that reads the type list and tells you what changed since last week is the most a scheduled run should do here. Prompts: - Use @StoreSEO to list my schema types and tell me which are on, which are off and which I can configure. - Read my organization schema settings, save them to a file first, then show me what a complete set would look like with my business details. - Turn on product schema and send the full settings, but show me the change and wait for my yes before you call the update. In Grok Bot: Because storeseo_update_schema_settings replaces the stored object, an Always allow rule in Grok Bot turns a mistaken edit into a silent overwrite of working settings. The client no longer asks, and StoreSEO's read-before-write instruction is the only check, so keep the tool on Allow once or an Ask first rule. If you need a rollback, ask the Bot to save the object it read before the update. Q: Can Grok Bot invent my business address for organization schema? A: It should not. StoreSEO's guidance is that business details come from you, so give the Bot the name, address and hours and have it show the complete object before it calls the update. Q: Why did the Bot say a schema type was refused? A: Some types need a plan feature or another schema on first. The FAQ types need the FAQ Schema feature, and review needs product schema enabled, so the Bot should name the missing piece instead of retrying. --- ## Manage sitemaps and indexing with Grok Bot URL: https://storeseo.com/agent/grok-bot/sitemaps (full page: https://storeseo.com/agent/grok-bot/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps with Grok Bot means asking a Bot to list which Shopify resources StoreSEO leaves out of the XML sitemap or marks noindex or nofollow, then changing those flags one resource at a time after you confirm each one. A single noindex flag can take a page out of search results, so this job rewards a careful reader. You attach StoreSEO to a Bot chat with @ and ask what is currently hidden from search. The Bot calls storeseo_start and then storeseo_list_resource_sitemaps for one resource type at a time, filtering on sitemap, noindex or nofollow. It returns the flags for each product, collection, page or article, and you can ask it to group them, for instance every noindex page and every product left out of the sitemap, so you can decide which of them were intended. Changes go through storeseo_update_resource_sitemap, one resource per call, and it writes Shopify metafields on that resource. In Grok Bot, that call is a write, so the Bot should name the resource and the flag, wait for your yes, and the client may add Allow once, Always allow or Deny. After each change the Bot should list the flags again before telling you the state. The HTML sitemap adds another thread: storeseo_update_html_sitemap generates it, you add the link to a menu in Shopify admin using the edit URL from storeseo_list_sitemap_menus, and the Bot then marks setup complete. A weekly routine suits the read side of this job well. Ask the Bot to list noindex and sitemap-excluded resources each Monday and tell you what is new compared with the last report, which catches an accidental flag early. Keep that routine report only. Changing a flag from a routine would remove a page from search with no one watching, and a Test run would perform the change for real. Prompts: - Use @StoreSEO to list every page and product marked noindex or excluded from the sitemap, and ask me which of them are intentional. - Remove my Summer Sale page from the sitemap with StoreSEO, name the flag you will change and wait for my yes first. - Set up the HTML sitemap, then give me the menu edit link and tell me what I have to do in Shopify admin before you mark it complete. In Grok Bot: A noindex change takes a page out of search results, and a Bot that is allowed to flag resources unattended could hide the wrong ones. Do not give a sitemap routine write permission. If you save Always allow on storeseo_update_resource_sitemap, the client stops asking, and an Ask first rule in Auto-review is the way to keep a pause on every flag change. The Bot also cannot add the sitemap link to your menu, because that is a manual Shopify admin step. Q: Can a Grok Bot routine flag pages as noindex for me? A: It can if you word it that way, but a scheduled run has nobody to approve the change and a Test run does the work for real. Keep routines to listing flags and reporting what is new. Q: Can Grok Bot add the HTML sitemap to my menu? A: No. After it generates the sitemap, you add the link to a menu in Shopify admin yourself, using the edit link the Bot gives you, and then the Bot can mark setup complete. --- ## Track keyword rankings with Grok Bot URL: https://storeseo.com/agent/grok-bot/keyword-tracking (full page: https://storeseo.com/agent/grok-bot/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Grok Bot means asking a Bot to add keywords to StoreSEO's Google rank tracker, read how positions move by country, language and device, and post a regular digest from a routine on its cloud computer. Rank tracking is a Pro feature, and it is the job where a Bot's habit of coming back pays off most. You attach StoreSEO to a chat with @ and say which keywords matter and in which market. The Bot calls storeseo_start, then looks up the exact country and language values with storeseo_list_rank_locations and storeseo_list_rank_languages, because storeseo_add_keywords needs those names verbatim and the Bot should not type them from memory. It asks you for the device, desktop or mobile, then adds the keywords. Positions arrive later, because the rank check is queued, so the first read may show no rank yet. Once keywords are tracked, reading them is cheap and safe. storeseo_list_keywords returns the latest rank and change, storeseo_get_keyword_history returns the dated history, and storeseo_list_competitors shows up to five competitor domains alongside. A Grok Bot routine fits this directly: every Monday, list the keywords that moved most, name any that fell out of the top ten, and post the result in the conversation. The run records under the Bot's conversation details let you page back through earlier digests and see the trend without keeping notes yourself. Deleting is the part to keep out of automation. storeseo_delete_keyword and storeseo_delete_competitor erase the history for good, and adding keywords counts against your plan's keyword limit. The tracker covers Google only, not AI answer engines, so the Bot should not report AI visibility from it. Keep the digest routine as read only and do additions and deletions in a chat where you can answer. Prompts: - Use @StoreSEO to add the keywords soy candles and scented candles for the United States, English, on mobile, then confirm the rank check was queued. - Show my tracked keywords that dropped the most this month and the history for the worst one. - Create a Monday routine that lists my keyword rank changes with @StoreSEO and posts a short digest here. Report only, no adding or deleting. In Grok Bot: Rank tracking needs the Pro plan, so on a lower plan the Bot should explain that instead of looping on refusals, and a routine would only report that failure each Monday. Be careful with deletes: storeseo_delete_keyword removes the entire rank history, and with Always allow saved on that tool the client no longer stops you. Keep the weekly digest report only, because a Test run performs real work. Q: Does a Grok Bot routine need my approval to read ranks? A: The read tools only fetch data, but Grok Bot may still prompt the first time with Allow once, Always allow or Deny. Saving Always allow for a read tool is reasonable if you keep write tools on Ask first. Q: Can the Bot track how I appear in ChatGPT or Gemini? A: No. StoreSEO's rank tracker checks Google only, and AI search engines are not tracked, so a Grok Bot digest covers Google positions and competitor domains. --- ## Read Search Console and Analytics with Grok Bot URL: https://storeseo.com/agent/grok-bot/google-reports (full page: https://storeseo.com/agent/grok-bot/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with Grok Bot means asking a Bot to pull your Search Console and Google Analytics 4 figures through StoreSEO and write the answer as a short summary, which makes it the most natural job to put on a weekly routine. Both Google tools are reads, so this job is where a Bot can run with the least friction. You attach StoreSEO with @ and ask a plain question, such as which queries bring the most clicks and which pages sit between positions eight and twenty. The Bot calls storeseo_start, then storeseo_get_search_console_insights with the dimension that fits, which can be date, query, page, country, device or search appearance, and storeseo_get_analytics_report for GA4 views such as traffic source or page performance. Both default to the last 28 days and compare with the previous period. The routine is the point. Ask the Bot to run both reports every Monday, name the three biggest gains and drops, and post them in the conversation. Because the data comes from Google through your StoreSEO connection, the Bot reports the figures as returned and should not project outcomes from them. The run records show earlier summaries, so a decline over three weeks is visible without you opening Search Console, and the Bot can follow a finding into an audit: a page with many impressions and a low position becomes a candidate for a meta preview in a later chat. Because these tools only read, you can reasonably save Always allow for them, though the saved rule matches by tool, so scope it to these two and leave write tools on Ask first. The setup itself lives in the StoreSEO app: a paid plan, Google connected, and a GA4 property selected for the Analytics report. If that is missing, the Bot cannot fix it and should hand you the link rather than loop. Prompts: - Use @StoreSEO to show Search Console queries from the last 28 days where we rank between 8 and 20, with impressions and CTR. - Compare this month's organic traffic by channel with last month using the GA4 report from @StoreSEO and tell me the biggest change. - Create a routine for every Monday at 9:00 AM that reads both Google reports with @StoreSEO and posts three gains and three drops here. Report only. In Grok Bot: These reports need a paid StoreSEO plan with Google connected, and the Analytics report also needs a GA4 property selected. A weekly routine without that setup will only report an error each Monday, so test the questions in a chat first. If you save Always allow for the two read tools, do not extend it to StoreSEO write tools, because an Always allow rule removes the client's check there. Q: Can a Grok Bot connect my Google account for me? A: No. Connecting Google and choosing the GA4 property happen in the StoreSEO app, and the Bot only reads the reports after that. If it is missing, the Bot should tell you to connect it there. Q: Is it safe to leave a Google report routine running every week? A: Both report tools are read-only, so a routine that only calls them changes nothing in your store. Keep write tools out of that routine and in a separate conversation. --- ## Write SEO blog post drafts with Grok Bot URL: https://storeseo.com/agent/grok-bot/blog-posts (full page: https://storeseo.com/agent/grok-bot/blog-posts/index.md). Job: Blog post drafts, described once above. Drafting blog posts with Grok Bot means asking a Bot to have StoreSEO write a complete draft article in one of your Shopify blogs, watch the background job until it finishes, and then read the draft's SEO checklist before you publish anything yourself. A blog post takes StoreSEO a few minutes to write, which is an odd fit for a chat that wants an answer now and a natural fit for a Bot that keeps working. You attach StoreSEO with @ and give a brief: the topic, the audience, a target word count and whether you want a featured image. The Bot calls storeseo_start, then storeseo_list_blogs to find the numeric id of the blog, since only blogs synced in the StoreSEO app are listed. It states the credit cost, which grows with word count and a featured image, and asks for your yes before storeseo_generate_blog_post. After that, the Bot polls storeseo_get_blog_post_job until the job is completed, failed or cancelled, and reports progress as steps completed out of total. The draft article exists from the start of the run, so a careful Bot does not describe it as finished while the job is still going. Waiting on the Bot's cloud computer means you can close the app and come back to a finished draft in the conversation. When the job completes, the Bot can call storeseo_get_resource_seo with the article's Shopify id to read how the draft scores. The post is only ever a draft. You review it and publish it from Shopify admin, which is a step the Bot cannot take. Keep generation out of routines: each run spends credit, and a scheduled Bot would produce drafts nobody asked for. A better routine is a monthly reminder to suggest three topics from your Search Console queries, with the generating done later in a chat where you approve the cost. Prompts: - Use @StoreSEO to list my blogs, then tell me the credit cost of a 1,200 word post about caring for linen bedding before you write it. - Write that post as a draft in the News blog with a featured image, wait until the job finishes, then tell me the SEO score of the draft. - Suggest three blog topics from my Search Console queries with @StoreSEO. Suggest only, do not generate anything. In Grok Bot: Each storeseo_generate_blog_post call spends credit, so a routine that generates posts would burn credit on every run, and a Test run performs real work too. With Always allow saved on that tool, the client no longer asks before each post, leaving StoreSEO's cost-first instruction as the only check. Keep generation in a conversation on Allow once or Ask first, and publish the draft yourself in Shopify admin. Q: Can Grok Bot publish the blog post for me? A: No. StoreSEO saves the post as a draft in your Shopify blog and never publishes it live. You review it and publish it from Shopify admin. Q: What happens if I close the app while the post is generating? A: The job runs in the background on StoreSEO's side, and the Bot works on a cloud computer, so it can keep polling. Ask for the job status when you return and treat the draft as unfinished until it completes. --- ## Audit your store's SEO with Muse URL: https://storeseo.com/agent/muse/seo-audit (full page: https://storeseo.com/agent/muse/seo-audit/index.md). Job: Store SEO audit, described once above. Running a store SEO audit with Muse means texting Meta's personal AI agent a question such as which of my products score worst, and letting it read your StoreSEO scores and checklists so a ranked worklist arrives in the same thread you use for everything else. You are in a thread with Muse, in its app, on muse.ai or in WhatsApp, and StoreSEO is already added as a custom connector. The first message can be a single line: audit my store and tell me where to start. Muse calls storeseo_start, which returns the conversation id every other StoreSEO tool needs and a snapshot of how many products, collections, pages and posts are optimized, need improvement or are not optimized. It then calls storeseo_list_resources with the lowest scores first and storeseo_get_resource_seo on the products it wants to explain. All three are reads, so the answer arrives without a question in between. A phone is where this format fits best. A reply that lands in a message thread has to be short enough to read on a small screen, so ask Muse for the worst five instead of the worst fifty, and for one line per product: the score, the failing criterion that costs the most and whether StoreSEO can fix it. Muse can then split the list into fixes you can ask for in the same thread, such as meta, FAQs and image alt text, and fixes only you can make in the StoreSEO app, such as setting a focus keyword. Products also carry a GEO score for AI readiness, so one pass shows which listings AI shopping agents will struggle with. Muse runs on its own Secure VM for each person, and Meta says a separate Sentinel agent decides what leaves that machine. That check sits around Muse and is separate from StoreSEO, so you may see Muse pause on a first call to the connector and ask you to confirm, even though an audit only reads. Treat that as the client doing its job and answer it. Because you control the thread, the audit is a conversation: follow up with only collections, or only products that are marked not optimized, and Muse reuses the session it already opened. Prompts: - Use StoreSEO to list my five lowest-scoring products, one line each: score, the biggest failing check and whether you can fix it from here. - Which of my collections and pages are not optimized yet? Read only, change nothing. - Which products have the weakest GEO score for AI shopping agents, and what is the first thing missing on each? In Muse: StoreSEO is a custom connector in Muse, and Meta says it does not review custom connectors, so the first audit is also the moment to check that the sign-in you approved was the one you started. Muse has no screen listing connected tools, which means you cannot switch the apply or generate tools off for an audit-only habit. If you want read-only behavior, say so in the message, and to cut off access entirely revoke the connection in StoreSEO under Settings, AI Toolkit. Q: Can I run a StoreSEO audit from Muse in WhatsApp? A: Yes, if Muse is available to you there. Muse works in the Muse app, on muse.ai and in WhatsApp, and the same StoreSEO connection serves all three. Ask for short answers, such as the worst five products, so the list reads well on a phone. Q: Does a Muse audit change anything in my store? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which only read. Muse or its Sentinel may still ask you to confirm the first connector call. --- ## Optimize meta tags and content with Muse URL: https://storeseo.com/agent/muse/optimize-meta (full page: https://storeseo.com/agent/muse/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags with Muse means asking Meta's personal AI agent to have StoreSEO draft a meta title, description, tags and URL handle for one Shopify resource, reading the draft in your thread, and saying yes only to the parts you want applied. Start with one resource and name it, for example write better meta for my best selling linen shirt, because every preview spends AI Content Optimizer credit. Muse calls storeseo_start, reads the current fields with storeseo_get_resource_seo and then calls storeseo_preview_resource_optimization. StoreSEO writes the copy on its own servers, so the words come from StoreSEO and not from Muse, and the response carries a draft id plus the projected score change. Muse pastes the old and new title and description into the thread, which on a phone means you read two short blocks and a score instead of opening the Shopify admin. The apply is a separate step. storeseo_apply_resource_optimization is a write, and StoreSEO tells the agent to wait for your yes. In Muse there is a second layer: Meta says Muse asks for permission on important actions and that a Sentinel agent approves what leaves its machine, so you may be asked once by Muse in its own words and once by StoreSEO's flow. You can answer yes to the title and description and no to the handle. When a handle changes, the apply creates a redirect from the old address, so ask Muse to show old and new addresses side by side before you agree. Previews are where the credit goes, and a thread makes it easy to lose count. A fair routine is to ask Muse for the three weakest products, preview them one at a time, and reply to each draft before the next arrives. If you dislike a draft, tell Muse what to change, but know that a new preview spends credit again. The product title and focus keyword are not written by this job, so a request to rewrite them gets a pointer to the StoreSEO app instead. Prompts: - Use StoreSEO to preview new meta for my Linen Shirt product and show me old and new title, description and handle in one message. Save nothing yet. - Preview meta for my three lowest-scoring products, one at a time, and wait for my yes after each before you apply anything. - Apply the meta title and description from that draft but skip the URL handle. In Muse: Do not treat Muse's own permission prompt as the review of the content. Muse decides when to ask, and the question it puts to you may be about sending a call out of its machine rather than about what the draft says. Read the title and description in the thread first, then answer. Because a custom connector is not reviewed by Meta, the only brake you fully control is the wording of your request plus revoking the connection in StoreSEO under Settings, AI Toolkit. Q: Does Muse write the meta text itself? A: No. StoreSEO generates the meta title, description, tags and handle on its own servers. Muse fetches the draft, shows it in the thread and passes along only the parts you approve. Q: What stops Muse from applying a draft I have not read? A: StoreSEO tells the agent to wait for your yes, and Muse and its Sentinel agent ask permission on important actions. Those are instructions and client behavior, not a server lock, so tell Muse to preview and ask before it applies and read each question. --- ## Optimize images and alt text with Muse URL: https://storeseo.com/agent/muse/image-seo (full page: https://storeseo.com/agent/muse/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing Shopify image SEO with Muse means asking Meta's personal AI agent to list the heaviest or unlabelled product images through StoreSEO and then optimize them one at a time, answering yes in your thread before each change. Image fixes suit a phone thread because each one is small and easy to judge. You ask Muse which product images are heaviest or have no alt text, and it calls storeseo_start and then storeseo_list_images, which sorts by file size by default and reports compression, alt text and file name status for each image. That listing is a read, so the first reply is a shortlist with sizes and a count, and nothing has been spent. Narrow it in plain words: only images on one collection, or only images with no alt text at all. The write step is storeseo_optimize_image, which takes one image per call and all wanted axes in the same call. Muse should not fire two calls at one image, because a file rename swaps the underlying Shopify media. The reply is a per-axis report, so Muse can tell you that compression applied while alt text spent AI credit, or that compression was refused because the Image Optimizer add-on is not installed. Compression counts against the add-on's monthly quota, and alt text and file names spend AI Content Optimizer credit, so ask Muse to state the number of images and the kind of spend before it starts. If a result looks wrong, say so in the thread and Muse can call storeseo_restore_image for that axis. Renaming is the one to watch: a renamed file gets a new storefront URL, and anything that hard-coded the old address keeps serving the old file. Muse's Secure VM and Sentinel sit between the agent and the internet, so a long run of image calls may be slowed by confirmations. Prefer batches of five over a catalog-wide request, and expect to answer yes more than once. Prompts: - Use StoreSEO to list my ten heaviest product images and tell me which have no alt text. Read only for now. - Optimize compression and alt text on the first five of those, one image at a time, and tell me the result for each axis. - Restore the file name on the last image you did. I do not want the new URL. In Muse: Muse has no settings screen for its connectors, so there is no allow-list to tighten around storeseo_optimize_image and no way to lock the tool for a batch. The limit is the sentence you type. Name how many images, which axes and whether renaming is allowed, and tell Muse to ask after each one. If a reply goes missing on a phone, ask Muse to list the image again instead of repeating the call, since a repeat can spend credit a second time. Q: Can Muse compress images if I do not have the Image Optimizer add-on? A: Compression is refused without the add-on, and Muse should report that for the compression axis. Alt text and file names can still apply because they spend AI credit instead of add-on quota. Q: How do I undo an image change made through Muse? A: Ask Muse to call storeseo_restore_image for that image and name the axis: compression, alt text or file name. Restoring a file name changes the image URL back as well. --- ## Get products ready for AI search with Muse URL: https://storeseo.com/agent/muse/ai-search-readiness (full page: https://storeseo.com/agent/muse/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Muse means letting Meta's personal AI agent read each product's GEO score through StoreSEO, then answering StoreSEO's questions about the product by text so it can write a description that AI shopping agents can understand. The GEO score covers products only, and it measures how well an AI shopping agent can tell what a product is. You ask Muse which products are weakest, and it calls storeseo_start, which includes a GEO overview, then storeseo_list_resources and storeseo_get_resource_seo to read the failing criteria. Some are fixable by tools, such as FAQs, the AI snippet and image alt text. Others, such as title, category and price, you change in Shopify or the StoreSEO app. Muse should say which is which, so you do not wait for a fix that has no tool. The most conversational part of this job is the description. storeseo_scan_product_description scores the current text on seven pillars, identity, materials, dimensions, features, use cases, what is included and care, and returns questions for you. StoreSEO's instructions tell the agent to put those questions to you word for word and in order, which suits a thread: Muse sends a question, you reply with a short answer, and you can do it in WhatsApp on the way to somewhere else. Some questions come with options or a prefilled value, and you can confirm, change or pick. Then comes the write. storeseo_write_product_description replaces the product's description in Shopify and StoreSEO immediately, with no undo and no revision history. The response returns the previous text, the only copy left, so ask Muse to paste it into the thread before you move on. StoreSEO writes the copy, so you cannot hand Muse your own text for it. The product needs a focus keyword set in the StoreSEO app first, and Muse cannot set one, so a scan on a product without one ends with a pointer back to the app. Prompts: - Which of my products have the lowest GEO score? Name the first thing missing on each. - Scan the description on my Linen Shirt product and send me StoreSEO's questions one at a time, word for word. - Write the new description from my answers, and paste the old description back to me in this thread afterwards. In Muse: Five or six questions in a message thread can turn into an interview you abandon halfway. If you stop answering, the write has not happened yet, but Muse may offer to skip the questions, and StoreSEO allows that only if you decline to answer or authorize a bulk run without them. Do not let it. Skipped questions mean guesses, and the write replaces your current description with no way back except the previous text Muse should keep for you. Q: Can I answer StoreSEO's product questions from WhatsApp? A: Yes, when you use Muse in WhatsApp. Muse relays each question in the thread and passes your reply to StoreSEO, which then writes the description from your answers. Q: Why did Muse refuse to scan one of my products? A: The scan and the rewrite need a focus keyword on the product, and no tool here can set one. Set it in the StoreSEO app, then ask Muse to scan again. --- ## Publish llms.txt and agents.md with Muse URL: https://storeseo.com/agent/muse/llms-txt-agents-md (full page: https://storeseo.com/agent/muse/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Muse means asking Meta's personal AI agent to read your store's discovery-file status in StoreSEO, then generate or reset the files that tell AI shopping agents what your store sells, after you confirm each live write. Begin with a question and no generation: is my llms.txt up to date, and is an agents.md live? Muse calls storeseo_start, then storeseo_get_llms_txt_status and storeseo_get_agents_md_status. The first reports saved settings, the latest generated version, whether a run is in progress and the public address where the file is served. The second says whether a custom agents.md is live or Shopify's native default still serves. Muse can add storeseo_list_llms_txt_versions to show which version is active, which matters because the live file is not always the newest one. Generating is a different order of act. storeseo_generate_llms_txt writes templates/llms.txt.liquid into your live theme and uploads a copy to Shopify Files, and storeseo_generate_agents_md publishes live straight away, spends AI credit and replaces Shopify's native file. StoreSEO's instructions tell the agent to say that before it calls, and to wait for your yes. In Muse, expect a layered check: the agent's own question, possibly a Muse permission request, and the Sentinel that Meta describes as approving what leaves its Secure VM. Say which resource types the file should include, because a request to include products and collections but not blog posts is saved as settings for later runs. Generation takes a minute or two, and Muse polls storeseo_get_llms_txt_status until generation_ongoing is false. On a phone, tell Muse to report back in the thread when it is done instead of waiting in the app. Afterward, ask it to read the public URL back to you so you can check the file reads like your store. Undoing is possible but destructive: deleting an llms.txt version or resetting agents.md cannot be undone, and deleting the active version promotes the next newest, so Muse should name the version that will go live. Prompts: - Is my llms.txt up to date, which version is live, and is a custom agents.md serving on my store? - Generate a new llms.txt with products and collections but no blog posts. Tell me what it writes to my theme before you start. - Generate an agents.md for my store. Say what it costs and that it goes live, and wait for my yes. In Muse: A live theme write is not something to approve from a lock screen notification. Muse can surface its own permission request while you are away from the thread, and a quick yes to a message you did not read is the one case where the layers fail you. Open the thread, read what the agent says it will write and where, and answer there. If you want an agents.md gone, name reset in your message so Muse does not confuse it with regenerating. Q: Does Muse itself write my llms.txt? A: No. StoreSEO composes llms.txt and agents.md from your store data on its own servers. Muse starts the run, watches its status and reports the result, and you confirm the live write first. Q: How do I remove a custom agents.md that Muse generated? A: Ask Muse to reset it with storeseo_reset_agents_md. That discards the custom content for good, with no version history, and Shopify's native agents.md serves again. --- ## Set up structured data with Muse URL: https://storeseo.com/agent/muse/structured-data (full page: https://storeseo.com/agent/muse/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Muse means asking Meta's personal AI agent to read which JSON-LD schema types StoreSEO has switched on for your store, and then to enable one or change its settings with details you give it in the thread. Structured data is a good first job to hand a conversational agent because the starting point is a plain list. You ask Muse which schema types are enabled, and it calls storeseo_start and storeseo_list_schema_types. Each row says whether a type is on and whether it is configurable: product, review, product merchant, organization and local business take settings, and the rest are simply on or off. Muse can read one type in detail with storeseo_get_schema_settings, which returns the current settings and a field template describing every field. The write is storeseo_update_schema_settings, and it replaces the stored settings object as a whole. That is why Muse must read the current settings first and send the edited copy back, rather than patching one field from memory. Business details come from you. If you ask for organization schema, Muse should ask for your business name, logo and contact email in the thread, and it must not invent an address or opening hours. Replying from a phone, you can paste a line at a time. Turning a type on or off changes the markup your storefront serves, so expect a confirmation question first. Dependencies surface as refusals you can act on. Review, sale dates and product merchant need product schema on first, and the FAQ types need the FAQ Schema feature in your plan, so a change to them can fail even when you say yes. Muse should name the missing piece, not retry. Sale dates also need an optional Shopify scope that only the app's own toggle can request. When a change goes through, ask Muse to read the type back so the thread ends with the stored state instead of an assumption. Prompts: - Which JSON-LD schema types are on for my store? Read only. - Read my organization schema settings, tell me what is missing, and ask me for the details one at a time. - Turn on product schema. Read the current settings first and tell me what changes on my storefront before you do it. In Muse: Details you type into a message thread can arrive in a form schema does not want, such as a logo described in words instead of an address. If StoreSEO rejects the settings, Muse should report the field and ask again, not guess. Because the update replaces the whole object, do not accept a shortened version that drops fields you set earlier. Muse has no per-tool switch for this connector, so confirm what the saved object will contain before you say yes. Q: Why did Muse read my schema settings before changing them? A: StoreSEO replaces the whole settings object on an update instead of merging, so the agent reads the current one, edits it and sends the full copy back. Skipping the read could erase fields you set before. Q: Can Muse turn on FAQ schema for my pages? A: Only if the FAQ Schema feature is in your StoreSEO plan. Without it any change to the four FAQ schema types, on or off, is refused, and Muse should tell you that. --- ## Manage sitemaps and indexing with Muse URL: https://storeseo.com/agent/muse/sitemaps (full page: https://storeseo.com/agent/muse/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps with Muse means asking Meta's personal AI agent which of your products, collections, pages and posts StoreSEO excludes from the sitemap or marks noindex, and then changing those flags one resource at a time after you confirm. Everything starts from a list, which suits a thread. You ask Muse which products are hidden from search, and it calls storeseo_start and storeseo_list_resource_sitemaps for one resource type at a time. Each row carries three flags: excluded from the sitemap, noindex and nofollow. The tool can filter by flag, so a question like which of my pages are noindex returns exactly those. Muse can read the HTML sitemap with storeseo_get_html_sitemap_settings, which returns whether it is enabled and its full configuration, and nothing in that read changes your store. Changing a flag is storeseo_update_resource_sitemap, which sets any of the three flags on one resource and writes the matching Shopify metafields. The danger is noindex: it removes a page from search results, so Muse should name the resource and wait for your yes, and you should check it picked the right one when two products have similar names. After a change, ask Muse to list the flags again in the same reply, so the thread ends with the stored state instead of the state it expected. A test product you hide today is a page you can lose by a mistyped name. The HTML sitemap is a two-person job. storeseo_update_html_sitemap generates it and later marks setup complete, but the step between, adding the sitemap link to a navigation menu, is yours in Shopify admin, and Muse hands you the menu addresses from storeseo_list_sitemap_menus. Settings replace the whole configuration, so Muse reads them before it edits. Because a phone is a poor place to edit a menu, ask Muse to send those links and finish the menu later at a desk, then come back to the thread to mark the setup done. Prompts: - Which of my products are set to noindex or excluded from the sitemap? Read only. - Exclude the product Gift Card Test from the sitemap and set it to noindex. Tell me exactly which product you mean before you do it. - Set up my HTML sitemap, send me the menu links I need, and tell me what I have to do myself. In Muse: A short thread invites a short instruction, and short instructions are how the wrong product gets hidden. Muse is not told your catalog, so give the exact title or ask it to list matches before it flags one. There is no list of connected tools to switch the sitemap writes off in Muse, so wording is your only control inside the client, and a noindex mistake is quiet: nothing breaks until the page drops from search. Re-list flags after every change. Q: Can Muse add my HTML sitemap to the store menu? A: No. Muse can generate the sitemap and send you the menu edit links, but adding the link to a navigation menu is done in Shopify admin. Muse can then mark setup complete. Q: What does noindex do to a product Muse changes? A: It asks search engines to drop the page from results. Muse should name the resource and wait for your yes before it sets the flag, and it can clear the flag again by calling the same tool. --- ## Track keyword rankings with Muse URL: https://storeseo.com/agent/muse/keyword-tracking (full page: https://storeseo.com/agent/muse/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keywords with Muse means telling Meta's personal AI agent which search terms matter for your store, so StoreSEO checks their Google positions and you can ask in a message thread how they and your competitors have moved. Rank tracking is a Pro plan feature, so a sensible first message to Muse is a look at what you already have: which keywords am I tracking and where do I rank. Muse calls storeseo_start, then storeseo_list_keywords, which returns the latest known Google position, the change since the last check, SERP features and competitor positions. For one keyword, storeseo_get_keyword_history gives the dated record, newest first. If your plan lacks rank tracking, Muse should explain the upgrade in the thread instead of calling the rank tools and failing. Adding keywords takes preparation that Muse can do for you. The country and language must be exact names from storeseo_list_rank_locations and storeseo_list_rank_languages, so Muse looks them up rather than typing them from memory. It then calls storeseo_add_keywords with keyword, language, location and device. Positions are checked in the background, so the reply confirms that the keywords were created and queued, and the first ranks show up on a later question. Every keyword counts against the plan limit. Competitors can be added with storeseo_add_competitors, to a maximum of five. Clusters are the heavier tool. storeseo_generate_keyword_cluster groups related keywords around a product's focus keyword with search volume and difficulty, spends AI credit and runs for a couple of minutes, and Muse polls storeseo_get_keyword_cluster until the job finishes. It fails if the product has no focus keyword, which only you can set in the StoreSEO app. Muse can state the estimated credit when you ask, and because you are on a phone, the natural pattern is to start the cluster, close the app and ask for the result in the thread later. Prompts: - Which keywords am I tracking in StoreSEO, and which moved the most this month? - Track linen shirts for men in the United States on mobile. Look up the exact country and language names first. - Build a keyword cluster for my Linen Shirt product. Tell me the estimated credit before you start. In Muse: Deleting a tracked keyword or a competitor erases its history for good, and a loosely worded message such as clean up my keywords is easy to misread in a quick thread. Name the keyword in full and make Muse repeat which one it will delete before you say yes. Muse also has no tool list to restrict, so a Pro-only habit like this leans on your wording and on revoking the connection in StoreSEO under Settings, AI Toolkit, if you want it to stop. Q: Do I need the Pro plan to track keywords with Muse? A: Yes. Rank and competitor tracking are Pro plan tools, and on another plan Muse should explain that instead of calling them. Keyword clusters also spend AI credit. Q: Why are my new keywords showing no position in Muse? A: Ranks are checked in the background after storeseo_add_keywords creates the keywords, so positions appear on a later storeseo_list_keywords call. Ask Muse again after a while. --- ## Read Search Console and Analytics with Muse URL: https://storeseo.com/agent/muse/google-reports (full page: https://storeseo.com/agent/muse/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with Muse means asking Meta's personal AI agent a question about search clicks, queries, pages or site traffic, and letting StoreSEO pull the Search Console and Google Analytics numbers behind the answer, compared with the previous period. This is the lightest job to run from a phone, because both tools are reads and the answer is a short list of numbers. You ask which queries brought the most Google clicks in the last 28 days, and Muse calls storeseo_start and storeseo_get_search_console_insights with a dimension: query, page, country, device, search appearance or date. The result carries clicks, impressions, click-through rate and average position, plus the previous equal-length period and per-metric changes unless you turn the comparison off. The default window is 28 days, and you can name another. For visitors rather than searches, Muse calls storeseo_get_analytics_report with a report type: search traffic, user activity, demographics, page performance, traffic source or key events. Time series come back as rows by date and the rest as paginated lists with totals. Both tools need a paid StoreSEO plan with Google connected, and GA4 also needs a property selected. Connecting Google and choosing the property happen in the StoreSEO app, and Muse hands you the link when either is missing instead of guessing a result. The format that works in Muse is one question, one table, one follow-up. Ask for the top ten queries, then ask which of them rank between positions 8 and 20, or which product pages have high impressions and low click-through. Muse reports numbers as Google returns them and should not project outcomes from them. Because Muse also runs in WhatsApp, you can ask the same question on the move and compare this month with the last without opening a dashboard. Prompts: - Which queries brought the most Google clicks in the last 28 days, and how does that compare with the 28 days before? - Which of my product pages have high impressions but a click-through rate under two percent? - Show organic search traffic from GA4 for the last three months, as a small table I can read on my phone. In Muse: Because both reports are reads, the risk here is not damage but wrong inferences from a small screen. Tables on a phone get trimmed, so ask Muse to state the date range, the dimension and the totals line at the top of each answer, and do not rely on a trimmed list when you compare two periods. Google data also lags, so a very recent window can look thin. If Muse says Google is not connected, that is a StoreSEO app setting, not something to approve in the thread. Q: Do I need a paid StoreSEO plan for Google reports in Muse? A: Yes. Both the Search Console and GA4 tools need a paid StoreSEO plan with Google connected, and Analytics also needs a GA4 property selected in the StoreSEO app. Q: Can Muse change my Google settings or submit pages to Google? A: No. The two Google tools only read reports. Connecting Google and choosing the GA4 property are done in the StoreSEO app. --- ## Write SEO blog post drafts with Muse URL: https://storeseo.com/agent/muse/blog-posts (full page: https://storeseo.com/agent/muse/blog-posts/index.md). Job: Blog post drafts, described once above. Writing blog posts with Muse means asking Meta's personal AI agent to start a StoreSEO job that drafts a full, SEO-optimized article in one of your Shopify blogs, checking on it in the thread, and publishing it yourself after you have read it. Tell Muse the topic, the blog and the length you want, and it calls storeseo_start and storeseo_list_blogs. The list returns each blog's numeric id, title and article count, and only blogs synced with StoreSEO appear, so an empty list means you sync them in the StoreSEO app first. Muse then calls storeseo_generate_blog_post with the blog id. The cost grows with word count and a featured image, so StoreSEO tells the agent to state the expected credit and wait for your yes, and the call fails before anything is spent if your credit is too low. The writing runs in the background for a few minutes, and the thread is a good place to wait. Muse polls storeseo_get_blog_post_job until the status is completed, failed or cancelled, and should report steps completed out of total along the way. The draft article exists from the first moment of the run, before any text is written, so Muse should never call it finished early. If you leave the thread, you can come back and ask how far along the post is. This is the one job where the agent's persistence matters more than its speed. When the job completes, the draft sits in your Shopify blog and nothing is live. Muse can call storeseo_get_resource_seo with resource type article to read its SEO score and failing checks, so you know what to fix before publishing. Publishing itself happens in Shopify admin, and Muse cannot do it. The copy comes from StoreSEO rather than from Muse, so do not ask Muse to paste its own text into the draft. If you want a different angle, start another job and accept that it spends credit again. Prompts: - List my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost first. - How far along is the blog post you started? Do not call it finished until the job says so. - Check the SEO score of that finished draft and tell me what to fix before I publish it in Shopify admin. In Muse: A job that runs for minutes may outlast the attention of a phone thread, and Muse's own limits on long tasks are not documented on the pages Meta publishes. Do not assume it will message you when the draft is ready. Ask for the status when you return, and read steps completed against total. If Muse says the job failed, the error message explains why and credits are settled automatically, but starting a new job is a new decision to confirm. Q: Will Muse publish the blog post to my store? A: No. StoreSEO saves the post as a draft in your Shopify blog, and you review and publish it in Shopify admin. Muse can read the draft's SEO score once the job completes. Q: Why can Muse not find my blog? A: Only blogs synced with StoreSEO are listed. If the list is empty, sync your blogs in the StoreSEO app, then ask Muse to list them again. --- ## Audit your store's SEO with Manus URL: https://storeseo.com/agent/manus/seo-audit (full page: https://storeseo.com/agent/manus/seo-audit/index.md). Job: Store SEO audit, described once above. Running a Shopify SEO audit with Manus means giving a general task agent your StoreSEO custom MCP server and one plain request, so it works through the listing on its own and returns a ranked worklist of the products, collections, pages and posts that score worst. You work in a Manus task, after adding StoreSEO once under Settings, Integrations, Custom MCP Servers with your access token as the Bearer credential. Because StoreSEO is not in Manus's connector directory, the task has to name it, so you open with something like use StoreSEO to audit my store and tell me where to start. Manus calls storeseo_start, which returns the conversation id every other tool needs plus a snapshot of how many resources are optimized, need improvement or are not optimized. It then calls storeseo_list_resources sorted by score and storeseo_get_resource_seo on the weakest ones. All three are reads, so the audit runs from start to finish without anyone saying yes. What suits Manus here is that it treats the request as a job rather than a chat turn. It can page through a large catalog, read the checklist for each low scorer and come back with one summary instead of a stream of tool output. For each product it names the failing criteria by their visible titles, shows the GEO score for AI readiness next to the SEO score, and separates the fixes StoreSEO can make, such as meta, FAQs, the AI snippet and alt text, from the ones only you can make in the StoreSEO app, such as the focus keyword. The audit is also the one job that fits a recurring Manus task without any adjustment, because every tool it uses only reads. A weekly run lets you compare this Monday's worst ten with last Monday's and see whether the list is shrinking. Treat the audit as the first half of a loop: read the report, pick two or three items, and start a separate task where you can preview and approve the fixes. Prompts: - Use StoreSEO to find my ten lowest-scoring products, name the biggest failing check on each and tell me which fixes only I can make in the app. - Use StoreSEO to audit my collections, pages and blog posts, worst first, and give me one summary with a suggested order of work. - Every Monday, use StoreSEO to read my score breakdown and my ten worst products, summarize what changed since last week, and do not change anything. In Manus: Name StoreSEO in the task. Manus does not list StoreSEO among its prebuilt connectors, so a request that only says audit my SEO may be answered from the open web instead of from your scores. If you turn the audit into a scheduled task, keep its wording at reading and summarizing and leave the skip confirmations setting off. A scheduled run has nobody to answer a question, and a prompt that later grows to include previews or fixes would be exactly the kind of task that setting is meant to be kept away from. Q: Can Manus run my StoreSEO audit on a schedule? A: Yes. Manus documents scheduled tasks that use your connected apps and run unattended. The audit tools only read, so a weekly report fits well. Keep the task report-only and leave the skip confirmations setting off. Q: Does the Manus audit spend any AI credit? A: No. The audit uses storeseo_start, storeseo_list_resources and storeseo_get_resource_seo, which only read your scores and checklists. Credit is spent later, when you ask for a preview or a generated file. --- ## Optimize meta tags and content with Manus URL: https://storeseo.com/agent/manus/optimize-meta (full page: https://storeseo.com/agent/manus/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing Shopify meta tags with Manus means asking a task agent to have StoreSEO generate a meta title, description, tags, URL handle, FAQs and AI snippet for several resources, then bringing the whole batch back so you decide what gets applied. The starting point is a task that names StoreSEO and the size of the job, for example use StoreSEO to preview new meta for my five lowest-scoring products. Manus calls storeseo_start, lists resources by score with storeseo_list_resources, reads each candidate with storeseo_get_resource_seo and then calls storeseo_preview_resource_optimization once per product. StoreSEO writes the copy on its own servers, so the wording comes from StoreSEO, not from Manus, and each response carries a draft id and the projected score change. Nothing is saved at this point, but every preview spends AI Content Optimizer credit, and the task should tell you that before it starts. Manus is at its best when it can return one consolidated answer. Instead of you opening five products, you get a single summary with the old and new title, description and handle for each, ranked by projected score gain. When a handle changes it should show the old and new URL, because applying that part creates a redirect. You answer in the same task with a decision such as apply all five, apply everything except the handles, or apply only the two biggest jumps. Only then does Manus call storeseo_apply_resource_optimization, passing the products and the fields you approved. Drafts expire after an hour, so answer while the task is still open. The apply is free because the credit went at preview time. If you also want StoreSEO to optimize future products on its own, that is a separate settings change through storeseo_update_ai_optimization_settings, and it should be proposed with the exact sections and a credit warning before it runs. Prompts: - Use StoreSEO to preview new meta for my five lowest-scoring products. Tell me the credit this will use first, then show me one summary of old and new text. - Apply the previewed meta titles and descriptions for all five products but skip every URL handle change. - Use StoreSEO to tell me what it optimizes automatically for new products right now, and do not change any setting. In Manus: A general agent will happily keep going, so put a number in the task. A request like optimize my store could turn into dozens of previews, and each one spends credit whether or not you apply it. Previews and applies also do not belong in a scheduled Manus task: nobody is there to approve, and the skip confirmations setting that lets a scheduled task send or publish without asking should stay off. Schedule the audit that finds weak products, then run the previews in a normal task you are watching. Q: Does Manus write the meta text itself? A: No. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet on its own servers. Manus fetches the draft, shows it to you and passes along only the parts you approve. Q: How long do I have to approve a Manus meta preview? A: A draft lasts one hour after the preview. If your task sits unanswered for longer, ask Manus to preview again, which spends credit again. --- ## Optimize images and alt text with Manus URL: https://storeseo.com/agent/manus/image-seo (full page: https://storeseo.com/agent/manus/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing Shopify image SEO with Manus means giving a task agent the StoreSEO server and a scope, such as the heaviest ten product images with no alt text, so it works through them one call at a time and reports each result. Image work is a loop, which is why it suits an agent that carries a task from start to finish. You name StoreSEO and the scope in one request, for example find my heaviest product images with no alt text and fix the first ten. Manus calls storeseo_start and then storeseo_list_images, which sorts by file size by default and shows the status of compression, alt text and file name for every image. That listing only reads, so the first thing you get back is a shortlist with a count, before any quota or credit is spent. The writing step is storeseo_optimize_image, one image per call with every axis you want in the same call. Compression uses your Image Optimizer add-on quota, while alt text and file names spend AI Content Optimizer credit, and asking for alt text and file name together counts as one generation. Each response is a per-axis report, so Manus can tell you that compression applied, alt text applied and the file name was left alone, or that compression was refused because the add-on is missing while the other axes went through. Where Manus helps is the summary at the end of a batch. You get one list of what changed per image, including any new file names, which matters because a rename changes the storefront URL and anything that hard-coded the old address keeps serving the old file. If you dislike a result, the same task can call storeseo_restore_image for that image and axis. Say the scope up front, such as product images only, alt text and compression but no renaming, so the batch ends where you meant it to. Prompts: - Use StoreSEO to list my twenty heaviest product images and show which have no alt text. Do not change anything yet. - Optimize the first ten of those images with compression and alt text only, one image per call, and give me a per-image result at the end. - Restore the file name on the last image you optimized, because I want the old URL back. In Manus: Manus can run a long list without you watching it, so state the number of images in the request and keep a stop point after the first one. The tool takes one image per call and a retry after a lost response regenerates and spends again, so if a call times out ask Manus to list the image again and check its status instead of repeating the call. Keep image optimization out of scheduled tasks. A scheduled run could count unlabelled images and report the number, but it should not write. Q: Can Manus optimize images on a schedule? A: It can run a scheduled task, but keep that task to listing and counting images that need work. Each storeseo_optimize_image call writes to your store and spends quota or credit, and a scheduled run has nobody to say yes. Q: What if Manus renames an image I wanted to keep? A: Ask it to call storeseo_restore_image for that image and name the axis, which can be compression, alt text or file name. Restoring the file name points the storefront URL back to the original. --- ## Get products ready for AI search with Manus URL: https://storeseo.com/agent/manus/ai-search-readiness (full page: https://storeseo.com/agent/manus/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with Manus means asking a task agent to read each product's GEO score in StoreSEO, fix what a tool can fix, and pause to put StoreSEO's questions to you before it rewrites a thin description. GEO is the AI readiness score for products, and it exists for products only. You start a Manus task that names StoreSEO, such as find my five products that AI shopping agents will understand worst and fix them. Manus calls storeseo_start, which includes a GEO overview, then storeseo_list_resources and storeseo_get_resource_seo to read each failing criterion. Some are fixable with tools, namely FAQs, the AI snippet, image alt text and product schema. Others, such as the product title, category and price, you change in Shopify or the StoreSEO app, and a good task report says which is which for every product. The tool fixes come in a sensible order. If product schema is off, it is a shop-wide setting, so turning it on through storeseo_update_schema_settings credits every product at once, and Manus should say so before it does it. FAQs and the AI snippet come from storeseo_preview_resource_optimization, which spends credit and saves nothing, and then storeseo_apply_resource_optimization after your yes. A multi-step task like this is a natural fit for Manus because it can carry the plan from the first read to the final apply while you stay in one place. The description is where an autonomous agent needs a hard stop. storeseo_scan_product_description scores the description on seven pillars and returns questions about materials, dimensions, what is in the box and care. Manus has to put those questions to you word for word and wait, because your answers become the facts in the new copy. Only then does storeseo_write_product_description run, and it saves immediately. The response holds the only copy of the old text, so ask Manus to paste it into the task before it moves on. Prompts: - Use StoreSEO to find my five products with the lowest GEO score and tell me which gaps tools can fix and which only I can fix. - Make my Merino Beanie product ready for AI search with FAQs, an AI snippet and product schema. Preview everything first and tell me the credit it will use. - Scan the description of my best seller with StoreSEO, ask me every question it returns word for word, and wait for my answers before you write anything. In Manus: Do not let a Manus task answer the description questions for you, and do not tell it to skip them to save time. Skipping questions is something only you decide, and a description written without your facts is a guess. The write also replaces the live description at once with no undo, so ask Manus to keep the previous text from the response in the task before you leave. The product needs a focus keyword set in the StoreSEO app first, and a scheduled task should never include any of these writes. Q: Can Manus rewrite my product descriptions without asking me questions? A: It should not. StoreSEO returns questions for facts it cannot infer, and the agent is told to put them to you word for word. Skipping them is only for when you decline to answer or authorize a bulk run without them. Q: Does Manus need a focus keyword before it scans a description? A: Yes. The product must already have a focus keyword, and no tool can set it. Set it in the StoreSEO app, then ask Manus to scan the description again. --- ## Publish llms.txt and agents.md with Manus URL: https://storeseo.com/agent/manus/llms-txt-agents-md (full page: https://storeseo.com/agent/manus/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with Manus means asking a task agent to check what your store serves to AI crawlers and shopping agents through StoreSEO, then generate or roll back those files once you have agreed what goes live. Begin with a question rather than a command. A task that says use StoreSEO to tell me what my store serves at llms.txt and agents.md makes Manus call storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status, and storeseo_list_llms_txt_versions if you want the version history. You get the saved settings, the latest version, whether a rebuild is due, the public URLs and whether agents.md is a custom file or Shopify's default. All of these are reads, so this first answer needs no approval and costs no credit. Generating is where a long task shape helps. storeseo_generate_llms_txt queues the run, and the same run writes a template into your live theme and uploads a copy to Shopify Files, so Manus should tell you that before it starts. The generation takes a minute or two and the status tool is the poll, which Manus can keep checking in the same task until generation_ongoing turns false. Then it can report the version that is now active. storeseo_generate_agents_md is different: it publishes live at once, spends credit and needs at least one published product. Rolling back is separate. storeseo_activate_llms_txt_version makes an older version the live one, and the live file serves the active version, which is not always the newest. Deleting a version and resetting agents.md to Shopify's default are permanent, so Manus should name exactly what it will remove and which version will take over. Ask it to paste the generated text into the task before a reset, because there is no history to restore it from. Prompts: - Use StoreSEO to tell me whether my llms.txt is up to date, which version is live and whether my store serves a custom agents.md. Read only. - Generate a new llms.txt with products and collections but not blog posts, and warn me about the live theme write before you start. Wait for my yes. - Every Monday, use StoreSEO to check my llms.txt and agents.md status and tell me if a rebuild is due. Do not generate, activate, delete or reset anything. In Manus: This is the job where a scheduled Manus task is most tempting and most risky. A weekly task that reads the status and reports whether a rebuild is due is a good use of it. A weekly task that also generates would write to your live theme and publish agents.md with nobody there to agree, and the skip confirmations setting would remove the one pause Manus has. Leave that setting off, schedule the status check only, and run any generation as a normal task you are watching. Q: Does a Manus task have to stay open while llms.txt generates? A: The generation runs in StoreSEO and takes a minute or two. Manus polls storeseo_get_llms_txt_status until generation_ongoing is false, so keep the task open until it reports the new version rather than assuming it finished. Q: Can a scheduled Manus task regenerate llms.txt for me? A: You could write it that way, but you should not. Generating writes to your live theme, and a scheduled run cannot wait for your yes. Schedule a status check and generate in a task you start. --- ## Set up structured data with Manus URL: https://storeseo.com/agent/manus/structured-data (full page: https://storeseo.com/agent/manus/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with Manus means asking a task agent to read which JSON-LD schema types StoreSEO has switched on, collect the business details only you have, and send back a complete settings object for the type you want enabled. You describe the outcome in a Manus task that names StoreSEO, for example check which schema types are on and set up organization schema for my store. Manus calls storeseo_start and then storeseo_list_schema_types, which returns every type with whether it is enabled and whether it takes settings. The configurable ones are product, review, sale dates, product merchant, organization and local business. For the one you want, it calls storeseo_get_schema_settings, which returns the stored settings and a field template that describes the full object. All of that is reading and runs without a pause. The next step is the one an autonomous agent can get wrong. A settings update through storeseo_update_schema_settings replaces the whole stored object, so Manus has to start from what it just read and send back the complete object with only your changes. The business name, address, phone number, logo and contact email come from you, not from the model. Give them in the task or let Manus ask, and tell it not to fill gaps from the web, because organization and local business share the same business information and a wrong address lands in both. Dependencies show up in the report. Review, sale dates and product merchant need product schema on first, and disabling product also disables product merchant. The four FAQ schema types need the FAQ Schema feature in your plan, so a refusal there is a plan answer, not a failure to retry. Turning a type on changes the markup on your live storefront, so Manus should name the type and the settings and wait for your yes before the call. Prompts: - Use StoreSEO to list my schema types and tell me which are on, which take settings and which are missing details. Read only. - Set up organization schema with my business name, logo URL and contact email from this message. Read the current settings first and show me the full object before you send it. - Turn on product schema and then review schema, and tell me what each one changes on my storefront before you start. In Manus: Manus is built to finish tasks, and with a half-filled form it may try to complete the missing fields itself from public pages. Give the exact business details in the task and say not to invent an address, phone number or opening hours. Also remember the settings object is replaced, not merged, so a task that skips the read can wipe fields you had set. A scheduled Manus task can safely list schema types and report which are off. It should not change them. Q: Why does Manus read my schema settings before it changes them? A: A settings update replaces the whole stored object instead of merging. Reading first and sending back the complete object is how one field changes without wiping the others. Q: Can Manus enable FAQ schema on any StoreSEO plan? A: No. The four FAQ schema types need the FAQ Schema feature in your plan. Without it, StoreSEO refuses any change to them and the message says the feature is not in your plan. --- ## Manage sitemaps and indexing with Manus URL: https://storeseo.com/agent/manus/sitemaps (full page: https://storeseo.com/agent/manus/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing with Manus means asking a task agent to read the sitemap, noindex and nofollow flags StoreSEO holds for your products, collections, pages and posts, and to change them one resource at a time on your say-so. A good request names one resource type and a question, for example use StoreSEO to show me every product that is excluded from the sitemap or set to noindex. Manus calls storeseo_start and then storeseo_list_resource_sitemaps, which takes the resource type and filters on sitemap, noindex or nofollow. The result is a table of resources with their three flags. Because noindex pulls a page out of search results, this read is worth doing before any change, and it runs without a pause, so Manus can walk several pages of a large catalog in one task and return a short list. Changes go through storeseo_update_resource_sitemap, which sets sitemap_disabled, no_index or no_follow on one resource and writes the matching Shopify metafields. Manus should name the resource and the flag it will set, wait for your yes, and then call the tool for that resource. Flags can also be changed in the StoreSEO app while the task runs, so it should list the resource again before it tells you the current state instead of relying on an earlier answer. The HTML sitemap is a multi-step setup that fits a task agent well, with one human step in the middle. Manus reads storeseo_get_html_sitemap_settings and generates the page with storeseo_update_html_sitemap, then calls storeseo_list_sitemap_menus, which returns your menus with edit links. You add the sitemap link to a menu in Shopify admin, because no tool does it, and then Manus marks setup complete. Any later change to the page settings starts with a read, because an update replaces them. Prompts: - Use StoreSEO to list every product and page that is set to noindex or excluded from the sitemap. Read only, and tell me which ones look accidental. - Set the product Gift Card Test to noindex and exclude it from the sitemap. Name the resource and wait for my yes before you change it. - Set up an HTML sitemap with StoreSEO, then tell me which of my menus to add the link to and wait while I do it. In Manus: Indexing flags are easy to get wrong in bulk, and a general agent can read a bulk instruction literally. Ask for one resource per change, name it, and have Manus list it again afterward. A scheduled Manus task can audit these flags and report which important products are noindex, which is a useful weekly check. It should not set flags, because a scheduled run has nobody to approve and a wrong noindex removes a page from search results until someone notices. Q: Can Manus add the HTML sitemap link to my menu? A: No. Manus can generate the HTML sitemap and list your menus with their edit links, but you add the link to a menu in Shopify admin. Then Manus can mark setup complete. Q: Why does Manus list my flags again before answering? A: The flags can also be changed in the StoreSEO app while the task is open, so an earlier listing may be out of date. Listing again keeps the answer accurate. --- ## Track keyword rankings with Manus URL: https://storeseo.com/agent/manus/keyword-tracking (full page: https://storeseo.com/agent/manus/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with Manus means asking a task agent to add Google keywords and competitor domains to StoreSEO, read where you rank by country and device, and turn the history into a plain summary on a schedule you set. Rank tracking is a Pro plan feature, so a Manus task that names StoreSEO should begin by finding out whether you have it. Manus calls storeseo_start, whose snapshot includes rank tracking, and if the plan lacks it the task should say so instead of calling the rank tools. When it is available, the agent first looks up the exact market names with storeseo_list_rank_locations and storeseo_list_rank_languages, because new keywords must use the country and language names the tracker accepts, word for word, and the device is desktop or mobile. Adding is a write. After your yes, storeseo_add_keywords sets the keyword, language, country and device, and storeseo_add_competitors takes up to five domains. Checks run in the background, so positions appear later and Manus should not report a rank on the same minute it added the keyword. Reading is where Manus helps most. storeseo_list_keywords returns the latest position, change, SERP features and competitor positions, and storeseo_get_keyword_history returns the dated history, so one task can answer which keywords dropped out of the top ten this month and where your competitors sit. Keyword clusters are a related step for one product. storeseo_generate_keyword_cluster needs a focus keyword set in the StoreSEO app, spends credit and runs for a couple of minutes, and storeseo_get_keyword_cluster is the poll. This job suits a recurring Manus task better than most, as long as the recurring part only reads. Deleting a keyword or competitor erases its history for good, so Manus should name what it will remove and wait for your yes. Prompts: - Use StoreSEO to check my plan includes rank tracking, then track linen shirt for men in the United States on mobile. Look up the exact country and language names first. - Use StoreSEO to show how my tracked keywords moved this month, which fell out of the top ten, and where my competitors rank. - Every Friday, use StoreSEO to read my tracked keywords and rank history and send me a short summary of the biggest changes. Do not add, delete or change anything. In Manus: A weekly ranking summary is a good scheduled Manus task, but write it as a report that reads. A scheduled run that adds or deletes keywords would count against your keyword limit or erase rank history with no one present to approve it, and leaving the skip confirmations setting off keeps Manus from sending or posting a result on its own. Do the adding and deleting in a task you are watching, and never type country or language names from memory. Q: Does Manus track rankings in AI search engines through StoreSEO? A: No. The StoreSEO rank tracker checks Google results by country, language and device. AI visibility is handled by the GEO score, llms.txt and agents.md jobs. Q: Why does Manus say my new keyword has no position yet? A: Rank checks run in the background. Adding a keyword queues the first check, and the position appears in storeseo_list_keywords once that check finishes, so ask Manus to read it again later. --- ## Read Search Console and Analytics with Manus URL: https://storeseo.com/agent/manus/google-reports (full page: https://storeseo.com/agent/manus/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics with Manus means asking a task agent to pull your clicks, impressions, queries, pages and traffic sources through StoreSEO and write up what moved, without you opening Google or the StoreSEO app. These reports need a paid StoreSEO plan with Google connected in the app, and a GA4 property selected for Analytics, so check that before you blame the agent. In a Manus task that names StoreSEO, a request such as which queries brought the most clicks over the last 28 days and how does that compare with the month before makes Manus call storeseo_start and then storeseo_get_search_console_insights. It picks the dimension that fits the question, whether query, page, country, device or search appearance, and the report returns the totals plus the previous period and the change per metric. For visitor behavior it calls storeseo_get_analytics_report with the report type that matches, for example traffic source or page performance. Both tools read only and never connect or configure Google, so if Google is missing the answer is a link into the app, not something Manus can fix. The strength of Manus here is combining the two reports into one brief, for instance pages that earn impressions in Search Console but a weak click rate, next to what visitors do once they arrive. This is the clearest fit for a scheduled task. A Monday task can read the last 28 days against the period before, name the five biggest movers and write a short brief to read with coffee. The numbers are reported as Google returns them and the task should not project outcomes from them. When a brief points at a page with low click-through, the follow-up, new meta for that page, is a different job and belongs in a task where you can approve the preview. Prompts: - Use StoreSEO to show my Search Console queries from the last 28 days where I rank between 8 and 20, compared with the previous 28 days. - Use StoreSEO to read my GA4 traffic sources and page performance for last month and tell me which channels led to purchases. - Every Monday, use StoreSEO to read my Search Console and GA4 reports for the last 28 days, write a short brief of what changed, and do not change anything. In Manus: Because every tool in this job reads, a scheduled Manus task is safe, but only for as long as the prompt stays that narrow. Do not extend a recurring brief to add apply new meta to low-CTR pages. The skip confirmations setting lets a scheduled task send, publish or post without asking each time, so leave it off for any task that touches StoreSEO. If Google is not connected or no GA4 property is selected, the tool returns a link to set that up in the app, and the schedule will keep returning the same gap until you do. Q: Can a scheduled Manus task read Search Console for me every week? A: Yes. Manus documents scheduled tasks that use your connected apps and run unattended, and both Google tools only read. Write the prompt as a read-and-summarize task and leave the skip confirmations setting off. Q: Why does Manus say Google is not connected? A: Connecting Search Console and Analytics, and choosing the GA4 property, is done in the StoreSEO app and needs a paid plan. Manus cannot do it, and the tool returns a link to the setting that is missing. --- ## Write SEO blog post drafts with Manus URL: https://storeseo.com/agent/manus/blog-posts (full page: https://storeseo.com/agent/manus/blog-posts/index.md). Job: Blog post drafts, described once above. Writing Shopify blog drafts with Manus means asking a task agent to have StoreSEO write a full SEO post into one of your synced blogs as a draft, then checking the finished article's score while you review and publish it yourself. The request carries the brief: use StoreSEO to write a 1,200 word draft on how to care for linen shirts in my News blog, with a featured image. Manus calls storeseo_start and then storeseo_list_blogs, which returns your synced blogs with their numeric ids. An empty list means your blogs have not been synced in the StoreSEO app yet, and no tool can do that. StoreSEO writes the post itself on its own servers, so the prose comes from StoreSEO, not from Manus, and Manus is the one who carries the brief, the cost and the waiting. Generation spends AI Content Optimizer credit, and the cost grows with the word count and a featured image. Manus should say that and wait for your yes before it calls storeseo_generate_blog_post, which fails before anything is queued if your credit is short. The post then writes in the background for a few minutes. This is the part Manus handles well: it keeps polling storeseo_get_blog_post_job with the job id, reports how many steps are done, and does not call the draft finished while the job runs, until the status is completed, failed or cancelled. When the job completes, the task can read the draft with storeseo_get_resource_seo using resource type article and the article id from the job result, and report its SEO score. Nothing is published. The article sits as a draft in Shopify admin, where you read it, fix what you want and publish it. If the job fails, the error message explains why and credits are settled automatically. Prompts: - Use StoreSEO to list my blogs and write a 1,200 word draft on how to care for linen shirts in the News blog. Tell me the credit cost before you start. - Use StoreSEO to write a draft targeting best gifts for coffee lovers with a featured image, then check its SEO score when it finishes. - How far along is the blog post job you started? Check it and do not present the draft as finished until it is. In Manus: Manus can make a recurring task out of anything, but a weekly write a post task would spend credit every run with nobody to agree to the cost, and it would pile up drafts you never asked for. Keep blog generation in a task you start, and keep any scheduled task to listing drafts or reporting. Leave the skip confirmations setting off. It lets a scheduled task publish or send without asking, which is the wrong default for anything that spends credit. Whatever Manus produces, you still review and publish from Shopify admin. Q: Will Manus publish the blog post to my store? A: No. StoreSEO saves the generated post as a draft in the Shopify blog you choose. You review and publish it from Shopify admin. Q: Can I schedule Manus to write a blog post every week? A: You can schedule a task, but you should not let it generate. Each post spends credit and a scheduled run cannot wait for your yes. Schedule a report of your drafts and start the writing yourself. --- ## Audit your store's SEO with Paperclip URL: https://storeseo.com/agent/paperclip/seo-audit (full page: https://storeseo.com/agent/paperclip/seo-audit/index.md). Job: Store SEO audit, described once above. Running a store SEO audit from Paperclip means assigning an agent on your org chart a task to read StoreSEO's product, collection, page and post scores, so the ranked worklist lands on the ticket as a written report. You do not type into a StoreSEO window here. You create a task in Paperclip, assign it to an agent that already has StoreSEO on its runtime, and describe the audit in a sentence or two: which resource types, how many, and that it is a report. When the agent wakes, its runtime, for example Claude Code launched by the claude_local adapter, calls storeseo_start, which returns the conversation id every later call needs and a snapshot of how many resources are optimized, need improvement or are not optimized. It then calls storeseo_list_resources sorted lowest score first and storeseo_get_resource_seo for each candidate. Paperclip does not make those calls itself; it records the run, so you read the trace on the ticket. What the agent posts back is the part Paperclip is good for. Each weak resource comes with its failing criteria in plain words, a GEO score for products, and a split between fixes a StoreSEO tool can make and fixes only you can make in the StoreSEO app, such as the focus keyword. Because the findings sit on a ticket, a teammate can comment on them, reassign the follow-up to another agent, or turn the top five items into separate tasks with their own budgets. Nothing in an audit writes to the store, which is why it is the safest first job to give a headless agent. Once the one-off run reads well, put it on a rhythm. A Paperclip routine fires on a cron schedule and creates the same task each week, so the audit becomes a baseline you compare against the previous report. Keep the routine's wording explicit that it is read only, because the agent runs unattended. Prompts: - Audit my StoreSEO scores. List the ten lowest-scoring products with their main failing criterion and post the table on this ticket. Read only, change nothing. - Weekly audit: report how many products, collections, pages and posts are optimized, need improvement or are not optimized, and compare with last week's comment on this task. - Split the fixes for my five worst collections into what StoreSEO can do and what I must set in the app, then open one follow-up task per group for me to approve. In Paperclip: The audit tools are reads, but the agent that runs them is not limited to reads. The Claude Code adapter runs headless with dangerouslySkipPermissions on by default, so any StoreSEO tool that the model decides to call will run without a prompt. For a scheduled audit, say in the task text that it is report only, and consider setting dangerouslySkipPermissions to false or using the Paperclip connector with write actions on Ask first, so a stray preview or apply cannot run unattended. Q: Can a Paperclip routine run the SEO audit every week? A: Yes. A routine fires on a cron schedule or a webhook, creates a task and wakes the agent you assigned. Word the task as a read-only audit so the weekly run reads scores and posts findings without trying to change anything. Q: Where do I see what the agent asked StoreSEO? A: Paperclip records the run against the task, so you can read the agent's tool calls and answers on the ticket. The calls themselves are made by the agent runtime, such as Claude Code, with StoreSEO attached to it. --- ## Optimize meta tags and content with Paperclip URL: https://storeseo.com/agent/paperclip/optimize-meta (full page: https://storeseo.com/agent/paperclip/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags from Paperclip means giving an agent a ticket to have StoreSEO draft new meta titles, descriptions, handles, FAQs and AI snippets, then reviewing the previews on the task before anyone approves an apply. The unit of work is a Paperclip task, not a chat message. You write the ticket, for instance preview better meta for the five lowest-scoring products and show me the old and new text, and assign it to an agent whose runtime has StoreSEO connected. The agent calls storeseo_get_resource_seo so it knows which fields fail, then storeseo_preview_resource_optimization for each product. StoreSEO writes the meta title, description, tags, URL handle, FAQs and AI snippet on its own servers and projects the new score; the agent only relays it. Those previews spend AI Content Optimizer credit even if you never apply them, so the ticket should name how many products you are willing to preview. The review happens in Paperclip's comment thread. The agent posts each draft next to the current text, flags any URL handle change with the old and new address, and stops with a question. Applying is a different kind of step: storeseo_apply_resource_optimization changes the live product and, for a handle, creates a redirect. Run that step in a session you are watching, or reply on the ticket with exactly which products and which parts to apply, for example the meta but not the FAQ, and let the agent make the call. Splitting the work this way fits how Paperclip is built. A cheaper, scheduled run can do the reading and drafting, and a supervised run does the writing, so the unattended part never holds the permission to change your store. Prompts: - Preview new meta titles and descriptions for the three lowest-scoring products. Post old and new side by side on this ticket and do not apply anything. - From the previews on this ticket, apply the meta for Linen Shirt but skip the FAQ, and keep its current URL handle. Confirm in a comment before you call apply. - Which of my pages have no meta description? Preview one for each, tell me the AI credit this will use, and wait for my reply. In Paperclip: A reply on a Paperclip ticket is not a permission prompt. Under the Claude Code adapter's default of dangerouslySkipPermissions, an agent that decides to apply a preview will do so without stopping, and the Paperclip connector review queue does not sit in front of a server added with claude mcp add. If you want a real hard stop before storeseo_apply_resource_optimization, set that flag to false so the tool does not run, or attach StoreSEO as a connector and leave the apply action on Ask first. Q: Can a scheduled Paperclip run apply meta changes by itself? A: It can if the agent is allowed to, but it should not. A routine has nobody to say yes, so keep it to reading and previewing, and run the apply in a task you start and watch. Previews still spend credit, so cap how many products the routine drafts. Q: How do I approve only part of a preview from Paperclip? A: Say which parts in your reply, for example the meta title and description but not the FAQ or the handle. The agent passes that subset to the apply step, and skipped fields stay on the same draft for one hour. --- ## Optimize images and alt text with Paperclip URL: https://storeseo.com/agent/paperclip/image-seo (full page: https://storeseo.com/agent/paperclip/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing image SEO from Paperclip means tasking an agent to find the heaviest or unlabelled images in StoreSEO, report the plan and its quota cost on the ticket, and compress or label them only once you start the run that does the writing. Image work has a natural two-task shape in Paperclip. The first task is a survey: you ask the agent to find product images with no alt text and the heaviest unoptimized files, and it calls storeseo_list_images, optionally scoped to one product by title. Because a title search can also match file names from other resources, a careful agent checks that each row belongs to the product you named before it reports. The result is a list on the ticket with file sizes, missing alt text and a count, and every call so far is a read. The second task does the writing, and it is where cost matters. storeseo_optimize_image works on one image per call and can run compression, file naming and alt text together. Compression uses the Image Optimizer add-on's monthly quota, while alt text and file names spend AI Content Optimizer credit, and a renamed file changes the image's storefront URL. So the ticket should state the number of images and which of the three axes to run, and the agent should say in a comment what that will spend before it starts. A long batch is a poor fit for an agent with a tight Paperclip budget, since each image is its own call. If a result is wrong, a follow-up task can call storeseo_restore_image for that one image, per axis. Keep the survey on a schedule if you like, and keep the optimizing in a run you supervise. Prompts: - Find my ten heaviest unoptimized product images and the products with no alt text. Post the list with file sizes on this ticket and change nothing. - Optimize the images on the product Canvas Tote: compression, alt text and file name. Tell me the quota and credit it will use in a comment first, then wait for my reply. - Restore the original alt text on the last image the previous task changed, and report which axis you reverted. In Paperclip: Do not let a headless run retry an image call on its own. If a response is lost, the right move is to re-list the image, because a retry regenerates alt text and spends credit again. Paperclip heartbeats can wake an agent mid-batch, so write the ticket to say one image per call and re-list before any retry. With the Claude Code adapter's default of skipping permission prompts, nothing else will stop a long optimizing loop, so set the batch size in the task text. Q: Can a Paperclip routine compress my product images every week? A: It is better not to. Compression counts against the Image Optimizer add-on's monthly quota and alt text spends AI credit, and a routine has nobody to approve the spend. Schedule the survey that lists images, and run the optimizing in a task you start. Q: What does the agent do if image compression is refused? A: StoreSEO reports the refused axis, for example compression without the add-on, and still applies the others you asked for. The agent should post that split on the ticket so you see what applied and what did not. --- ## Get products ready for AI search with Paperclip URL: https://storeseo.com/agent/paperclip/ai-search-readiness (full page: https://storeseo.com/agent/paperclip/ai-search-readiness/index.md). Job: AI search readiness, described once above. Preparing products for AI search from Paperclip means tasking an agent to read StoreSEO's GEO checklist for a product, preview FAQs and an AI snippet on the ticket, and hold the schema switch and description rewrite until you start a supervised run. A GEO task starts like an audit. You assign the ticket, say which product or how many of the lowest GEO scores to look at, and the agent calls storeseo_get_resource_seo to read the product's GEO block: the score and each AI-readiness criterion that passes or fails. It tells you in a comment which gaps a tool can fix, namely FAQs, the AI snippet, alt text, product schema and the description, and which only you can fix in the app or Shopify admin, such as the product title, category or barcode. That first pass is all reads and suits a scheduled routine. The second pass previews. storeseo_preview_resource_optimization generates the FAQs and an AI snippet with summary, benefits and best for, and projects the new SEO and GEO scores without saving, spending credit. The agent posts it on the ticket. The risky steps come after your reply. Turning on product schema through storeseo_update_schema_settings affects every product at once, and storeseo_write_product_description replaces the live description immediately with no undo, returning the old text once. A description scan also produces questions about materials, dimensions and care, and those are yours to answer: the agent should relay them word for word as a comment and wait, which is slow in a task queue but gives the copy real facts. Because a rewrite has no undo, treat it as its own supervised task, and ask the agent to paste the old description into the ticket before it writes the new one. Prompts: - Read the GEO score and failing criteria for Merino Beanie and post them on this ticket. Say which are fixable by StoreSEO tools. Change nothing. - Preview FAQs and an AI snippet for the three lowest GEO products and post the projected scores. Do not apply, and tell me the AI credit used. - Scan the description of Ceramic Pour Over and post StoreSEO's questions for me word for word. Do not write anything until I answer in a comment. In Paperclip: A description rewrite and the shop-wide product schema switch are the two steps here you cannot easily take back, and a headless Claude Code agent with dangerouslySkipPermissions left on will run them if it decides to. Keep scheduled GEO routines to reading and previewing. Run storeseo_write_product_description only in a task you are watching, after you have answered the scan questions and the agent has pasted the old description onto the ticket. Q: Can a Paperclip agent answer the description questions for me? A: It should not. StoreSEO returns questions about facts such as materials and what is in the box, and the agent is meant to relay them to you word for word. Answer in a comment on the ticket so the copy states real facts. Q: Is a GEO check a good routine to schedule in Paperclip? A: Reading the GEO score and the failing criteria is. Previews spend credit and apply, schema and description changes write to your store, so leave those to a task you start and review. --- ## Publish llms.txt and agents.md with Paperclip URL: https://storeseo.com/agent/paperclip/llms-txt-agents-md (full page: https://storeseo.com/agent/paperclip/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Managing llms.txt and agents.md from Paperclip means letting an agent read StoreSEO's status for both files on a schedule and keeping generation, version switches and resets for a task you start, because each of those publishes or removes a live file. These two files are a good example of reads and writes that look similar and are not. A status check is cheap and safe: the agent calls storeseo_get_llms_txt_status and storeseo_get_agents_md_status and reports whether llms.txt is current, which version serves, whether a rebuild is due, and whether the store serves a custom agents.md or Shopify's default. That fits a Paperclip routine well, since it fires on a cron schedule, creates a task, wakes the agent and leaves the answer as a comment you read on Monday. Generation is different. storeseo_generate_llms_txt writes a template into your live theme and uploads a copy to Shopify Files in the same run, then takes a minute or two, so the agent polls storeseo_get_llms_txt_status until generation is finished. storeseo_generate_agents_md publishes live at once, replaces Shopify's default and spends credit. Deleting an llms.txt version and resetting agents.md cannot be undone. For those, write a task that names the exact action, assign it to an agent, and have it ask in a comment before the call. Switching versions with storeseo_activate_llms_txt_version is milder, and storeseo_list_llms_txt_versions shows which of the latest ten is active. Polling suits a heartbeat-driven agent badly if the wake interval is long, so ask it to poll within one run and report when generation ends. Also ask it to state the public /llms.txt and /agents.md URLs in its final comment, so a reviewer can open the live file and check it against the report instead of trusting the summary. Prompts: - Check whether my llms.txt is current and which version is live, and whether agents.md is custom or Shopify's default. Report on this ticket and change nothing. - Generate a new llms.txt with products and collections but not blog posts. Tell me it writes to my live theme and wait for my reply before you start. - List my llms.txt versions and tell me which one is active. Do not switch or delete anything. In Paperclip: llms.txt generation takes one to two minutes and publishes to the live theme in the same run, and a Paperclip agent woken by a routine will not wait for a yes. With the Claude Code adapter's dangerouslySkipPermissions on by default, a scheduled agent could generate, delete a version or reset agents.md. Keep the routine to the two status tools and the version list, and run generation, deletion and reset as tasks you start yourself. Q: Can Paperclip keep my llms.txt fresh on a schedule? A: StoreSEO can regenerate llms.txt on its own saved schedule, and settings passed in a run are saved for those scheduled runs. A Paperclip routine is better used to check status and report, so you decide when the live file is regenerated. Q: What if I do not like a generated agents.md? A: Reset it and Shopify's default serves again within seconds, but the custom text is discarded with no history. Ask the agent to paste the generated markdown onto the ticket before you reset, if you might want it back. --- ## Set up structured data with Paperclip URL: https://storeseo.com/agent/paperclip/structured-data (full page: https://storeseo.com/agent/paperclip/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data from Paperclip means tasking an agent to list which StoreSEO JSON-LD types are on and what each is missing, then switching a type on only when you have supplied the business details in a task you approve. A schema ticket begins with an inventory. The agent calls storeseo_list_schema_types, which shows every type, whether it is on and whether it takes settings, then storeseo_get_schema_settings for the configurable ones: product, review, sale dates, product merchant, organization and local business. It posts what is missing, for example organization schema with no contact email, or review schema blocked because product schema is off. That is a read-only report and a reasonable weekly routine for a store that keeps changing plans or themes. The change is a separate task because storeseo_update_schema_settings replaces the whole settings object rather than merging, so the agent has to read first and send back every field. The details are yours: business name, logo, address, phone and opening hours should come from the ticket text or a Paperclip secret or file you attach, not from the model's guess. Organization and local business share business information, so editing one edits the other. Turning a type on changes the markup on your live storefront, and the FAQ types need the FAQ Schema feature in your plan. A Paperclip ticket is a good place to keep the before and after: ask the agent to paste the current settings before it writes, so a reviewer can see what changed. If a type is refused for a missing dependency or a plan feature, the agent should name which one on the ticket and stop, rather than retrying the call or switching on other types to get around it. Prompts: - List every JSON-LD schema type on my store, whether it is on, and which configurable ones have missing fields. Post it on this ticket and change nothing. - Set up organization schema using the business name, logo URL and contact email in this ticket's description. Read the current settings first and paste them here before writing. - Review schema was refused last time. Check why and tell me which dependency is off, without turning anything on. In Paperclip: Because a schema update replaces the whole stored object, an agent that skips the read can wipe fields you set earlier. The Claude Code adapter that Paperclip uses runs without permission prompts by default, so nothing but the task text and the agent's habits enforce read-then-write. Put read the current settings first and paste them on the ticket in every schema task, and keep business details in the ticket or a secret so the agent never invents an address. Q: Can a Paperclip routine switch schema types on for me? A: It can only if you let an unattended agent write to your live storefront markup, which is a poor trade. Use the routine to list types and report gaps, and switch a type on in a task you start with the business details attached. Q: Where should I keep my business address for the agent? A: In the ticket description or in a Paperclip secret or attached file, so the agent copies it verbatim. StoreSEO's tools expect complete settings for organization, local business and product merchant, and the agent is meant not to invent any detail. --- ## Manage sitemaps and indexing with Paperclip URL: https://storeseo.com/agent/paperclip/sitemaps (full page: https://storeseo.com/agent/paperclip/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing from Paperclip means letting an agent report from StoreSEO which products, collections, pages and posts are noindex, nofollow or left out of the sitemap, and changing a flag only in a task that names the exact resource. The sitemap flags are quiet settings with loud consequences, which makes them a good fit for a watching routine. The agent calls storeseo_list_resource_sitemaps for products, collections, pages or articles and reports which ones are left out of the XML sitemap, set to noindex or set to nofollow. A Monday task that lists everything flagged, with a note on anything that looks accidental, such as a bestselling product marked noindex, catches a mistake before search traffic shows it. It is all reads, and because flags can also change in the StoreSEO app, the agent should re-list in the same run rather than trusting last week's comment. Changing a flag is one resource per call through storeseo_update_resource_sitemap, which writes Shopify metafields. A noindex flag removes a page from search results, so the task text should name the resource and the flag, and the agent should ask in a comment before it writes. The HTML sitemap is a second, longer task: the agent reads storeseo_get_html_sitemap_settings, generates and enables the page with storeseo_update_html_sitemap, then lists your menus with storeseo_list_sitemap_menus and gives you the edit links. That last step stays with you. Adding the sitemap link to a menu happens in Shopify admin, and no tool does it, so the ticket stays open until you confirm. Leave the choice of menu to a person too: StoreSEO does not store which menu you picked, so the comment on the task is the only record that the link exists. Prompts: - List every product, collection, page and post that is noindex, nofollow or excluded from the sitemap. Flag any that look accidental and change nothing. - Set the product Gift Card Test to noindex and exclude it from the sitemap. Ask me in a comment first, then re-list to confirm the flags. - Set up an HTML sitemap for my store and list my menus with their edit links so I can add it to one. In Paperclip: The one flag that can hurt overnight is noindex, and a routine has nobody to say no. With the Claude Code adapter's default of skipping permission prompts, an agent that decides a page should be hidden can write that flag unattended. Make the scheduled task read only in so many words, keep storeseo_update_resource_sitemap for a task you start, and have the agent re-list flags in the same run before it reports the current state. Q: Can Paperclip alert me when a page becomes noindex? A: A routine can list the flags each week and the agent can post any resource that is noindex or excluded. It reports what it reads at that moment, so it catches a change at the next run rather than the moment it happens. Q: Can the agent add the HTML sitemap to my navigation menu? A: No. It can generate the page and list your menus with edit links, but you add the link in Shopify admin. Then the agent can mark setup complete, and the Paperclip task is the place to record that you did it. --- ## Track keyword rankings with Paperclip URL: https://storeseo.com/agent/paperclip/keyword-tracking (full page: https://storeseo.com/agent/paperclip/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings from Paperclip means giving an agent a recurring task to read your StoreSEO Google positions and competitor movement, while adding or deleting keywords stays in a task you start and approve. Rank tracking is the job most natural to put on a rhythm. A Paperclip routine fires on a cron schedule, creates a task and wakes the agent, which calls storeseo_list_keywords for the latest Google position, change and competitor positions, then storeseo_get_keyword_history for any keyword that moved. It posts a short note: what rose, what dropped out of the top ten, where tracked competitors sit. Because each run is a task with a trace, a month of them gives you a log on the ticket board you can compare week to week. Rank and competitor tracking are Pro plan tools, so on another plan the agent should say so. Ask it to keep the report to the keywords that moved, so the comment stays short enough to read. Adding keywords is a write that touches limits. storeseo_add_keywords needs the exact country and language names from storeseo_list_rank_locations and storeseo_list_rank_languages, counts against your plan's keyword limit and queues a background check, so positions appear on a later run, not in the same reply. Competitors are capped at five and cannot be edited, only deleted and re-added. Deleting a keyword with storeseo_delete_keyword erases its whole history for good. Keyword clusters through storeseo_generate_keyword_cluster spend AI credit, need a focus keyword set in the app and run for a couple of minutes, so they belong in a supervised task as well. The agent then polls storeseo_get_keyword_cluster until the background job is done and posts the related keywords with their search volume and difficulty on the ticket, which is the point at which you decide what a product page should target next. Prompts: - Weekly rank report: list my tracked keywords with position change, mark any that left the top ten, and compare competitor positions. Read only. - Look up the exact United States and English names the tracker accepts, then ask me before adding linen shirt and linen shirt for men on mobile. - Show the rank history for organic cotton tote since July and say whether the drop started before or after the last theme change I noted on this ticket. In Paperclip: storeseo_delete_keyword and storeseo_delete_competitor erase history permanently, and the Claude Code adapter's default of skipping permission prompts means an unattended routine could call them. Word every recurring rank task as read only and name the delete tools as off limits. If you attach StoreSEO as a Paperclip connector instead, set the delete and add actions to Ask first or Off, since the connector classifies each action. Q: Does a Paperclip rank report need a Pro StoreSEO plan? A: Yes for the rank and competitor tools, which are Pro plan features. On another plan the agent should explain that instead of calling them, and Search Console reports from the Google reports job can still give you positions. Q: Why is my new keyword missing from the report? A: Rank checks run in the background. Adding a keyword queues the first check, so its position appears in the list on a later run, usually the next weekly task, rather than in the run that added it. --- ## Read Search Console and Analytics with Paperclip URL: https://storeseo.com/agent/paperclip/google-reports (full page: https://storeseo.com/agent/paperclip/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Search Console and Analytics from Paperclip means scheduling an agent to pull your clicks, impressions, queries and traffic sources through StoreSEO each week and post the movers on a task you can skim. This is the cleanest job to hand to an unattended agent, because both tools write nothing. A Paperclip routine fires on its cron schedule, which Paperclip evaluates in the browser's timezone, creates a task and wakes the agent. The agent calls storeseo_get_search_console_insights for the last 28 days, broken down by query, page, country, device or search appearance, and storeseo_get_analytics_report for a GA4 bundle such as search_traffic, traffic_source or page_performance. Both compare against the previous equal-length period by default, so the comment arrives with the change per metric already worked out. The useful asks are the ones with a threshold: queries ranking between positions eight and twenty, pages with impressions but a low click rate, a traffic source that fell. Report the numbers as Google returns them and do not let the agent project outcomes from them. Where a finding points at a fix, such as a page with weak CTR needing new meta, the agent can open a follow-up task rather than doing it, which keeps the report run harmless. Setup is not something the routine can do. Connecting Google and choosing the GA4 property happen in the StoreSEO app, and a paid plan is required, so when the tool returns a link saying what to set up, the agent should post that link on the task for you. A good weekly note also keeps its length in check: five movers per report, each with its before and after, is easier to act on than a dump of every query. Prompts: - Weekly report: pull Search Console clicks, impressions, CTR and position for the last 28 days against the previous 28, and list the ten biggest query movers. Read only. - Which product pages have high impressions but a CTR under two percent? Post them with their queries and open one follow-up task for me to review, without changing anything. - Pull GA4 traffic_source for the last three months and note which channels grew, with the previous-period comparison. In Paperclip: A routine can only report what Google data StoreSEO can reach, and the setup is outside the agent's reach. If Google is not connected, or no GA4 property is selected, the tool returns a link to fix it in the StoreSEO app and every weekly run will repeat that message until you do. Also remember the agent itself is not limited to reads: with the Claude Code adapter's default of skipping permission prompts, keep the routine's instructions read only. Q: Is a Search Console report safe to schedule in Paperclip? A: Yes, as far as StoreSEO goes: both Google tools are read-only and never connect or configure Google. Still word the routine as read only, since the agent running it could call other StoreSEO tools if the instructions allow it. Q: Which StoreSEO plan does a Paperclip Google report need? A: A paid StoreSEO plan with Google connected in the app. Analytics reports also need a GA4 property selected. Paperclip itself is open source and self-hosted, so the plan question is only about StoreSEO. --- ## Write SEO blog post drafts with Paperclip URL: https://storeseo.com/agent/paperclip/blog-posts (full page: https://storeseo.com/agent/paperclip/blog-posts/index.md). Job: Blog post drafts, described once above. Drafting blog posts from Paperclip means giving an agent a content ticket that has StoreSEO write a full SEO post into a Shopify blog as a draft, with the credit cost agreed on the task and publishing left to you in Shopify admin. Blog drafting suits Paperclip's way of working because it is slow, queued and reviewed by a person. The ticket carries the brief: topic, target keyword, word count and whether to add a featured image. The agent calls storeseo_list_blogs to find the numeric id of the blog the post should go into. An empty list means your blogs are not synced yet, which you fix in the StoreSEO app. Then it states the expected AI credit, which grows with length and a featured image, and asks. Only after your reply does it call storeseo_generate_blog_post, and if credit is short the call fails before anything is queued or spent. Writing runs in the background for a few minutes, so the agent polls storeseo_get_blog_post_job and comments progress as steps completed out of steps total. It should not call the post finished while the job runs, since the draft article row exists from the start. This is a place where Paperclip's heartbeats help: a run that polls within one wake can post the completed status on the ticket without you watching. When the job completes, the agent reads the draft's SEO score with storeseo_get_resource_seo and links it. Publishing is not part of this job. The post is a draft in your Shopify blog, and the person who reads the draft and publishes it from Shopify admin is you or an editor on your team. Name that person on the ticket, so the finished task has a clear next owner. Prompts: - List my synced blogs, then write a 1,200 word draft on how to care for linen shirts in the News blog. Post the expected credit on this ticket and wait for my reply before you start. - Check the blog-writing job from the last task. If it has finished, read the draft's SEO score and post it with what still needs a human edit. - Suggest three post ideas for my store as comments on this ticket, without generating anything, so I can choose one. In Paperclip: A routine that writes a post on every run will spend AI credit on every run, and with the Claude Code adapter's default of skipping permission prompts nobody is asked first. Do not schedule storeseo_generate_blog_post. Schedule a routine that suggests topics or checks an earlier job's status, and start each real draft as a task where you state the word count and image choice, so the cost is one you chose. Q: Will a Paperclip agent publish the blog post to my store? A: No. StoreSEO saves the generated post as a draft in the Shopify blog you choose, and nothing goes live. You or an editor reviews it and publishes from Shopify admin. Q: How do I keep blog drafting inside the AI credit I have? A: Put the word count and image choice in the ticket and have the agent state the expected credit before it starts. If your credit is not enough, the call fails before anything is queued or spent, and the agent should report that on the task. --- ## Audit your store's SEO with NanoClaw URL: https://storeseo.com/agent/nanoclaw/seo-audit (full page: https://storeseo.com/agent/nanoclaw/seo-audit/index.md). Job: Store SEO audit, described once above. Auditing your Shopify SEO with NanoClaw means texting your own self-hosted agent a question in the chat app you already use, so it reads your StoreSEO scores in its container and replies with a ranked list of what to fix first. You are in a chat thread, not a dashboard. Open WhatsApp, Telegram, Slack or Discord, write the agent's trigger word, which is @Andy unless you renamed it, and ask where your store stands. The agent calls storeseo_start, which returns the conversation id every other StoreSEO tool needs plus a snapshot of how many resources are optimized, need improvement or are not optimized. It then calls storeseo_list_resources sorted by score, lowest first, and storeseo_get_resource_seo on the worst few. The reply arrives as a short chat message, so ask for a top five rather than a full catalogue, because a long listing is hard to read on a phone. All three of those calls only read. Whether your fork stops to ask before running any tool at all depends on how you and your admins configured it, so on a fresh install you may see no prompt, and on a locked down one you may see a request every time. Either way the audit itself changes nothing in your store. Ask the agent to split its answer into fixes StoreSEO can generate, such as meta text and alt text, and fixes only you can make in the StoreSEO app, such as setting a focus keyword. That split tells you whether the next message should be a preview request or a trip to the app. The audit also fits NanoClaw's scheduled tasks. Create one with ncl tasks create and a cron recurrence such as every Monday at 8:00, and write the prompt as audit only. A scheduled task has no chat attached, so the prompt must name where the result goes, for example send it to Telegram. Because nobody is present to say yes when the task runs, keep preview and apply out of it. Read the Monday message, pick two or three products, and start a normal chat to fix them. Prompts: - @Andy use StoreSEO to list my five lowest-scoring products, name the biggest failing check on each and tell me which ones only I can fix in the app. - @Andy give me a one-message summary of how many of my products, collections and pages are optimized, need work or are not optimized. - @Andy which of my pages and blog posts score lowest? Show the three worst and what is missing. In NanoClaw: NanoClaw adds the StoreSEO server to one agent group at a time, so an audit asked in a chat that belongs to a different group will not find the tools. Add the server to every group you message from, and restart each group's container with ncl groups restart before expecting the audit to work. Also remember the token sits in that group's container configuration on your host, so treat the machine like a password. Q: Can I run the StoreSEO audit from my phone with NanoClaw? A: Yes. Message your agent with its trigger word from WhatsApp, Telegram, Slack or Discord, and it calls the StoreSEO read tools and replies in the same thread. The audit tools only read your scores, so ask for a short top five that fits a phone screen. Q: Will a NanoClaw scheduled audit change anything in my store? A: Not if you word the prompt as read only. StoreSEO tells the agent to wait for your yes before writing, but a scheduled task has nobody to answer, so keep preview and apply out of it and name the chat destination in the prompt. --- ## Optimize meta tags and content with NanoClaw URL: https://storeseo.com/agent/nanoclaw/optimize-meta (full page: https://storeseo.com/agent/nanoclaw/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags with NanoClaw means asking your self-hosted agent, in a chat message, to have StoreSEO draft a meta title, description, tags and URL handle for one Shopify resource and to save only the parts you answer yes to. The whole exchange happens in one chat thread. You message the agent with its trigger word and name a single product, for instance @Andy improve the meta on my Soy Candle. The agent calls storeseo_get_resource_seo to read the current title and description, then storeseo_preview_resource_optimization. StoreSEO writes the draft on its own servers, so the words come from StoreSEO and not from the model behind NanoClaw. The response carries a draft id and a projected score change, and the agent pastes the old and new text into the chat so you can compare them on your phone. Reply in plain words. If you say apply the title and description but skip the handle, the agent passes only those fields to storeseo_apply_resource_optimization. A changed URL handle creates a redirect from the old address, which is why the agent should show both addresses before that call. The preview spends AI Content Optimizer credit even if you decline, so name one product at a time. Your yes in the thread is an instruction the agent follows. Whether NanoClaw itself pauses before the tool runs depends on your fork and your admin approval settings, so the chat confirmation may be the only check you have. Do the previews in the conversation and keep them out of scheduled tasks. A task that previews on a cron schedule would spend credit every run and has no one to approve the result. A better pattern is a report only task that lists the lowest scoring products each week, followed by a chat where you pick one and walk through its meta. Prompts: - @Andy use StoreSEO to preview new meta for my Soy Candle product and show the old and new title, description and handle before saving anything. - @Andy preview meta for my three lowest-scoring products, one at a time, and wait for my reply after each one. - @Andy apply only the meta title and description from that draft and skip the URL handle. In NanoClaw: On a chat app a yes can be misread. If you send a quick thumbs up or a reply that lands on a different message in a busy Slack or Discord channel, the agent may take it as consent for the wrong draft. Reply in a full sentence that names the product and the fields, such as yes, apply the Soy Candle title and description. If your fork has no tool approval, that sentence is the only brake before the write. Q: Does the model behind NanoClaw write my meta text? A: No. StoreSEO generates the meta title, description, tags, handle, FAQs and AI snippet on its own servers. NanoClaw's agent fetches the draft, shows it in the chat and passes along only the fields you approve. Q: How do I make NanoClaw ask before it saves meta? A: StoreSEO tells the agent to preview first and wait for your yes. That is an instruction, not a lock, so the hard stop is whatever approval behavior you set in your own fork. Test it once with a harmless product before trusting it. --- ## Optimize images and alt text with NanoClaw URL: https://storeseo.com/agent/nanoclaw/image-seo (full page: https://storeseo.com/agent/nanoclaw/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing Shopify image SEO with NanoClaw means telling your self-hosted agent in a chat which product images are too heavy or have no alt text, and letting StoreSEO optimize them one at a time while you confirm in the thread. Start from the chat, with a question rather than a command: @Andy which product images are heaviest and have no alt text? The agent calls storeseo_start and storeseo_list_images, which sorts by file size by default and reports compression, alt text and file name status for each image. That is a read, so you get a short shortlist with sizes and a count in a single message. Filters work in plain words, such as only images with no alt text, or only one collection, and the agent turns them into the listing's filters. The write is storeseo_optimize_image, one image per call, with every axis you want in the same call. Each reply reports per axis: compression applied, alt text written with AI credit, file renamed, or a refusal because the Image Optimizer add-on is missing. Run them in sequence, not in parallel, because a file rename swaps the image's underlying Shopify media. If you dislike a result, message the agent to restore that image and name the axis, and it calls storeseo_restore_image. A renamed file gets a new storefront URL, so ask the agent to list which ones it renamed. Chat suits this job because each image is a short exchange you can answer when you have a minute, and NanoClaw keeps the thread's memory between messages. State a scope in the first message: five images, product images only, alt text and compression but no renaming. Scheduling is the wrong tool for the optimizing itself. A report only task that counts unlabelled images each week is fine, and the optimizing belongs in a thread where you can say yes or stop. Prompts: - @Andy use StoreSEO to list my ten heaviest product images and tell me which have no alt text. Read only for now. - @Andy optimize the alt text and compression of those first five images, one at a time, and tell me the result for each axis. - @Andy restore the file name on the last image you optimized, I do not want the new URL. In NanoClaw: A chat thread invites you to reply go on for the rest of the list, and a long run of storeseo_optimize_image calls can outlive a restart of your agent's container. If the container restarts mid batch, do not assume the last image finished. Ask the agent to list the images again and read the status of each axis, rather than repeating a call, because each repeat can spend quota or credit a second time. Q: Can NanoClaw optimize images from a WhatsApp message? A: Yes. Message the agent with its trigger word, name the scope, and it calls the StoreSEO image tools one image per call. Each result is reported back in the thread, so you can restore any image you do not like. Q: Why did NanoClaw say compression was refused? A: Compression needs the Image Optimizer add-on and uses its quota. The agent should report the refusal per axis instead of retrying, and it can still write alt text or a file name for that image if you want those. --- ## Get products ready for AI search with NanoClaw URL: https://storeseo.com/agent/nanoclaw/ai-search-readiness (full page: https://storeseo.com/agent/nanoclaw/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search with NanoClaw means asking your self-hosted agent in a chat which products score lowest on StoreSEO's GEO check, fixing what tools can fix, and answering its questions to rewrite a thin description. The GEO score describes how well AI shopping agents can understand a product, and it applies to products only. In your chat app you ask @Andy which products are weakest for AI readiness. The agent calls storeseo_start, which includes a GEO overview, then storeseo_get_resource_seo on a product to read the failing criteria. Some are fixable with tools, such as FAQs, the AI snippet, alt text and product schema, and others, such as title, category or price, you change in Shopify or the StoreSEO app. Ask the agent to label each criterion so you know which message to send next. The description rewrite becomes a conversation. storeseo_scan_product_description scores seven pillars and returns gap questions, and the agent must put them to you word for word and in order. In a chat that means a few short messages where you reply with materials, dimensions and care instructions from your phone, then the agent calls storeseo_write_product_description with your answers. That write saves immediately and cannot be undone. The response returns the previous description once, so ask the agent to paste the old text into the thread before it writes, which gives you a copy in your own chat history. The scan spends AI credit but saves nothing, so it is a safe place to look before you commit. Keep the write to one product per exchange. A scheduled task can read GEO scores each week and send you the weakest five, worded as report only, but the question and answer part cannot be scheduled, because no one is there to answer. Whether NanoClaw stops before the write depends on your fork, so the thread may be the only check. Prompts: - @Andy use StoreSEO to list my five products with the lowest GEO score and say which failing criteria a tool can fix. - @Andy scan the description of my Linen Throw product, ask me the gap questions word for word and paste the current description here before writing anything. - @Andy preview FAQs and an AI snippet for my three weakest products and wait for my reply before applying each. In NanoClaw: The description write has no undo, and a chat answer can be short enough to be ambiguous. If you reply only yes to a long list of questions, the agent may fill the gaps itself. Answer each question in its own words, and tell the agent to wait for them. A product also needs a focus keyword set in the StoreSEO app before a scan, which no chat message to NanoClaw can set for you. Q: Does my NanoClaw agent answer the description questions for me? A: It should not. StoreSEO tells the agent to put the scan's questions to you word for word and in order. If you answer them in the chat, the written description reflects your facts instead of inferred guesses. Q: How do I keep the old description after NanoClaw rewrites it? A: StoreSEO keeps no history, and the write response returns the old text only once. Ask the agent to paste the current description into the chat before writing, so a copy stays in your thread. --- ## Publish llms.txt and agents.md with NanoClaw URL: https://storeseo.com/agent/nanoclaw/llms-txt-agents-md (full page: https://storeseo.com/agent/nanoclaw/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md with NanoClaw means asking your self-hosted agent in a chat which AI discovery files your Shopify store serves, then having StoreSEO generate the missing ones into your live theme after you reply yes. Two files tell AI crawlers and shopping agents about your store: llms.txt, a curated map of your content, and agents.md, a description of who you are and how to buy. In a chat you message @Andy and ask if both are live. The agent calls storeseo_start, storeseo_get_llms_txt_status and storeseo_get_agents_md_status, all reads, and replies with what is served, when llms.txt was last generated and whether your settings changed since. The public URLs come back in the message, so you can tap them and read the files yourself. Generation needs a clear heads up in the thread. storeseo_generate_llms_txt writes a template into your live theme and uploads a copy to Shopify Files in the same run, and it takes about one to two minutes, so the agent should say so, wait for your yes, then poll storeseo_get_llms_txt_status until generation is no longer ongoing. storeseo_generate_agents_md also publishes live and spends AI credit, and storeseo_reset_agents_md hands the file back to Shopify's default with no history. NanoClaw's container keeps running while you close the chat, so the agent can finish polling and message you the result. A quiet scheduled task suits this job. Create one with ncl tasks create that reads both statuses every Monday and sends you a message only when something is missing, stale or due for a rebuild. StoreSEO can regenerate llms.txt on its own saved schedule, so the task does not need to generate anything. Keep generation, activation, deletion and reset in a chat where you can answer, because a scheduled run has nobody to say yes. Prompts: - @Andy use StoreSEO to tell me whether my llms.txt and agents.md are live, when llms.txt last generated and which version is active. - @Andy explain what generating llms.txt will write to my theme and Shopify Files, and wait for my reply before you start. - @Andy list my last llms.txt versions and tell me which one is serving at my store's /llms.txt. In NanoClaw: A scheduled task prompt written as keep llms.txt up to date can be read as permission to call storeseo_generate_llms_txt on every run, publishing to your live theme while nobody is watching. NanoClaw tasks run as the agent, with whatever approval your fork applies, so word the prompt as read both statuses and report, and say explicitly that nothing is to be generated, activated, deleted or reset. Q: Will a NanoClaw scheduled task regenerate llms.txt by itself? A: Only if you word the task that way, and it is better not to. StoreSEO has its own saved schedule for llms.txt, so a task can simply read the status and message you when a rebuild is due. Q: How do I undo agents.md from a NanoClaw chat? A: Ask the agent to call storeseo_reset_agents_md, which makes Shopify's default file serve again within seconds. The custom text is gone for good, so the agent should confirm with you before it runs. --- ## Set up structured data with NanoClaw URL: https://storeseo.com/agent/nanoclaw/structured-data (full page: https://storeseo.com/agent/nanoclaw/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data with NanoClaw means asking your self-hosted agent in a chat to list your StoreSEO schema types, read the current settings of the one you care about, and send back a complete edited object after you confirm. Structured data rewards a careful order of steps, and a chat thread is a good place to keep them visible. You message @Andy and ask which schema types are on. The agent calls storeseo_start and storeseo_list_schema_types, which shows every JSON-LD type, whether it is enabled and whether it is configurable. For one type, say organization, it calls storeseo_get_schema_settings, which returns the current values and a field template describing the full object to send back. These are reads, so a single reply can summarize the whole set. storeseo_update_schema_settings replaces the stored settings wholesale, so a half filled object can wipe good values. Before it edits anything, ask the agent to paste the settings it just read into the chat. That message becomes your rollback copy, kept in your own history. Then give it the business details in the thread, such as legal name, address and opening hours, and have it show the complete object back before the update. It should not invent any of those details, and the agent should tell you if a required field is still missing. Some types depend on others. Review and product merchant schema need product schema on first, and the four FAQ types need the FAQ Schema feature in your plan, so a refusal is possible and the agent should name the missing piece instead of retrying. Each update changes the markup your live storefront serves, so keep it in a conversation. A scheduled task that only lists the type states and sends you any change since last week is as far as automation should go here. Prompts: - @Andy use StoreSEO to list my schema types and tell me which are on, which are off and which I can configure. - @Andy read my organization schema settings, paste them here as a backup, then show me what a complete set looks like with my business details. - @Andy turn on product schema and send the full settings, but show me the change and wait for my reply before you call the update. In NanoClaw: Pasting your address and phone number into a WhatsApp or Telegram thread stores them in that chat's history and in NanoClaw's memory for the agent group. For organization and local business schema that is usually public business information, but do not paste anything you would not publish. Keep tokens out of the thread entirely, since the StoreSEO access token belongs in the container configuration, not in a message. Q: Can NanoClaw make up my business details for organization schema? A: It should not. StoreSEO's guidance is that business details come from you, so message the agent your name, address and hours and have it show the whole object before it calls the update. Q: Why did NanoClaw report that a schema type was refused? A: Some types need a plan feature or another type enabled first. FAQ types need the FAQ Schema feature, and review needs product schema on, so the agent should name what is missing and stop. --- ## Manage sitemaps and indexing with NanoClaw URL: https://storeseo.com/agent/nanoclaw/sitemaps (full page: https://storeseo.com/agent/nanoclaw/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps with NanoClaw means asking your self-hosted agent in a chat which Shopify resources StoreSEO leaves out of the sitemap or marks noindex, then changing those flags one resource at a time after you reply yes. One stray noindex flag can remove a page from search results, so this job is best done in a thread you can scroll back through. Message @Andy and ask what is hidden from search. The agent calls storeseo_start and then storeseo_list_resource_sitemaps for one resource type at a time, filtered on sitemap, noindex or nofollow. Each row shows the flags for a product, collection, page or article. Ask it to group the result, for instance every noindex page and every product left out of the sitemap, and reply with the ones that were intentional. Changes go through storeseo_update_resource_sitemap, one resource per call, and it writes Shopify metafields on that resource. The agent should name the resource and the flag in the thread, wait for your yes, make the call, and then list the flags again to confirm the state. The HTML sitemap adds a manual step: storeseo_update_html_sitemap generates it, you add the link to a menu in Shopify admin using the edit URL from storeseo_list_sitemap_menus, and only then does the agent mark setup complete. The agent cannot add the menu link for you. NanoClaw scheduled tasks work well on the read side. Create one that lists noindex and sitemap-excluded resources every Monday and tells you what is new since last time, which catches an accidental flag early. Keep it report only. Changing a flag from a scheduled run would hide a page from search with nobody watching, and your fork may or may not stop to ask before the tool runs, so the prompt itself has to forbid it. Prompts: - @Andy use StoreSEO to list every page and product marked noindex or excluded from the sitemap, and ask me which are intentional. - @Andy remove my Summer Sale page from the sitemap with StoreSEO, name the flag you will change and wait for my reply first. - @Andy set up the HTML sitemap, give me the menu edit link and tell me what I must do in Shopify admin before you mark it complete. In NanoClaw: Group chats are the risk here. In a shared Slack or Discord channel anyone who can message your agent with the trigger word can ask for a flag change, and the StoreSEO token behind it is yours. Add the StoreSEO server only to an agent group whose chat has just the people you trust with your store, and check how your fork restricts who may trigger the agent before allowing noindex changes. Q: Can a NanoClaw scheduled task flag pages as noindex for me? A: It can if the prompt says so, but nobody is there to confirm, and a wrong flag hides a page from search. Keep scheduled tasks to listing flags and reporting what is new, and make changes in a chat. Q: Can NanoClaw add the HTML sitemap to my menu? A: No. After it generates the sitemap, you add the link to a menu in Shopify admin yourself, using the edit link the agent sends you, and then the agent can mark setup complete. --- ## Track keyword rankings with NanoClaw URL: https://storeseo.com/agent/nanoclaw/keyword-tracking (full page: https://storeseo.com/agent/nanoclaw/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings with NanoClaw means telling your self-hosted agent in a chat which keywords matter, letting StoreSEO check Google positions, and having a scheduled task send you a short digest of what moved. Rank tracking is a Pro feature, and it is the job where a recurring message pays off most. You message @Andy with the keywords and the market, for example soy candles for the United States, on mobile. The agent calls storeseo_start, then looks up the exact values with storeseo_list_rank_locations and storeseo_list_rank_languages, because storeseo_add_keywords needs those names verbatim and typing them from memory fails. It asks for the device, then adds the keywords. The rank check is queued, so the first read may show no position yet, and the agent should say so. Reading is cheap. storeseo_list_keywords returns the latest rank and change, storeseo_get_keyword_history returns the dated history, and storeseo_list_competitors shows up to five competitor domains. This suits a NanoClaw scheduled task: every Monday morning, list the keywords that moved most, name any that left the top ten, and send the result to your chat. Because a scheduled task has no chat attached, the prompt has to name the destination, and it should say read only so the run never adds or deletes a keyword on its own. Deletes are the part to keep in a conversation. storeseo_delete_keyword and storeseo_delete_competitor remove the history for good, and adding keywords counts against your plan's keyword limit. The tracker covers Google only, not AI answer engines, so ask the agent not to describe AI visibility from it. A digest in a chat thread also gives you a scroll back timeline, which is a simple way to see a trend without keeping notes. Prompts: - @Andy use StoreSEO to add the keywords soy candles and scented candles for the United States, English, on mobile, and confirm the rank check was queued. - @Andy show my tracked keywords that dropped the most this month and the history for the worst one. - @Andy list my tracked competitors and tell me which of them outranks me on my top three keywords. In NanoClaw: On a lower plan the rank tools refuse, and a Monday task would just send you the same error each week. Test the question in a chat first. Also note that a scheduled task has no memory of what you said in other chats unless your agent group shares it, so the task prompt must spell out the keywords or say list all tracked keywords, and must say read only so a deletion is never part of it. Q: Can NanoClaw send me a weekly rank digest? A: Yes. Create a scheduled task with a cron recurrence, write the prompt as read only and name the chat where the digest should go, because a scheduled task has no conversation attached. Q: Can NanoClaw track how I appear in ChatGPT or Gemini? A: No. StoreSEO's rank tracker checks Google only and does not track AI search engines, so a NanoClaw digest covers Google positions and the competitor domains you added. --- ## Read Search Console and Analytics with NanoClaw URL: https://storeseo.com/agent/nanoclaw/google-reports (full page: https://storeseo.com/agent/nanoclaw/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports with NanoClaw means asking your self-hosted agent in a chat for Search Console and Google Analytics 4 figures through StoreSEO, and getting a short written summary, which also makes it the safest job to schedule. Both Google tools only read, so this job asks the least of your approval settings. In your chat app you message @Andy something plain, such as which queries bring the most clicks and which pages sit between positions eight and twenty. The agent calls storeseo_start, then storeseo_get_search_console_insights with the dimension that fits, which can be date, query, page, country, device or search appearance, and storeseo_get_analytics_report for GA4 views such as traffic source or page performance. Both default to the last 28 days and compare with the previous period. Chat is a good size for the answer. Ask for the top five rows and the change, not the whole table, and follow up with a second message to dig into one query. A finding can lead straight into another job: a page with many impressions and a low average position becomes a candidate for a meta preview in a later thread. The agent reports the figures as returned and should not project outcomes from them. The natural automation is a NanoClaw scheduled task. Create it with ncl tasks create, a Monday cron recurrence, and a prompt that reads both reports, names the three biggest gains and drops and sends the summary to a named chat. Because these tools only read, an unattended run changes nothing in your store. The setup lives in the StoreSEO app: a paid plan, Google connected, and a GA4 property selected for the Analytics report. If any of that is missing, the agent cannot fix it from chat and should tell you where to go. Prompts: - @Andy use StoreSEO to show Search Console queries from the last 28 days where we rank between 8 and 20, with impressions and CTR. - @Andy compare this month's organic traffic by channel with last month using the StoreSEO GA4 report and tell me the biggest change. - @Andy which five pages lost the most clicks in Search Console against the previous period? In NanoClaw: These reports need a paid StoreSEO plan with Google connected, and the Analytics report also needs a GA4 property chosen in the StoreSEO app. A weekly NanoClaw task without that setup will only message you an error each Monday, so ask the same question in a chat first. Also give the task a destination, since a scheduled run has no chat attached and may otherwise have nowhere to send the summary. Q: Can my NanoClaw agent connect my Google account? A: No. Connecting Google and choosing the GA4 property happen in the StoreSEO app. After that the agent only reads the reports, and if the connection is missing it should tell you to finish it there. Q: Is it safe to leave a NanoClaw Google report task running weekly? A: Both report tools are read only, so a task that calls just those changes nothing in your store. Keep write tools out of that task's prompt and use a separate chat for any fixes. --- ## Write SEO blog post drafts with NanoClaw URL: https://storeseo.com/agent/nanoclaw/blog-posts (full page: https://storeseo.com/agent/nanoclaw/blog-posts/index.md). Job: Blog post drafts, described once above. Drafting blog posts with NanoClaw means messaging your self-hosted agent a topic, letting StoreSEO write a complete draft article in one of your Shopify blogs in the background, and getting a chat message when the draft is ready to review. A post takes StoreSEO a few minutes to write, which suits an agent that keeps running in its container after you put your phone down. In your chat app you message @Andy with a brief: the topic, the audience, a target word count and whether you want a featured image. The agent calls storeseo_start, then storeseo_list_blogs to find the numeric id of the blog, since only blogs synced in the StoreSEO app are listed. It tells you the credit cost, which grows with word count and a featured image, and waits for your reply before storeseo_generate_blog_post. After the call the agent polls storeseo_get_blog_post_job until the status is completed, failed or cancelled, and reports progress as steps completed out of total. The draft article exists from the start of the run, so it should not call the post finished while the job is still going. When it completes, ask the agent to call storeseo_get_resource_seo with the article's Shopify id and send you the draft's score and failing checks in the same thread. The result is a draft only. You review it and publish it in Shopify admin, a step the agent cannot take. Keep generation out of scheduled tasks, because each run spends credit and would produce drafts you never asked for. A better scheduled task is a monthly message suggesting three topics from your Search Console queries, worded as suggest only. Then you pick one in a chat and approve the cost there, where your yes is part of the thread. Prompts: - @Andy use StoreSEO to list my blogs, then tell me the credit cost of a 1,200 word post about caring for linen bedding before you write it. - @Andy write that post as a draft in the News blog with a featured image, follow the job until it finishes, then tell me the SEO score of the draft. - @Andy suggest three blog topics from my Search Console queries with StoreSEO. Suggest only, do not generate anything. In NanoClaw: Polling depends on your NanoClaw container staying up. If the host restarts or you run ncl groups restart while a post is generating, the StoreSEO job carries on, but the agent may lose track of the job id. Ask it to start from storeseo_list_resources for articles, or give it the job id from the thread, and treat the draft as unfinished until the job reports completed. Q: Can NanoClaw publish the blog post for me? A: No. StoreSEO saves the post as a draft in your Shopify blog and does not publish it live. You review the draft and publish it from Shopify admin. Q: What if I stop chatting while the post is generating? A: The job runs in the background on StoreSEO's side, and your agent keeps running in its container, so it can keep polling. When you return, ask for the job status and wait for completed before judging the draft. --- ## Audit your store's SEO with Kimi Code URL: https://storeseo.com/agent/kimi/seo-audit (full page: https://storeseo.com/agent/kimi/seo-audit/index.md). Job: Store SEO audit, described once above. Running a store SEO audit in Kimi Code means typing one question into Moonshot AI's terminal agent, letting it read your StoreSEO scores and per-criterion checklists through MCP, and getting back a ranked worklist in the same session where you edit your theme. You start kimi in the folder you normally work from, which for many Shopify developers is the theme repository, and ask something plain such as which products score lowest and what each one is missing. Kimi Code calls storeseo_start first, because every other StoreSEO tool needs the conversation id it returns, and that call also hands back a snapshot of how many products, collections, pages and articles are optimized. It then calls storeseo_list_resources with the lowest scores first and storeseo_get_resource_seo on the worst candidates. Kimi Code starts in Always Ask mode, where read-only tool calls run without a prompt, so an audit scrolls past with no approval panel at all. The terminal changes what you can do with the result. Because the agent already has the repository open, you can ask it to write the worklist to a file, for example seo-worklist.md at the project root, or to split the findings into what StoreSEO can generate and what needs a Liquid change. Products also carry a GEO score for AI shopping agents, so a good follow-up is which listings are weakest for AI search. Criteria that no tool can fix, such as choosing a focus keyword, come back as items for you to handle in the StoreSEO app. If the tools are missing on the first try, the cause is almost always timing. A server added to ~/.kimi-code/mcp.json while Kimi Code is open registers only in the next session, so quit, start kimi again and run /mcp-config login storeseo before you ask. Once the audit has given you a worklist, the next step is a preview in the same session, and that is where the approval behavior starts to matter. Prompts: - Use StoreSEO to list my 15 lowest-scoring products, name the biggest failing check on each and mark which ones only I can fix in the app. - Audit my collections and pages, worst first, and save the result as seo-worklist.md in this repository. - Which products have a weak GEO score, and what would make them easier for AI shopping agents to read? In Kimi Code: Kimi Code runs read-only calls straight away in its default mode, which suits an audit, but remember the setting when the audit turns into fixes. If you switched to /yolo or /auto earlier in the day, StoreSEO's preview and apply tools stop prompting too, and the only check left is the confirmation StoreSEO tells the agent to ask for. Switch back to the default mode before you move from reading scores to changing them, or keep a permission rule that leaves the write tools on ask. Q: Does a StoreSEO audit in Kimi Code ask me to approve each call? A: Not in Always Ask mode. The audit tools storeseo_start, storeseo_list_resources and storeseo_get_resource_seo only read, and Kimi Code runs read-only calls without a prompt. Prompts appear later, when a tool writes or spends credit. Q: Can Kimi Code save the audit into my theme repository? A: Yes, it can write a Markdown or CSV file into the folder it was started in, since it edits project files. The StoreSEO calls themselves change nothing in your store, but file writes follow Kimi Code's own approval setting. --- ## Optimize meta tags and content with Kimi Code URL: https://storeseo.com/agent/kimi/optimize-meta (full page: https://storeseo.com/agent/kimi/optimize-meta/index.md). Job: Meta tags and on-page content, described once above. Optimizing meta tags in Kimi Code means asking Moonshot AI's terminal agent to have StoreSEO generate a meta title, description, tags, URL handle and FAQs for one resource, reading the draft in the terminal and applying only what you approve. You work in the kimi prompt and name one resource to begin with, for example rewrite the meta for my linen shirt product. Kimi Code reads the current state with storeseo_get_resource_seo, then calls storeseo_preview_resource_optimization. StoreSEO writes the copy on its own servers, so the wording comes from StoreSEO rather than from the Kimi model, and the response carries a draft id and a projected score change. The preview spends AI Content Optimizer credit and saves nothing, which is why starting with one product before a larger batch is the sensible pattern. Reading the draft is easier in a terminal than it sounds. Ask Kimi Code to print the old and new title and description as a short before and after block, and to show the old and new URL when the handle changes, because applying a handle creates a redirect from the old address. You then reply with which fields you accept. storeseo_apply_resource_optimization takes a subset, so you can approve the title and description and skip the handle, and the draft stays usable for an hour if you want the rest later. Kimi Code can also compare the draft against the theme. If a template builds the title tag from a metafield or appends a store name, ask it to check the Liquid before you apply so the final title does not repeat the brand. For a larger run, ask for previews in batches of five and apply after each batch. Each preview costs credit whether or not you apply it, so keep the scope tight. Prompts: - Use StoreSEO to preview new meta for my Linen Shirt product and show the old and new title, description and handle before saving anything. - Preview meta for my three lowest-scoring products, one at a time, and wait for my yes before applying each one. - Apply only the meta title and description from that draft and skip the URL handle. In Kimi Code: Kimi Code's Approve for this session button is the one to watch here. Choosing it on storeseo_apply_resource_optimization lets every later apply in that session go through without a panel, so a batch of ten products becomes ten silent writes. Approve the first apply once, read what comes back, and keep a permanent rule in config.toml only for tools you are comfortable running unattended. Preview calls spend credit, so they deserve the same care even though they write nothing. Q: Does the Kimi model write the meta text? A: No. StoreSEO generates the meta title, description, tags, handle and FAQs on its own servers when Kimi Code calls the preview tool. Kimi Code reads, presents and applies the draft, and any extra edits you want go through your own instructions. Q: What happens if I apply a URL handle change from Kimi Code? A: StoreSEO updates the live URL and creates a redirect from the old address. The agent should show you the old and new URL before the apply call, and you can skip the handle while still applying the title and description. --- ## Optimize images and alt text with Kimi Code URL: https://storeseo.com/agent/kimi/image-seo (full page: https://storeseo.com/agent/kimi/image-seo/index.md). Job: Image SEO and alt text, described once above. Fixing image SEO in Kimi Code means asking Moonshot AI's terminal agent to list the product, collection and article images StoreSEO still has work for, then run compression, alt text and file names one image per call after you agree the scope. You ask for a slice of the catalog, such as the heaviest unoptimized product images or every image with no alt text, and Kimi Code calls storeseo_list_images with the matching filters. The listing shows each image's three axes, compression, alt text and file name, so the agent can tell you how many images need which kind of work before anything is spent. Reading the list is a plain read and runs without a prompt in Kimi Code's default Always Ask mode. The writing step is storeseo_optimize_image, and it works on one image per call with every wanted axis passed together. Compression counts against the Image Optimizer quota and needs that add-on, while alt text and file names spend AI Content Optimizer credit. In a terminal this becomes a loop you can watch: the agent calls the tool, prints a one-line report per axis, and moves to the next image. Kimi Code shows an approval panel for each call, which is useful on the first few images because you see the real alt text before you let it continue. A file name change gives the image a new storefront URL, so Kimi Code can grep the theme repository for the old file name before you rename. If something hard-codes the old URL, you find out in the same session. If you dislike a result, storeseo_restore_image reverts compression, alt text or the file name, and the agent can restore one image without touching the rest of the batch. Prompts: - Use StoreSEO to list product images with no alt text, tell me how many there are and what writing alt text for all of them would spend. - Optimize the heaviest five unoptimized images, compression and alt text together, one image per call, and show me each result. - Restore the file name on the last image you renamed and search this theme for the old file name first. In Kimi Code: A bulk image run is where Kimi Code's session approval does the most damage if misused. Choosing Approve for this session on storeseo_optimize_image removes the panel for every remaining image, and each call can spend credit or quota. If a response is lost, for example a dropped connection, ask Kimi Code to re-list the image rather than retry the call, because a retry regenerates alt text and spends again. Agree the number of images before you let a loop start. Q: Can Kimi Code compress images without the Image Optimizer add-on? A: Compression needs the Image Optimizer add-on and counts against its monthly quota. Without it, StoreSEO reports that axis as refused while alt text and file names, which spend AI Content Optimizer credit, can still apply in the same call. Q: Why does Kimi Code optimize images one at a time? A: The StoreSEO tool works on one image per call because the work is synchronous and the axes share a lock. Parallel calls on the same image can collide, so the agent passes every axis you want in a single call and then moves on. --- ## Get products ready for AI search with Kimi Code URL: https://storeseo.com/agent/kimi/ai-search-readiness (full page: https://storeseo.com/agent/kimi/ai-search-readiness/index.md). Job: AI search readiness, described once above. Getting products ready for AI search in Kimi Code means using Moonshot AI's terminal agent to read each product's GEO checklist in StoreSEO, preview FAQs and an AI snippet, and then answer the description scan so an agent-ready description can be written. You open the session and ask which products AI shopping agents are most likely to misread. Kimi Code calls storeseo_get_resource_seo, which returns a GEO block for products with its own score and per-criterion checklist, and reports the failing items by their visible titles. If product schema is off, the agent can propose turning it on with storeseo_update_schema_settings, which changes live storefront markup, so it should ask first. Then it previews FAQs and the AI snippet through storeseo_preview_resource_optimization, spending credit and saving nothing. The distinctive part is the description scan. storeseo_scan_product_description scores seven pillars, from identity and materials to care, and returns gap questions that the agent must put to you word for word and in order. In a terminal that is a short interview inside the kimi prompt: Kimi Code prints each question, you type the facts about the fabric, dimensions or what is in the box, and the answers go to storeseo_write_product_description. Answered questions turn guesses into stated fact, so the copy is better than a run with the questions skipped. Writing the description replaces the current one at once with no undo, and the response returns the previous text as the only remaining copy. Because Kimi Code works in your repository, ask it to save that previous description into a file before you move to the next product. The product also needs a focus keyword set in the StoreSEO app, and no tool here can set one, so the agent should hand you that fix first. Prompts: - Use StoreSEO to read the GEO checklist for my Linen Shirt product and tell me which items need me, which need schema and which need a description. - Scan the description on that product, ask me the gap questions exactly as StoreSEO words them, then write the description from my answers. - Before you write, save the current description to descriptions-backup.md in this repository. In Kimi Code: storeseo_write_product_description saves immediately and cannot be undone, so a Kimi Code session set to /auto or covered by a permanent allow rule has no panel left between your answers and a live replacement. Keep that one tool on ask even if everything else is relaxed. The scan's questions are meant to reach you unedited, so if the agent starts answering them itself from the product page, stop it and have it ask you. Q: Who answers the description scan questions in Kimi Code? A: You do. StoreSEO tells the agent to put each question to you word for word and in order, then pass your answers to the writing tool. The agent only skips them if you decline or authorize a bulk run without questions. Q: Can Kimi Code undo a description it wrote? A: StoreSEO keeps no backup and no revision history of the old description. The write response returns the previous description once, so ask Kimi Code to save it to a file in your project before moving on. --- ## Publish llms.txt and agents.md with Kimi Code URL: https://storeseo.com/agent/kimi/llms-txt-agents-md (full page: https://storeseo.com/agent/kimi/llms-txt-agents-md/index.md). Job: llms.txt and agents.md, described once above. Publishing llms.txt and agents.md from Kimi Code means asking Moonshot AI's terminal agent to check what your store serves now, generate the discovery files with StoreSEO after you agree they go live, and manage versions if one needs rolling back. You begin with a question rather than a command: is my llms.txt current, and does my store serve an agents.md. Kimi Code calls storeseo_get_llms_txt_status and storeseo_get_agents_md_status, both reads, and reports the public URLs, the active version and whether your settings changed since the last generation. That gives you a picture of the live state before anyone writes to the theme. Generation is where the warnings matter. storeseo_generate_llms_txt writes templates/llms.txt.liquid into your live theme and uploads a copy to Shopify Files, with no separate publish step, and it takes a minute or two. storeseo_generate_agents_md spends AI credit, needs at least one published product and publishes live straight away. Kimi Code should say both things before it calls either tool, and you reply in the prompt. After the call it polls storeseo_get_llms_txt_status until generation is no longer ongoing, so you see a finished result rather than a queued one. A terminal agent adds one useful check: because llms.txt lands in the live theme, a local copy of the theme in your repository will not contain that template until you pull it. Ask Kimi Code to remind you before you push an older theme over it. Version tools cover the rest, and they matter because only the latest ten versions are retained. storeseo_list_llms_txt_versions shows the retained versions, and activating an earlier one is a single call if the newest generation reads badly. Deleting a version cannot be undone, and deleting the active one promotes the next newest, so the agent should say which version will go live before it deletes anything. Prompts: - Use StoreSEO to check my llms.txt and agents.md status and tell me what each one currently serves. - Regenerate llms.txt, tell me before you start that it writes to the live theme, and report when generation has finished. - List my llms.txt versions and activate the previous one, telling me which version goes live before you do. In Kimi Code: Generation and deletion tools here publish to a live theme or erase versions, so they should not share a permission rule with the status reads. In Kimi Code you can allow the two status tools permanently in config.toml and leave storeseo_generate_llms_txt, storeseo_generate_agents_md, storeseo_delete_llms_txt_version and storeseo_reset_agents_md on ask. A session granted Approve for this session on a generate tool will also run a second generation without a panel. Q: Does generating llms.txt from Kimi Code change my live theme? A: Yes. The generation run writes templates/llms.txt.liquid into the live theme and uploads a copy to Shopify Files. StoreSEO tells the agent to warn you before it starts, and Kimi Code's own approval setting is the hard stop. Q: How do I undo agents.md from Kimi Code? A: Ask it to reset agents.md. That removes the custom file so Shopify's native default serves again, but the custom content is discarded with no history, so the agent should confirm first. llms.txt has retained versions you can activate instead. --- ## Set up structured data with Kimi Code URL: https://storeseo.com/agent/kimi/structured-data (full page: https://storeseo.com/agent/kimi/structured-data/index.md). Job: Structured data (JSON-LD), described once above. Setting up structured data in Kimi Code means asking Moonshot AI's terminal agent to list the JSON-LD schema types StoreSEO manages, read each type's settings before editing, and switch on the markup you want for your Shopify storefront. You ask which schema types are on and which are off, and Kimi Code calls storeseo_list_schema_types, which says for every type whether it is enabled and whether it is configurable. Product, review, organization and local business types take settings, while the rest only switch on or off. The agent then calls storeseo_get_schema_settings for the type you care about, and that response includes a field template describing the full settings object. That template is the center of the job. storeseo_update_schema_settings replaces a type's settings wholesale instead of merging, so the agent must read first, edit the whole object and send it back. In a terminal you can keep that object in a scratch file, review it as JSON and ask Kimi Code to diff it against what StoreSEO returned. Business details such as address, phone and opening hours come from you, since the agent does not invent them, and organization and local business share the same basic information, so changing it in one updates both. Some switches have prerequisites. Review and product merchant markup need product schema enabled first, and the four FAQ schema types need the FAQ Schema feature in your plan. If a change is refused, Kimi Code should report the reason and not retry. Because schema changes alter live markup, ask the agent to name the type and the new state before it calls the write, then confirm in your own tab with a structured data test tool if you want an outside check. Prompts: - Use StoreSEO to list my schema types and tell me which are off, which take settings and which need a plan feature. - Read the organization schema settings, show me the field template and ask me for each missing business detail. - Turn on product schema, then tell me whether review and product merchant schema can be switched on after it. In Kimi Code: Because the settings update replaces the stored object, a stale copy is the main risk in a long Kimi Code session. If you saved a schema object to a scratch file an hour ago and someone changed the setting in the StoreSEO app since, sending the old file back overwrites their change. Have the agent call storeseo_get_schema_settings again right before each write, and keep the write tool on ask so you see the exact object before it goes. Q: Why does Kimi Code read schema settings before every update? A: StoreSEO replaces a type's settings with whatever object it receives, with no partial merge. Reading first, editing the whole object and sending it back keeps fields you did not mention from being wiped out. Q: Can Kimi Code turn on FAQ schema for my store? A: Only if your plan includes the FAQ Schema feature. Without it StoreSEO rejects any change to the four FAQ types, on or off, and the agent should tell you which feature is missing instead of retrying. --- ## Manage sitemaps and indexing with Kimi Code URL: https://storeseo.com/agent/kimi/sitemaps (full page: https://storeseo.com/agent/kimi/sitemaps/index.md). Job: Sitemaps and indexing, described once above. Managing sitemaps and indexing in Kimi Code means asking Moonshot AI's terminal agent to list the sitemap, noindex and nofollow flags StoreSEO holds for each resource, change them one at a time, and set up an HTML sitemap for your Shopify store. You ask something like which products are excluded from my sitemap or marked noindex. Kimi Code calls storeseo_list_resource_sitemaps for a resource type, product, collection, page or article, and can filter by sitemap, noindex or nofollow state. Each row carries the flags and a resource id, which the write tool takes. The list is a read, so in Always Ask mode it runs without a panel and you can page through a large catalog quickly. Changes are one resource per call. storeseo_update_resource_sitemap sets sitemap_disabled, no_index or no_follow for a single product, collection, page or article, writing the matching Shopify metafields. A noindex flag removes a page from search results, so the agent should name the resource and wait for your yes, and after the write it should list the flags again in the same turn rather than assume the change took. In a terminal that makes a neat loop: change, re-list, report the current state. The HTML sitemap has its own pair of tools. Kimi Code reads the current state with storeseo_get_html_sitemap_settings, generates it with storeseo_update_html_sitemap, and fetches your navigation menus with storeseo_list_sitemap_menus. Adding the link to a menu is a step you do in Shopify admin, because no tool does it, so the agent hands you the edit URL and the sitemap address and then marks setup complete once you tell it the link is in. If you only want a page out of the XML sitemap and not out of search, say so, because sitemap exclusion and noindex are separate flags and the agent should set only the one you name. Prompts: - Use StoreSEO to list every collection and page marked noindex and tell me which ones look like they should be indexed. - Exclude my old Summer Sale page from the sitemap, then list its flags again to confirm. - Set up the HTML sitemap and give me the menu edit link and sitemap address so I can add it myself. In Kimi Code: Sitemap flags are single-resource writes with search consequences, and a long catalog tempts you toward Approve for this session. In Kimi Code that choice turns a careful one-page change into a loop that can flip many flags without a panel, and a wrong noindex only shows up weeks later in search traffic. Approve each write on its own, or keep a rule that leaves storeseo_update_resource_sitemap on ask while the list tools run freely. Q: Can Kimi Code add the HTML sitemap link to my menu? A: No. Adding the link is a step you do in Shopify admin. Kimi Code can fetch your menus and their edit links with StoreSEO, give you the sitemap address, and mark setup complete after you tell it the link is in place. Q: Why does Kimi Code re-list flags after changing one? A: StoreSEO tells the agent to re-read the flags in the same turn before reporting their state, so what you hear back is what the store now holds, not what the agent expected the write to do. --- ## Track keyword rankings with Kimi Code URL: https://storeseo.com/agent/kimi/keyword-tracking (full page: https://storeseo.com/agent/kimi/keyword-tracking/index.md). Job: Keyword rank tracking, described once above. Tracking keyword rankings in Kimi Code means asking Moonshot AI's terminal agent to add keywords to StoreSEO's Google rank tracker, read their position and history, compare competitor domains and build a cluster around a product's focus keyword. You start with the market. Kimi Code calls storeseo_list_rank_locations and storeseo_list_rank_languages, because StoreSEO needs the exact country and language names from those lookups and the agent should not type them from memory. With those in hand it calls storeseo_add_keywords, one item per keyword with a language, a country and a device. The response only confirms the check was queued; positions arrive later, so the next question you ask, how are my keywords ranking, goes to storeseo_list_keywords. Rank tracking and competitor tools are Pro plan tools, so on another plan the agent should explain the upgrade instead of calling them. On Pro, storeseo_get_keyword_history gives a dated rank history per keyword, and storeseo_add_competitors takes up to five competitor domains so their positions appear beside yours. A terminal is a good place to hold the results: ask Kimi Code to write the current ranks to a CSV in your project, then rerun the question next week and have it compare against the file. Clusters are a separate, credit-spending step. storeseo_generate_keyword_cluster builds related keywords around a product's existing focus keyword, with search volume and difficulty, and runs in the background for a couple of minutes. Kimi Code polls storeseo_get_keyword_cluster until the job is done, and if the product has no focus keyword it stops and hands you that fix, since only the StoreSEO app can set one. Deleting a keyword or competitor erases its history for good, so the agent should confirm each removal by name, and the five competitor limit means a sixth domain needs one deleted first. Prompts: - Use StoreSEO to find the exact country and language names for the United Kingdom in English, then add the keyword linen shirt for mobile. - Show my tracked keywords sorted by position, and save today's ranks to rankings.csv in this repository. - Build a keyword cluster for my Linen Shirt product, tell me the estimated credit first and poll until it finishes. In Kimi Code: Kimi Code can run a polling loop for a keyword cluster, but a loop of repeated reads is also a loop where one mistyped command can delete things. Keep storeseo_delete_keyword and storeseo_delete_competitor on ask in your Kimi Code rules, because each removal erases rank history that no tool can bring back. If your plan lacks rank tracking, the lookup tools will be refused, and the agent should say so before trying anything else. Q: Do I need the Pro plan to track keywords from Kimi Code? A: Rank tracking and competitor tracking are Pro plan tools in StoreSEO. Keyword clusters are a separate credit-spending job that also needs a focus keyword on the product, which you set in the StoreSEO app. Q: Why are my new keywords missing ranks in Kimi Code? A: Adding a keyword only queues the first rank check. Ask for the list again a little later and the position and rank change appear once the check finishes, with history building on later checks. --- ## Read Search Console and Analytics with Kimi Code URL: https://storeseo.com/agent/kimi/google-reports (full page: https://storeseo.com/agent/kimi/google-reports/index.md). Job: Search Console and Analytics reports, described once above. Reading Google reports in Kimi Code means asking Moonshot AI's terminal agent for Search Console and GA4 numbers through StoreSEO, comparing them with the previous period, and turning queries and pages into a prioritized list of fixes. You ask a question the data can answer, such as which queries rank between positions eight and twenty, or which pages lost clicks this month. Kimi Code calls storeseo_get_search_console_insights with a dimension, query, page, country, device or date, and gets clicks, impressions, click-through rate and average position, plus the previous equal-length period and the change in each. For traffic questions it calls storeseo_get_analytics_report with a report type such as search_traffic, page_performance or traffic_source. The default range is the last 28 days. Both tools are reads, so in Kimi Code's default mode the whole analysis runs without an approval panel. What the terminal adds is a place to keep the work. Ask the agent to save the table of near-page-one queries to a file, and to map each query to a product or collection in your theme repository by searching the templates and content. The agent reports numbers as Google returns them and does not project outcomes from them. These reports have prerequisites that come up before the first answer. Google reports need a paid StoreSEO plan with Google connected in the app, and Analytics reports need a GA4 property selected. If either is missing, Kimi Code should give you the link into the StoreSEO app and stop. A good habit is to pair each report with a follow-up: take the low-click, high-impression pages and hand them back to the optimize job as candidates for new meta. Prompts: - Use StoreSEO to show Search Console queries from the last 28 days where we rank between 8 and 20, with the change from the previous period. - Which pages lost the most clicks compared with the previous 28 days? Save them to lost-pages.md in this repository. - Pull the GA4 traffic source report for last month and tell me how organic search compares with the other channels. In Kimi Code: A report can be large, and Kimi Code carries whatever a tool returns into its context for the rest of the session. Ask for a narrow dimension and a modest top row count, and write the result to a file you can reopen, instead of pulling queries, pages and countries all at once. If the tools answer that Google is not connected or no GA4 property is selected, that is a setting in the StoreSEO app, not something to fix with a different command. Q: Does reading Google reports in Kimi Code need an approval? A: The two tools, storeseo_get_search_console_insights and storeseo_get_analytics_report, only read, and Kimi Code runs read-only calls without a prompt in its default mode. Nothing is written to your store or to Google. Q: Why does Kimi Code say Google is not connected? A: Search Console and Analytics reports need a paid StoreSEO plan with Google connected in the app, and GA4 reports also need a property selected. Connect Google or pick the property in StoreSEO, then ask again in a new request. --- ## Write SEO blog post drafts with Kimi Code URL: https://storeseo.com/agent/kimi/blog-posts (full page: https://storeseo.com/agent/kimi/blog-posts/index.md). Job: Blog post drafts, described once above. Drafting blog posts in Kimi Code means asking Moonshot AI's terminal agent to have StoreSEO write an SEO-optimized article into one of your Shopify blogs as a draft, following the background job until it finishes and checking the draft. You describe the post in the kimi prompt: a topic, who it is for and a rough length. Kimi Code calls storeseo_list_blogs to find the numeric blog id, because only blogs synced with StoreSEO appear, and an empty list means you sync in the app first. It then agrees the brief with you, including the word count and whether to add a featured image, since credit use grows with both. StoreSEO writes the article itself on its own servers, so the text is not the Kimi model's. The write is storeseo_generate_blog_post, and it saves a draft article in the blog you chose. Nothing is published live; you review and publish from Shopify admin. The job runs for a few minutes in the background, so Kimi Code polls storeseo_get_blog_post_job and reports steps completed out of steps total until the status is completed, failed or cancelled. The draft row exists from the start, so a good agent does not call the article finished while the job is still running. When the job completes, the agent reads the draft with storeseo_get_resource_seo using resource type article and the Shopify article id from the job result, and reports the score and anything still missing. A terminal helps with planning. Ask Kimi Code to scan your theme or content folder for product handles, then include them in the brief as internal link targets, or to keep a posts-plan.md file of topics so a series stays consistent without repeating titles. Prompts: - Use StoreSEO to list my blogs, then write a 1,200 word post on how to care for linen shirts as a draft in the News blog, after telling me the credit cost. - Check the blog post job and tell me how many steps are done. Do not call the draft finished until the job is. - Read the SEO checklist for the draft that job produced and list what I still need to fix before publishing. In Kimi Code: Polling is where a terminal session can mislead you. Kimi Code may stop after a poll that shows the job still running, and the draft article already exists in Shopify at that point with partial content. Ask the agent to keep checking storeseo_get_blog_post_job until a terminal status, and open the draft only after it reports completed. If the job fails, StoreSEO settles the credits automatically, so do not start a second job before reading the error. Q: Will Kimi Code publish the blog post it writes? A: No. StoreSEO saves the article as a draft in your Shopify blog and nothing goes live from this job. You review the draft and publish it yourself in Shopify admin when you are ready. Q: Why does Kimi Code say I have no blogs to write into? A: Only blogs synced with StoreSEO are listed, so an empty result means you need to sync your blogs in the StoreSEO app. No tool in the MCP server can sync them for you. --- # Documentation --- # How StoreSEO Will Help Optimize SEO For Shopify Products? Source: https://storeseo.com/docs/optimize-seo-for-shopify-products [**StoreSEO**](/) helps [**optimize SEO for Shopify products**](/docs/what-is-search-engine-optimization) so that they have more online visibility among your potential customers. StoreSEO ensures everything is in order by showing an SEO score and all the fields that need to be fixed to bring a good SEO score for your product page. Learn how StoreSEO will help Optimize SEO for Shopify products from the guidelines below. Before you start, make sure you have created or added a product and [**installed StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) to your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Then, follow these step-by-step guidelines to find out how to optimize the product page with StoreSEO. ## Step 1: Navigate To StoreSEO Product Optimization Log in to your Shopify store. Then, from the search bar on the top or the ‘**Apps**’ option in the left sidebar, search for the StoreSEO app and click on it. From the StoreSEO dashboard, go to the ‘**Optimize SEO**’ tab. Now click on the ‘**Products**’ option. Make sure to sync all products to see the product list on this page. Choose the product you want to optimize, and click on the ‘**Fix issue**’ button. 
## Step 2: Optimize Shopify Product Now, it is time to optimize the [**Shopify product page SEO**](/shopify-product-page-seo-tips-and-best-practices). As you can see in the image below, the SEO details of your selected product are shown by StoreSEO. From the right side panel, you will also find **Basic SEO Analysis** and **Detailed SEO Analysis** to find the key factors that have to be addressed to improve the SEO Score. ![](../../assets/content/body/2021/08/image-12.webp) ### Basic SEO Analysis Let us start with **Basic SEO Analysis**. Here, we have to ensure various criteria such as placing a unique focus keyword in the required places, maintaining an optimized length of the meta title and description and others. We will now show you how to do these one by one. ![](../../assets/content/body/2021/08/image-13.webp) #### Focus keyword is unique First, you need to choose the proper **focus keyword**. This is crucial because the product is indexed on search engines based on this keyword. To put it simply, your potential customers will have a higher likelihood of finding your product if their search keyword and your selected focus keyword match. Therefore, you need to be very careful and do [**proper keyword research**](/keyword-research-for-shopify) when choosing the focus keyword. Here are a few things to consider: **Keyword relevance:** Make sure the keyword is relevant to your product, business, audience and niche. Suppose, you are selling apple cider vinegar. Your focus keyword should be relevant to this product. Now, you select a high search volume keyword containing the term ‘**fresh apple**’ or ‘**apple iphone’**. These keywords are outside your product niche. So although it might have a higher search volume, it will not be a relevant option.  **Search volume:** Keywords with higher search volume have more searches every month. If you select keywords with higher search volume, you are more likely to face more competition. The more specific keywords you use, the more precisely you can reach your target audience. Now, if you choose a long focus keyword such as ‘**original unfiltered organic apple cider vinegar**’, you have a unique keyword but there is less possibility for customers to use this entire keyword during a search. This is why it does not have a good search volume as shown in the image below. ![](../../assets/content/body/2021/08/image-14.webp) Now, if you choose a short focus keyword like ‘**apple cider vinegar**’, you will find it has a high search volume as shown in the image below. However, such popular keywords have too much traffic and will be a lot harder to rank and reach out to your audience. ![](../../assets/content/body/2021/08/image-15.webp) On the other hand, if you select the focus keyword ‘**organic apple cider vinegar**’, you have a good search volume (not too high) and there is a higher possibility to index without facing much competition. This is why this is the more appropriate focus keyword option in this particular case. ![](../../assets/content/body/2021/08/image-16.webp) **Cost per click:** If you are running ads with particular keywords, the cost per click determines the approximate budget for the ad campaign. So, choosing keywords with a higher cost per click means you need to have a bigger campaign budget. Please note that it is an estimated value and you do not have to worry about it if you are not running any ads**.**  **Paid difficulty:** This represents the demand for keywords when used in paid ads. Higher paid difficulty indicates the keyword also has more competition in organic search. So, it is an important metric during keyword research. **Related keywords:** Remember to check the related keywords for a particular keyword. These related keywords will give you additional keyword suggestions. So, you might come across a better focus keyword option. Make sure the related keywords suit your product, business, audience and niche. Back to product optimization on StoreSEO, place your preferred keyword in the ‘**Focus Keyword**’ field. Click on the ‘**Keyword Analytics**’ button to instantly get the **Search Volume**, **Cost Per Click**, **Paid Difficulty** and **related keywords** for your keyword. Here, we have selected ‘**organic apple cider vinegar**’ as the focus keyword. It is recommended that different focus keywords be used for different products. However, if your products are all similar, the same focus keyword can be used for them. Please choose a keyword according to your business. Do [**keyword research for Shopify**](/keyword-research-for-shopify) for this.
#### Focus keyword is used in the meta title Your product’s meta title and meta description are the first things that customers see when exploring search engines. Therefore, both of these are very important and should be written carefully and properly. Now, make sure that the focus keyword is used in the meta title. As you can see in the image below, our meta title is the following:  **Organic Apple Cider Vinegar** 100% Pure Natural Hand Pressed So, it contains our selected focus keyword (marked bold). ![](../../assets/content/body/2021/08/image-17.webp) #### Meta description is within 120-165 characters Your [**product’s meta description**](/shopify-homepage-title-and-meta-descriptions) should be within 120-165 characters. It should also contain the focus keyword that you have selected. Our meta description is the following:  **Discover the health benefits of our premium organic apple cider vinegar. Ideal for detox, weight loss, and overall wellness. Get it now at an amazing price.** As you can see, the meta description is 160 characters long so it meets the requirement of being within 120-165 characters. ![](../../assets/content/body/2021/08/image-19.webp) #### Meta title is within 50-70 characters Keep your meta title within 50-70 characters. As you can see, our meta title is the following:  **Organic Apple Cider Vinegar 100% Pure Natural Hand Pressed** It is 58 characters long so it meets the requirement. ![](../../assets/content/body/2021/08/image-20.webp) #### Product title is unique Make sure the product title is unique. This means do not use the same product title for more than one product in your Shopify store. Remember this when adding a new product. Also, place your selected focus keyword in the product title. When you do not have your focus keyword in the product title, it will show a cross mark in StoreSEO detailed SEO analysis. So, click on the ‘**Edit product**’ button and edit the product title from the ‘**Title**’ field. Make sure the product title contains the focus keyword in a sensible manner. Click on the ‘**Save**’ button once you are done.
As you can see, one of our product titles is ‘**Organic Apple Cider Vinegar With The Mother**’ which cannot be found in the title of any other product. It also contains our selected focus keyword. ![Product title is unique 2](../../assets/content/body/2021/08/11-storeseo-product-optimization.webp) ### Detailed SEO Analysis We will now proceed to **Detailed SEO Analysis**. For this, we have to make sure that the focus keyword is placed in the image alt text, product title, meta description and URL and meet a few other requirements. Let us show you how to do all these. ![](../../assets/content/body/2021/08/image-21.webp) #### Focus keyword is used in image alt text Scroll down and find the ‘**Alt text’** fields. Here, make sure that the focus keyword is used in the [**image alt text**](/docs/how-to-add-an-image-alt-text-for-shopify-products-with-storeseo). As you can see in the image below, all our Alt text fields contain **Organic Apple Cider Vinegar** (our selected focus keyword). ![](../../assets/content/body/2021/08/image-22.webp) #### Focus keyword is used in the product title When you do not have your focus keyword in the product title, it will show a cross mark in StoreSEO detailed SEO analysis. So, click on the ‘**Edit product**’ button and edit the product title from the ‘**Title**’ field. Make sure the product title contains the focus keyword in a sensible manner. Click on the ‘**Save**’ button once you are done.
As shown below, our product title contains ‘**Organic Apple Cider Vinegar’** which is our selected focus keyword. ![Focus keyword is used in the product title 2](../../assets/content/body/2021/08/11-storeseo-product-optimization-1.webp) #### Focus keyword found in meta description As we have shown previously, you have to place the focus keyword in the meta description. So, you can check it again to make sure that you have done it. ![](../../assets/content/body/2021/08/image-23.webp) #### Focus keyword is at the beginning of meta title Now, make sure that the focus keyword lies at the beginning of the meta title. As you can see in the GIF below, when we have done this the SEO score has noticeably improved. This is our meta title: Organic Apple Cider Vinegar 100% Pure Natural Hand Pressed As you can see, the focus keyword ‘**Organic Apple Cider Vinegar**’ has been placed at the beginning. ![](../../assets/content/body/2021/08/image-24.webp) #### Focus keyword density is 3-6 times for product description For this, from the Shopify dashboard, click on the ‘**Edit** **Products**’ tab in the detailed analysis tab. Now, make sure that the focus keyword density is 3-6 times in the product description without any keyword stuffing. We have used our focus keyword 6 times in the example below:
#### Focus keyword is used in the URL Finally, make sure that the product URL contains the focus keyword. As you can see, the URL field contains ‘**Organic Apple Cider Vinegar**’ or our selected focus keyword.  **Note:** If your product has already been indexed and you want to change its URL, it is recommended to create a URL redirection. We are assuming that you are optimizing a new or your first store. In that case, it is best not to create any unnecessary redirections. ![](../../assets/content/body/2021/08/image-25.webp) #### Write a product description of 50 to 300 words Make sure that the product description shown previously contains 50 to 300 words. As you can see, we have written our product description accordingly to ensure the best SEO results. Read [**this blog**](/guide-to-write-product-descriptions) to find out how to write unique product descriptions in Shopify.
#### Add alt text to all images If you have multiple images for your product, ensure that the Focus Keyword is used in the alt text of all images. As you can see in the image below, we have used our selected Focus Keyword in the alt text of all our product images. We have used the following image alt text: **Organic Apple Cider Vinegar Machine Pressed 100 ml**, **Organic Apple Cider Vinegar Hand Pressed 100 ml** and **Organic Apple Cider Vinegar Diluted 100 ml**. All contain ‘**Organic Apple Cider Vinegar**’ or our selected Focus Keyword. ![](../../assets/content/body/2021/08/image-26.webp) #### All Images are optimized Remember to optimize all the images used for the product. With StoreSEO, you can [**optimize images on Shopify**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) with ease. When optimizing a product with StoreSEO, click on the ‘**Optimize**’ button beside each image to optimize it individually. You can also click on the ‘**Optimize All**’ button to optimize all of the images at once. ![](../../assets/content/body/2021/08/image-27.webp) To optimize all your store’s images in one place, select the ‘**Image Optimizer**’ tab under StoreSEO from the left sidebar. Here, you can optimize images, restore optimized images to the original and compare the restored and original versions under **Action** as shown below.
For additional settings, click on the ‘**Settings**’ tab in the left sidebar. Then, choose the ‘**Image Optimizer**’ option. Turn on ‘**Auto Image Optimization**’ option to automatically have your images optimized upon upload. You can also choose your desired image compression and image format settings along with the output dimension for the resized image. Click on the ‘**Save**’ button when you are done.
By following these steps, you can address every important SEO factor for the best possible optimization score for your product page. If you need help to optimize your content for SEO, you can also use the [**StoreSEO AI Content Optimizer**](/docs/how-to-use-storeseo-ai-content-optimizer).  Once you are done with your product optimization and happy with the overall SEO Score, click on the ‘**Save**’ button. ![](../../assets/content/body/2021/08/image-28.webp) This is how easily you can optimize Shopify **Product Pages** with StoreSEO. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Add An Image Alt Text For Shopify Products With StoreSEO? Source: https://storeseo.com/docs/how-to-add-an-image-alt-text-for-shopify-products-with-storeseo With the **StoreSEO app**, you can **[add an Image alt text](https://www.google.com/url?q=/docs/generate-and-add-image-alt-text-with-ai)** that will describe the product picture to visitors who may not be able to see the product, in case of any image loading error on your store or any accessibility needs. After conducting your keyword research, you can incorporate the relevant keywords into the alt text and rank your product image in Google searches. ## **How To Add An Image Alt Text For Shopify Products** Before you start, make sure you have created a product list inside your [**Shopify product store**](https://shopify.pxf.io/storeseo) and **installed the StoreSEO app** successfully. Then, follow the step-by-step guidelines below to find out how to add an image alt text for your Shopify products. ### **Step 1: Select StoreSEO From The Installed Apps** First, you need to [**install and activate the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify shop. Then go to the **‘Apps’** option from your dashboard and select the ‘**StoreSEO’** app.  ### **Step 2: Go To Shopify Product List**  Now, click on the ‘**Optimize SEO’** tab from the side panel and choose **‘Products’**, where all your Shopify products are listed with details as shown below. For the product you want to add an **Image Alt text**, click on the **‘Fix issues’** button from the product details under ‘**Actions’**. ![How To Add An Image Alt Text For Shopify Products With StoreSEO?](../../assets/content/body/2021/08/image-6.webp) ### **Step 3: Add Image Alt Text For Shopify Products** Scroll down to the **‘Image Alt Text’** section to add and edit the alt text of each of your Shopify product images.  Here, you can add descriptive text or the focus keyword to describe your Shopify product. Then, click on the ‘**Save**’ button that appears at the top bar to save the progress. 
How To Add An Image Alt Text For Shopify Products With StoreSEO?
You can see the changes and improvements in the SEO score immediately. This is how easily you can **add image alt text** for Shopify products using the StoreSEO app. ![How To Add An Image Alt Text For Shopify Products With StoreSEO?](../../assets/content/body/2021/08/image-7.webp) Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # How To Add Image Alt Tag Or Text To Blog Posts And Pages On Shopify? Source: https://storeseo.com/docs/add-image-alt-tag-or-text-to-blog-posts-and-page With StoreSEO, you can easily [**add alt text to your product images**](/docs/how-to-add-an-image-alt-text-for-shopify-products-with-storeseo). You can also add alt text to images in the blog posts and pages of your Shopify store website. In this documentation, we will show you how to easily add image alt tag or text to blog posts and pages on Shopify. **Note:** The image alt text can help describe the picture of the product to the visitors who are not able to see the product for any reason, such as an image loading error. After you have done your keyword research, you can add suitable keywords in the alt text to increase the chances of ranking your product image in Google search. ## **How To Add Image Alt Tag Or Text To Blog Posts And Pages?** First, make sure you have successfully [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify). Then, follow these step-by-step guidelines to find out how to add alt text for the images of your Shopify blog posts and pages using the StoreSEO app.  ### **Step 1: Add Alt Text To Your Blog’s Featured Image With StoreSEO** You can directly add the alt text to any blog post’s featured image using StoreSEO. This is very useful if you missed adding the alt text to any featured image when adding the blog post. To do this, open the StoreSEO app and navigate to the ‘**Optimize SEO**’ option under StoreSEO in the left side panel. Under the ‘**Blog Posts**’ tab, click on your desired blog. 
Now, under the ‘**Images**’ section, add the alt text for your featured image. Click on the ‘**Save**’ button in the top right once you are done.
### **Step 2: Add Image Alt Text When Adding Your Blog Posts & Pages** Adding the image alt text with your selected focus keyword is an important component of [**Shopify blog post and page SEO**](/docs/configure-seo-for-blog-posts-and-pages). Now, we will show you how to add image alt text when adding your blog posts and pages. To add a new blog, navigate to the ‘**Online Store**’ option under ‘**Sales Channels**’ on the left side of your Shopify dashboard. Under the ‘**Online Store**’ option, click on the ‘**Blog posts**’ button. Next, click on the ‘**Add blog post**’ button in the top right. 
Once you have added the blog post, you will find it in the list in the ‘**Blog posts**’ option. When adding or editing your blog post, remember to place the suitable alt text for all the images. To add the alt text to the featured image, click on the featured image or select the ‘**Edit image**’ option under ‘**Update’** on the right side. Paste your copied alt text or write it in the designated field and click on the ‘**Save**’ button.
To add the alt text to any internal image, double-click on any image inside the ‘**Content**’ section. Then, add the alt text and click on the ‘**Edit image**’ button. Click on the ‘**Save**’ button to save all the changes.
To add a new page, navigate to the ‘**Online Store**’ option under ‘**Sales Channels**’ on the left side of your Shopify dashboard. Under the ‘**Online Store**’ option, click on the ‘**Pages**’ button. Next, click on the ‘**Add page**’ button in the top right. 
Once you have added the page, you will find it in the list in the ‘**Pages**’ option. When adding or editing your page, remember to place the suitable alt text for all the images. To add the alt text to any image, double-click on any image inside the ‘**Content**’ section. Then, add the alt text and click on the ‘**Edit image**’ button. Click on the ‘**Save**’ button to save all the changes.
This is how easily you can **add image alt tags or text** to blog posts and pages on Shopify. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) with any type of query. --- # How To Add Image Optimizer To Your Subscription Plan Source: https://storeseo.com/docs/how-to-add-image-optimizer-to-your-subscription-plan StoreSEO has an **[Image Optimizer](/docs/how-to-use-storeseo-image-optimizer-on-shopify)** add-on feature that can help you resize and reduce your image sizes. Follow these steps to add Image Optimizer to your subscription plan. ## **How To Add Image Optimizer To Subscription Plan** First of all, [**install StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) in your Shopify store to add Image Optimizer to your plan before following the guidelines below.  ### **Step 1: Navigate To The ‘Subscription Plans’ Tab** Click on the ‘**Subscription Plans**’ tab where you will see all StoreSEO subscription plans. Now, select any pricing plan, and then click on the ‘**Choose**’ button. 
### **Step 2: Choose Your Image Optimizer **Subscription** Plan** Now, choose any Image Optimizer plan you want to subscribe to and then hit the ‘**Checkout**’ button. If you choose the ‘**Free**’ package this button will redirect you to the dashboard. 
If you choose any premium plan, you will be redirected to the checkout page. Add your preferred payment method and click on the ‘**Approve**’ button. Instantly your subscription plan will be upgraded.  ![How To Add Image Optimizer To Subscription Plan](../../assets/content/body/2024/01/screely-1705051533004.webp) This is how easily you can add Image Optimizer to your subscription plan.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. ### Image Optimizer Related Docs: - **[What Happens If I Reach The Monthly Usage Limit For Image Optimizer?](/docs/monthly-usage-limit-for-image-optimizer)** - [**How To Use StoreSEO Image Optimizer On Shopify**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) --- # How To Add Keywords And Tags For Shopify Products With StoreSEO Source: https://storeseo.com/docs/how-to-add-keywords-and-tags-shopify [**StoreSEO**](/) lets you add keywords and tags to individual Shopify products, helping them rank better on Search Engines. After you have done your keyword research for your products, you can easily add the targeted keywords and tags to each of your products using this feature.  ## How To Add Keywords And Tags For Shopify Products Before you start, make sure you have created a product list inside your Shopify store and installed StoreSEO from the [**Shopify Apps Store**](https://apps.shopify.com/storeseo) page in your store.  Then, follow these step-by-step guidelines to find out how to add keywords and tags for your Shopify products.  ### Step 1: Go To The StoreSEO Product List Page Click on the ‘**Optimize SEO**’ tab from the side panel of your dashboard and head over to ‘**Products**’ to find all your Shopify products listed with details. ![How To Add Keywords And Tags For Shopify Products With StoreSEO](../../assets/content/body/2021/08/image-8.webp) For the product you want to add keywords and tags to, click on the fix or edit icon from the product details under ‘**Actions’**.   ### Step 2: Add Focus Keyword And Tags For Shopify Products Scroll down to add your targeted keyword in the ‘**Focus Keyword**’ input field, as shown below. You will find some keyword suggestions right below it. ![How To Add Keywords And Tags For Shopify Products With StoreSEO](../../assets/content/body/2021/08/image-9.webp) Afterwards, scroll further down and type your tags that you want to add in the ‘Tags’ box. You can add multiple tags to each of your products. Then, click on the ‘**Save**’ button to save your changes.  ![How To Add Keywords And Tags For Shopify Products With StoreSEO](../../assets/content/body/2021/08/image-10.webp) This is how easily you can add keywords and tags to your Shopify products using the StoreSEO app. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # How To Add Meta Description To Products With StoreSEO? Source: https://storeseo.com/docs/add-meta-description-to-shopify-products [**StoreSEO app**](/) lets you add meta descriptions to Shopify products to improve your products’ visibility on Search Engines so that your products can have a high click-through rate in Google searches. Once you have done your keyword research for your product, you can add the keyword in the meta description using our AI SEO Agent, which will raise its SEO score instantly.  ## How To Add Meta Description To Shopify Products? Before you start, make sure you have created a product list inside your **Shopify store** and [**installed StoreSEO in your store**](/docs/how-to-install-storeseo-app-on-shopify). Then, follow these step-by-step guidelines to find out how to add meta descriptions to your Shopify products.   ### Step 1: Go To StoreSEO Product List  Go to the **‘Apps’** option from your dashboard and select the ‘**StoreSEO’** app. Now, click on the ‘**Optimize SEO’** tab from the side panel and choose **‘Products’**, where all your Shopify products are listed with details.
To add meta description to the product of your choice, click on the **‘Fix issues’** button from the product details under ‘**Actions’** next to that product.
### Step 2: Add Meta Description To Shopify Products You can now write your meta description for your Shopify product that describes the product well and increases your click-through rate. In the **‘Meta Description’** box, write a short description for your product in 120 -165 characters (150+ characters for best practice) and click on the **‘Save’** button to save your progress or change.  Afterwards, you will see the change in your SEO score immediately. ![How To Add Meta Description To Products With StoreSEO?](../../assets/content/body/2021/11/image-16.webp) This is how easily you can add meta descriptions with StoreSEO for your [**Shopify products**](https://shopify.pxf.io/storeseo).  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # How to Backup and Restore SEO Data of Your Shopify Store with StoreSEO? Source: https://storeseo.com/docs/backup-and-restore-seo-data-with-storeseo [**StoreSEO**](/) lets you backup and restore the SEO data of your Shopify store with a few clicks. With this feature, you can back up your products’ important SEO data and restore it later when needed.  This functionality can be very useful for recovering the SEO data of your store’s products whenever you want, if it is corrupted or lost due to any issues. In this documentation, we will show you how this functionality works.  # How to Backup and Restore SEO Data Using StoreSEO in Shopify? StoreSEO lets you backup your products’ crucial SEO data like Meta Titles, Meta Descriptions, Focus Keywords, URL handles, Tags & Image ALT Text. Follow these steps to learn how it works: ### Step 1: Navigate to the ‘Backup & Restore’ Option Open the StoreSEO app and go to the **‘Settings’** tab in the left side panel on your dashboard. Then, navigate to the **‘Backup & Restore’** option in the StoreSEO settings tab. Click on the **‘Create backup’** button.
How to Backup and Restore SEO Data of Your Shopify Store with StoreSEO?
### Step 2: Create a Backup of Your SEO Data A **‘Create backup’** module will appear. Select where you want to backup from ‘**Collections**’, ‘**Products**’ and ‘**Blog Posts**’ and hit the ‘**Start backup**’ button.  StoreSEO will now start the process of creating the backup. Once it is over, you will see a success message. You will find the backups that are in progress or newly created in the list within the **‘Data backups’** section on your dashboard. 
How to Backup and Restore SEO Data of Your Shopify Store with StoreSEO?
### Step 3: Restore Your SEO Data Hit the **‘Restore’** button beside the backup in the list that you want to restore. In the following pop-up window, click on the **‘Restore’** button again.  StoreSEO will then start restoring the SEO data according to the specific backup. A notification will pop up to show that the backup process is in action. You will get a success message once the data is restored.
How to Backup and Restore SEO Data of Your Shopify Store with StoreSEO?
**Note:** To keep your backup list organized, you can remove any backup in the list by hitting the **‘Delete’** button beside the **‘Restore’** button. This is how easily you can backup and restore SEO data of your Shopify store’s products using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support). --- # How to Cancel Your Subscription Plan on StoreSEO? Source: https://storeseo.com/docs/cancel-your-subscription-plan-on-storeseo **[StoreSEO](/)** allows you to pick from several subscription plans so that you can optimize your store’s products according to your requirements and get access to more advanced SEO features.  However, you can cancel your subscription plan on StoreSEO in a few simple steps if needed. We will cover these steps in this quick documentation. If you have been using a paid StoreSEO plan and would like to stop your subscription, the cancellation is handled by downgrading your current plan to the **Free plan.** This guide will walk you through the process of downgrading to the Free plan, which is the recommended approach if you still want access to StoreSEO’s basic SEO features at no cost. _**Note:** Cancellation takes effect at the end of your current billing cycle. Your store data and previously optimized products will be adjusted according to the limits of the Free plan once the downgrade is processed._ ## **Step 1: Open StoreSEO & Go to Plans & Add-ons** Log in to your Shopify Admin and open the StoreSEO app from the left sidebar under **Apps**. Once inside the StoreSEO dashboard, look for the **Plans & Add-ons** option in the left navigation menu and click on it. The Plans & Add-ons page shows all available subscription tiers including Lite, Essential, Growth, Advanced, and Free. Your currently active plan will be highlighted with a **Current Plan** badge so you can identify it at a glance. ![The store is currently on the Lite plan at $14.99/month, which is highlighted with a red border for reference. The Plans & Add-ons menu item in the left sidebar is also highlighted to show exactly where to navigate.](../../assets/content/body/2024/08/image-12.webp) ## **Step 2: Scroll Down And Locate the Free Plan** On the Plans & Add-ons page, scroll down past the paid plan cards. You will find the **Free plan** section listed separately below the paid tiers. The Free plan is labeled as a “Limited time” offer and is completely free at **$0.00**. The Free plan includes: - Up to **25 products** (onetime) - Optimize **25 images** (onetime) - **200 AI content credits** (onetime) - **5 Keywords Rank Tracking** Click the **Subscribe now** button next to the Free plan. An arrow indicator in the screenshot below points to this button to help you locate it quickly. ![The screenshot above shows the Free plan highlighted with a red border, confirming this is the plan you need to select to downgrade. The red arrow points to the Subscribe now button you need to click to initiate the downgrade process.](../../assets/content/body/2024/08/image-9.webp) ## **Step 3: Review the Checkout Details And Confirm the Downgrade** After clicking Subscribe now on the Free plan, a **Checkout: Free** popup window will appear on your screen. This window displays the full pricing details for the plan you are switching to, so you can review everything before confirming. The checkout summary will show: - **Plan Name:** Free - **Plan Price:** $0.00 - **Sub Total:** $0.00 - **Total:** $0.00 The checkout popup also displays a **What you’ll get** section that outlines the features included with the Free plan, such as optimizing 25 products, 25 images, 200 AI credits and 5 Keywords Rank Tracking. _**Note:** Pay close attention to the warning message at the bottom of the popup: ‘You’re about to downgrade your current subscription. Your store data will be deleted according to the new plan.’ This means products and optimizations that exceed the Free plan’s limits may be removed. Make sure to back up any important SEO data before confirming._ Once you have reviewed all the details, click the **Subscribe now** button at the bottom right of the popup to complete the downgrade. The red arrow in the screenshot below points directly to this button. ![the full Checkout: Free popup with the pricing breakdown, the feature summary and the important data warning highlighted in a yellow notice box. The red arrow points to the Subscribe now button you need to click to finalize your cancellation.](../../assets/content/body/2024/08/image-10.webp) ## **Step 4: Verify Your Plan Has Been Downgraded** After completing the checkout, you will be redirected back to the Plans & Add-ons page. Your plan should now show **Free** with a **Current Plan** badge next to it, confirming that the downgrade was successful and your paid subscription has been cancelled. ![a successful downgrade. The Free plan now shows the green Current Plan badge and the page displays $0.00 as your active plan price. Your paid subscription has been effectively cancelled and you will not be billed further for the previous pl](../../assets/content/body/2024/08/image-11.webp) ## **Need More Help?** If you run into any issues while cancelling or downgrading your StoreSEO subscription, our **[support team](/support)** is ready to help. Visit **[our documentation](/docs)** for more guides and troubleshooting articles. ## **Frequently Asked Questions** ### **Will I be charged after downgrading to the Free plan?** No. Once you switch to the Free plan, your billing will stop at the end of your current billing period. You will not be charged for the next cycle. ### **What happens to my SEO data when I cancel?** Some of your optimized product data may be removed if it exceeds the Free plan’s limits (25 products and 25 images). It is recommended to export or take note of any critical SEO configurations before downgrading. ### **Can I re-subscribe to a paid plan after cancelling?** Yes. You can re-subscribe to any paid plan at any time from the Plans & Add-ons page inside StoreSEO. Your account will be updated immediately upon subscribing. ### **What is the difference between downgrading and uninstalling?** Downgrading to the Free plan cancels your paid subscription but keeps StoreSEO installed and accessible with limited features. Uninstalling the app removes StoreSEO completely from your store, including all configurations and optimizations. --- # How To Change Language In StoreSEO Source: https://storeseo.com/docs/how-to-change-language [**StoreSEO**](https://apps.shopify.com/storeseo) is now available in the following languages: **_English, हिन्दी, Français, Español, Dutch, Italiano, Deutsch, 中国人._** Follow this guide to change the language in StoreSEO.  ## How To Change Language In StoreSEO On Your Shopify Store To change the Language in StoreSEO to your native or preferred language for a better user experience, follow the step-by-step guidelines below. ### Step 1: Install The StoreSEO App in Your Shopify Store First, go to the StoreSEO app from your Shopify dashboard. To do this, type in and look for the app in the search bar at the top. Then, click on the [**‘Install’**](/docs/how-to-install-storeseo-app-on-shopify) button. Alternatively, go to the **‘Apps’** option in the left side panel and select the **‘StoreSEO’** app.  ### Step 2: Change The Language on Your Dashboard Now, scroll down to the bottom of your dashboard screen to the **‘Language Selector’** as shown below. Choose your preferred language from the dropdown, and it will instantly change the language on StoreSEO. 
How To Change Language In StoreSEO
Once you have switched to your preferred language, the change will be reflected on all tabs, pages, and screens on the StoreSEO app.
How To Change Language In StoreSEO
This is how easily you can change the language in StoreSEO. Need assistance? Feel free to reach out to our [dedicated support team](/contact) for any type of query. --- # How To Choose Products To Include In Sitemap For Your Shopify Store? Source: https://storeseo.com/docs/how-to-choose-products-to-include-in-sitemap StoreSEO app lets you choose which products to include in the sitemap, helping search engine crawlers to find your store and content better. This helps improve your products’ visibility on the search results page and rank higher, so potential customers can find your [Shopify store](https://shopify.pxf.io/storeseo) quickly. ## How To Choose Which Products To Include In Your Sitemap? With StoreSEO, you can choose which products to include in the sitemap and which to exclude. This gives you more flexibility over which products you want Search Engines to discover, identify and crawl from your store. To do this, [**install the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify store and then click on the ‘**Sitemap**’ tab as shown below. Here,  you can see all your products listed on your store. To add or remove products from your sitemap, simply check or uncheck the box next to each product under the **‘Sitemap’** column. A module will appear on your dashboard screen. Click on the ‘**Enable**’ button to confirm that you want to add this product to your Shopify store’s sitemap. 
This is how easily you can add products to the sitemap using StoreSEO.  Need assistance? Feel free to reach out to our [dedicated support team](/contact) for any type of queries. --- # How To Collect GA4 Google Analytics Property ID Source: https://storeseo.com/docs/how-to-collect-ga4-google-analytics-property-id You can integrate Google Analytics with StoreSEO to easily track your store traffic and activities.  To integrate StoreSEO with Google Analytics, you need to collect the GA4 Google Analytics Property ID first. Follow this documentation to learn how to collect the Property ID following the simple step-by-step. **_Note:_** Before following this guide, please complete the first 6 steps from [**this documentation**](/docs/how-to-integrate-storeseo-with-google-analytics) to smoothen the whole process.  ## **How To Collect GA4 Google Analytics Property ID**  First of all, you need to have a Google Analytics account to collect your property ID. Follow this documentation to open your GA4 Google Analytics and retrieve the Property ID. ### **Step 1: Set Up Your Google Analytics Account**  First, go to [**Google Analytics**](https://analytics.google.com/) and log in with your Gmail account. After that, Click on the ‘**Start measuring**’ button. ![Google Analytics Property ID](../../assets/content/body/2023/09/image-8.webp) Now, provide your account name there and check the ‘**Google products & services**’ option before clicking on the ‘**Next**’ button. 
From the ‘**Property Creation**’ section provide the ‘**Property name**’ and then click on the ‘**Next**’ button. ![Google Analytics Property ID](../../assets/content/body/2023/09/image-9.webp) From the next step, provide your business details before hitting the ‘**Next**’ button as shown below. ![Google Analytics Property ID](../../assets/content/body/2023/09/image-10.webp) Now, you need to provide your objectives for using Google Analytics. Check all the options that are relevant to your business and click on the ‘**Create**’ button. After that, from the popup window, checkmark all the terms and conditions before clicking on the ‘**I Agree**’ button. 
After that, click on the ‘**Web**’ button from the next tab and provide your website URL and a Stream Name before hitting the ‘**Create stream**’ button.
Now, navigate to the ‘**Google tag**’ and click the ‘**View tag instruction**’ button to connect your website with Google Analytics.  ![Google Analytics Property ID](../../assets/content/body/2023/09/image-11.webp) From the ‘**Install manually**’ option copy the Google tag that you have to paste in the code of every page of your website, right after the element. ![Google Analytics Property ID](../../assets/content/body/2023/09/image-12.webp) ### **Step 2: Paste The Google Tag In Theme Liquid** Now, log in to your Shopify account dashboard. From the ‘**Sale Channel**’ section, go to the ‘**Online Store’** → ‘**Themes**’ and then click on the 3 dot button at the top right corner of the screen. Now, click on the ‘**Edit Code**’ which will open your Shopify site code editor. 
Now, paste the code that you have copied previously, right after the element. Then click on the ‘**Save**’ button.  ![Google Analytics Property ID](../../assets/content/body/2023/09/image-13.webp) ### **Step 3: Add Your Search Console Links** Now, you have to go back to the [**Google Analytics**](https://analytics.google.com/) dashboard. Click on the Gear icon to open the admin section. After that, navigate to the ’**Search Console Links**’ option. 
Now, click on the ‘**Link**’ button and then hit the ‘**Choose account**’ button. This will open a new window for you.
Now, check the mark before the Property Name to link with Google Analytics. Afterward, simply, click on the ‘**Confirm**’ button.  ![Google Analytics Property ID](../../assets/content/body/2023/09/image-14.webp) ### **Step 4: Collect GA4 Google Analytics Property ID**  Now, click on the Gear icon to open the admin section. After that, navigate to the ’**Property Settings**’ option. Copy the ‘**PROPERTY ID**’ from there to add to the StoreSEO app. ![Google Analytics Property ID](../../assets/content/body/2023/09/image-15.webp) Alternatively, you can search for the property ID from the search bar and copy it from there without any hassle.  ![Google Analytics Property ID](../../assets/content/body/2023/09/image-16.webp) Now, come back to your StoreSEO dashboard. Navigate to ‘**Settings**‘ → ‘**Google Analytics**‘ tab.  Paste your **GA4 Google Analytics property ID** on the blank field and then click on the ‘**Update**’ button to set up analytics on your store.  ![Google Analytics Property ID](../../assets/content/body/2023/09/image-39.webp) This is how you can collect your GA4 Google Analytics Property ID and add it to the StoreSEO app to track your data from your Shopify dashboard. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Configure & Enable SEO Schema Markups In Shopify Using StoreSEO? Source: https://storeseo.com/docs/local-seo-to-enable-schema-markups-in-shopify By easily configuring [**StoreSEO**](/)‘s new feature, **‘SEO Schema,’** you can enable schema markups to optimize your Shopify store for better search engine rankings and results. Let us explore each schema markup, how they work and how they can bring you more effective results for your store’s growth. Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Then, follow these step-by-step guidelines to find out how to configure and enable schema markups using StoreSEO. ## Breadcrumb Schema As a Shopify store owner, you can significantly enhance your technical SEO and user experience by implementing proper breadcrumb schema. StoreSEO allows you to do this with just one click using its SEO Schema feature.  ### Enable the Breadcrumb Schema From your StoreSEO app, click on the **‘SEO Schema’** option from the left navigation bar. This will display a list of all the possible schemas you can enable for your Shopify store using StoreSEO. Select **‘Breadcrumb Schema’** and click on the **‘Turn on’** button. You won’t see any direct changes on your Shopify store. Once you turn on the schema, it sends a code to Google, and that’s how it technically works in the backend. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Collection Schema The collection schema allows structured data for Shopify store owners to display multiple products or content items in a visually appealing carousel format directly in Google search results.  ### Enable the Collection Schema Open your StoreSEO app and select **‘SEO Schema’** from the left navigation menu. You will find a list of schemas for your Shopify store. Pick **‘Collection Schema’** and click on **‘Turn on’**. Once you enable the schema, it’ll help search engines show your products in search results in the carousel or other engaging format that would directly make the customer take better and direct actions from the search results page. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications.
## Product Schema Product schema markup for Shopify stores provides structured data that helps search engines **understand and display product information** more effectively in search results. This includes details like product name, price, availability, and reviews. ### Enable the Product Schema From your StoreSEO app click on **‘SEO Schema’** in the left-hand navigation. You’ll find a selection of schemas for the Shopify store. Choose **‘Product Schema’** and select **‘Turn on’**. After that, select the review integratoin from, LAI product review, and click on the save button. Product Schema markup enhances search engine understanding, making products eligible for rich snippets, which can increase visibility and attract more clicks. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Product Merchant Schema Adding product merchant markup makes your page eligible for merchant listings on Google Search, displaying detailed product information like price, availability, and shipping details in various formats such as the **shopping knowledge panel and product snippets**. ### Enable the Product Merchant Schema In your StoreSEO app, find **‘SEO Schema’** in the left navigation bar. You’ll see a list of schemas designed for your Shopify store. Select **‘Product Merchant Schema’** and click **‘Turn on’**. Then fill out the information as per your own preferences according to your business including shipping information, and return policy, then click on, the ‘Save’ button. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Local Business Schema Implementing the Local Business schema on a Shopify store enhances search visibility by **providing Google with detailed business information**, such as location, hours, and contact details.  ### Enable the Local Business Schema Open your StoreSEO app and go to **‘SEO Schema’** in the left menu. You’ll come across several schemas for your Shopify store. Select **‘Local Business Schema’** and click **‘Turn on’**. Then, fill out the information according to your own preferences and your business’s basic information, upload your business logo, working hours, and much more, and then click on the **‘Save’** button. This can lead to improved SEO Schema performance, potentially increasing foot traffic and online engagement. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Article Schema The article schema helps Shopify store owners enhance their content’s visibility in Google Search through rich results. It improves click-through rates by providing users with more information directly in search results, such as headlines, images, and publication dates. ### Enable the Article Schema From your StoreSEO app, click on the **‘SEO Schema’** from the left navigation bar.  From the list of schemas to use for your Shopify store, Select **‘Article Schema’** and click on **‘Turn on.’** You can upload images for your article, such as the banner image, so help Google show it in rich snippets along with the headline when users search for it. **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Blog Schema The Blog schema helps Shopify store owners by enhancing their blog visibility in Google Search through rich results.  ### Enable the BlogSchema Launch your StoreSEO app and navigate to **‘SEO Schema’** on the left sidebar. From the list of various schemas. Pick **‘Article Schema’** and click **‘Turn on.’** **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
## Organization Schema Organization schema helps search engines understand and display business details more effectively. This can enhance search visibility by providing rich snippets, such as logos and contact information, directly in search results. Improved visibility can lead to increased click-through rates and potentially higher sales for Shopify store owners. ### Enable the Organization Schema In the StoreSEO app, navigate to **‘SEO Schema’** on the left side. You’ll see a list of schemas you can use for Shopify. Choose **‘Organization Schema’** and press **‘Turn on’.** **_Note:_** For further detailed information, click on **‘Know more’** to see how the schema works. This will provide you with an in-depth understanding of the framework and its applications. 
This is how easily you can use the **SEO Schema feature** to instantly implement the essential schema markups for your Shopify store, increase the rich search visibility and generate more sales with the StoreSEO app. Need assistance? Feel free to reach out to our [dedicated support team](/contact) with any type of query. --- # How To Configure Keyword Analytics With StoreSEO Source: https://storeseo.com/docs/how-to-configure-keyword-analytics **Keyword Analytics** is the newest feature in [**StoreSEO**](https://apps.shopify.com/storeseo) where you can easily research keywords and use them to optimize your products, pages or posts on Shopify. This feature helps you find the right keywords, enabling your products to rank higher in search engines and drive a higher volume of targeted traffic to your site. ## **How To Configure Keyword Analytics With StoreSEO?** ### **Step 1: Select And Install StoreSEO App** First, go to the ‘**Apps**’ option on Shopify and select the StoreSEO app. Then follow the instructions in this guide to [**install the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify). ![](../../assets/content/body/2022/07/storeseo-ai-seo-agent-storeseo-smart-ai-seo-zight-optimizer-ranking.-shopify-app-store-2025-07-16-at-1.18.09-pm.webp) ### **Step 2: Configure Keyword Analytics For Products** Now browse through any of your store products and click the **‘Fix’** button on the right panel of your dashboard.  ![](../../assets/content/body/2022/07/ollivanders---storeseo---shopify-2025-07-16-at-3.01.13-pm.webp) After clicking the button you will be redirected to the ‘**General SEO**’ tab of your product. Under this tab, you will find the ‘**KEYWORD ANALYTICS**’ option. Click this option and you will see all the relevant information based on the keyword research. 
**‘Search Volume’** indicates the number of times a specific keyword is searched for in a given time period. It shows the estimated number of searches for this specific product.  **‘Cost Per Click’** refers to the price you will pay for placing an ad for that keyword. The cost represents the amount you would have to pay if a potential customer clicks on your ad**.** Every click on your product represents attention from a person who is searching for something that you offer. Keep in mind that this is only relevant if you are placing ads for that keyword using Google Adwords. ‘**Paid difficulty**’ means that if you create an ad on Google Adword or a marketing campaign for that keyword, you will have to face the mentioned level of difficulty. You can also check these keyword metrics and choose the right keyword for your product from the keyword suggestions and add them as **Focus Keyword** instantly by simply clicking the ‘**Insert**’ option under the‘ **Related Keywords**’ tab. 
This is how easily you can easily optimize Shopify products, pages and posts with **Keyword Analytics** using the StoreSEO app.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Configure SEO For Blog Posts And Pages? Source: https://storeseo.com/docs/configure-seo-for-blog-posts-and-pages [**StoreSEO**](/) helps configure [**SEO for blog posts and pages**](/docs/what-is-search-engine-optimization) on Shopify so that they have more online visibility among your potential customers. StoreSEO makes sure everything is in order by showing an SEO score and all the issues that need to be fixed in order to bring a good SEO score for your Shopify blog post. This documentation will cover all these. Before you start, make sure you have added a blog post and [**installed StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Then, follow these step-by-step guidelines to find out how to optimize the blog post with StoreSEO. ## **Step 1: Navigate To StoreSEO Blog Post Optimization** Log in to your Shopify store. Then, from the search bar on the top or the ‘**Apps**’ option in the left sidebar, search for the StoreSEO app and click on it. From the StoreSEO dashboard, go to the ‘**Optimize SEO**’ tab. Now click on the ‘**Blog Posts**’ option. Once your blog posts are synced properly, you will see your blog post list on this page. Choose the one you want to optimize, and click on the ‘**Fix issue**’ button. 
## **Step 2: Optimize Your Blog Post On Shopify** Now, it is time to optimize [**SEO for your Shopify blog post**](/blog-seo-for-shopify). As you can see in the image below, the SEO details of your selected blog are shown by StoreSEO. From the right side panel, you will also find **Basic SEO Analysis** and **Detailed SEO Analysis** to find the key factors that have to be addressed to improve the SEO Score. ### **Basic SEO Analysis** Let us start with **Basic SEO Analysis**. Here, we have to ensure various criteria such as placing a unique focus keyword in the required places, maintaining an optimized length of the meta title and description and others. We will now show you how to do these one by one.
RickMortdev · StoreSEO · Shopify 2025-07-16 at 11.58.19 AM
#### **Focus keyword is unique** First, you need to [**choose the proper focus keyword**](/docs/what-is-a-focus-keyword). This is crucial because the blog is indexed on search engines based on this keyword. To put it simply, your potential readers or audience will have a higher likelihood of finding your blog post if their search keyword and your selected focus keyword match. Therefore, you need to be very careful and do [**proper keyword research**](/keyword-research-for-shopify) when choosing the focus keyword. Here are a few things to consider: **Keyword relevance:** Make sure the keyword is relevant to your blog topic, business, audience and niche. Suppose, you are selling leather formal shoes. Your focus keyword should be relevant to this product. Now, you select a high search volume keyword containing the term ‘**leather bag**’ or ‘**formal shirt’**. These keywords are outside your product niche. So although it might have a higher search volume, it will not be a relevant option.  **Search volume:** Keywords with higher search volume have more searches every month. If you select keywords with higher search volume, you are more likely to face more competition. The more specific keywords you use, the more precisely you can reach your target audience. Now, if you choose a long focus keyword such as ‘**Water Resistant Synthetic Leather Formal Shoes**’, you have a unique keyword but there is less possibility for customers to use this entire keyword during a search. This is why it does not have a good search volume as shown in the image below. ![SEO For Blog Posts](../../assets/content/body/2022/04/rickmortdev---storeseo---shopify-2025-07-16-at-11.59.58-am-1.webp) Now, if you choose a short focus keyword like ‘**Leather Shoes**’, you will find it has a high search volume as shown in the image below. However, such popular keywords have too much traffic and will be a lot harder to rank and reach out to your audience.
RickMortdev · StoreSEO · Shopify 2025-07-16 at 12.01.26 PM
On the other hand, if you select the focus keyword ‘**Synthetic Leather Shoes**’, you have a good search volume (not too high) and there is a higher possibility to index without facing much competition. This is why this is the more appropriate focus keyword option in this particular case. ![](../../assets/content/body/2022/04/rickmortdev---storeseo---shopify-2025-07-16-at-12.02.24-pm.webp) **Cost per click:** If you are running ads with particular keywords, the cost per click determines the approximate budget for the ad campaign. So, choosing keywords with a higher cost per click means you need to have a bigger campaign budget. Please note that it is an estimated value and you do not have to worry about it if you are not running any ads**.**  **Paid difficulty:** This represents the demand for keywords when used in paid ads. Higher paid difficulty indicates the keyword also has more competition in organic search. So, it is an important metric during keyword research. **Related keywords:** Remember to check the related keywords for a particular keyword. These related keywords will give you additional keyword suggestions. So, you might come across a better focus keyword option. Make sure the related keywords suit your product, business, audience and niche. Back to blog post optimization on StoreSEO, place your preferred keyword in the ‘**Focus Keyword**’ field. Click on the ‘**Keyword Analytics**’ button to instantly get the **Search Volume**, **Cost Per Click**, **Paid Difficulty** and **related keywords** for your keyword. Here, we have selected ‘**Synthetic Leather Shoes**’ as the focus keyword. It is recommended that different focus keywords be used for different blog posts. Please choose a keyword according to your business and the blog topic. Do [**keyword research for Shopify**](/keyword-research-for-shopify) for this.
#### **Focus keyword is used in the introduction** Next, make sure the focus keyword is used in the introduction of the blog post. To edit your blog post, navigate to Sales Channels in the left side panel of your Shopify dashboard. Then, select ‘**Online Store**’ and click on the ‘**Blog posts**’ option. Select the blog you want to edit. Make sure to place the focus keyword in the introduction. As you can see in the image below, our blog’s introduction contains ‘**Synthetic Leather Shoes**’ or our selected focus keyword.
#### **Content should be more than 800 words** Include more than 800 words in the blog content to further optimize the SEO score. Make sure that your blog content contains more than 800 words while maintaining relevance and meaningfulness as shown in the image below.
#### **Focus keyword is used 1-2% times of the blog content** Your focus keyword should be used 1-2% times of the blog content. But avoid keyword stuffing or placing the keyword forcefully or unnaturally. As you can see in the image below, we have placed our selected focus keyword to maintain the ideal frequency mentioned above.
### **Detailed SEO Analysis** We will now proceed to the **Detailed SEO Analysis**. For this, we have to make sure to place the focus keyword in the image alt text, subheadings, meta description and URL and meet a few other requirements. Let us show you how to do all these. ![SEO For Blog Posts](../../assets/content/body/2022/04/rickmortdev---storeseo---shopify-2025-07-16-at-12.05.37-pm.webp) #### **Focus keyword is used in image alt text** Make sure the focus keyword is used when adding the image alt text of the blog’s featured image. For this, navigate to StoreSEO blog post optimization for your desired blog. Under the ‘**Images**’ section, you will find the ‘**Alt text’** field. Place your desired alt text containing the focus keyword in this field.  As you can see in the image below, the Alt text field ‘Top **Synthetic Leather Shoes** for Business Events Blog Featured Image’ contains our selected focus keyword (marked bold).
RickMortdev · StoreSEO · Shopify 2025-07-16 at 12.07.21 PM
#### **Focus keyword found in subheadings** Make sure your blog’s subheadings contain the focus keyword. You can add this to your subheading when editing your blog from the Shopify dashboard as discussed previously. As you can see in the image below, our subheading ‘Classic Oxford **Synthetic Leather Shoes**’ contains our selected focus keyword (marked bold). ![Focus keyword found in subheadings](../../assets/content/body/2022/04/13-storeseo-blog-post-optimization.webp) #### **Focus keyword found in meta description** Your meta description should contain your selected focus keyword. From StoreSEO blog post optimization, you can add the focus keyword in the meta description field. Our meta description is the following:  Discover the top **Synthetic Leather Shoes** for business events that make a lasting impression. Our collection offers a variety of styles to complement your attire. So, it contains our selected focus keyword (marked bold)
#### **Focus keyword used in the URL** Make sure that the Shopify blog post URL contains the focus keyword. You can edit the URL from under the ‘**URL handle**’ field when optimizing your blog post with StoreSEO. As you can see, the URL field contains ‘**Synthetic Leather Shoes**’ or our selected focus keyword. **Note:** If your blog post has already been indexed and you want to change its URL, it is recommended to [**create a URL redirection**](/how-to-create-and-manage-shopify-redirections). ![SEO For Blog Posts](../../assets/content/body/2022/04/rickmortdev---storeseo---shopify-2025-07-16-at-12.16.14-pm.webp) #### **Meta description must be within 160 characters** Your [**blog post’s meta description**](/shopify-homepage-title-and-meta-descriptions) should be ideally within 160 characters. As you can see in the image below, our meta description is 159 characters long, which meets this requirement.
RickMortdev · StoreSEO · Shopify 2025-07-16 at 12.17.20 PM
#### **Focus keyword is used in the meta title** Make sure your meta title contains your selected focus keyword. As you can see in the image below, our meta title is the following: Top **Synthetic Leather Shoes** for Any Business Events or Formal Settings So, the meta title contains our selected focus keyword (marked bold).  ![SEO For Blog Posts](../../assets/content/body/2022/04/rickmortdev---storeseo---shopify-2025-07-16-at-12.18.03-pm.webp) #### **1 internal link found in blog content** Your blog content should contain 1 internal link. Internal link means linking to another blog, page or product of your Shopify store. To add this, proceed to edit your blog post from the Shopify dashboard as described previously. Select the words you want to set as anchor text and click on the three-dot option in the ‘**Content**’ section. Next, select the ‘**Insert link**’ option. Enter the internal link in the designated field. You also get to choose if the link will open in the same or new window upon clicking. You can set a link title as well. Click on the ‘**Insert link**’ button when you are done. Try to link to a page or blog post of your store relevant to the blog topic or the product or service discussed in your blog. As you can see in the image below, we have internally linked our blog to our FAQs page.
#### **1 external link found in blog content with do-follow** Keep one external link in the blog content with do-follow. External link means linking to external pages outside your Shopify store. Include it when editing your blog post as shown in the previous point.  Try to link to a high authority external link of a blog post or page. Also, the external link should be relevant to the blog topic or the product or service discussed in your blog. As you can see in the image below, we have included 1 external link in our blog content.
#### **Add alt text to all images** If you have multiple images in the blog post, ensure that the focus keyword is used in the alt text of all images. You have to do this when editing your blog post from Shopify dashboard as discussed before. Under the ‘**Content**’ section when editing a blog post, Double click on any image within the blog content. You will then be able to add the alt text in the ‘**Image alt text**’ field.  The images used in our blog post contain relevant alt text as shown in the image below.
By following these steps, you can address every important SEO factor for the best possible optimization score for your blog post. If you need help to optimize your content for SEO with the power of AI, you can also use the [**StoreSEO AI Content Optimizer**](/docs/how-to-use-storeseo-ai-content-optimizer).  Once you are done with your blog post optimization and happy with the overall SEO Score, click on the ‘**Save**’ button.
RickMortdev · StoreSEO · Shopify 2025-07-16 at 12.19.01 PM
This is how easily you can configure **SEO for blog posts and pages** on Shopify using StoreSEO. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Connect StoreSEO MCP to Claude Source: https://storeseo.com/docs/connect-storeseo-to-claude Claude is Anthropic's AI assistant. You use it in a browser at claude.ai, in the Claude Desktop app and in mobile apps, and it can answer questions, write and analyze, and take actions through connected tools. This guide connects Claude to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Claude can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Claude](/agent/claude). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** You sign in when you add the connector: Claude opens a browser window where you pick your Shopify store and approve access. StoreSEO supports dynamic client registration, so there is no token to paste. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Claude ### Step 1: Add StoreSEO as a custom connector In Claude, open Customize, then Connectors, choose + Add and then Add custom connector. Name it StoreSEO and paste this URL as the remote MCP server URL. ```text https://mcp.storeseo.com/mcp ``` ### Step 2: Approve your store Claude opens the StoreSEO sign-in in your browser. Pick your Shopify store and approve access. Then start a new chat and ask this to confirm the connection works. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Add it as an owner on Team or Enterprise Only if you are on a Team or Enterprise plan: an owner adds the same URL under Organization settings, then Connectors. Members then connect their own store from Customize, then Connectors. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new chat and send this prompt. Claude should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Claude - Use StoreSEO to find my five lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for the ones on my best-selling collection. - Is my llms.txt live? If not, tell me what generating one will do before you run it. - Which Search Console queries bring impressions to my store but few clicks in the last 28 days? - Check whether product schema is switched on for my store and tell me what it covers. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Claude - The Free plan allows one custom connector, so StoreSEO takes that slot. Pro, Max, Team and Enterprise plans allow more. - On Team and Enterprise, members cannot add a custom connector themselves: an owner adds it under Organization settings, Connectors, and members then authenticate from Customize, Connectors. - Claude's own prompts for write actions are the hard stop. Anthropic advises using Allow always only for servers you trust, and for StoreSEO it is safer to keep the apply, optimize, generate and delete tools on ask. - Start each conversation by checking that the connector is switched on for it in the Search and Tools menu, and turn off tools you do not need for that chat. ## Disconnect Remove the StoreSEO server in Claude, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Claude: jobs, prompts and FAQs](/agent/claude) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Anthropic: custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Your AI Assistant And Optimize Shopify Store With Simple Prompts? Source: https://storeseo.com/docs/connect-storeseo-mcp-to-your-ai-assistant The **StoreSEO MCP (Model Context Protocol) Server** can be easily connected to any AI platform (e.g., Claude) that supports custom MCP servers. This new feature lets you connect your Shopify store directly to your favorite AI assistants. Instead of managing SEO tasks manually in your admin panel, you can provide simple prompts and let AI agents handle optimizations on your behalf. This document covers everything on how you can connect StoreSEO MCP to your AI assistant. Once configured, you will be able to optimize product listings, generate alt text, fix meta descriptions, and more, all from your favorite AI tool. ## Guide to Connect StoreSEO MCP Server to Your AI Assistant Follow the step-by-step guide to connect StoreSEO MCP with **[Claude](https://claude.ai/)** or other AI assistants and **[optimize your store](/docs/ai-toolkit-ai-agent-friendly-solution)** with simple prompts directly from your favorite AI platforms. ### **Step 1: Go To Claude Settings & Add Connectors** First, log in to your Claude account and go to the Settings tab. Here, we will show how to connect **Claude** with StoreSEO. For that, go to **‘Settings’** → **‘Connector’** and click on the **‘Customize’** button. 
You will see the list of connectors. Now, to connect the StoreSEO MCP Server, click on the **‘+’** icon, and then proceed by clicking on the **‘Add Custom Connector’** button. ![Connect StoreSEO MCP to Your AI Assistant](../../assets/content/body/2026/06/storeseo-mcp-3-scaled.webp) Here, a pop-up window will appear. Give the custom connector an appropriate name. Now, you need to place the **StoreSEO MCP Server URL (** [**https://mcp.storeseo.com/mcp**](https://mcp.storeseo.com/mcp) **)** to connect it to your preferred AI tool. So, place the remote **MCP server URL.** Click on the **‘Add’** button to proceed.  ![Connect StoreSEO MCP to Your AI Assistant](../../assets/content/body/2026/06/image.webp) After that, you will be asked to connect your store that you want to optimize with simple prompts. Click on **‘Connect’** and choose the store you want to connect. 
After you choose the store, you will see this pop-up where you can control when and what Claude is allowed to do.
You can now go back to the conversation and type simple prompts to perform SEO tasks across your store. The AI will connect to your store and identify the relevant products or content. You can carry out the optimizations automatically from the prompt interface as well.
This is how easily you can connect **StoreSEO with Claude AI assistant** and optimize with simple prompts. Need assistance? Reach out to the [**StoreSEO support team**](/contact) for any questions. --- # How To Connect StoreSEO With Google Analytics Source: https://storeseo.com/docs/how-to-connect-storeseo-with-google-analytics [**StoreSEO PRO**](/#pricing) comes with an amazing feature that will help you connect StoreSEO with Google Analytics. Using this feature, you can track your store traffic and understand how people find and interact with your store. 
_Note: You need to have a Google Analytics account to connect it with StoreSEO. So, before following this guideline, create your_ [**_Google Analytics_**](https://support.google.com/analytics/answer/9304153) _and_ [**_Google Cloud Console_**](https://console.cloud.google.com/) _accounts first._  ## **How To Connect StoreSEO With Google Analytics**  Before you start, make sure you have [**upgraded to any StoreSEO Premium**](/docs/upgrade-your-storeseo-plan) plan to connect your Shopify store with Google Analytics.  ### **Step 1: Navigate To Google Analytics Set Up** First, from your dashboard, navigate to the ‘**Settings**’ → ‘**Google Analytics’** tab. Now, you need to authenticate your store with Google from there. 
### **Step 2: Enable The Analytics API** Afterward, click on the ‘**Enable API**’ button to enable analytics API in your Google Cloud Console project.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-120.webp) Now, you will find a window where you have to log in to your Google Cloud Console account. Login to your account from there.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-106.webp) Now, If you already have a project on Google Cloud Console then can skip the following part. If you don’t have a project created click on the ‘**CREATE PROJECT**’ part.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-107.webp) Provide the project name and then click on the ‘**CREATE**’ button. Once you have created the project now then you can easily enable the analytics API.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-108.webp) Click on the ‘**ENABLE**’ button to provide access to Analytics configuration and report data. ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-109.webp) ### **Step 3: Enable Analytics Data API** Once you have enabled Analytics API, now, you need to enable the Analytics Data API. Click on the ‘**Enable**’ button now, and wait for the next window to come up. ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-121.webp) You will be redirected to the Google Cloud again. Now, click on the ‘ENABLE’ button to provide necessary access to Google Analytics report data.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-111.webp) ### **Step 4: Enable Search Console API** Click on the ‘**Enable API**’ button to Enable search console API in your Google Cloud Console project.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-122.webp) Now, click on the ‘**ENABLE**’ button from the next window. It will provide the necessary access to connect StoreSEO with Google Analytics.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-113.webp) ### **Step 5: Setup Service Account** After that, click on the ‘**Setup service account**’ button to set up your service account in Google Cloud Console.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-123.webp) Now, you will find a window where you have to choose your project. Choose your created project and click on it.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-115.webp) From the next window, fill up the necessary details in the ‘**Service account details**’ section and then click on the ‘**Done**’ button afterward.  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-116.webp) Now, copy the service email address from there. Afterward, you need to visit Google Analytics to provide property access. ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/11/image.webp) Go to your [**Google Analytics**](https://analytics.google.com/) account and then navigate to the ‘**Admin**’ → ‘**Property access management**’ option. 
Now, click on the ‘**+**’ icon and hit the ‘**Add users**’ button. This will open a new window where you have to add your service email address. ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/11/image-1-2.webp) Enter your service email address on the ‘**Email addresses**’ field and keep the roles as ‘**Viewer**’ before clicking on the ‘**Add**’ button. ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/11/image-2-2.webp) ### **Step 6: Upload Service JSON Config File** Afterward, go back to your Google Cloud service account to download the service JSON Config file. Click on the ‘**Action**’ button and hit the ‘**Manage Keys**’ option. Then click on the ‘**Add Key**’ button and select the ‘**Create a new key’** option. You will find a window where you have to choose the key type. Select ‘**JSON**’ and then hit the ‘**Create**’ button. Your service JSON config file will be downloaded to your device. 
Now, from the StoreSEO dashboard, click on the ‘**Add file**’ button from the ‘**Upload service JSON config file**’ section. Choose the downloaded JSON file and upload it. 
### **Step 7: Add Google Analytics Property ID** After that, provide your **GA4 Google Analytics property ID** and then click on the ‘**Update**’ button to set up analytics on your store.  _Note: Follow_ [**_this documentation_**](/docs/how-to-collect-ga4-google-analytics-property-id) _to collect your GA4 Google Analytics property ID._  ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-124-scaled.webp) ### **Final Outcome** After successfully following these previous steps, click on the ‘**Analytics**’ button to track your user data from your Shopify store. This is how easily you can integrate StoreSEO with Google Analytics on your Shopify store.  _Note: It may take up to 24 hours before analytics data is reported by Google after the setup is complete._ ![Connect StoreSEO With Google Analytics](../../assets/content/body/2023/10/image-125-scaled.webp) Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Customize JSON-LD Of Your Shopify Store Using StoreSEO Source: https://storeseo.com/docs/customize-json-ld StoreSEO allows you to **customize JSON-LD** of your Shopify store, which allows you to display structure data on search engines so that your visitors can see all of your store’s information at a glance.  ## **How To Customize JSON LD Using StoreSEO?** After you have [**installed StoreSEO**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify account, you need to upgrade to any premium plan as this is a **PRO feature**. Now you need to go to your Store SEO app settings. From the **‘General Settings’** section, you will see the checkmark option of JSON-LD.  ![Customize JSON-LD](../../assets/content/body/2023/10/image-49.webp) Now you need to click on the **‘Local SEO’** tab to configure the local SEO setup. From this part, you can change the logo, name, URL, and other options. Click on the **‘Save Changes**’ button from the bottom. 
## **How To Add Multiple Locations In Local SEO?** You can also add multiple locations for your business by managing your local SEO. Simply scroll down to the **‘Local SEO’** tab. From there, you can easily switch between your store’s primary address or add a different address by clicking on the **‘New Location’** option as shown below. ![Customize JSON-LD](../../assets/content/body/2023/10/image-50.webp) This is how you can easily customize the JSON-LD of your Shopify store using the StoreSEO app.  Need assistance? Feel free to reach out to our [**dedicated support team**](https://wpdeveloper.net/support/) for any type of query. --- # How To Customize Social Media Preview Of Shopify Products With StoreSEO? Source: https://storeseo.com/docs/customize-social-media-preview With the [**StoreSEO app**](https://apps.shopify.com/storeseo), you can now **customize social media previews** of your Shopify products that will help you efficiently show detailed information on Facebook, Twitter and Google search results pages. You can even choose which images should be displayed on Facebook and Twitter, by using this feature from StoreSEO. ## **How To Customize Social Media Preview Of Products?** Before you start, make sure you have created a product list inside of your Shopify product store. Follow these step-by-step guidelines to find out how to **customize social media preview** with StoreSEO.  ### **Step 1: Select And Install StoreSEO App** First, go to the **‘Apps’** option on Shopify and select the ‘StoreSEO’ app from there. Then, follow the instructions in this guide to install the [**StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify). **Step 2: Go To The Product List From StoreSEO Dashboard**  Now click on the ‘**Products’** tab from the StoreSEO dashboard and find all your Shopify products listed down with details. Beside the product details, click on the ‘**Fix**’ button to customize the social media preview text. ![Customize social media preview](../../assets/content/body/2023/10/image-60.webp) You can add a new description or change the meta description of your product from the ‘**General SEO**’ tab. ![Customize social media preview](../../assets/content/body/2023/10/image-61.webp) After adding the meta description, you will see the immediate effect of an boosted number of the overall SEO score. ![Customize social media preview](../../assets/content/body/2023/10/image-62.webp) Click on the ‘**Save Changes**’ button to save the progress. Now  come to the ‘**Preview Snippet**’ tab and see the changes you just have made.  ![Customize social media preview](../../assets/content/body/2023/10/image-63.webp) ### **Step 3: Customize Social Media Preview With StoreSEO**  After clicking the ‘**Fix**’ button, you need to navigate to the **‘ Preview Snippet’** tab to customize social media previews.
From here, you can toggle buttons to see the mobile view and desktop view as well. You can also check the Google search preview, Facebook link preview, and Twitter link preview texts.  If you want to change the preview images for the **‘Facebook Link Preview’** and ‘**Twitter Link Preview**’ options, you can do it from the same ‘Preview Snippet’  section by clicking on the **‘Change Facebook Image’ and ‘Change Twitter Image’** buttons respectively. 
This is how easily you can easily customize social media previews using the StoreSEO app.Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Customize Social Media Previews For Pages And Blog Posts? Source: https://storeseo.com/docs/social-media-previews-for-pages-and-blogs With the [**StoreSEO**](/) app, you can now customize [**social media previews for blog posts and pages**](/docs/customize-social-media-preview), allowing you to provide detailed information on Facebook, Twitter, and Google search results pages more effectively. You can even decide which images should be displayed on Facebook and Twitter by using this StoreSEO feature.  ## How To Customize Social Media Preview For Pages? Before you start, ensure you have created a product list inside your Shopify store. Follow these step-by-step guidelines to find out how to customize social media previews for pages.   ### Step 1: Select And Install The StoreSEO App First, go to the **‘Apps’** option on Shopify and select the [StoreSEO app](https://apps.shopify.com/storeseo) from there. Then follow the instructions in this guide to [install the StoreSEO app](/docs/how-to-install-storeseo-app-on-shopify). ![Social Media Previews For Pages](../../assets/content/body/2023/10/image-24.webp) ### Step 2: Configure Social Media Previews For Pages Now, from the StoreSEO dashboard, move to the ‘Pages’ tab as shown below. You will find all the pages of your store enlisted here. ![Social Media Previews For Pages](../../assets/content/body/2023/10/image-25.webp) Now click on the **‘Fix’** button to customize your social media previews for pages.
You can add a new description or change the [**meta description**](/docs/add-meta-description-to-shopify-products) of your product from the **‘General SEO’** section. ![Social Media Previews For Pages](../../assets/content/body/2023/10/image-26.webp) After customizing the necessary changes, you can see your modifications once you scroll down and see the changes in the **‘Preview Snippets’** section.
## How To Configure Social Media Preview For Blog Posts? With **StoreSEO for Shopify**, you can customize the social media previews of your blog posts very easily. The process is exactly the same as the social media previews for pages.  To configure social media preview for blog posts, go to the **‘Blog Post’** tab. Click on ‘**Fix**’ to make the necessary changes to the blog post, and scroll down. You will see the social media previews accordingly in the **‘Preview Snippets’** section.
If you want to change the preview images for the **‘Facebook Link Preview’** and **‘X Link Preview’** options, you can do it from the same **‘Preview Snippet’**  section by clicking on the **‘Change Facebook Image’** and **‘Change X Image’** buttons respectively. 
This is how you can easily customize your social media preview for pages and blog posts easily.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Easily Add FAQ Schema to Your Shopify Products with StoreSEO Source: https://storeseo.com/docs/add-faq-schema-to-your-shopify-products With StoreSEO, you can easily win FAQ snippets in **LLM-powered AI search engines** like ChatGPT, Perplexity, Grok, and others, as well as organic **Google search** results! This detailed documentation will walk you through activating and adding the FAQ schema using **[StoreSEO](https://apps.shopify.com/storeseo)** manually or with AI help in just a few simple steps. Perfect for making your product pages more informative and clickable. Let’s get started!  Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify store. Then, follow these step-by-step guidelines to find out organic search insight for your **Shopify** store using **StoreSEO**. ### Step 1: Go to SEO Schema From your **StoreSEO dashboard**, select the SEO schema option. Here you will find a list of essential schemas for your Shopify store to use and enhance the search visibility. ![FAQ Schema](../../assets/content/body/2026/01/image.webp) ### Step 2: Turn on the FAQ Schema From this list of available SEO Schemas, pick the FAQ Schema, and click on the Turn on button to activate the schema for your store.
### Step 3: Add Schema Next, click on ‘**Optimize SEO** ‘ and choose the product you want to add FAQs to. This will help those FAQs show up right in the search results.
### Step 4: Write FAQ Questions And Answers Once you are on the product page, scroll down to the FAQ schema section and click ‘**Add Schema**.’
Here, you can type in your questions and answers, and make them relevant to the product, of course! When you’re happy with them, just hit ‘**Save.**‘ ![faq schema](../../assets/content/body/2026/01/image-1.webp) ### Step 6: Generate with AI You can also use our AI feature to generate FAQ questions and answers for your product. Click on the ‘ **Generate with AI**‘ option and provide how many FAQ you want to generate, and select the ‘**Generate FAQs with A**I’ button.
### Step 7: Review the Response After the AI generates the FAQs, take a look. If you like one (or all), click the ‘**Keep**‘ button to add it. If you want to adjust the response to your preference, just click again to edit or tweak the questions and answers until they’re perfect for your product.
This is how easily you can use the FAQ Schema feature for your Shopify store using StoreSEO. To bring your FAQs on storefront and further customization **[follow this documentaion](/docs/how-to-show-faqs-on-your-storefront)**. If you need any assistance, feel free to contact [**our support team**](/support). --- # How To Enable ‘Redirect On Out Of Stock’ In StoreSEO? Source: https://storeseo.com/docs/redirect-on-out-of-stock [**StoreSEO**](https://apps.shopify.com/storeseo) comes up with a **‘Redirect On Out Of Stock’ feature** that allows you to automatically redirect customers to a particular page when the desired products are out of stock. Let’s learn how you can enable ‘Redirect On Out Of Stock’ in **StoreSEO.** **Note:** Before getting started, make sure that you have the [**StoreSEO app installed**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify store. To enable this feature, you need [**StoreSEO PRO**](/#pricing)**.** ## **How To Enable ‘Redirect On Out Of Stock’ Feature?** First, go to your Shopify store and open the StoreSEO app inside it. From the Shopify Dashboard, go to ‘**Apps’ → ‘StoreSEO’**.  ![Redirect On Out Of Stock](../../assets/content/body/2022/06/image-1.webp) You will find yourself in the StoreSEO dashboard. Now, click on the **‘Settings’** option as shown below.
From the ‘**General**’ settings, select the **‘Redirect On Out Of Stock’** button. Put the URL of your store in the relevant field and update. 
This is how easily you can redirect visitors to another page when a product is out of stock. Getting stuck? Feel free to communicate with our [**dedicated support team**](/contact) for further assistance. --- # How To Enable Auto AI Optimization For StoreSEO’s AI Content Optimizer? Source: https://storeseo.com/docs/how-to-enable-auto-ai-optimization [**StoreSEO**](https://apps.shopify.com/storeseo) lets you enable Auto AI Optimization on Shopify for the [**AI Content Optimizer**](/docs/how-to-use-storeseo-ai-content-optimizer) add-on. With this functionality, you do not have to manually enter meta titles, meta descriptions & tags for your products. You can also configure this feature to automatically generate the image alt text. Once enabled, this feature will help you optimize the SEO content and generate image alt text automatically when you add products to your store. ## How to Enable Auto AI Content Optimizer In StoreSEO Follow this step-by-step guide to enable the Auto AI Optimization feature in your shop.  **_Note:_** The AI Auto Optimization feature is only available when you have unlocked an AI Content Optimizer credit bundle. Before going through the guidelines below, you need to add the AI Content Optimizer add-on by following [**this documentation**](/docs/credits-ai-content-optimizer)**.**  ## Step 1: Navigate To & Enable Auto AI Optimization From the StoreSEO app dashboard, navigate to the **‘View more’ → ‘Settings’ → ‘Auto AI Optimization’** option from the side panel. Click on the ‘Manage Settings’ button on the top right-hand corner of the screen, as shown below, and hit the **‘Turn on’** button to enable the AI Content Optimizer.
How To Enable Auto AI Optimization For StoreSEO’s AI Content Optimizer?
## Step 2: Manage Settings for Auto AI Optimization Once enabled, click on the checkbox to select the type of content that you want to optimize automatically with AI from between the options **‘Meta Titles & Descriptions’** and **‘Tags’,** or both by default. Select the image type, **‘Featured image’** or **‘Product images’,** for which you want AI-generated image alt text as well. Afterward, click on the **‘Save’** button. **Note:** Auto optimization for content and image alt text will be done automatically using AI based on your provided focus keywords.
How To Enable Auto AI Optimization For StoreSEO’s AI Content Optimizer?
StoreSEO will then start the AI auto optimization process. To check, navigate to **‘Optimize SEO’ → ‘Products’**. Here, the optimized product will have the **‘AI Optimized’** icon on the right side of its title, as shown below. 
How To Enable Auto AI Optimization For StoreSEO’s AI Content Optimizer?
This is how easily you can enable StoreSEO’s Auto AI Optimization on Shopify stores.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Enable Auto Image Optimization On Shopify Source: https://storeseo.com/docs/how-to-enable-auto-image-optimization-on-shopify **[StoreSEO](https://apps.shopify.com/storeseo)** brings you an Auto Image Optimization feature with the [**Image Optimizer**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) add-on. You no longer have to manually optimize your product images; once enabled, this feature will help you optimize and resize your images automatically when you upload them to your Shopify store. Follow this step-by-step guide to enable the Auto Image Optimization feature for your store.  _**Note:** The Auto Image Optimization feature is only available on the premium Image Optimizer subscription plans. Before going to the guidelines below, you need to add the Image Optimizer add-on to your subscription plan by following [**this documentation**](/docs/how-to-add-image-optimizer-to-your-subscription-plan)._  ## **Step 1: Navigate To Auto Image Optimization Settings**  From the StoreSEO app dashboard, navigate to the ‘**Image Optimizer**’ → ‘**Manage Settings**’ option. 
Alternatively,  you can go to the ‘**Settings**’ tab from your StoreSEO dashboard and navigate to the ‘**Image Optimizer**’ option.
## **Step 2: Enable Auto Image Optimization On Your Store** Choose the default settings for Image Compression, Image Format, and Image Resizer as you want. Afterward, click on the ‘**Save**’ button.
Now, click on the ‘**Turn on**’ button from the ‘**Auto Image Optimization**’ section. This will instantly enable auto image optimization for your store. **_Note:_** _After optimizing an image with StoreSEO, if you replace this image and then restore it, StoreSEO will restore the previously replaced image that you optimized and your newly added image will be deleted from your Shopify store._
This is how easily you can enable Auto Image Optimization on Shopify stores.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query.  ### **Image Optimizer Related Docs:** - [**What Happens If I Reach The Monthly Usage Limit For Image Optimizer?**](/docs/monthly-usage-limit-for-image-optimizer) - [**How To Add Image Optimizer To Your Subscription Plan**](/docs/how-to-add-image-optimizer-to-your-subscription-plan) - [**How To Use StoreSEO Image Optimizer On Shopify**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) --- # How To Enable Or Disable Email Notification in StoreSEO? Source: https://storeseo.com/docs/enable-disable-email-notification-in-storeseo With [**StoreSEO**](https://apps.shopify.com/storeseo), you can get weekly updates or reports of your Shopify store’s SEO performance through an email notification. This feature helps you to stay informed about your store’s product optimization report via email without having to log into your store dashboard. Read the documentation below to learn how to enable or disable Email notifications from StoreSEO. ## How To Enable Or Disable Email Notification in StoreSEO? To get weekly updates on the overall product optimization report, follow the simple step-by-step guidelines to find out how to enable or disable email notifications in StoreSEO. ### Step 1: Turn on Email Notifications in StoreSEO First, log in to your [**Shopify store**](https://shopify.pxf.io/storeseo). From the app search panel, search for StoreSEO. Then, from the StoreSEO dashboard, go to **‘Settings’ → ‘Email Notification’.** Now, click on the **‘Turn On’** button to get performance reports of your store delivered to your inbox and the email notification will be enabled.
How To Enable Or Disable Email Notification in StoreSEO?
### Step 2: Enable Email Notifications for Overall Product Optimization Report To receive email notifications for **‘Overall Product Optimization Report’** for your Shopify store, click on the **‘Enable’** button as shown below. Then, you can select a specific **‘Day’ and ‘Time’** when you prefer to get the email notification on overall product optimization. Click on the **‘Save’** button at the bottom of the page once done.
How To Enable Or Disable Email Notification in StoreSEO?
### Step 2: Enable Email Notifications for Auto AI Content Optimization Report Similarly, you can also enable email notifications for the **Auto AI Content Optimization Report** on your store and set the **Day and Time** when you want to receive the reports via email. Click on the **‘Save’** button to save your changes.
How To Enable Or Disable Email Notification in StoreSEO?
**Note:** You can enable email notifications for both reports at once or choose which one of the two emails you want to receive with StoreSEO.  If you do not want to get any weekly notifications, simply click on **‘Turn Off’** and then hit the **‘Save’** button; the email notification will be disabled.
How To Enable Or Disable Email Notification in StoreSEO?
This is how easily you can enable or disable email notifications from the StoreSEO app and get performance reports of your store delivered to your inbox.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Generate AI Snippets To Shopify Products With StoreSEO Source: https://storeseo.com/docs/how-to-generate-ai-snippets-to-shopify-products With our AI-powered [**StoreSEO app,**](/) you can now generate AI snippets for the products of your store. The AI Snippet form aims to provide a “concise identity” for your product. Instead of a long, rambling description, it breaks down the product into specific attributes that AI models prefer for quick extraction. ## **How to Generate AI Snippets for Shopify Products?** When you generate AI Snippets for your product with **StoreSEO****,** it helps you weave keywords naturally into your content to improve your search engine rankings. Follow these simple steps onward: ### **Step 1: Go To StoreSEO Product List**  Go to the **‘Apps’** option from your dashboard and select the ‘**StoreSEO’** app. Now, click on the ‘**Optimize SEO’** tab from the side panel and choose ‘**Products**,’ where all your Shopify products are listed with details. ![Generate AI Snippet](../../assets/content/body/2026/02/image-17.webp) ### **Step 2: Generate AI Snippet** Here, you can simply generate a full AI snippet centered on your focus keyword. Simply click on the “generate” button as shown below.
The content for “**Product** **Summary**,” “**Best** **For**,” “**Key** **Benefits**,” “**Unique** **Features**,” and “**Use** **Cases**” will be generated automatically. Click on the “**Keep** **It**” button under the criteria. If you want to make changes to the generated content, click on the “**Revert**” button and add your preferred content as shown below. 
Alternatively, you can click on the “**Re-Generate**” button to get another version of AI-generated content for the product snippets.
### **Step 3: Keep All Generated Content & Save Progress** If you are satisfied with all the generated content, you can click on the “**Keep All**” button to keep all the generated content for Product Snippets. 
Once you’re done filling up all the required fields with your preferred content, make sure you click on the **‘Save’** button to keep all the information.
This is how easily you can Generate AI Snippet with StoreSEO for your [**Shopify products**](https://shopify.pxf.io/storeseo).  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # How To Generate And Add Image Alt Text With AI For Shopify Products Using StoreSEO? Source: https://storeseo.com/docs/generate-and-add-image-alt-text-with-ai With the [**StoreSEO app**](/), you can **generate and add Image alt text with AI** for your Shopify products and boost your user experience, SEO, and other important factors. Using this AI-powered feature, you can easily create and add image alt text, boost your Shopify store and help rank your product images in Google search results. ## **How To Generate And Add Image Alt Text With AI Using StoreSEO?** Before you start, make sure you have set up your [**Shopify store**](https://www.shopify.com/) and added your products with their images. Next, [**install StoreSEO app on the store**](/docs/how-to-install-storeseo-app-on-shopify). Once you have done all these, follow these step-by-step guidelines to find out how to use an image alt text generator for your Shopify products.  ### **Method 1: Use Image Alt Text Generator Feature From StoreSEO Dashboard** In this method, we will navigate to the ‘**Image Alt Text Generator**’ feature directly from the StoreSEO dashboard. Let us see how to do it. #### **Step 1: Navigate To StoreSEO App From Shopify Dashboard** First, go to StoreSEO app from your Shopify dashboard. To do this, type in and look for the app in the search bar at the top. Alternatively, go to the **‘Apps’** option in the left side panel and select the ‘**StoreSEO’** app. 
#### **Step 2: Select Image Alt Text Generator Option**  Now, click on the **‘Image Alt-text Generator**’ tab in the left side panel under StoreSEO. This will take you to the list of all the product images on your Shopify store where you can generate their image alt text.
#### **Step 3: Generate And Add Alt Text For Any Product Image** To generate the image alt text for any product, click on the ‘**Generate**’ button under **Action** on the right side of the image. Click on the ‘**Generate with AI**’ button in the window that will appear. This will instantly generate the image alt text with AI. You can then either re-generate the alt text or keep it by clicking on the corresponding buttons.
You can also generate image alt text in bulk. To do this, select the images you want to generate image alt text for by clicking on the checkboxes for the products on the left side. Then, click on the ‘**Generate with AI**’ button on the top right. Click on the ‘**Restore**’  button under **Action** on the right side of the image to revert to the original alt text. You can also select multiple images by clicking on their checkboxes and restore them in bulk.
When we do bulk optimization, you will notice a ‘**Pending**‘ status for some time. This status indicates that the process of reading the image with AI is running in the background. After a short while, the image alt text will be generated. **Note:** The user’s AI credits will be deducted when generating the image alt text with AI using StoreSEO. ### **Method 2: Use Image Alt Text Generator Feature When Optimizing Product With StoreSEO** In this method, we will show you how to use the ‘**Image Alt Text Generator**’ when you are optimizing any product using StoreSEO. #### **Step 1: Optimize SEO For Your Desired Product** Navigate to the ‘**Optimize SEO**’ option in the left side panel under StoreSEO. Click on the ‘**Fix**’ button beside the product you want to optimize. 
#### **Step 2: Select Generate Alt-Text With AI Option** Now, go to the ‘**Images**’ section of the product. Then, click on the ‘**Generate alt-text with AI**’ option. This will take to the list of all the images of that product for which you can generate alt text.
#### **Step 3: Generate And Add Alt Text For The Product’s Images** To generate the image alt text for any image of the product, click on the ‘**Generate**’ button under **Action** on the right side of the image. Click on the ‘**Generate with AI**’ button in the window that will appear. This will instantly generate the image alt text with AI. You can then either re-generate the alt text or keep it by clicking on the corresponding buttons.
You can also generate image alt text in bulk. To do this, select the images you want to generate image alt text for by clicking on the checkboxes for the products on the left side. Then, click on the ‘**Generate with AI**’ button on the top right. Click on the ‘**Restore**’  button under **Action** on the right side of the image to revert to the original alt text. You can also select multiple images by clicking on their checkboxes and restore them in bulk.
When we do bulk optimization, you will notice a ‘**Pending**‘ status for some time. This status indicates that the process of reading the image with AI is running in the background. After a short while, the image alt text will be generated. _**Note:** The user’s AI credits will be deducted when generating the image alt text with AI using StoreSEO._ This is how easily you can use the AI-powered image alt text generator feature to instantly generate and add alt text for your Shopify product images with the StoreSEO app. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) with any type of query. --- # How to Generate Blog Posts with StoreSEO AI Blog Generator Source: https://storeseo.com/docs/how-to-generate-blog-posts-with-ai-in-storeseo The Generate blog posts with AI is a powerful new feature in [**StoreSEO**](/) introduced to generate blogs for your Shopify store. Easily make SEO-friendly blog posts that bring in more visitors, connect with customers, and improve your store’s search rankings. This documentation shows you the simple steps to use StoreSEO’s AI to create clear, engaging blog content that fits your brand.
Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://shopify.pxf.io/storeseo) store. Then, follow these step-by-step guidelines to find out how to generate blogs with AI for your Shopify store using StoreSEO. ### Step 1: Navigate to AI Blog Generator To start, first go to your Shopify store and select the StoreSEO app. Click on the ‘**AI Blog Generator**’ from the side menu.
### Step 2: Choose To Generate To generate blog content for your Shopify store using AI, from the side menu, locate and select the ‘**AI Blog Generator**’ option.  ![AI Blog Generator](../../assets/content/body/2025/07/image-2-2.webp) ### Step 3: Sync The Blogs If you have not synced your blog with StoreSEO before, you must sync your blog to use the AI blog generation feature. For the syncing, just click on the ‘**Please sync your blogs**’ button. ![AI Blog Generator](../../assets/content/body/2025/07/image-3-2.webp) ### Step 4: Fill out the Instructions To create a blog post with AI, ensure you enter all the details displayed in the panel, such as the blog topic, primary target keyword, adding a feature image, and any additional relevant fields or settings.
### Step 5: Edit the Blog After the blog generation, you can edit the blog post according to your preferences. Simply click on the ‘**Edit blog post**’ button. ![AI Blog Generator](../../assets/content/body/2025/07/orbitculture---storeseo---shopify-2025-07-28-at-12.20.16-pm-1.webp) Here you can adjust the necessary changes on your blog like headings, adding images, tags, etc, according to your preferences.
Your AI generated blogs will be on draft. You can edit the blog further from the ‘**AI Blog Generator**’ dashboard, and regenerate your blog as well, just by clicking on the **‘Regenerate your blog**’ icon.
### Step 6: SEO Optimize Your Blog Your AI generated blog is already SEO optimized. However, for further advance SEO optimization of your AI generated blogs with the best SEO practices, go to the ‘**Optimize SEO**’ option, click on ‘**Blog Posts**’ and select the blog post you want to optimize following the [**SEO analysis**](/docs/configure-seo-for-blog-posts-and-pages).
This is how easily you can generate blogs with AI for your Shopify store using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support). --- # How To Generate HTML Sitemap For Your Store? Source: https://storeseo.com/docs/how-to-generate-html-sitemap With [**StoreSEO**](/), you can easily create an HTML Sitemap for your store, which helps users find all the pages of your Shopify store. Additionally, it tells search engines about all the different pages on your website. This boosts SEO optimization for your store products and increases your user experience. Follow this step-by-step guide on how to generate HTML Sitemap for your Shopify store with StoreSEO.
## **Step 1: Go To HTML Sitemap In StoreSEO** Log in to your Shopify store and go to the ‘**Sitemap**’ tab from the StoreSEO app dashboard. Click on the ‘**HTML Sitemap**’ button to generate a Sitemap.
Hit the ‘**Generate Now**’ button, and then the sitemap will be automatically generated. Then copy the URL to your sitemap.
After copying the URL, click the **‘Edit Navigation’** button. This will redirect you to the Navigation tab of your Shopify store.
### Customize Sitemap URL StoreSEO app automatically generates a sitemap URL by default for the users. However, you can customize the URL and set it as you prefer or need. To customize the sitemap URL, navigate to **Manage Apps –> Apps and Sales Channels –> App Proxy** from the ‘**Sitemap**‘ tab. Hit the ‘**Customize URL**‘ button, modify the sitemap URL, and save changes.
## **Step 2: Add Sitemap URL To A Menu** Add a new menu or click an existing one where you want to add the HTML Sitemap. For example, we are adding the Sitemap to the **‘Footer’**. Click on the menu to add an item. Set a name and paste the Sitemap URL link. Once done, save the menu settings.
After adding the HTML Sitemap link to your store, you can hit the ‘**Confirm**’ button to go to the next step of customizing it.
## **Step 3: Customize HTML Sitemap** After adding, you can customize the HTML Sitemap for your store. Here, you can customize the Heading, Description, the Section Heading, and pick colors to match your brand identity. You can also modify Styling and set Visibility for specific groups. After customizing all the preferences, hit the ‘Save’ button.
## **Final Outcome** After adding and customizing the HTML Sitemap, your store’s sitemap page will look like the one below. ![generate HTML Sitemap](../../assets/content/body/2023/12/screely-1702275960642.webp) This is how easily you can generate an HTML Sitemap for your Shopify store with StoreSEO. Getting stuck? Feel free to communicate with our [**dedicated support team**](/contact) for further assistance. --- # How to Generate LLMs.txt With StoreSEO  Source: https://storeseo.com/docs/how-to-generate-llms-txt-with-storeseo The LLMs.txt Generator is a powerful new feature in [**StoreSEO**](/) introduced to optimize your Shopify store for AI-driven search and indexing.  By creating a structured LLMs.txt file, this feature enables large language models (LLMs) like ChatGPT, Claude, and Gemini to efficiently discover, categorize, and utilize your store’s content. This enhances your store’s visibility in AI-powered search results, ensuring accurate representation and improved discoverability. ## How To Generate LLMs.txt For Your Shopify Products With StoreSEO To get started, make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://shopify.pxf.io/storeseo) store. Then, follow these step-by-step guidelines to find out how to generate LLMs.txt for your Shopify store using StoreSEO. ### Step 1: Navigate to Settings Option From your Shopify dashboard, go to the **‘Apps’** section and choose **‘StoreSEO.’** Once you are in the StoreSEO panel, click on **‘Settings,’** then scroll down and select the **‘LLMs.txt Generator’** option.
How to Generate LLMs.txt With StoreSEO 
### Step 2: Generate Your LLMs.txt File Now, simply click the **“Generate”** button. You will see a message saying **“Generation queued,”** which means the process has begun. Just wait a moment, and soon the generated file will appear in the list panel for you to view. You can also choose specific Resource types for creating your file. The options include products, collections, pages, articles, and generation schedules. For schedules, you can pick from daily, weekly, or monthly to suit your needs. This makes it easy to customize the generation process based on what you are working with.
How to Generate LLMs.txt With StoreSEO 
### Step 3: View The Generated File Once your LLMs.txt file is ready, you can easily view it by clicking the eye or preview icon, located as shown below. This will open the text file for you to check out.
How to Generate LLMs.txt With StoreSEO 
**Note:** We keep the latest 10 generated files for you to view. When you generate a new file, the oldest one will be automatically removed, so you will always see the most recent 10 files in the list panel. This helps keep things organized and up to date.
How to Generate LLMs.txt With StoreSEO 
This is how easily you can generate the LLMs.txt file for your Shopify store using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support)**.** --- # How to Increase Your AI Content Credit Limit in StoreSEO? Source: https://storeseo.com/docs/increase-ai-content-credit-limit-in-storeseo StoreSEO **[AI Content Optimizer](/features)** add-on feature helps to optimize your Meta Title, Meta Description, and Tags within the power of AI in a few seconds. If you have reached your limit and need more credits you can buy an AI Credit bundle to increase your usage limit.  ## How to Increase Your AI Content Credit Limit in StoreSEO? First, make sure you have successfully [**installed the StoreSEO app**](https://apps.shopify.com/storeseo). Then, follow these step-by-step guidelines to learn how to increase your AI content credit limit for your Shopify store using the StoreSEO app.  ### Step 1: Increase the AI Content Credit Limit Go to the AI content optimizer option to increase the AI credit limit for optimizing your Shopify store products.
### Step 2: Select the AI Credit Bundle Once you get here, you will get to see different credit bundles for increasing your AI content credit limit. Select the one you prefer and proceed to the checkout by simply clicking on the ‘**Confirm**’ button. ![AI Content Credit](../../assets/content/body/2025/03/image-22.webp) **Note:** after purchasing the increased AI credit bundle. Go to the StoreSEO main dashboard. you will get to see the additional credit along with your current AI content credit limit. ![AI Content Credit](../../assets/content/body/2025/03/image-23.webp) This is how easily you can increase the AI credit limit for your Shopify store. Need assistance? Feel free to reach out to our [dedicated support team](/contact) with any type of query. --- # How To Install StoreSEO App On Your Shopify Store? Source: https://storeseo.com/docs/how-to-install-storeseo-app-on-shopify [**StoreSEO**](/) app can help you optimize your Shopify products for a better SEO rank in search engine results. This will help potential customers find your products and drive more traffic to your business. You can learn more about [**how StoreSEO works**](/docs/what-is-storeseo-and-how-does-it-work) here. ## **How To Install StoreSEO App On Shopify Store** To install the StoreSEO app, first, you need to [**create a Shopify store account**](https://help.shopify.com/en/manual/intro-to-shopify). Then follow the steps given below. ### **Step 1: Click On The ‘Apps’ Option From Shopify Admin** First, you need to click on the ‘**Add**‘ option from your Shopify navigation bar under the App option. Click on the **‘Shopify App Store**‘. It will redirect you to the [**Shopify Apps Store**](https://apps.shopify.com/) page.
### **Step 2: Install StoreSEO App To Your Store** From here, search for **‘StoreSEO**‘ in the search bar and click on the app. Afterward, hit the **‘Install’** button. ![Install StoreSEO App](../../assets/content/body/2021/08/storeseo-ai-seo-agent-storeseo-smart-ai-seo-zight-optimizer-ranking.-shopify-app-store-2025-07-15-at-11.03.04-am-2.webp) You will be redirected to the Shopify dashboard from where you can install the app by clicking on the **‘Install’** button. ![Install StoreSEO App](../../assets/content/body/2021/08/image.webp) ### Step 3: Start the Onboarding Setup Process  In this section, you will be on the first step out of four. Here you will see the perks and benefits of StoreSEO: AI SEO Agent. The features and how they will help your store to grow. Click on the ‘**Confirm**’‘ button to proceed further. ![](../../assets/content/body/2021/08/image-1.webp) In this section, you will be on the first step out of four. Here you will see the perks and benefits of StoreSEO: AI SEO Agent. The features and how they will help your store to grow. Click on the ‘**Confirm**’‘ button to proceed further. ![](../../assets/content/body/2021/08/image-2.webp) Here, you can select the preferred SEO plan for your Shopify store. Click on the **‘Subscribe Plan’** button to proceed further. _**Note**: If you prefer to take a trial plan, simply click on the_ **_Start 7-Day Free Trial,_** _and your trial plan will be active with its features._
Once your trial plan is activated, proceed to the next step, where your store will be automatically scanned. ![Install StoreSEO App](../../assets/content/body/2021/08/image-4.webp) Now, at the third step, your store will be automatically scanned and after a while you will get the SEO score based on the analysis. Once you are done checking the suggested areas of improvement,hit the ‘**Confirm**‘ button to proceed further. ![Install StoreSEO App](../../assets/content/body/2021/08/image-3.webp) This is how easily you can install the **StoreSEO** app on your Shopify store. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # How To Integrate StoreSEO With Google Analytics Source: https://storeseo.com/docs/how-to-integrate-storeseo-with-google-analytics Integrating your Shopify store to Google Analytics through StoreSEO takes just a few minutes and unlocks powerful insights: real-time organic traffic data, sessions, event counts, and much more, all directly inside your StoreSEO analytics dashboard. ## **How To Integrate StoreSEO With Google Analytics** This step-by-step documentation will walk you through the entire process, from initial authentication to viewing your Google Analytics data. You’ll only need to sign in with your Google account a few times (**Google requires separate permissions for each service**). ### **Step 1: Authenticate Your Store With Google** First, from your dashboard, navigate to the **‘Settings’ → ‘Google Integrations’** tab. Now, you need to authenticate your store with Google by clicking on the ‘**Sign-in Now**’ button. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-15.webp) Now, select the Google account that is configured with your Google Analytics account from the next window to connect with your store. After logging in, the authentication will be done immediately. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-97.webp) ### **Step 2: Verify Site With Google** Afterward, click on the ‘**Verify Now**’ button to verify your Shopify store website. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-16.webp) Now, you will find a window where you have to log in to your same Gmail account again. Once you log in to your Gmail, your store verification will be done.  ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-98.webp) ### **Step 3: Add Site To Google Search Console** After verifying your site with Google, click on the ‘**Link Now**’ button to integrate StoreSEO with Google Search Console. Now, wait for the next window to come up. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-20.webp) Now, you will find a window where you have to log in to your same Gmail account again. Login to your Gmail, then click on the ‘**Allow**’ button. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-99.webp) ### **Step 4: Submit Sitemap To Google** Click on the ‘**Permit Now**’ button to submit your Shopify Store sitemap to Google. Once you hit the ‘**Permit Now**’ button, your store sitemap will be automatically submitted to Google. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-17.webp) Now, you will find a window where you have to log in to your same Gmail account again. Once you login to your Gmail account, your store sitemap will be automatically submitted to Google. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-100.webp) ### **Step 5: Grant Permission For Instant Indexing** After that, click on the ‘**Allow Now**’ button to provide the necessary permissions to index your store instantly. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-21.webp) Now, you will find a window where you have to log in to your same Gmail account again. Login to your Gmail, then click on the ‘**Allow**’ button.  ### Step 6: Grant Google Analytics Permission Access Here you need to grant permission to move onward for enabling the Google Analytics feature on your store. Click on the ‘**Grant** **Access**’ button. Allow the necessary accesses to complete the permission processes.  ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-22.webp) Now, you will find a window where you have to log in to your same Gmail account again. Once you login to your Gmail account, you need to select the account you want to grant access for an allow other essential access onward. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-101.webp) ### **Step 7: Google Analytics Property**  After that, grant permission now you need to **Select Google Analytics Property**. Click on the **‘Select a property…’**  ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-18.webp) Choose the property you would like to connect to see the organic google analytics data for that property on your StoreSEO dashboard. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-24.webp) ### **Step 7: Connect to Google Analytics** After selecting the Google Analytics property click on the **‘Connect’** button. ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-23.webp) ### **Final Outcome** After successfully following these previous steps, click on the ‘**Analytics**’ button to track your user data from your Shopify store. From the analytics dashboard, click on the, “**Analytics”** button. It will show your Google Analytics data and other measures. This is how easily you can integrate StoreSEO with Google Analytics on your Shopify store.  _Note: It may take up to 24 hours before analytics data is reported by Google after the setup is complete._ ![Integrate StoreSEO With Google Analytics](../../assets/content/body/2023/08/image-19-2.webp) Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Integrate StoreSEO With Google Search Console Source: https://storeseo.com/docs/storeseo-with-google-search-console StoreSEO PRO comes with a unique feature that lets you integrate [**StoreSEO with Google Search Console**](/) so that you can optimize your products with search analytics and see which products are bringing your customers to your shop.  ## **How To Integrate StoreSEO With Google Search Console?** Before you start, make sure you have created a product list inside of your Shopify store and **[installed StoreSEO](/docs/how-to-install-storeseo-app-on-shopify)** in your store.  ### **Step 1: Sign In To Your Google Account** First, navigate to **StoreSEO Settings** →  **Google Integrations**. Now, you need to sign in to Google by hitting on the **‘Signing In’** button. ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-18.webp) After clicking the **‘Sign-In Now’** button a window will open where you need to select your preferred Google account which you want to connect with your store. Log in to your Gmail account then the authentication will be done immediately. ![StoreSEO With Google Search Console](../../assets/content/body/2023/10/image-84.webp) ### **Step 2: Verify Site With Google** Once the authentication is done, now you need to verify your site with Google. Click on the **‘Verify Now’** button to verify your **[Shopify](https://www.shopify.com/)** store website. ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-19.webp) Now, you will find a window where you have to log in to your Gmail account. Login to your Gmail then click on **‘Continue’**. Once you have finished these steps, your site will be verified automatically. ![StoreSEO With Google Search Console](../../assets/content/body/2023/10/image-86.webp) ### **Step 3: Link To Google Search Console** After verifying your site with Google, you can easily link StoreSEO to Google Search Console.  ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-20.webp) Just click on the **‘Link Now’** button and wait for the next window to come up. ![StoreSEO With Google Search Console](../../assets/content/body/2023/10/image-88.webp) ### **Step 4: Permit For Google Sitemap Submission** Now, click on the **‘Permit Now’** button to submit your Shopify Store sitemap to Google. Once you hit the button your store sitemap will be automatically submitted to Google. ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-21.webp) ### **Step 5: Permit For Instant Indexing** In this step, you will be asked to give permission for instant indexing. Click on the **‘Allow Now’** button to proceed. ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-22.webp) ## **How To Reset Google Integration Settings In StoreSEO?** If you want to reset all of your Google Integration settings, you can do that with one click in StoreSEO. Simply click on the **‘Disconnect Now’** button and you will be able to reset Google Integration settings in StoreSEO. ![StoreSEO with Google Search Console](../../assets/content/body/2021/11/image-23-scaled.webp) This is how you can easily integrate Store SEO with Google Search Console without any hassle. Need assistance? Feel free to reach out to our [dedicated support team](/contact) for any type of query. --- # How To Migrate Data To StoreSEO From Any Shopify SEO App? Source: https://storeseo.com/docs/how-to-migrate-data-to-storeseo With the [**StoreSEO**](https://apps.shopify.com/storeseo) app, you can now easily migrate data to StoreSEO from any Shopify app without any difficulty. This way, you can make sure that all of your previous store data can be migrated easily. For this tutorial, we will migrate data to the StoreSEO app from **Smart SEO**. _Note: Make sure you have_ [**_installed the StoreSEO App_**](/docs/how-to-install-storeseo-app-on-shopify) _on your Shopify store._  ## **How To Migrate Data To StoreSEO From Smart SEO?** With some very simple steps, you can migrate your data from the Smart SEO app. From the Smart SEO app, you can migrate **Meta Titles**, **JSON-LD**, **Meta Descriptions,** and much more. Let’s go through the step-by-step process.   ### **Step 1: Install StoreSEO On Your Shopify Store** Check if you have another SEO app for Shopify installed. Next, you have to [**install and activate the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify store. ![Migrate data to StoreSEO](../../assets/content/body/2023/10/image-65.webp) ### **Step 2: Easily Migrate Data To StoreSEO**  After you have successfully installed the StoreSEO app, you can now migrate data from Smart SEO to StoreSEO. Simply go to the ‘**SEO Settings**’ from the StoreSEO app, and you can see which data you can migrate. You can uncheck or check the boxes as per your preference.  ![Migrate data to StoreSEO](../../assets/content/body/2023/10/image-66.webp) Now, press the ‘**Next**’ button, and you will be redirected to the ‘**Analysis**’ tab. From this page, you can get an idea of how many products have been synched, analyzed, and much more. Under the ‘**Optimization**’ tab, StoreSEO will help you with various issues that you need to fix for your store.  ![Migrate data to StoreSEO](../../assets/content/body/2023/10/image-67.webp) ### **Step 3: Finalize Data Migration From Smart SEO** If you have followed the previous steps, then you have successfully migrated your Smart SEO data to your Shopify store with **StoreSEO.** From the ‘**Success**’ tab, you can navigate to the StoreSEO’s documentation section, follow it on X, and join the Facebook community. Plus, there’s a button to go to the dashboard. ![Migrate data to StoreSEO](../../assets/content/body/2023/10/image-68.webp) This is how you can easily migrate data to StoreSEO from any SEO app effortlessly.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Optimize Agentic Discovery Score in Shopify with StoreSEO Source: https://storeseo.com/docs/how-to-optimize-agentic-discovery-score-storeseo Agentic Discovery Score is a powerful new feature in [**StoreSEO**](/) introduced to optimize your Shopify store for AI-driven search and shopping experiences for your customers.  The Agentic Discovery Score (ADS) in StoreSEO measures how well your product pages are optimized for AI-driven discovery in addition to traditional search engines. This guide walks you through every step to understand, access, and improve your Agentic Discovery Score from inside the StoreSEO app on Shopify. ## Step 1: Open StoreSEO in Your Shopify Admin Log in to your Shopify store admin panel. From the left-hand navigation menu, click on **‘Apps**‘. Find and click on StoreSEO to open the app dashboard. ![Agentic Discovery Score](../../assets/content/body/2026/04/image-5.webp) ## Step 2: Navigate to the Product List Inside the StoreSEO dashboard, click on Optimize SEO from the main navigation. You will see a list of all products in your Shopify store. Each product row displays its current SEO Score and Agentic Discovery Score at a glance, making it easy to identify which products need optimization. ![](../../assets/content/body/2026/04/image-12-2.webp) ## Step 3: Select a Product to Optimize Click on any product from the list to open its product optimization screen. On this screen you will see two score cards at the top: the SEO Score card and the Agentic Discovery Score card. The Agentic Discovery Score card shows your current ADS rating. The right-hand sidebar displays the ADS Analysis panel with a detailed checklist of all optimization areas. ![](../../assets/content/body/2026/04/image-8.webp) ## Step 4: Review the ADS Analysis Checklist In the right-hand ADS Analysis panel, review each category. The checklist is organized into the following sections: ![](../../assets/content/body/2026/04/image-9-2.webp) - **Title Quality** – checks whether the product title meets length and content requirements. - **Description Quality** – checks whether the product description is complete and well-written and AI agent friendly. - **Taxonomy & Classification** – checks whether the product has type, tag, and category assigned. - **Media & Visuals** – checks whether a product image is present and properly optimized. - **Availability & Status** – checks whether the product is published, in stock, and tracked. - **Structured** **Data** – checks whether schema markup and structured data signals are in place. - **AI Content Readiness** – checks whether the content is optimized for AI-driven interpretation. ## Step 5: Optimize the Product Title In the product editor on the left side of the screen, locate the Title field. Update the product title so that it meets all three Title Quality checks shown in the ADS Analysis panel:
- Keep the title within 50 characters to ensure it displays cleanly in search results and AI surfaces. - Include the brand name so AI systems can correctly attribute the product to its manufacturer. - Include the product type so the item can be classified correctly by AI discovery engines.
After editing the title, the corresponding checklist items in the ADS Analysis panel will turn green to confirm they pass. ## Step 6: Make Your Product Description AI Agent-Friendly To make your product description ai agent-friendly accroding to our Agent Discovery Score suggestion, click on the **Edit product** button and select the **Generate Text** icon on the left. Now, provide the instrucitons for the changes you’d like to make for your products, for example: _Rewrite this product description to be AI agent-friendly. Keep the marketing copy but also add structured specs including: materials, dimensions/measurements, care instructions, compatibility/fit, and any other key product attributes. Format specs clearly, either inline or as a list._ AI will generate the copy accordingly, later, click on the **save** button to keep the product description.
## Step 7: Set the Taxonomy and Classification The Taxonomy & Classification section evaluates whether your product is correctly categorized within Shopify and StoreSEO. Proper classification allows AI agents to place the product in the right context, making it easier to surface in category-based queries and AI-powered recommendations. StoreSEO checks three conditions in this category.  - Whether Shopify product category is assigned - Whether Product type is set - Whether Product has at least one tag Each criteria has the edit product option along with them. Click on it, and set the product type accordingly, and the product type and tag would be assigned.
For the product category select one relevant catgory that fits the product, and click on the save button. Once done, click on the sync prodcut button, and your assigned product category would be updated.
## Step 8: Verify Availability and Publication Status Check the Availability & Status section in the ADS Analysis panel. StoreSEO checks three conditions in this section: - The product must be active and published in your Shopify store. - Inventory tracking must be enabled for the product. - The product must be in stock.
If any of these conditions fail, go to the Shopify product editor and update the product status, enable inventory tracking, and confirm the stock level. Return to StoreSEO to verify these checks now pass in the ADS Analysis panel. ## Step 9: Add a Product Image In the product optimization screen, check the Media & Visuals section of the ADS Analysis panel. If the product is missing an image, go to the Shopify product editor and upload a high-quality product photo. After uploading, return to the StoreSEO product screen and confirm that the Media & Visuals check now passes. A clear, relevant product image is essential for AI systems to connect visual data with the correct product listing. ![Add a Product Image](../../assets/content/body/2026/04/image-6.webp) ## Step 10: Enable Structured Data and FAQ Schema In the StoreSEO, for enabling the Product JSON-LD schema, look for the SEO Schema and enable the Product Schema and FAQ Schema options. Enabling structured data allows search engines and AI systems to parse your product information in a machine-readable format. 
To add FAQ schema, switch to the FAQ section and add relevant question-and-answer pairs about the product. Each FAQ entry should address a common customer question clearly and concisely. Once saved, these additions will positively affect the Structured Data and AI Content Readiness scores in the ADS Analysis panel. ![](../../assets/content/body/2026/04/image-10.webp) ## Step 11: Save the Updated Agentic Discovery Score After completing all the above steps, click the Sync Product button. The Agentic Discovery Score card at the top of the page will update to reflect your improvements. Each resolved item in the ADS Analysis panel will be marked as passed. The overall ADS score will increase as more checklist items are completed. Aim to resolve all checklist items to achieve the highest possible Agentic Discovery Score for the product. ![](../../assets/content/body/2026/04/image-11.webp) This is how easily you can optimize the **Agentic Discovery Score** in Shopify with StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support)**.** --- # How To Optimize BetterDocs For Shopify Docs With StoreSEO? Source: https://storeseo.com/docs/optimize-betterdocs-for-shopify-docs [**StoreSEO**](/) helps optimize BetterDocs for Shopify docs so that your store’s individual docs and documentation landing page get more online visibility among potential customers. During optimization, StoreSEO makes sure everything is in order by showing an SEO score. These scores tell you about all the issues that need to be fixed in order to bring a good SEO score for your docs. So, this documentation will show you how it works. **_Note:_** Before you start, make sure you have installed [**BetterDocs for Shopify**](https://apps.shopify.com/betterdocs-knowledgebase) and **StoreSEO** apps on your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Next, [**add docs using BetterDocs for Shopify**](https://betterdocs.co/docs/how-to-add-new-documentation-with-betterdocs/). Then, follow these step-by-step guidelines to find out how to optimize a doc with StoreSEO. ## **Step 1: Navigate To StoreSEO Doc Optimization** Log in to your Shopify store. Then, from the search bar on the top or the ‘**Apps**’ option in the left sidebar, search for the StoreSEO Shopify SEO app and click on it. From the StoreSEO dashboard, go to the ‘**Optimize SEO**’ tab. Now click on the ‘**Docs**’ option. Once your documentations are synced properly, you will see your doc list on this page. Choose the one you want to optimize, and click on the ‘**Fix issue**’ button. 
## **Step 2: Optimize Your Individual Doc On Shopify** Now, it is time to optimize **your Shopify doc for SEO**. As you can see in the image below, the SEO details of your selected doc are shown by StoreSEO. From the right side panel, you will also find **Basic SEO Analysis** and **Detailed SEO Analysis** to find the key factors that have to be addressed to improve the SEO Score. ![Step 2: Optimize Your Doc On Shopify](../../assets/content/body/2024/10/2-storeseo-betterdocs-for-shopify-integration.webp) ### **Basic SEO Analysis** Let us start with **Basic SEO Analysis**. Here, we have to ensure various criteria such as placing a unique focus keyword in the required places, maintaining an optimized length of the meta title and description and others. We will now show you how to do these one by one. ![Basic SEO Analysis](../../assets/content/body/2024/10/3-storeseo-betterdocs-for-shopify-integration.webp) #### **Focus keyword is unique** First, you need to [**choose the proper focus keyword**](/docs/what-is-a-focus-keyword). This is crucial because the doc is indexed on search engines based on this keyword. To put it simply, your potential readers or audience will have a higher likelihood of finding your doc if their search keyword and your selected focus keyword match. Therefore, you need to be very careful and do [**proper keyword research**](/keyword-research-for-shopify) when choosing the focus keyword. Here are a few things to consider: **Keyword relevance:** Make sure the keyword is relevant to your doc topic, business, audience and niche. Suppose, you are selling online streaming subscriptions through your store. Your focus keyword should be relevant to this service. Now, you select a high search volume keyword like ‘**leather bag**’ or ‘**formal shirt’**. These keywords are outside your service niche. So although it might have a higher search volume, it will not be a relevant option.  **Search volume:** Keywords with higher search volume have more searches every month. If you select keywords with higher search volume, you are more likely to face more competition. The more specific keywords you use, the more precisely you can reach your target audience. Now, if you choose a long focus keyword such as ‘**How to Buy a Subscription on Our Website**’, you have a unique keyword but there is less possibility for customers to use this entire keyword during a search. This is why it does not have a good search volume as shown in the image below. ![Focus keyword is unique 1](../../assets/content/body/2024/10/4-storeseo-betterdocs-for-shopify-integration.webp) Now, if you choose a short focus keyword like ‘**subscription**’, you will find it has a high search volume as shown in the image below. However, such popular keywords have too much traffic and will be a lot harder to rank and reach out to your audience. ![Focus keyword is unique 2](../../assets/content/body/2024/10/5-storeseo-betterdocs-for-shopify-integration.webp) On the other hand, if you select the focus keyword ‘**buy subscription**’, you have a good search volume (not too high) and there is a higher possibility to index without facing much competition. This is why this is the more appropriate focus keyword option in this particular case. ![Focus keyword is unique 3](../../assets/content/body/2024/10/6-storeseo-betterdocs-for-shopify-integration.webp) **Cost per click:** If you are running ads with particular keywords, the cost per click determines the approximate budget for the ad campaign. So, choosing keywords with a higher cost per click means you need to have a bigger campaign budget. Please note that it is an estimated value and you do not have to worry about it if you are not running any ads**.**  **Paid difficulty:** This represents the demand for keywords when used in paid ads. Higher paid difficulty indicates the keyword also has more competition in organic search. So, it is an important metric during keyword research. **Related keywords:** Remember to check the related keywords for a particular keyword. These related keywords will give you additional keyword suggestions. So, you might come across a better focus keyword option. Make sure the related keywords suit your product, business, audience and niche. Back to doc optimization on StoreSEO, place your preferred keyword in the ‘**Focus Keyword**’ field. Click on the ‘**Keyword Analytics**’ button to instantly get the **Search Volume**, **Cost Per Click**, **Paid Difficulty** and **related keywords** for your keyword. Here, we have selected ‘**buy subscription**’ as the focus keyword. It is recommended that different focus keywords be used for different docs. Please choose a keyword according to your business and the doc’s topic. Do **keyword research for Shopify** for this.
#### **Focus keyword is used in the introduction** Next, make sure the focus keyword is used in the introduction of the doc. To edit your doc, open the **BetterDocs for Shopify app** in your store. Then, click on the ‘**Docs**’ tab in the left side panel. Select the doc you want to edit from the list. Make sure to place the focus keyword in the introduction. As you can see in the image below, our doc’s introduction contains ‘**buy subscription**’ or our selected focus keyword.
#### **Content should be more than 200 words** Include more than 200 words in the doc’s content to further optimize the SEO score. Make sure that your doc’s content contains more than 200 words while maintaining relevance and meaningfulness as shown in the image below.
#### **Focus keyword is used 1-2% times of the doc content** Your focus keyword should be used 1-2% times of the doc content. But avoid keyword stuffing or placing the keyword forcefully or unnaturally. As you can see in the image below, we have placed our selected focus keyword to maintain the ideal frequency mentioned above.
### **Detailed SEO Analysis** We will now proceed to **Detailed SEO Analysis**. For this, we have to make sure to place the focus keyword in the image alt text, subheadings, meta description and URL and meet a few other requirements. Let us show you how to do all these. ![Detailed SEO Analysis](../../assets/content/body/2024/10/11-storeseo-betterdocs-for-shopify-integration.webp) #### **Focus keyword is used in image alt text** Make sure the focus keyword is used when adding the image alt text of the doc’s featured image. For this, navigate to StoreSEO doc optimization for your desired doc. Under the ‘**Images**’ section, you will find the ‘**Alt text’** field. Place your desired alt text containing the focus keyword in this field.  As you can see in the image below, the Alt text field ‘How to **Buy Subscription** on Our Website Doc Featured Image’ contains our selected focus keyword (marked bold). ![Focus keyword is used in image alt text](../../assets/content/body/2024/10/12-storeseo-betterdocs-for-shopify-integration.webp) #### **Focus keyword found in subheadings** Make sure your doc’s subheadings contain the focus keyword. You can add this to your subheading when editing your doc with BetterDocs for Shopify. As you can see in the image below, our subheading ‘Step 3: **Buy Subscription**’ contains our selected focus keyword (marked bold). ![Focus keyword found in subheadings](../../assets/content/body/2024/10/13-storeseo-betterdocs-for-shopify-integration.webp) #### **Focus keyword found in meta description** Your meta description should contain your selected focus keyword. From StoreSEO doc optimization, you can add the focus keyword in the meta description field. Our meta description is the following:  **Buy subscription** in our website by following the quick and easy steps in this doc. If you face any difficulties, contact our support team for a very fast solution. So, it contains our selected focus keyword (marked bold)
#### **Focus keyword used in the URL** Make sure that the Shopify doc URL contains the focus keyword. You can edit the URL from under the ‘**URL handle**’ field when optimizing your doc with StoreSEO. As you can see, the URL field contains ‘**buy subscription**’ or our selected focus keyword. **Note:** If your doc has already been indexed and you want to change its URL, it is recommended to [**create a URL redirection**](/how-to-create-and-manage-shopify-redirections). ![Focus keyword used in the URL](../../assets/content/body/2024/10/15-storeseo-betterdocs-for-shopify-integration.webp) #### **Meta description must be within 160 characters** Your [**doc’s meta description**](/seo-for-shopify-store-meta-descriptions) should be ideally within 160 characters. As you can see in the image below, our meta description is 158 characters long, which meets this requirement. ![Meta description must be within 160 characters](../../assets/content/body/2024/10/16-storeseo-betterdocs-for-shopify-integration.webp) #### **Focus keyword is used in the meta title** Make sure your meta title contains your selected focus keyword. As you can see in the image below, our meta title is the following: How to **Buy Subscription** on Our Website? So, the meta title contains our selected focus keyword (marked bold).  ![Focus keyword is used in the meta title](../../assets/content/body/2024/10/17-storeseo-betterdocs-for-shopify-integration.webp) #### **1 internal link found in doc content** Your doc content should contain 1 internal link. Internal link means linking to another doc, blog, page, product or collection of your Shopify store. To add this, proceed to edit your doc with BetterDocs for Shopify as described previously. Select the words you want to set as anchor text and click on the ‘**Insert/edit link**’ button. Enter the internal link in the ‘**URL**’ field. You also get to choose if the link will open in the same or new window upon clicking. You can set the ‘**Title**’ and ‘**Text to display**’ options as well. Click on the ‘**Save**’ button when you are done. Try to link to a page or doc of your store relevant to the doc’s topic or the product or service discussed in your doc. As you can see in the image below, we have internally linked our doc to our FAQs page.
#### **1 external link found in doc content** Keep one external link in the doc content. External link means linking to external pages outside your Shopify store. Include it when editing your doc as shown in the previous point.  Try to link to a high authority external link of a doc or page. Also, the external link should be relevant to your doc’s topic or the product or service discussed in your doc. As you can see in the image below, we have included 1 external link in our doc content.
#### **Add alt text to all images** Ensure that the alt text containing your selected focus keyword is added to the doc’s featured image. So, remember to place the image alt text in the ‘**Image alt text**’ field when optimizing your doc with StoreSEO.  The featured image of our doc contains the relevant alt text as shown in the image below: ![Add alt text to all images](../../assets/content/body/2024/10/20-storeseo-betterdocs-for-shopify-integration.webp) By following these steps, you can address every important SEO factor for the best possible optimization score for your doc. Once you are done with your doc optimization and happy with the overall SEO Score, click on the ‘**Save**’ button. ![Hit the save button once you are done with doc optimization](../../assets/content/body/2024/10/21-storeseo-betterdocs-for-shopify-integration.webp) ## **Step 3: Optimize Your Documentation Landing Page On Shopify** To optimize your documentation landing page, navigate to the ‘**Optimize SEO**’ tab and go to the ‘**Pages**’ option. In the list of your store’s pages, you will find the ‘**Doc Page**’ label beside your documentation landing page. Click on the ‘**Fix issue**’ button on the right side of this page to optimize it.
Now, you can start optimizing your documentation landing page. The SEO Score and Basic SEO Analysis on the right side will guide you during optimization. ![Step 3: Optimize Your Documentation Landing Page On Shopify 2](../../assets/content/body/2024/11/2-new-doc-storeseo-betterdocs-for-shopify-integration.webp) The Basic SEO Analysis will cover the following criteria: - Focus keyword used in page title - Page title within 70 characters - Focus keyword used in meta description - Meta description within 165 characters Once you are done with the optimization, hit the ‘**Save**’ button. This is how easily you can configure **SEO for Shopify docs** using StoreSEO. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Optimize Homepage In Shopify With StoreSEO? Source: https://storeseo.com/docs/optimize-homepage-in-shopify With StoreSEO app, you can now easily **optimize homepage** in Shopify to improve the SEO score. Generally, Shopify does not list the homepage as an individual page in its page list. Using this app, you can now list your Shopify homepage and optimize it for SEO accordingly.  ## How To Optimize Homepage In Shopify For SEO? Before you start, make sure to [**install and activate the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) in your Shopify store. Follow these step-by-step guidelines to find out how to optimize homepage in Shopify.  ### Step 1: Select StoreSEO From The Installed Apps First, you need to **install and activate** the StoreSEO app on your Shopify shop. Then go to the **‘Apps’** option and select the ‘**StoreSEO’** app from there.  ![Optimize homepage](../../assets/content/body/2022/06/storeseo-ai-seo-agent-storeseo-smart-ai-seo-zight-optimizer-ranking.-shopify-app-store-2025-07-16-at-1.18.09-pm.webp) ### **Step 2: Optimize Your Homepage**  Now, from the left sidebar, go to ‘Sales channels’ → ‘Online Store’ → ‘Preferences’. From the **‘Preferences’** page, you can change ‘**Homepage Title**’ and ‘**Homepage meta description**’. You can also update ‘**Social sharing image preview**’ here. Now, click on the ‘**Save**’ button to save the progress.  ![Optimize homepage](../../assets/content/body/2022/06/rickmortdev---preferences---shopify-2025-07-16-at-2.22.16-pm.webp) ### **Step 3: Sync Your Homepage In Shopify With StoreSEO** Now click on the ‘**Apps**’ option from the left-hand side. From your StoreSEO dashboard, click on the ‘Pages’ tab and choose the ‘Pages’ option from the dropdown menu. Click on the ‘Sync pages’ button. 
Now, you can see the ‘**Home page**‘ on the page. Hit the ‘**Fix**’ button beside the homepage. When you are inside the homepage, all the issues will be shown on the right-hand side with an overall SEO score. Based on the issues shown, you can make the necessary changes. Then, click on the ‘**Save changes**’ button.
This is how easily you can use StoreSEO to optimize your homepage in Shopify for SEO purposes. Need assistance? Feel free to reach out to our [**dedicated support team**](https://wpdeveloper.net/support/) for any type of query. --- # How To Optimize Shopify Collection Page With StoreSEO? Source: https://storeseo.com/docs/optimize-shopify-collection-page-with-storeseo [**StoreSEO**](/) helps optimize [**Shopify collection pages for SEO**](/docs/what-is-search-engine-optimization) so that they have more online visibility among your potential customers. StoreSEO makes sure everything is in order by showing an SEO score and all the issues that need to be fixed in order to bring a good SEO score for your Shopify collection page. This documentation will cover all these. Before you start, make sure you have added a Shopify collection page and [**installed StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Then, follow these step-by-step guidelines to find out how to optimize the collection page with StoreSEO. ## **Step 1: Navigate To StoreSEO Collection Optimization** Log in to your Shopify store. Then, from the search bar on the top or the ‘**Apps**’ option in the left sidebar, search for the StoreSEO app and click on it. From the StoreSEO dashboard, go to the ‘**Optimize SEO**’ tab. Now click on the ‘**Collections**’ option. Once your collection pages are synced properly, you will see your collection list on this page. Choose the one you want to optimize, and click on the ‘**Fix issue**’ button. 
## **Step 2: Optimize Your Collection Page On Shopify** Now, it is time to optimize the **Shopify collection page SEO**. As you can see in the image below, the SEO details of your selected collection are shown by StoreSEO. From the right side panel, you will also find **Basic SEO Analysis** and **Detailed SEO Analysis** to find the key factors that have to be addressed to improve the SEO Score. ![](../../assets/content/body/2024/03/image-56.webp) ### **Basic SEO Analysis** Let us start with **Basic SEO Analysis**. Here, we have to ensure various criteria such as using a unique focus keyword, placing it in the meta title and meet a few other requirements. We will now show you how to do these one by one. ![](../../assets/content/body/2024/03/image-57.webp) #### **Focus keyword is unique** First, you need to [**choose the proper focus keyword**](/docs/what-is-a-focus-keyword). This is crucial because the collection page is indexed on search engines based on this keyword. To put it simply, your potential customers will have a higher likelihood of finding your collection page if their search keyword and your selected focus keyword match. Therefore, you need to be very careful and do [**proper keyword research**](/keyword-research-for-shopify) when choosing the focus keyword. Here are a few things to consider: **Keyword relevance:** Make sure the keyword is relevant to your collection, product, business, audience and niche. Suppose, you are selling leather formal shoes. Your focus keyword should be relevant to this product. Now, you select a high search volume keyword containing the term ‘**leather bag**’ or ‘**formal shirt’**. These keywords are outside your product niche. So although it might have a higher search volume, it will not be a relevant option.  **Search volume:** Keywords with higher search volume have more searches every month. If you select keywords with higher search volume, you are more likely to face more competition. The more specific keywords you use, the more precisely you can reach your target audience. Now, if you choose a long focus keyword such as ‘**Water Resistant Synthetic Leather Formal Shoes**’, you have a unique keyword but there is less possibility for customers to use this entire keyword during a search. This is why it does not have a good search volume as shown in the image below. ![](../../assets/content/body/2024/03/image-58.webp) Now, if you choose a short focus keyword like ‘**Leather Shoes**’, you will find it has a high search volume as shown in the image below. However, such popular keywords have too much traffic and will be a lot harder to rank and reach out to your audience. ![](../../assets/content/body/2024/03/image-59.webp) On the other hand, if you select the focus keyword ‘**Synthetic Leather Shoes**’, you have a good search volume (not too high) and there is a higher possibility to index without facing much competition. This is why this is the more appropriate focus keyword option in this particular case. ![](../../assets/content/body/2024/03/image-60.webp) **Cost per click:** If you are running ads with particular keywords, the cost per click determines the approximate budget for the ad campaign. So, choosing keywords with a higher cost per click means you need to have a bigger campaign budget. Please note that it is an estimated value and you do not have to worry about it if you are not running any ads**.**  **Paid difficulty:** This represents the demand for keywords when used in paid ads. Higher paid difficulty indicates the keyword also has more competition in organic search. So, it is an important metric during keyword research. **Related keywords:** Remember to check the related keywords for a particular keyword. These related keywords will give you additional keyword suggestions. So, you might come across a better focus keyword option. Make sure the related keywords suit your product, business, audience and niche. Back to collection page optimization on StoreSEO, place your preferred keyword in the ‘**Focus Keyword**’ field. Click on the ‘**Keyword Analytics**’ button to instantly get the **Search Volume**, **Cost Per Click**, **Paid Difficulty** and **related keywords** for your keyword. Here, we have selected ‘**Synthetic Leather Shoes**’ as the focus keyword. It is recommended that different focus keywords be used for different collection pages. Please choose a keyword according to your business and the collection page. Do [**keyword research for Shopify**](/keyword-research-for-shopify) for this.
#### **Focus keyword is used in the meta title** Make sure your meta title contains your selected focus keyword. As you can see in the image below, our meta title is the following: **Synthetic Leather Shoes** – Special Summer Collection For Men & Women So, the meta title contains our selected focus keyword (marked bold). ![](../../assets/content/body/2024/03/image-61.webp) #### **Meta description is within 80-165 characters** Make sure your meta description is within 80-165 characters. As you can see in the image below, our meta description is the following: Explore our special summer collection of Synthetic Leather Shoes for men and women. Our high-quality shoes are not only comfortable but also come with stunning looks. So, the meta description is 161 characters long and so it meets the length requirement. ![](../../assets/content/body/2024/03/image-62.webp) #### **Collection Meta title is within 30-70 characters** Make sure your collection meta title is within 30-70 characters. As you can see in the image below, our meta title is the following: Synthetic Leather Shoes – Special Summer Collection For Men & Women So, the meta description is 67 characters long and so it meets the length requirement. ![](../../assets/content/body/2024/03/image-63.webp) ### **Detailed SEO Analysis** We will now proceed to **Detailed SEO Analysis**. For this, we have to make sure to place the focus keyword in the Collection title, meta description and URL. Let us show you how to do all these. ![](../../assets/content/body/2024/03/image-64.webp) #### **Focus keyword is used in the Collection title** Your collection title should contain the focus keyword. If the focus keyword is missing, you can add it when editing your collection page on Shopify. To do this, click on the ‘**Products**’ button in the left side panel on your Shopify dashboard. Then, click on the ‘**Collections**’ button below and you will find the list of all the collection pages of your store. Then, you can edit the collection title and include your selected focus keyword in the ‘**Title**’ field. Click on the ‘**Save**’ button once you are done. As you can see in the image below, our collection title contains ‘**Synthetic Leather Shoes**’ or our selected focus keyword.
#### **Focus keyword found in meta description** Make sure the meta description of your collection page contains the focus keyword. As you can see in the image below, the meta description of our collection page contains our selected focus keyword. ![](../../assets/content/body/2024/03/image-65.webp) #### **Focus keyword is used in the URL** Finally, make sure that the collection page URL contains the focus keyword. As you can see, the ‘**URL handle**’ field contains ‘**Synthetic Leather Shoes**’ or our selected focus keyword.  **Note:** If your collection page has already been indexed and you want to change its URL, it is recommended to [**create a URL redirection**](/how-to-create-and-manage-shopify-redirections). ![](../../assets/content/body/2024/03/image-66.webp) By following these steps, you can address every important SEO factor for the best possible optimization score for your collection page. If you need help to optimize your content for SEO with the power of AI, you can also use the [**StoreSEO AI Content Optimizer**](/docs/how-to-use-storeseo-ai-content-optimizer).  Once you are done with your collection page optimization and happy with the overall SEO Score, click on the ‘**Save**’ button. ![](../../assets/content/body/2024/03/image-67.webp) This is how easily you can **optimize Shopify collection pages** with StoreSEO. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Optimize Shopify Pages With StoreSEO? Source: https://storeseo.com/docs/how-to-optimize-shopify-pages [**StoreSEO**](/) helps [**optimize Shopify pages for SEO**](/docs/what-is-search-engine-optimization) so that they have more online visibility among your potential customers. StoreSEO makes sure everything is in order by showing an SEO score and all the issues that need to be fixed in order to bring a good SEO score for your Shopify page. This documentation will cover all these. Before you start, make sure you have added a Shopify page and [**installed StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify**](https://www.shopify.com/free-trial/3s?term=shopify&adid=566143522046&campaignid=15433369425&branded_enterprise=1&BOID=brand&utm_medium=cpc&utm_source=google&gad_source=1&gclid=Cj0KCQjwk6SwBhDPARIsAJ59GwcpTl5ifpGWLRPQ-e_SpErjVBbm7eMycfBziwtfQjiKsXX9UU68pRoaAj8GEALw_wcB&cmadid=516585705;cmadvertiserid=10730501;cmcampaignid=26990768;cmplacementid=324494758;cmcreativeid=163722649;cmsiteid=5500011) store. Then, follow these step-by-step guidelines to find out how to optimize the page with StoreSEO. ## **Step 1: Navigate To StoreSEO Page Optimization** Log in to your Shopify store. Then, from the search bar on the top or the ‘**Apps**’ option in the left sidebar, search for the StoreSEO app and click on it. From the StoreSEO dashboard, go to the ‘**Optimize SEO**’ tab. Now click on the ‘**Pages**’ option. Once your pages are synced properly, you will see your page list. Choose the one you want to optimize, and click on the ‘**Fix issue**’ button. 
## **Step 2: Optimize Your Page On Shopify** Now, it is time to optimize the [**Shopify page SEO**](/mastering-shopify-on-page-seo). As you can see in the image below, the SEO details of your selected page are shown by StoreSEO. From the right side panel, you will also find **Basic SEO Analysis** and **Detailed SEO Analysis** to find the key factors that have to be addressed to improve the SEO Score. ![Step 2: Optimize Your Page On Shopify](../../assets/content/body/2022/03/2-storeseo-page-optimization.webp) ### **Basic SEO Analysis** Let us start with **Basic SEO Analysis**. Here, we have to ensure various criteria such as placing a unique focus keyword in the introduction, maintaining an ideal focus keyword frequency and meeting a few other requirements. We will now show you how to do these one by one. ![Basic SEO Analysis](../../assets/content/body/2022/03/3-storeseo-page-optimization.webp) #### **Focus keyword is unique** First, you need to [**choose the proper focus keyword**](/docs/what-is-a-focus-keyword). This is crucial because the page is indexed on search engines based on this keyword. To put it simply, your potential visitors will have a higher likelihood of finding your page if their search keyword and your selected focus keyword match. Therefore, you need to be very careful and do [**proper keyword research**](/keyword-research-for-shopify) when choosing the focus keyword. Here are a few things to consider: **Keyword relevance:** Make sure the keyword is relevant to your page topic, business, audience and niche. Suppose, you are selling apple cider vinegar. Your focus keyword should be relevant to this product. Now, you select a high search volume keyword containing the term ‘**apple iphone**’ or ‘**apple juice’**. These keywords are outside your product niche. So although it might have a higher search volume, it will not be a relevant option.  **Search volume:** Keywords with higher search volume have more searches every month. If you select keywords with higher search volume, you are more likely to face more competition. The more specific keywords you use, the more precisely you can reach your target audience. Now, if you choose a long focus keyword such as ‘**original unfiltered organic apple cider vinegar**’, you have a unique keyword but there is less possibility for customers to use this entire keyword during a search. This is why it does not have a good search volume as shown in the image below. ![Focus keyword is unique 1](../../assets/content/body/2022/03/4-storeseo-page-optimization.webp) Now, if you choose a short focus keyword like ‘**apple cider vinegar**’, you will find it has a high search volume as shown in the image below. However, such popular keywords have too much traffic and will be a lot harder to rank and reach out to your audience. ![Focus keyword is unique 2](../../assets/content/body/2022/03/5-storeseo-page-optimization.webp) On the other hand, if you select the focus keyword ‘**Organic Apple Cider Vinegar**’, you have a good search volume (not too high) and there is a higher possibility to index without facing much competition. This is why this is the more appropriate focus keyword option in this particular case. ![Focus keyword is unique 3](../../assets/content/body/2022/03/6-storeseo-page-optimization.webp) **Cost per click:** If you are running ads with particular keywords, the cost per click determines the approximate budget for the ad campaign. So, choosing keywords with a higher cost per click means you need to have a bigger campaign budget. Please note that it is an estimated value and you do not have to worry about it if you are not running any ads**.**  **Paid difficulty:** This represents the demand for keywords when used in paid ads. Higher paid difficulty indicates the keyword also has more competition in organic search. So, it is an important metric during keyword research. **Related keywords:** Remember to check the related keywords for a particular keyword. These related keywords will give you additional keyword suggestions. So, you might come across a better focus keyword option. Make sure the related keywords suit your product, business, audience and niche. Back to page optimization on StoreSEO, place your preferred keyword in the ‘**Focus Keyword**’ field. Click on the ‘**Keyword Analytics**’ button to instantly get the **Search Volume**, **Cost Per Click**, **Paid Difficulty** and **related keywords** for your keyword. Here, we have selected ‘**Organic Apple Cider Vinegar**’ as the focus keyword. It is recommended that different focus keywords be used for different pages. Please choose a keyword according to your business and the page topic. Do [**keyword research for Shopify**](/keyword-research-for-shopify) for this.
#### **Focus keyword is used in the introduction** Next, make sure the focus keyword is used in the introduction of the Shopify page. To edit your page, navigate to Sales Channels in the left side panel of your Shopify dashboard. Then, select ‘**Online Store**’ and click on the ‘**Pages**’ option. Select the page you want to edit. Make sure to place the focus keyword in the introduction. As you can see in the image below, our Shopify page’s introduction contains ‘**Organic Apple Cider Vinegar**’ or our selected focus keyword.
#### **Content should be more than 300 words** Include more than 300 words in the page content to further optimize the SEO score. Make sure that your page content contains more than 300 words while maintaining relevance and meaningfulness as shown in the image below.
#### **Focus keyword is used 1-2% times of the page content** Your focus keyword should be used 1-2% times of the page content. But avoid keyword stuffing or placing the keyword forcefully or unnaturally. As you can see in the image below, we have placed our selected focus keyword to maintain the ideal frequency mentioned above.
### **Detailed SEO Analysis** We will now proceed to **Detailed SEO Analysis**. For this, we have to make sure to place the focus keyword in the image alt text, subheadings, meta description and URL and meet a few other requirements. Let us show you how to do all these. ![Detailed SEO Analysis](../../assets/content/body/2022/03/11-storeseo-page-optimization.webp) #### **Focus keyword is used in image alt text** Make sure the focus keyword is used when adding image alt text to your Shopify page’s images. To do this, when editing your Shopify page, double click on the image. Then, make sure that the focus keyword is included in the ‘**Image alt text**’ field. Click on the ‘**Edit image**’ button to save any changes. As you can see in the image below, the Image alt text field ‘**Organic Apple Cider Vinegar** From The Best Sources – The Story Behind Our Team’ contains our selected focus keyword (marked bold).
#### **Focus keyword found in subheadings** Make sure your page’s subheadings contain the focus keyword. You can add this to your subheading when editing your page from the Shopify dashboard as discussed previously. As you can see in the image below, our subheading ‘The Craft of Making **Organic Apple Cider Vinegar**’ contains our selected focus keyword (marked bold). ![Focus keyword found in subheadings](../../assets/content/body/2022/03/13-storeseo-page-optimization.webp) #### **Focus keyword found in meta description** Your meta description should contain your selected focus keyword. From StoreSEO page optimization, you can add the focus keyword in the meta description field. Our meta description is the following:  Explore the story behind our **Organic Apple Cider Vinegar**. Learn how our team ensures that you only get the best products from the purest sources at the best prices. So, it contains our selected focus keyword (marked bold)
#### **Focus keyword used in the URL** Make sure that the Shopify page URL contains the focus keyword. You can edit the URL from under the ‘**URL handle**’ field when optimizing your page with StoreSEO. As you can see, the URL field contains ‘**Organic Apple Cider Vinegar**’ or our selected focus keyword. **Note:** If your page has already been indexed and you want to change its URL, it is recommended to [**create a URL redirection**](/how-to-create-and-manage-shopify-redirections). ![Focus keyword used in the URL](../../assets/content/body/2022/03/15-storeseo-page-optimization.webp) #### **Meta description must be within 160 characters** Your [**page’s meta description**](/shopify-homepage-title-and-meta-descriptions) should be ideally within 160 characters. As you can see in the image below, our meta description is 157 characters long, which meets this requirement. ![Meta description must be within 160 characters](../../assets/content/body/2022/03/16-storeseo-page-optimization.webp) #### **Focus keyword is used in the meta title** Make sure your meta title contains your selected focus keyword. As you can see in the image below, our meta title is the following: **Organic Apple Cider Vinegar** From The Best Sources – Our Great Story So, the meta title contains our selected focus keyword (marked bold).  ![Focus keyword is used in the meta title](../../assets/content/body/2022/03/17-storeseo-page-optimization.webp) #### **1 internal link found in page content** Your page content should contain 1 internal link. Internal link means linking to another blog, page or product of your Shopify store. To add this, proceed to edit your page from the Shopify dashboard as described previously. Select the words you want to set as anchor text and click on the three-dot option in the ‘**Content**’ section. Next, select the ‘**Insert link**’ option. Enter the internal link in the designated field. You also get to choose if the link will open in the same or new window upon clicking. You can set a link title as well. Click on the ‘**Insert link**’ button when you are done. Try to link to a page or blog post of your store relevant to the topic or the product or service discussed on your page. As you can see in the image below, we have internally linked our example page to our FAQs page.
#### **1 external link found in page content with do-follow** Keep one external link in the page content with do-follow. External link means linking to external pages outside your Shopify store. Include it when editing your page as shown in the previous point.  Try to link to a high authority external link of a blog post or page. Also, the external link should be relevant to the topic or the product or service discussed on your page. As you can see in the image below, we have included 1 external link in our page content.
#### **Add alt text to all images** If you have multiple images on the page, ensure that the focus keyword is used in the alt text of all images. You have to do this when editing your page from Shopify dashboard as discussed before. Under the ‘**Content**’ section when editing a page, Double click on any image within the page content. You will then be able to add the alt text in the ‘**Image alt text**’ field.  The images used in our page contain relevant alt text as shown in the image below.
By following these steps, you can address every important SEO factor for the best possible optimization score for your page. If you need help to optimize your content for SEO with the power of AI, you can also use the [**StoreSEO AI Content Optimizer**](/docs/how-to-use-storeseo-ai-content-optimizer).  Once you are done with your page optimization and happy with the overall SEO Score, click on the ‘**Save**’ button. ![click on the save button when you are done optimizing the Shopify page](../../assets/content/body/2022/03/21-storeseo-page-optimization.webp) This is how easily you can **optimize Shopify pages** for SEO using StoreSEO. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Optimize Translated Shopify Products with StoreSEO? Source: https://storeseo.com/docs/optimize-translated-shopify-products [**StoreSEO**](https://apps.shopify.com/storeseo) offers Multilingual SEO Support for product optimization. So, you can now optimize translated Shopify products with StoreSEO. In this documentation, we will show you how to use StoreSEO to optimize your product pages translated into another language for SEO. **Note:** The Multilingual SEO Support feature is available in [StoreSEO’s Growth and Advanced plans](/#pricing) only. ## How to Optimize Translated Shopify Products Using StoreSEO Multilingual SEO Support? Let us find out how to optimize translated Shopify products for SEO using the StoreSEO app. We will cover every important step, from translating the products to optimization. Follow the step-by-step guidelines below to get started. ### Step 1: Translate Your Store’s Products Make sure you have added translations for your Shopify store’s products. Shopify offers the Translate & Adapt app for this purpose.  If you add a new language to your store, this app will be installed automatically. If you already have languages published, [**install the app from the Shopify app store**](https://apps.shopify.com/translate-and-adapt) or in **‘Settings’ → ‘Languages’** from your Shopify dashboard. Then, you can manually or automatically translate your product pages’ content using the app as you want.
How to Optimize Translated Shopify Products with StoreSEO?
**Note:** To learn more about managing languages in Shopify, read [**this documentation**](https://help.shopify.com/en/manual/international/languages/manage-languages#manage-markets-languages). And to find out how translations work using the Translate & Adapt app, check out [**this documentation**](https://help.shopify.com/en/manual/international/languages/translate-adapt-app). ### Step 2: Enable Multilingual SEO Support in StoreSEO Open the StoreSEO app and click on the ‘**View more’** button in the left side panel. Then, go to **‘Settings’ → ‘Multilingual SEO’.** Click on the **‘Turn on’** button to enable StoreSEO’s Multilingual SEO Support as shown below.  Under the **‘Published Language’** section, hit the **‘Sync’** button beside the published language into which your products have been translated, for example, ‘German’ here. This will start the synchronization process. You will get a message once the sync is complete.
How to Optimize Translated Shopify Products with StoreSEO?
### Step 3: Select Your Preferred Language For All Products From the left side panel of the StoreSEO app, navigate to **‘Optimize SEO’ → ‘Products’**. Select the language in which you want to optimize products from the language selector dropdown at the top, as shown below. 
How to Optimize Translated Shopify Products with StoreSEO?
Then, choose the translated product from the list that you want to optimize for SEO by clicking on the **‘Fix issue’** button on the right side under the **‘Action’** column. Now, you can optimize the translated product for SEO following the ‘**SEO Score’, ‘Basic SEO Analysis’ and ‘Detailed SEO Analysis’** information provided by StoreSEO on the right side.  Alternatively, you can also access the language selector dropdown from the optimization screen of any single product. To do so, navigate to the product by clicking on the **‘Fix Issue’** icon and then choose your preferred language from the Language Selector dropdown menu, as shown below. **Note:** To learn more about optimizing Shopify products using StoreSEO, follow [**this documentation**](/docs/optimize-seo-for-shopify-products).
How to Optimize Translated Shopify Products with StoreSEO?
This is how easily you can optimize translated Shopify products for SEO with StoreSEO Multilingual SEO Support. Need any help? Contact [**our dedicated support team**](/contact) if you have any queries. --- # How to Set up Your Store for AI Brand Visibility And Get AI Insights Using StoreSEO? Source: https://storeseo.com/docs/set-up-your-store-for-ai-brand-visibility AI answer engines like ChatGPT, Gemini, Perplexity, and Claude are becoming a primary research channel for shoppers. Traditional rank trackers cannot tell you whether your brand is being mentioned or recommended inside AI-generated answers. Using StoreSEO’s latest **‘AI Insights’** feature, you can set up your store for **AI Brand Visibility** that allows you to configure your brand to track how ChatGPT, Claude, Gemini, and Perplexity see you. Follow the step-by-step guide below. ### **What AI Brand Visibility Measures** Once configured, you can track four key metrics: - **AI Visibility Index:** A composite score from 0–100 showing your overall AI search health - **Mention Rate:** How often your brand appears in AI-generated answers - **Share of Voice (SoV):** How your brand visibility compares against your top competitors on a leaderboard - **Sentiment:** Whether AI platforms mention your brand in a positive, neutral, or negative context ## How to Set Up Your Store for AI Brand Visibility Using StoreSEO Follow the step-by-step guide to learn how to set up your store for **AI Brand Visibilit**y with StoreSEO and make your store appear in AI search answers. ### **Step 1: Set Up Your Brand** From the dashboard, go to **‘AI Insights’.** Here you can configure your brand to track how ChatGPT, Claude, Gemini, and Perplexity see you. First, click on **‘Set Up Your Brand’**  and start the setup wizard.  Here you need to add your brand name, store domain, and location, etc. Confirm the details, add a brief brand description if needed, and click **Continue**. ![AI Brand Visibility in StoreSEO](../../assets/content/body/2026/06/ai-brand-visibility-1-scaled.webp) ### **Step 2: Add Your Competitors** Enter up to 5 competitors manually, or import them directly from your Keyword Rank Tracker.
If you skip this step, StoreSEO will automatically detect the top brands appearing in AI answers and populate a discovered competitor set for your dashboard. ### **Step 3: Generate Your AI Queries & Review**  Now, in the next step, a popup will appear asking you to generate the AI queries for those you want to appear in the AI answers. Simply click on ‘Generate Queries’ to proceed. ![AI Brand Visibility in StoreSEO](../../assets/content/body/2026/06/image-8.webp) StoreSEO automatically generates up to 12 targeted queries based on your store data. These are grouped into four categories: - Branded queries - Category queries - Problem-solving queries - Comparison queries You can review, edit, or regenerate any of these queries before running the analysis. When you are satisfied with the query set, click **Continue**. ![AI Brand Visibility in StoreSEO](../../assets/content/body/2026/06/ai-brand-visibility-5-scaled.webp) ### **Step 4: Select Platforms and Run the Analysis** Select the AI platforms you want to query. StoreSEO supports **ChatGPT, Gemini, Perplexity, and Claude**. You will see a live estimate of the AI credits required before you proceed. When ready, click **Run Analysis**. StoreSEO will query each selected platform and collect the results automatically.
If your score needs improvement, use the **Remediation Panel** on the dashboard to jump directly into StoreSEO’s optimization tools, including LLMs.txt generation, AI Snippets, and schema markup. StoreSEO will run the analysis automatically in the background and send you an email when your latest results are ready. ![AI Brand Visibility in StoreSEO](../../assets/content/body/2026/06/image-6.webp) Once the analysis is complete, your dashboard displays: - **AI Visibility Index**: Your overall score across all platforms - **Competitor Leaderboard**: Your Share of Voice compared to the competitors you added - **Trend Charts**: Your score over time across multiple runs ![AI Brand Visibility in StoreSEO](../../assets/content/body/2026/06/image-5.webp) This is how you can set up and use AI Brand Visibility in StoreSEO to monitor and improve how your **[Shopify](https://www.shopify.com/)** brand appears across AI search platforms. Need assistance? Reach out to the [StoreSEO support team](/contact) for any questions. --- # How to Set Up Your StoreSEO Keyword Rank Tracker Source: https://storeseo.com/docs/how-to-set-up-your-storeseo-keyword-rank-tracker **Keyword Rank Tracker** is a powerful new StoreSEO feature that allows you to monitor exactly where your Shopify store stands in search engine results. Instead of guessing your progress, you can track the real-time performance of your most valuable keywords and see how your optimization efforts are paying off. This documentation walks you through the simple steps to set up tracking for your target keywords, interpret your ranking reports, and use data-driven insights to climb higher in search results.
Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify** **store**](https://apps.shopify.com/storeseo?search_id=aab4df77-20c3-4fcd-a46e-2ad302d1e31f&shallow_install_type=search&surface_detail=storeseo&surface_inter_position=1&surface_intra_position=1&surface_type=search). Then, follow these step-by-step guidelines to start using the keyword rank tracker feature for your Shopify store using StoreSEO. ### Step 1: Access the Keyword Rank Tracker From your Shopify admin dashboard, open the **StoreSEO** app. Locate and select **Keyword Rank Tracker** from the features panel on the left sidebar. ![keyword rank tracker](../../assets/content/body/2026/03/image-9.webp) ### Step 2: Choose Your Tracking Action The dashboard provides two main configuration options: **Add Competitors** and **Add Keywords**. Start by clicking the **Add Competitors** button to establish a baseline for keyword ranking comparison. ![](../../assets/content/body/2026/03/image-10.webp) ### Step 3: Configure Your Competitors Enter the valid URLs of up to five competitors within your niche into the provided fields. Use this section to delete or adjust your list as your keyword strategy evolves. Click the **Save** button once your list is finalized to lock in these comparisons.
### Step 4: Add and Configure Keywords Click the **Add Keywords** button to input the specific search terms you wish to monitor. ![StoreSEO Keyword Rank Tracker](../../assets/content/body/2026/03/image-11.webp) Use the dropdown menus to select the specific location, device type, search engine, and language that align with your target audience. Click the **Add** button to move these terms into your tracking list.
### Step 5: Utilize Keyword Suggestions Look to the keyword suggestion box on the right side of the screen to discover relevant terms based on your niche. Click on the Add button.
Review the AI-powered suggestions, which highlight high-performing terms already driving traffic in your industry. Click the **Add** button next to any suggested keyword to include it in your list, then click **Save** to confirm your selection.
### Step 6: Analyze Performance Data Return to the main dashboard to review the performance of your added keywords. Examine the status of each term, paying close attention to **SERP features** such as image rankings, featured snippets, and AI overviews. Compare your **current ranking** against the competitors you added earlier to identify where you are winning and where you need to improve. ![StoreSEO Keyword Rank Tracker](../../assets/content/body/2026/03/image-12-2.webp) ### Step 7: Analyze Detailed Keyword Analytics For individual keyword here’s the performance graph and detailed daily ranking cards. By tracking visibility across five critical search verticals: **Overall, Links, Video, Shopping Results, and AI Search**, the page provides you with a clear understanding of exactly where their traffic is originating and how they stack up against specific competitors like ![StoreSEO Keyword Rank Tracker](../../assets/content/body/2026/03/image-13-2.webp) This is how easily you can utilize the keyword rank tracker feature for your Shopify store using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support). --- # How to Show FAQs on Your Storefront Using StoreSEO Source: https://storeseo.com/docs/how-to-show-faqs-on-your-storefront Displaying FAQs on your storefront with the customization feature allows you to showcase product-specific questions and answers directly on your Shopify product pages. Once you create FAQs for your products in StoreSEO, they are presented in a clean, well-organized section that is easily visible to customers. This helps shoppers find the information they need quickly, make confident purchase decisions, and reduces the need to contact support. This documentation walks you through every step to show FAQs on your storefront, from the StoreSEO Dashboard all the way to your live storefront. ## Step 1: Activate the Product FAQs Block from Dashboard To get started, open the **StoreSEO app** from your Shopify admin. Once inside, navigate to the **Dashboard** and scroll down to the **Integrations & Tracking** section. Here you will find a card labeled **Product FAQs Block**.  Click the **Activate** button on this card. This action registers the StoreSEO FAQ block with your Shopify theme, making it available for you to add to any product page template. You will not see FAQs on your store until this step is completed.
## Step 2: Open the Shopify Theme Editor and Locate Your Product Page After activating the block, inside the Theme Editor, use the page selector at the top to navigate to a **product page**. At this stage, if no FAQ block has been added yet, you will see a message in the product page preview that reads **“No FAQs to Display”, “You need to add FAQs for this product to display in the storefront.”** This is expected and confirms that the block needs to be added to the template, Click on the **Generate FAQ** button. ![Show FAQs on Your Storefront](../../assets/content/body/2026/04/image-33.webp) To continue, click ‘**Ok**’ on the popup notification. This is necessary because the link will launch in a separate window, and granting permission is required for the process to move forward. ![Show FAQs on Your Storefront](../../assets/content/body/2026/04/image-10-2.webp) **Note**: Be sure to click the “**Save**” button to ensure your changes are updated on the page. ![Show FAQs on Your Storefront](../../assets/content/body/2026/04/image-34.webp) ## Step 3: Add the StoreSEO Product FAQs Block to Your Product Page In the left-hand panel of the Theme Editor, scroll down to the **Apps** section, which is located near the bottom of the template sections list. Click **Add block** and from the list of available blocks, select **StoreSEO Product FAQs**.  Once added, the block will appear in the left panel under Apps, and the product page preview in the center will immediately display a “**Frequently Asked Questions**” section with the FAQ questions and answers that were created for that product in StoreSEO. If no FAQs have been added to the product yet in StoreSEO, the section will remain empty. Make sure to create FAQs under the “**FAQ Schema**” tab of the product inside the StoreSEO app first.
## Step 4: Verify Your FAQs Are Displaying on the Storefront Once the **StoreSEO Product FAQs** block has been successfully added to your product page template, you will see the **Frequently Asked Questions** section appear in the live preview of the Theme Editor, highlighted with a blue border.  The section will show all FAQ questions associated with the selected product, confirming that the block is working correctly. At this point, click the “**Save**” button at the top right of the Theme Editor to publish the changes to your live store. Your customers will now be able to see the FAQs section directly on the product page every time they visit. ![Show FAQs on Your Storefront](../../assets/content/body/2026/04/image-4-1.webp) ## Step 5: Customize the FAQ Display (Optional) With the FAQ block selected in the left panel, the right-hand side of the Theme Editor displays a range of styling and layout options that allow you to tailor the appearance of the FAQ section to match your store’s branding. The “**Layout**” setting lets you choose between “**Accordion**“, “**List**“, or “**Plain list**” display formats. The Accordion layout is recommended for stores with many FAQs, as it keeps the page clean and lets customers expand only the questions they are interested in. 
You can also toggle “**Expand first item by default**” to have the first FAQ open automatically when the page loads and choose between a **Plus** or **Arrow** icon as the trigger icon. Under the **Heading** section, you can edit the section title and adjust its alignment, font size, color, and heading tag (H2 or H3). The **Questions** and **Answers** settings allow you to control the font size, font weight, and color for the question and answer text, respectively. When you are satisfied with the styling, click **Save** to apply all changes. ## Important Tips and Troubleshooting on How to Show FAQs on Your Storefront If the FAQ section is not appearing on your product page after following the steps above, the most common reason is that FAQs have not yet been created for that product inside StoreSEO. To resolve this, go back to the **StoreSEO app**, open the relevant product, navigate to the **FAQ Schema** tab, and create your FAQ questions and answers there before returning to the Theme Editor. It is also important to note that the **StoreSEO Product FAQs** block is added to a product page **template**, which means it will apply to all products that use that template. Each product will display its own unique set of FAQs as long as those FAQs are created individually within StoreSEO. If a product has no FAQs assigned, the block will remain empty for that product even though the block exists in the template. If you see a popup saying **“Override recent theme customizations”** when saving, click “**Override and save changes”** to confirm and apply your latest settings. This message may appear if another user or session made changes to the theme editor around the same time. This is how easily you can use the FAQ schema on you storefront and customization feature for your Shopify store using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support). --- # How To Sync Pages In Shopify with StoreSEO? Source: https://storeseo.com/docs/how-to-sync-pages-in-shopify With [**Store SEO**](/), you can optimize the pages of your **Shopify store** for better SEO ranking in search engine results. You can examine your page’s SEO score as well. Let’s learn how you can **sync pages** in Shopify with **StoreSEO.** **_Note:_** _Before getting started, make sure that you have the_ [**_StoreSEO app installed_**](/docs/how-to-install-storeseo-app-on-shopify) _on your Shopify store._ ## Sync Pages In Shopify with Store SEO First, go to your Shopify store and open the **StoreSEO** app inside it. From the [**Shopify Dashboard**](https://www.shopify.com/), go to **Apps → Store SEO.**  ![sync pages](../../assets/content/body/2022/03/image-2-2.webp) You will find yourself in the **Store SEO** dashboard. Now, move to the ‘**Pages’** tab and click on ‘**Sync Pages’**. You will get a pop-up to confirm the process. Click on ‘**Confirm**’.
This feature will make sure that all of your pages can be optimized for SEO. When more pages are added to your store, you can use the ‘**Sync’** button to make sure that all the pages have been listed here for optimization. This is how you can easily **sync Shopify pages with the Store SEO app**. Getting stuck? Feel free to communicate with our [**dedicated support team**](/contact) for further assistance. --- # How To Sync Shopify Products In StoreSEO Source: https://storeseo.com/docs/sync-storeseo-products [**StoreSEO app**](/) lets you easily sync products in your Shopify store to keep all your products up to date on the app. If you add more products to your store, you can use the **‘Sync’** button to make sure all your products are SEO optimized.  ## How To Sync Shopify Products In StoreSEO Before you get started, you need to [**install and activate the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your Shopify shop. Then, follow these steps to easily sync all Shopify products with one click in StoreSEO. Go to the **‘Apps’** option from your dashboard and select the ‘**StoreSEO’** app. Click on the ‘**Optimize SEO’** tab from the side panel and choose **‘Products’**, where all your Shopify products are listed with details. Then, click on the ‘Sync Products’ button on the top right side of your dashboard screen as shown below. ![How To Sync Shopify Products In StoreSEO](../../assets/content/body/2021/11/image-17.webp) This is how you can easily sync Shopify Store products in your StoreSEO app. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Track Orders & Revenue by Traffic Channel in Shopify Using StoreSEO? Source: https://storeseo.com/docs/how-to-track-orders-revenue-using-storeseo **[StoreSEO](/)** is the AI SEO Agentic App for Shopify that helps your store get discovered on Google, Bing and inside AI answer engines like ChatGPT, Gemini, Perplexity and Claude. With the newly released **‘Order & Revenue’** feature, StoreSEO now closes the loop between discovery and sales by showing you the money behind every visit your store receives. Knowing that AI assistants send you traffic is only half the story. What actually matters is whether those visitors buy. The **Order & Revenue** report inside **AI Insights** attributes orders, revenue, conversion rate and average order value to each channel that sends you visitors, so you can see exactly which discovery source is driving real sales and which one is only driving clicks. This documentation walks you through enabling the tracking pixel, opening the Order & Revenue report, reading each metric correctly and turning those numbers into decisions for your **[Shopify](https://www.shopify.com/)** store. ## **How to Track Orders & Revenue by Channel Using StoreSEO** Make sure you have **[installed the StoreSEO app](/docs/how-to-install-storeseo-app-on-shopify)** on your Shopify store. Then follow the step-by-step guidelines below to enable order tracking and read your revenue data channel by channel. ### **Step 1: Confirm Web Pixel Is Enabled on Your Dashboard** Order and revenue attribution depends on the StoreSEO **Web Pixel**, which records the checkout events that let StoreSEO connect a sale back to the channel that referred the shopper. Without it, your dashboard will show sessions but no orders or revenue. From your Shopify admin, open the **StoreSEO** app and stay on the **Dashboard**. Scroll down to the setup checklist and look for the **‘Web Pixel Enabled’** item. When it shows a checkmark with the green **Active** badge, tracking is live and StoreSEO has started collecting order data for your store. ![Confirm Web Pixel Is Enabled on Your Dashboard](../../assets/content/body/2026/08/image-3.webp) If the item is not marked as Active yet, enable it. Attribution only begins from the moment the pixel goes live, so orders placed before this step will not appear in your report. ### **Step 2: Open the ‘Orders & Revenue’ Tab From AI Insights** All of your channel performance data lives inside the AI Insights section of StoreSEO. From the left menu of your Shopify admin, click ‘**AI Insights’** under the StoreSEO app. On the AI Insights screen, you will find four tabs: **Overview, AI Traffic, Orders & Revenue** and **Brand Visibility**. Click the **‘Orders & Revenue’** tab to load the report. ![Navigate to AI Insights and open the Orders & Revenue tab to view your headline metrics.](../../assets/content/body/2026/08/image-2.webp) The top of the report shows your four headline cards for the selected period: **Orders, Revenue, Conversion rate** and **Avg. order value**. Use the date range selector on the right to switch between periods such as the last 30 days and click **‘Run analysis’** to refresh the data on demand.  ![click Run analysis to refresh the data on demand. ](../../assets/content/body/2026/08/image.webp) The **‘Manage settings’** button lets you adjust how your tracking and analysis are configured. **_Note:_** _If you are a free user, you can only check Orders, Revenue, Conversion Rate, and Avg. Order Value as shown. You need_ **_Pro access_** _to view the rest of the advanced details._ ### **Step 3: Analyze Channel Performance in Detail** Scroll down to see channel-level details and discover which sources perform best. **The report has three sections:** - **Share of sessions vs. revenue:** A side-by-side comparison that shows where a channel punches above its traffic weight. A channel with a small share of sessions but a large share of revenue is one worth investing in. Note that Paid and Direct carry no tracked sessions here. - **Channel breakdown:** A full table listing AI Search, Organic Search, Paid Search and Direct against Sessions, Orders, Revenue, Conv. rate, AOV and Rev. share, so you can compare every channel on the same terms. - **Sessions vs. revenue over time:** A trend chart that plots visits against earnings across the selected period, making it easy to spot the days when traffic and sales moved together or drifted apart. ![Review share of sessions vs. revenue, the channel breakdown table and the sessions vs. revenue trend chart.](../../assets/content/body/2026/08/image-1.webp) ## **Channels StoreSEO Attributes** Every visitor who lands on your store is grouped into one of four channels so that revenue can be attributed cleanly. Understanding what each label covers keeps your reading of the report accurate: - **AI Search:** Visitors referred by ChatGPT, Gemini, Perplexity, Claude and other AI assistants that cited or recommended your store. - **Organic Search:** Visitors who found you through unpaid results on search engines such as Google and Bing. - **Paid Search:** Visitors arriving from paid search campaigns pointing to your store. - **Direct:** Visitors who reached your store without a referrer, for example by typing your URL or using a bookmark. If you need any assistance, feel free to contact **[our support team](/support)**. --- # How To Upgrade Your StoreSEO Plan Source: https://storeseo.com/docs/upgrade-your-storeseo-plan You can take your Shopify experience to the next level if you [**upgrade your StoreSEO plan**](/#pricing) to get access to powerful features and benefits. Follow these steps to upgrade your StoreSEO plan for Shopify easily.  ## **How To Upgrade StoreSEO Plan** After you have [**installed the StoreSEO**](/docs/how-to-install-storeseo-app-on-shopify) app, you can upgrade **StoreSEO** with our subscription plans. Navigate to the StoreSEO app after opening your Shopify admin dashboard before going to the next steps. ### **Step 1: Navigate To The ‘Subscription Plans’ Tab** Click on the ‘**Subscription Plans**’ tab from the navigation bar. You will see all StoreSEO plans in one place. StoreSEO is currently offering 4 types of subscription plans.  
### Step 2: Select the Payment Method and Upgrade Your StoreSEO Plan After clicking the ‘**Subscribe Now**’ button, it will bring you to the next page. From here, select the payment method and click on the ‘Approve’ button to upgrade your StoreSEO plan. It will instantly upgrade your StoreSEO plan. ![upgrade your StoreSEO plan](../../assets/content/body/2021/11/image-4.webp) This is how easily you can upgrade your StoreSEO plan and enjoy the premium features. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Use Google Search Console Insights From StoreSEO for Your Shopify Store Source: https://storeseo.com/docs/how-to-use-google-search-console-insights Insight is a powerful new StoreSEO feature that brings Google Search Console insights to your Shopify store. It shows how shoppers find you, which pages perform best, and where to focus next to grow organic traffic. With clear metrics and plain-language takeaways, Insight highlights top queries, top pages and posts, rising content, and key traffic sources in one place. Use it to quickly understand what is working, spot new opportunities, and improve visibility, engagement, and rankings.
This documentation walks you through the simple steps to use Insight, interpret the reports, and turn recommendations into actions that fit your brand and eCommerce goals. Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your **Shopify** **store**. Then, follow these step-by-step guidelines to find out organic search insight for your Shopify store using StoreSEO. ### Step 1: Connect to Google From the StoreSEO app, go to Analytics. Here you’ll be asked to connect to Google to get the insights on your StoreSEO dashboard. Now, click on the ‘**Connect** **to** **Google**’ button. ![Google Search Console Insight](../../assets/content/body/2025/09/1.webp) ### **Step 2: Sign In To Your Google Account** Complete the first three steps during the integration to get the insight feature enabled on your store. First, navigate to **StoreSEO Settings** →  **Google Integrations**. Now, you need to sign in to Google by hitting the **‘Sign In Now’** button. ![Google Search Console Insight](../../assets/content/body/2025/09/2.webp) After clicking the **‘Sign In Now’** button, a window will open where you need to select your preferred Google account, which you want to connect with your store. Log in to your Gmail account, and then the authentication will be done immediately. ![Google Search Console Insight](../../assets/content/body/2025/09/3.webp) ### **Step 3: Verify Site With Google** Once the authentication is done, you need to verify your site with Google. Click on the **‘Verify Now’** button to verify your [**Shopify**](https://www.shopify.com/) store website. ![Google Search Console Insight](../../assets/content/body/2025/09/4.webp) Now, you will find a window where you have to log in to your Gmail account. Log in to your Gmail, then click on **‘Continue’**. Once you have finished these steps, your site will be verified automatically. ![Google Search Console Insight](../../assets/content/body/2025/09/5.webp) ### **Step 4: Link To Google Search Console** After verifying your site with Google, you can easily link StoreSEO to Google Search Console.  ![Google Search Console Insight](../../assets/content/body/2025/09/6.webp) Just click on the **‘Link Now’** button and wait for the next window to come up. ![Google Search Console Insight](../../assets/content/body/2025/09/7.webp) ### Step 5: Open Analytics To Get Google Search Console Insight Now, click Analytics on the left menu from your StoreSEO app to reach the Insights screen for the organic data metrics.
### Step 6: Check the overview The top of the page shows two big cards with trend lines: **Clicks** and **Impressions**. Use these two cards to spot quick trends, like drops or spikes during the selected dates. You can set the timeline accordingly. **Clicks** means people clicked a Google result and landed on the site, and the card shows the count, a small graph, and how it changed from the last period.  **Impressions** means the site showed up in Google results, even if no one clicked, and this card also shows the count, a small graph, and the change. ![Google Search Console Insight](../../assets/content/body/2025/09/10.webp) ### Step 7: Review Content Scroll to “**Your** **Content**” to see the main pages that got clicks in this time range, with simple tags like **Top**, **Trending** **up**, and **Trending** **down**. ![Google Search Console Insight](../../assets/content/body/2025/09/11.webp) Each row shows the page title, its link, total clicks, and the percent change so it’s easy to see winners and pages that slipped. Click “**View** **more**” to expand the list and pick pages to update or promote based on the movement shown here. ### Step 8: Check queries  Go to “**Queries** **leading** **to your site**” to see search terms people used, with click counts and whether each term is moving up or down.
Look at “**Top** **countries**” to learn where clicks are coming from; in the example, the United States leads, followed by Japan, India, Bangladesh, and Italy.  See “**Other** **traffic** **sources**” for additional sources, such as Image Search, so performance can be viewed in context, not just from standard web results. ![Google Search Console Insight](../../assets/content/body/2025/09/13.webp) This is how easily you can utilize the Insight feature for your Shopify store using StoreSEO. If you need any assistance, feel free to contact [**our support team**](/support). --- # How To Use StoreSEO AI Content Optimizer Source: https://storeseo.com/docs/how-to-use-storeseo-ai-content-optimizer **[StoreSEO](https://apps.shopify.com/storeseo)** has an AI Content Optimizer add-on feature that can help you optimize your content for SEO. From now on, you can optimize your store for better search engine ranking with just a single click. All you need is to enter your focus keyword and this feature will optimize your Meta Title, Meta Description, and Tags accordingly. 
## **How To Use AI Content Optimizer For Shopify** First of all, [**install StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) in your Shopify store before following the guidelines below.  ### **Step 1: Buy Credits For AI Content Optimizer** From the StoreSEO dashboard, click on the ‘**AI Content Optimizer**’ tab where you will see all credit plans. Now, select your preferred plan and click on the ‘**Confirm**’ button. Afterward, approve the payment to buy credits for the AI Content Optimizer. 
### **Step 2: Optimize Your Product Page**  Go to the product that you want to optimize by navigating to ‘**Apps**’ → ‘**StoreSEO’** → ‘**Optimize SEO’** → ‘**Products**’ → Click the ‘**Fix**’ button. 
Afterward, provide your focus keyword and click on the ‘**Generate**’ button. Your Meta Title, Meta Description, and Tags will be automatically optimized instantly. Click on the ‘**Keep**’ button to add the AI-generated content or you click on the ‘**Re-generate**’ button to rewrite the content automatically.  You can also click on the ‘**Revert**’ button to go back to the previous version or click on the ‘**Revert all**’ button to revert all changes.  Once you are satisfied with the result click on the ‘**Save**’ button.
### **Step 3: Optimize Your Collection Pages** Go to the page that you want to optimize by navigating to ‘**Apps**’ → ‘**StoreSEO’** → ‘**Optimize SEO’** → ‘**Collections**’ → Click the ‘**Fix**’ button.
Similarly, provide your focus keyword and just click on the ‘**Generate**’ button. Your Meta Title and Meta Description will be automatically optimized instantly. Click on the ‘**Keep**’ button to add the AI-generated content or you click on the ‘**Re-generate**’ button to rewrite the content automatically as previously shown and save it.  
### **Step 3: Optimize Your Pages** Go to the page that you want to optimize by navigating to ‘**Apps**’ → ‘**StoreSEO’** → ‘**Optimize SEO’** → ‘**Pages**’ → Click the ‘**Fix**’ button.
Similarly, provide your focus keyword and just click on the ‘**Generate**’ button. Your Meta Title and Meta Description will be automatically optimized instantly. Click on the ‘**Keep**’ button to add the AI-generated content or you click on the ‘**Re-generate**’ button to rewrite the content automatically as previously shown and save it.  
### **Step 4: Optimize Your Blog Posts** Go to the page that you want to optimize by navigating to ‘**Apps**’ → ‘**StoreSEO’** → ‘**Optimize SEO’** → ‘**Blog Posts**’ → Click the ‘**Fix**’ button.
Similarly, provide your focus keyword and just click on the ‘**Generate**’ button. Your Meta Title and Meta Description will be automatically optimized instantly. Click on the ‘**Keep**’ button to add the AI-generated content or you click on the ‘**Re-generate**’ button to rewrite the content automatically as previously shown.   Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How To Use StoreSEO Image Optimizer On Shopify Source: https://storeseo.com/docs/how-to-use-storeseo-image-optimizer-on-shopify StoreSEO comes with an advanced Image Optimizer add-on to help optimize and resize your store image without any hassle. Follow this step-by-step guideline to use the image optimizer feature for your store.  Note: first of all, you need to enable the Image Optimizer add-on on your store to get this feature. Follow [**this documentation**](/docs/how-to-add-image-optimizer-to-your-subscription-plan) to add the Image Optimizer to your subscription plan before following this guideline below. 
## **Step 1: Set Up Default Image Optimizer Settings**  StoreSEO image optimizer allows you to set your default image optimization settings. To configure the default settings, go to the ‘**Settings**’ tab from your StoreSEO dashboard. Now, navigate to ‘**Image Optimizer**’ where you will find all the options to set the default optimization settings. 
Now, choose the default settings for Image Compression, Image Format, and Image Resizer as you want. Afterward, click on the ‘**Save**’ button. 
## **Step 2: Navigation To The ‘Image Optimizer’ Tab** From the StoreSEO dashboard, click on the ‘**Image Optimizer’** tab. Now, go to the image you want to optimize and then click on the **‘Optimize**’ button. Instantly your image will be optimized according to your default optimization settings.
Here, to compare the changes, click on the ‘**Compare**‘ button. If you want to reoptimize the image, adjust the settings as you need and hit the ‘**Re-optimize’** button.
Now, if you want to revert the image optimization, just click the ‘**Restore**’ button. It will instantly change back to the original image.  **_Note:_** _After optimizing an image with StoreSEO, if you replace this image and click the ‘_**_Restore_**_‘ button here, StoreSEO will restore the previously replaced image that you optimized and your newly added image will be deleted from your Shopify store._
## **Step 3: Enable Auto Image Optimization** From the ‘**Image Optimizer**’ in sidebar click on the ‘**Manage Settings**’ button. A default settings tab with ‘**Auto Image Optimization**’ option will appear.
Now, click on the ‘**Turn on**’ button from the ‘**Auto Image Optimization**’ section. This will instantly enable auto image optimization for your Shopify store. _To learn more, follow this_ [**_documentation_**](/docs/how-to-enable-auto-image-optimization-on-shopify)_._
This is how easily you can use StoreSEO Image Optimizer on your Shopify store.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query.  ### Image Optimizer Related Docs: - **[What Happens If I Reach The Monthly Usage Limit For Image Optimizer?](/docs/monthly-usage-limit-for-image-optimizer)** - **[How To Add Image Optimizer To Your Subscription Plan](/docs/how-to-add-image-optimizer-to-your-subscription-plan)** - **[How To Enable Auto Image Optimization On Shopify](/docs/how-to-enable-auto-image-optimization-on-shopify)** --- # How To Use StoreSEO In Any Language With Multilingual Option? Source: https://storeseo.com/docs/enable-multilingual-support You can use [**StoreSEO**](https://apps.shopify.com/storeseo) in any language by using the multilingual Option feature. This will help you use this Shopify SEO app in your own preferred language. ## **How To Enable Multilingual Option With Store SEO?** Before you start, ensure you have created a product list inside your Shopify store. Follow these step-by-step guidelines to find out how to **configure multilingual Option with** **StoreSEO**. ### **Step 1: Select And Install StoreSEO App** First, go to the ‘Apps’ option on Shopify and select the StoreSEO app. Then follow the instructions in this guide to [**install the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify). ![Multilingual Option](../../assets/content/body/2022/07/image-31.webp) ### **Step 2: Configure & Choose Your Preferred Language** Now from the dashboard, scroll down. Here, you can find the option to choose the language. Simply click your preferred one and it will be automatically activated for the whole app. 
This is how easily you can easily change the language with the **Multilingual Option** feature from the StoreSEO app. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Use StoreSEO Keyword Cluster Generator for Product Optimization Source: https://storeseo.com/docs/storeseo-keyword-cluster-generator The StoreSEO **Keyword Cluster Generator** is a powerful feature, helping you to optimize your product descriptions with relevant keywords, which is a crucial step in improving your store’s search visibility and conversion performance. Our new feature simplifies this process by helping you automatically generate keyword clusters and SEO-optimized product descriptions tailored to your specific product. This documentation will guide you through each step of the process, from selecting your product to generating and refining keyword clusters, creating product descriptions, and making final edits. By following these instructions, you’ll be able to produce high-quality, keyword-rich product content that aligns with best SEO practices and drives more organic traffic to your store. Make sure you have [**installed the StoreSEO app**](/docs/how-to-install-storeseo-app-on-shopify) on your [**Shopify** **store**](https://apps.shopify.com/storeseo?search_id=aab4df77-20c3-4fcd-a46e-2ad302d1e31f&shallow_install_type=search&surface_detail=storeseo&surface_inter_position=1&surface_intra_position=1&surface_type=search). Then, follow these step-by-step guidelines to start using the keyword rank tracker feature for your Shopify store using StoreSEO. ## How to Use StoreSEO Keyword Cluster Generator for Product Optimization ### Step 1: Select the Product Start by navigating to the Optimize SEO dashboard. From there, choose the product for which you want to generate a keyword cluster. Selecting the correct product ensures that the generated keywords and descriptions are perfectly aligned with your product’s niche and audience intent.
### Step 2: Generate a Keyword Cluster Once the product is selected, look for the Generate Keyword Cluster option within the Keyword Research panel. Click on the Generate Keyword Cluster button to begin the process. ![How to Use StoreSEO Keyword Cluster Generator for Product Optimization](../../assets/content/body/2026/03/image-21.webp) **StoreSEO** will automatically analyze your product and pull together a set of relevant keywords based on search intent and topical relevance. After a short while, your keyword cluster will appear along with a visual cluster map. You’ll also see a list of related keywords that can help shape the product’s SEO-friendly description.
### Step 3: Generate the Product Description With your keyword cluster ready, it’s time to create your product description. Click on the Generate Product Description button. The system will use your selected keyword cluster to craft a compelling, SEO-optimized product description tailored to your product.
If you’d like a different variation, simply click on the Regenerate icon. Each regenerated version will still be based on the same keyword cluster, so every variant remains SEO-relevant. Once you’ve found the version you prefer, click Keep and Replace to finalize your choice. Don’t forget to click Save to secure all your progress.
### Step 4: Edit the Product If you need to fine-tune your description or adjust other details, click on the Edit Product button. This will allow you to make additional edits to the product description, metadata, or other attributes before finalizing your updates.
That’s how easily you can use the StoreSEO Keyword Cluster Generator feature. Following these steps ensures that your products are backed by accurate keyword research and written with strong SEO optimization for better visibility and conversions. If you need any assistance, feel free to contact [**our support team**](/support). --- # How to Use StoreSEO with Shopify Sidekick to Find And Fix SEO Issues Faster Source: https://storeseo.com/docs/how-to-use-storeseo-with-shopify-sidekick StoreSEO is now available inside [**Shopify Sidekick**](https://www.shopify.com/sidekick). This means you can ask Sidekick questions about your store’s SEO performance and get instant answers, without leaving your Shopify admin. So now, when a merchant asks Sidekick something that relates to that app, Sidekick can pull the answer from there. This document explains how to use Shopify Sidekick to identify SEO issues across your store through simple conversations. ### What You Can Do with StoreSEO in Sidekick Once you open a Sidekick conversation, you can ask about your store’s SEO in plain language. Here are some of the things you can get done: - Check the SEO score of your products, collections, or pages - Find out which products are missing meta titles or meta descriptions - Identify images with missing or weak alt text - Get instant recommendations to fix specific SEO issues Follow the step-by-step guide below to use StoreSEO inside Shopify Sidekick. ### **Step 1: Find the Sidekick Conversation from Shopify Admin** Log in to your Shopify store and go to your admin dashboard. Look for the **purple glasses icon** in the top bar of your admin. Click on it to open the Sidekick panel. ![Use StoreSEO with Shopify Sidekick](../../assets/content/body/2026/06/image-4.webp) You can also access Sidekick from the left side panel, where you will find the **‘Sidekick Conversations’**. Simply click on it to open a new conversation. ![Use StoreSEO with Shopify Sidekick](../../assets/content/body/2026/06/storeseo-sidekick-2-scaled.webp) ### **Step 2: Start a Conversation** Once the Sidekick panel opens, you will see a chat interface. You can type your question or use voice mode to ask. No configuration or special commands are needed. Simply describe what you want to know about your store’s SEO. For example: - _“Which of my products have missing meta descriptions?”_ - _“Show me the SEO score for my homepage.”_ - _“Which product images are missing alt text?”_ - _“How can I improve SEO for my best-selling products?”_ Sidekick will use StoreSEO to find the relevant data and bring it into the conversation. ![Use StoreSEO with Shopify Sidekick](../../assets/content/body/2026/06/sidekick-1-1.webp) ### **Step 3: Review the SEO Insights Sidekick Surfaces** After that, Sidekick will pull the relevant SEO data from StoreSEO and display it directly in the chat. You will see things like SEO scores, missing fields, and specific recommendations tied to your actual store data. Review the information to understand where your store needs attention.
Once Sidekick identifies the issues, you can ask it for further help. For example: - _“Write a meta description for this product.” or_ - _Can you write the meta description for the products you identified?_ Sidekick will prepare the changes and show them to you for review. Once you are satisfied with the suggested fix, apply the changes to your store through StoreSEO.  You can continue the conversation to address more issues, run a broader SEO audit, or ask follow-up questions, all without leaving the Shopify admin. This is how you can use StoreSEO with Shopify Sidekick to fix SEO issues faster. Need assistance? Reach out to the [StoreSEO support team](/contact) for any questions. --- # How To View SEO Reports For All Products, Blogs & Pages Source: https://storeseo.com/docs/view-seo-reports With the [**StoreSEO**](/) app, you can now see detailed SEO reports of all your products, pages & blog posts. From there you can find out your _Overall Score_, _Total Issues_, etc. This feature will help you understand if your Shopify store is properly optimized for SEO at a glance. ## **How To View SEO Reports For Your Shopify Store?** Head over to your StoreSEO dashboard. Then click on the **‘Report’** tab and find your in-depth report on products, pages & blog posts. 
SEO Reports
You can switch between the tabs to see your detailed report on respective categories. 
This is how you can easily see your SEO reports using [**StoreSEO**](https://apps.shopify.com/storeseo) and easily see if your Shopify store is properly optimized.   Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Write Product Descriptions with StoreSEO AI Description Generator Source: https://storeseo.com/docs/write-descriptions-with-ai-description-generator StoreSEO’s AI Product Description Generator is a powerful feature that helps Shopify store owners automatically generate high-quality, optimized product descriptions. Instead of writing descriptions manually, you can use StoreSEO’s AI to produce compelling, structured content for your products in seconds, saving time and improving your store’s search engine visibility.
## Step-by-Step on How to Write Product Descriptions with StoreSEO AI Description Generator ### Step 1: Navigate to Optimize SEO From the StoreSEO sidebar, click on Optimize SEO. Select the Products tab to view your full product list. Each product displays its **Focus Keyword, Optimize Status (Optimized or Needs Improvement), SEO Score**, and the date it was last updated. Find the product you want to generate a description for and click the Fix button next to it. This opens the individual Product SEO page.
### Step 2: Go to the Description Tab On the Product SEO page, you will see several tabs at the top: General SEO, Description, Image, H2C Schema, and AI Snippet. Click on the Description tab to access the product description editor and the AI Description Generator panel. ![Go to the Description Tab](../../assets/content/body/2026/04/image-23.webp) The page is divided into two main sections:  - **Left panel:** The product description editor, where you can view and edit the description content directly. - **Right panel:** The AI Description Generator with customizable filters and generation options. ![Go to the Description Tab](../../assets/content/body/2026/04/image-24.webp) ### Step 3: Generate Keyword Cluster Before generating a description, it is recommended to define a Keyword Cluster for the product. Click View Keyword Cluster next to the Keyword Cluster field. The Keyword Cluster panel will open and allow you to discover semantically related keywords to include in your description for better topical authority. The Keyword Cluster feature organizes keywords into clusters so the AI can naturally weave in SEO-friendly terms. If no cluster has been generated yet, click the “**Generate Keyword Cluster”** button to start.
### Step 4: Generate the Product Description Click the Generate button in the AI Description Generator panel. StoreSEO will use its AI to create a structured product description based on the product name, keyword cluster, and the Agent Discovery (7-pillar) inputs you provided. The generation process typically takes a few seconds.
You can select “**Leave empty**” if you want to skip information for your product that’s not required. ![AI Description Generator](../../assets/content/body/2026/04/image-25.webp) ### Step 5: Review and Rescan the Generated Description However, if you want to rescan and regenerate the product description, simply click the Rescan button. It will revert the agent discovery pillar criteria and start the rescan from scratch.
Then, click on the regenerate button to provide the inputs accordingly to get the regenerated product description. ![Regenerate the AI Description Generator](../../assets/content/body/2026/04/image-26.webp) This is how easily you can generate AI product descriptions for your Shopify store with StoreSEO for your [**Shopify products**](https://shopify.pxf.io/storeseo).  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # StoreSEO AI Toolkit: AI-Agent Friendly Solution for Your Shopify Store Source: https://storeseo.com/docs/ai-toolkit-ai-agent-friendly-solution The StoreSEO AI Toolkit is a feature that connects your Shopify store to AI coding assistants like Claude Code, Cursor, and VS Code through an **MCP (Model Context Protocol) server**. Think of it as a bridge that lets AI agents talk directly to your StoreSEO account. Instead of clicking around in the admin panel, you can simply type commands in your terminal and let AI handle SEO optimizations for you.
## Before You Begin Make sure you have the following before you start: - An active Shopify store with StoreSEO installed on it - A Shopify store connected to StoreSEO - Claude Code installed on your computer (download from [claude.ai](http://claude.ai)) - Basic familiarity with using a terminal/command line ### Step 1: Navigate to the AI Toolkit First, open your StoreSEO app. From the left sidebar, click on Settings. In the Settings menu, look for the AI Toolkit option and click on it. You will see a section that lets you generate API keys to connect StoreSEO with AI agents like Claude Code, Cursor, and VS Code.
### Step 2: Generate Your API Token Once you are on the AI Toolkit page, you will see a prominent button labeled Generate Token. Click this button. A popup window will appear asking you to name your token. Give it a meaningful name so you can identify it later.  For example, if you are using Claude Code, name it Claude Code. This name is just a label for your reference and does not affect functionality. Click the Generate button after naming it.
### Step 3: Add the StoreSEO MCP Server to Claude Code Open your terminal on your computer. Paste the command you copied in Step 3 and press Enter. The command will add StoreSEO as an MCP server to your Claude Code configuration. You will see a confirmation message showing that the HTTP MCP server was added successfully along with your StoreSEO URL. 
### Step 4: Verify the Connection In your terminal, run the following command: ``` claude mcp list ``` This will show all connected MCP servers. Look for “**StoreSEO**” in the list. If it shows Connected with a checkmark, your connection is working properly. ![AI Toolkit](../../assets/content/body/2026/05/image-14.webp) You should also see your StoreSEO URL connection confirmation next to it. If you see an error or the connection says disconnected, double-check that your token was copied correctly and try regenerating it from the AI Toolkit page.
### Step 5: Start Using the Toolkit Now you are ready to use the StoreSEO AI Toolkit. In your terminal, type claude and press Enter to start Claude Code. You can now give commands in plain English to perform SEO tasks. For example, you can type: SEO optimize all my least scoring products. Claude will connect to StoreSEO, identify your low-performing products, and automatically optimize them. The AI will work through the optimization process and show you a summary of what was fixed and what still needs attention. **Sample Commands You Can Use:** - SEO optimize all my least scoring products - SEO optimize all my least scored collections - Generate alt text for all product images - Fix meta descriptions for all products
### Step 6: Final Overview Once the AI completes the optimization task, it will provide you with a summary table showing each product’s SEO score before and after the changes. The table includes the product name, the original score, the new score, and the score improvement. Claude will also list what was fixed on each product, such as meta titles being keyword-optimized, meta descriptions updated, URL handles cleaned up, image alt text being added, FAQ schema being added, and AI snippets being added to all products. ![Final Overview](../../assets/content/body/2026/05/final-overview.webp) This is how easily you can use StoreSEO AI Toolkit for your Shopify store. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # What Do I Do If I Need More Credits For AI Content Optimizer? Source: https://storeseo.com/docs/credits-ai-content-optimizer StoreSEO AI Content Optimizer add-on feature helps to optimize your Meta Title, Meta Description, and Tags within the power of AI in a few seconds. If you have reached your limit and need more credits you can buy another AI Credit bundle to increase your usage limit. Choose according to your needs from our 3 different packages: - 5,000 AI Credits: $10 - 25,000 AI Credits: $50 - 50,000 AI Credits: $100 Please reach out to our [**support team**](/contact) if you have any other queries. --- # What Happens If I Reach The Monthly Usage Limit For Image Optimizer? Source: https://storeseo.com/docs/monthly-usage-limit-for-image-optimizer **[StoreSEO Image Optimizer](/docs/how-to-use-storeseo-image-optimizer-on-shopify)** add-on feature helps to optimize and resize your store images without any hassle. If you have reached your monthly limit you can not optimize any images before monthly usage reThe [**StoreSEO Image Optimizer**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) add-on enables seamless optimization and resizing of your Shopify store images to enhance page speed and SEO performance.  If you reach your monthly image optimization limit, you will be unable to optimize additional images until the limit resets at the start of the next billing cycle.  To increase your monthly limit, you can upgrade to a higher subscription plan that better suits your store’s needs, and the following subscription tiers with their respective Image Optimizer limits: - **Lite Plan** ($19.99/month): - Up to 25 products - Image Optimizer: 50 images/month included - **Essential Plan** ($39.99/month): - Up to 100 products - Image Optimizer: 500 images/month included - **Growth Plan** ($99.99/month): - Up to 250 products - Image Optimizer: 2,000 images/month included - **Advanced Plan** ($249.99/month): - Up to 10,000 products - Image Optimizer: 50,000 images/month included If your store requires optimizing more than 50,000 images per month, please contact the [**StoreSEO support team**](/support). We’d love to help you with your query ### Image Optimizer Related Docs: - **[How To Add Image Optimizer To Your Subscription Plan](/docs/how-to-add-image-optimizer-to-your-subscription-plan)** - [**How To Use StoreSEO Image Optimizer On Shopify**](/docs/how-to-use-storeseo-image-optimizer-on-shopify) - **[How To Enable Auto Image Optimization On Shopify](/docs/how-to-enable-auto-image-optimization-on-shopify)** --- # What Is A Focus Keyword? How To Choose A Focus Keyword? Source: https://storeseo.com/docs/what-is-a-focus-keyword When you launch a brand new website, your website will not rank high in search engine results automatically. To do this, you need to know **what is a focus keyword**, and how to use them. That’s where **SEO optimization** comes in. By using good SEO practices and using the right focus keywords, you can make your products more visible on search engines. ## **What Is A Focus Keyword?** A focus keyword is the search term that you want your **[Shopify products](https://www.shopify.com/)** to be ranked for. When a potential customer searches for a specific term or phrase, your product will come up as a result. **StoreSEO** app will help you to optimize your Shopify products for better SEO ranking in major search engines like Google or Bing.  For example, Karen is searching for “_Hudderton Backpack”_ on Google and similar kinds of pages will appear. Suppose you sell Hudderton Backpack too and want to appear on her search result. You need to optimize your page with _these exact keywords_ that will eventually pull your website before Karen. For better ranking in the search result, you need to use the correct keywords that your potential customers might use while searching. ![What Is A Focus Keyword](../../assets/content/body/2021/11/image.webp) ## **How To Choose A Focus Keyword?**  Using keywords and creating content around specific keywords can help your website drive more traffic organically. That’s why choosing the right focus keyword for your Shopify products is important. Let’s learn how to choose the right focus keyword for your content.  - Choose the words that **explain** your product right and have a good search volume. Take help from keyword research tools to see the volume of each keyword.  - Using specific keywords will help your SEO health more than anything.  Avoid “Keyword stuffing” by constantly repeating the keyword in your content. Keyword stuffing is penalized by search engines. A good ratio for keywords in your content is 1-2%. For example, if your product description is around **100-150 words**, your keywords should be used **_once or twice_** in the description. ## **Where To Put Keywords For Shopify Products?** As you are interested in inserting keywords in your Shopify content, you can use them in several ways.  **Meta title:** Use the keyword in the meta title so that your product’s click-through rate increases instantly.  ![What Is A Focus Keyword](../../assets/content/body/2021/11/image-1.webp) **Meta description:** When your product appears on someone’s search result, they can have the idea from the meta description that you have written for the product. Inserting your keyword in the meta description is a useful SEO tactic. ![What Is A Focus Keyword](../../assets/content/body/2021/11/image-2.webp) **Image alt text:** Image alt text helps the search engine crawlers understand what the image is about. That’s why to index the correct product picture, you need to input a keyword in the image alt text.  ![What Is A Focus Keyword](../../assets/content/body/2021/11/image-3.webp) In addition to that, you can add your focus keyword on your product title, product description, and permalink as well. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # What Is Search Engine Optimization? Why Do I Need It? Source: https://storeseo.com/docs/what-is-search-engine-optimization **Search engine optimization** or SEO refers to a process of optimizing your website so that it can get to the top of organic web search results. So more potential customers can find your website on Google and other search engines and increase your website traffic and sales. **[StoreSEO](/)** is an advanced SEO solution that is specifically designed for **[Shopify stores](https://www.shopify.com/)**. If you want your products to be more visible on search engines, you can use StoreSEO to improve SEO of the products. Want to learn more about StoreSEO and how it helps with search engine optimization for your Shopify store and products? Then read this documentation on [‘**What Is StoreSEO?’**](/docs/what-is-storeseo-and-how-does-it-work).  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries. --- # What Is StoreSEO And How Does It Work? Source: https://storeseo.com/docs/what-is-storeseo-and-how-does-it-work [**StoreSEO**](/) app helps your [**Shopify store**](https://www.shopify.com/) have a better [**SEO ranking**](/docs/what-is-search-engine-optimization) so that your potential customers can find your shop through Google search or other search engines. This app will help you to set SEO tactics both in product descriptions or titles and images easily. The app’s main features are listed below. ### **1\. Set Focus Keyword For Products** StoreSEO lets you set focus keywords for all of your Shopify products. You can pick keywords from the ‘**Keyword Suggestions**’ section or do keyword research on your own and add the keywords for your products easily to rank them better. See **how to set focus keywords for Shopify products** easily using StoreSEO. ![What is StoreSEO](../../assets/content/body/2021/08/image-5.webp) ### **2\. Set Meta Title & Meta Description**  With this app, you can add meta titles and descriptions to describe the product clearly. You can add more suitable keywords in the description for better SEO ranking. Check out **how to set the meta title & meta description** using the StoreSEO app on your Shopify store. ![What is StoreSEO](../../assets/content/body/2021/08/image-6-2.webp) ### **3\. Add Image Alt Text** You can also add image alt text to describe the images of your product to the search engines. It will boost your SEO score immediately. Learn **how to add image alt text using the StoreSEO** app. ![What is StoreSEO](../../assets/content/body/2021/08/image-7-2.webp) ### **4\. View Overall SEO Score With Basic SEO Analysis** You can view your SEO score and optimize each of your products accordingly. StoreSEO will also show you a list of each product that needs to be fixed immediately. ![What is StoreSEO](../../assets/content/body/2021/08/image-8-2.webp) For more information on how to use these features, reach out to our [**dedicated support team**](/contact)**.** --- # What Will Happen If I Uninstall StoreSEO? Source: https://storeseo.com/docs/what-happens-if-you-uninstall-storeseo-app When you **uninstall StoreSEO** app, there won’t be any changes on any of your products or your store. The only change will be that you won’t be able to see the SEO score ranking like before and you won’t get to use any of the StoreSEO features anymore. StoreSEO helps to **optimize SEO for Shopify products**, so once you uninstall it you can’t see where to improve anymore to get a higher SEO score for your products. Need assistance? Feel free to **[reach out to our support team](/contact)**. --- # Why Do I Not See My Shopify Store In Google Search Results? Source: https://storeseo.com/docs/store-not-appearing-in-google-search-results **[StoreS](/)[EO](/)** app helps your Shopify store score higher in search engine results by using all of the best practices. If you are new to the **StoreSEO app**, it is possible to you might need to wait few weeks to see **any immediate results** for Google search. This is because Google takes time to update, you’ll have to wait a few weeks or more to notice changes for your **[Shopify store](https://www.shopify.com/)**, even if you are using the SEO tool to improve your SEO score. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of queries --- # Why Has My Store’s Google Ranking Not Improved? Source: https://storeseo.com/docs/why-has-google-ranking-not-improved If you are wondering why has your store’s **Google ranking not improved** even after setting up StoreSEO, it is because Google might take more time to update your web page rankings. And it will take even more time for the SEO to start making noticeable and significant changes to your website sales and traffic. And that is just how Google and other search engines work. Search engine optimization is always a long-term project.  Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # Why Is My Meta Description Not Showing In Google? Source: https://storeseo.com/docs/meta-description-not-showing-in-google When you first set up **[StoreSEO](/)** on your Shopify store, it takes time for Google to update your store’s information in their listings. This is why you may notice your **meta description not showing in Google**. You may need to wait for few weeks before the app’s changes appear in Google search results. You can expect to wait a few weeks if your **[Shopify store](https://www.shopify.com/)** is completely brand new, Google takes time to index fresh content on your site. Need assistance? Feel free to reach out to our [**dedicated support team**](/contact) for any type of query. --- # How to Connect StoreSEO MCP to ChatGPT Source: https://storeseo.com/docs/connect-storeseo-to-chatgpt ChatGPT is OpenAI's AI assistant. You talk to it in a chat on the web or in an app, and it can answer questions, write and analyze, and take actions through connected tools. This guide connects ChatGPT to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, ChatGPT can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for ChatGPT](/agent/chatgpt). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** You sign in with StoreSEO in the browser: pick your Shopify store and approve access. ChatGPT can register itself through dynamic client registration, so there is no token to create or paste. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to ChatGPT ### Step 1: Turn on developer mode In ChatGPT on the web, open Settings, then Security and login, and turn on Developer mode. If your menu shows Apps and Connectors with Advanced settings instead, the same switch lives there. OpenAI lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. ```text https://chatgpt.com/plugins ``` ### Step 2: Create the app with the StoreSEO MCP URL Open Plugins and select the plus button to create a developer-mode app. Name it StoreSEO, paste this as the MCP server URL and choose OAuth. Sign in when the browser asks: pick your Shopify store and approve access. ```text https://mcp.storeseo.com/mcp ``` ### Step 3: Use it in a chat The app appears in the composer's Developer mode tool during conversations. Select StoreSEO there, then ask this to confirm the connection works. ```text Using the StoreSEO app, show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Ask a workspace admin to publish it On Business, Enterprise and Edu workspaces OpenAI's help center puts the controls under Settings, Apps, so an admin or owner may need to create and publish the app for you. Give them the same URL. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new chat and send this prompt. ChatGPT should call StoreSEO and answer with your store's real SEO score. ```text Using the StoreSEO app, show me my store's SEO score breakdown. ``` ## What to Ask ChatGPT - Using the StoreSEO app, what is my store's SEO score and what are the three biggest problems? - Which products have meta descriptions that are too short? Preview new ones for the first five. - Show me the images with no alt text on my best-selling collection. - Is my store ready for AI shopping assistants? Check my llms.txt status and AI settings. - Show my Search Console clicks and impressions for the last 28 days. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About ChatGPT - ChatGPT connects from OpenAI's servers, not from your computer. The StoreSEO MCP URL is public, so there is nothing to open on your side. - Write actions depend on your ChatGPT plan. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web, while OpenAI's help center, as we read it, puts full MCP support with write actions in beta on Business, Enterprise and Edu workspaces and gives Pro accounts read access only. On a plan without write actions, use ChatGPT for audits and reports and apply fixes from the StoreSEO app or another agent. - OpenAI marks developer mode as higher risk because a connected server can read and, with write actions, change your data. Read each payload before you approve, and be careful with the option that remembers approval for the rest of a conversation. - If you do not see developer mode under Settings, Security and login, check your plan and, in a workspace, ask an admin whether it is enabled. ## Disconnect Remove the StoreSEO server in ChatGPT, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for ChatGPT: jobs, prompts and FAQs](/agent/chatgpt) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [OpenAI: developer mode in ChatGPT](https://developers.openai.com/api/docs/guides/developer-mode) (vendor documentation these steps were checked against) - [OpenAI help center: developer mode and full MCP connectors](https://help.openai.com/en/articles/12584461-developer-mode-apps-and-full-mcp-connectors-in-chatgpt-beta) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Claude Code Source: https://storeseo.com/docs/connect-storeseo-to-claude-code Claude Code is Anthropic's agentic coding tool. It runs in your terminal, and in IDE extensions and a desktop app, reads the project in the folder you start it in, and works through multi-step tasks by calling tools and showing each call as it happens. This guide connects Claude Code to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Claude Code can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Claude Code](/agent/claude-code). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Claude Code signs you in with OAuth: run /mcp, pick your Shopify store in the browser and approve. Claude Code stores the credentials it receives and refreshes them. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it with --header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Claude Code ### Step 1: Add StoreSEO to Claude Code Run this in any terminal. The user scope makes StoreSEO available in every project on your machine. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp ``` ### Step 2: Sign in to your store Start Claude Code, run /mcp, choose storeseo and follow the browser window: pick your Shopify store and approve access. ```text /mcp ``` ## Other Ways to Connect ### Option: Connect with an access token If you cannot use a browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add the server with the token as a header instead. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp \ --header "Authorization: Bearer YOUR_TOKEN" ``` ## Check the Connection Start a new session and send this prompt. Claude Code should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Claude Code - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Write it for the ones on my top collection, one image at a time. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Turn on product schema and tell me how many products that fixes. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Claude Code - Without --scope user the server is added to the current project only, so it disappears when you start Claude Code in another folder. - Claude Code asks before a tool runs unless you have allowed it. If you add an allow rule for the StoreSEO tools, the confirmation StoreSEO asks for in the chat becomes the only check, so keep apply, write and generate tools on ask. - Claude Code warns when one tool result is larger than 10,000 tokens. Ask for smaller pages of products or images on a big catalog rather than raising the limit. ## Disconnect Remove the StoreSEO server in Claude Code, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Claude Code: jobs, prompts and FAQs](/agent/claude-code) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Claude Code: connect to tools via MCP](https://code.claude.com/docs/en/mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Claude Cowork Source: https://storeseo.com/docs/connect-storeseo-to-claude-cowork Claude Cowork is Anthropic's agent for delegated knowledge work. Instead of a back-and-forth chat, you describe a task, hand it off and let Claude work through it using your connected apps, plugins and skills, then you review the result. This guide connects Claude Cowork to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Claude Cowork can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Claude Cowork](/agent/claude-cowork). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** You sign in when you add the connector: a browser window opens where you pick your Shopify store and approve access. StoreSEO supports dynamic client registration, so there is no token to paste. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Claude Cowork ### Step 1: Add StoreSEO as a custom connector In Cowork, open Customize, then Connectors, choose + Add and then Add custom connector. If you already added StoreSEO to your Claude account, look for it in that list first. Use this URL. ```text https://mcp.storeseo.com/mcp ``` ### Step 2: Approve your store and check it Sign in when the browser opens: pick your Shopify store and approve access. Then start a task and ask this first, so you know the connector works before you delegate real work. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Add it as an owner on Team or Enterprise Only if you are on a Team or Enterprise plan: an owner adds the same URL under Organization settings, then Connectors, and members connect their own store from Customize, then Connectors. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new session and send this prompt. Claude Cowork should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Claude Cowork - Use StoreSEO to audit my store's SEO and give me a ranked list of fixes. Do not change anything. - Preview better meta for my ten lowest-scoring products and wait for me to review them before you apply any. - Check which product images have no alt text and summarize the damage by collection. - Pull my Search Console insights for the last 28 days and list queries where I rank between 8 and 20. - Check whether my llms.txt and agents.md are live and tell me what is missing. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Claude Cowork - A scheduled task runs without you in the conversation, so nobody is there to say yes. Keep scheduled StoreSEO tasks to reads and reports, and tell the task in plain words not to apply, generate, publish or delete anything. - Cowork shares the custom connector model of Claude: the Free plan allows one custom connector, and on Team and Enterprise an owner adds the connector under Organization settings, Connectors. - On Team and Enterprise plans an admin toggle controls access to Cowork, so you may need your admin to turn it on. - If StoreSEO does not appear in Cowork, check Customize, Connectors inside Cowork and add it there rather than assuming a connector added elsewhere carried over. ## Schedule a weekly SEO check in Cowork Cowork documents scheduled tasks that run on a cadence and can use your connected apps. For StoreSEO, make the task report-only: it should read scores and reports and write you a summary. Do not ask a scheduled task to apply content, generate files or delete anything, because you are not there to say yes. ```text Every Monday, use StoreSEO to read my store's SEO score breakdown, my lowest-scoring products and last week's Search Console insights. Summarize what changed. Do not apply, generate, publish or delete anything. ``` ## Disconnect Remove the StoreSEO server in Claude Cowork, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Claude Cowork: jobs, prompts and FAQs](/agent/claude-cowork) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Anthropic: custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) (vendor documentation these steps were checked against) - [Anthropic: scheduled tasks in Claude Cowork](https://support.claude.com/en/articles/13854387-schedule-recurring-tasks-in-claude-cowork) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Cursor Source: https://storeseo.com/docs/connect-storeseo-to-cursor Cursor is an AI code editor made by Anysphere. Its agent reads the project you have open, edits files and calls tools, and shows each tool call in the chat so you can see what ran. This guide connects Cursor to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Cursor can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Cursor](/agent/cursor). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Cursor supports OAuth for remote MCP servers, so you sign in by picking your Shopify store in the browser and approving. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it as an Authorization header in mcp.json. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Cursor ### Step 1: Add StoreSEO to mcp.json Put this in ~/.cursor/mcp.json to use StoreSEO in every project, or in .cursor/mcp.json inside one theme project. In the StoreSEO app's setup guide the same JSON goes into Settings, Tools and MCP, New MCP server. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Sign in to your store Open Cursor's MCP settings (the Customize page in current versions), find storeseo and sign in when Cursor asks. A browser window opens: pick your Shopify store and approve access. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ### Step 3: Switch off four StoreSEO tools StoreSEO has 44 tools and Cursor stops at about 40 active ones across all servers. In the storeseo entry in Cursor's MCP settings, turn off at least four you do not need. The four rank and competitor lookups are a good first choice if you are not on Pro. ```text storeseo_list_rank_locations storeseo_list_rank_languages storeseo_add_competitors storeseo_delete_competitor ``` ## Other Ways to Connect ### Option: Connect with an access token If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Cursor can read it from an environment variable so the token stays out of the file. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer ${env:STORESEO_TOKEN}" } } } } ``` ## Check the Connection Start a new session and send this prompt. Cursor should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Cursor - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest ones first. - Check my product schema settings and tell me what turning on Review and Sale dates would add. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Cursor - StoreSEO has 44 tools and Cursor allows about 40 active tools across all your servers. Turn off at least four StoreSEO tools in Cursor's MCP settings, and more if you run other MCP servers. - A server added to .cursor/mcp.json belongs to that project only. Use ~/.cursor/mcp.json if you want StoreSEO in every project. - If you set a StoreSEO tool to run without asking, the confirmation StoreSEO asks for in chat is the only check left. Keep the apply, generate, publish and delete tools on ask. - Admins on Teams and Enterprise plans can restrict which MCP servers run, so a server you add may be blocked until an admin allows its URL. ## Disconnect Remove the StoreSEO server in Cursor, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Cursor: jobs, prompts and FAQs](/agent/cursor) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Cursor: Model Context Protocol (MCP)](https://cursor.com/docs/mcp) (vendor documentation these steps were checked against) - [Anysphere](https://anysphere.inc/) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Codex Source: https://storeseo.com/docs/connect-storeseo-to-codex Codex is OpenAI's coding agent. It runs as a command-line tool, as an IDE extension and inside the ChatGPT desktop app, reads the repository you start it in and works through multi-step tasks by calling tools. This guide connects Codex to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Codex can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Codex](/agent/codex). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Codex signs in with OAuth: run codex mcp login storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and let config.toml read it from an environment variable. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Codex ### Step 1: Add StoreSEO to Codex Run this in any terminal. It writes a [mcp_servers.storeseo] table to ~/.codex/config.toml, which the Codex CLI, the IDE extension and the ChatGPT desktop app all read. ```bash codex mcp add storeseo --url https://mcp.storeseo.com/mcp ``` ### Step 2: Sign in to your store Run the login command. Your browser opens: pick your Shopify store and approve access. Then run codex mcp list, or type /mcp inside Codex, to confirm storeseo is there. ```bash codex mcp login storeseo codex mcp list ``` ## Other Ways to Connect ### Option: Write it in config.toml If you would rather edit the file, add this table to ~/.codex/config.toml, or to .codex/config.toml in a trusted project. In the desktop app or the IDE extension you can add the same URL under MCP servers, Add server, Streamable HTTP. ```toml [mcp_servers.storeseo] url = "https://mcp.storeseo.com/mcp" ``` ### Option: Connect with an access token Generate a token in StoreSEO under Settings, AI Toolkit, export it as STORESEO_TOKEN in your shell, and point config.toml at that variable. Codex sends it as an Authorization Bearer header, so the token never sits in the file. ```toml [mcp_servers.storeseo] url = "https://mcp.storeseo.com/mcp" bearer_token_env_var = "STORESEO_TOKEN" ``` ## Check the Connection Start a new session and send this prompt. Codex should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Codex - Use StoreSEO to list products scoring under 50 and preview better meta for the first five. - Is my agents.md live? If it is the Shopify default, tell me what generating a custom one will publish. - Which images have no alt text? List the ten heaviest first. - Show me my Search Console queries for the last 28 days, grouped by page. - Check which schema types are enabled and tell me what Product schema would add. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Codex - The Codex docs show codex mcp add for local commands, and the --url form appears in the OAuth section. If your version rejects it, use the config.toml table instead. - A project-level .codex/config.toml only loads in trusted projects. Use ~/.codex/config.toml for StoreSEO in every project. - Set the server's default_tools_approval_mode to writes or prompt, not approve. Approve skips the prompt, and then the confirmation StoreSEO asks for in chat is the only check. - Codex waits 60 seconds for a tool to finish by default. Content generation can take longer, so raise tool_timeout_sec for the storeseo server if a generation times out. ## Disconnect Remove the StoreSEO server in Codex, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Codex: jobs, prompts and FAQs](/agent/codex) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [OpenAI Codex: Model Context Protocol](https://developers.openai.com/codex/mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Gemini CLI Source: https://storeseo.com/docs/connect-storeseo-to-gemini-cli Gemini CLI is Google's open-source command-line AI agent. You run it in a terminal, it reads the folder you start it in, and it works through tasks by calling built-in tools and tools from MCP servers. This guide connects Gemini CLI to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Gemini CLI can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Gemini CLI](/agent/gemini-cli). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Gemini CLI finds the StoreSEO sign-in on its own: the first time it calls the server it opens your browser, you pick your Shopify store and approve, and it stores the tokens. On a machine with no browser, add a personal access token from StoreSEO under Settings, AI Toolkit as a header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Gemini CLI ### Step 1: Add StoreSEO to Gemini CLI Run this in any terminal. The user scope writes the server to ~/.gemini/settings.json so every project sees it. Without --scope user the command writes to the current project's .gemini/settings.json. ```bash gemini mcp add --scope user --transport http storeseo https://mcp.storeseo.com/mcp ``` ### Step 2: Sign in to your store Start gemini. The first time it calls StoreSEO it opens your browser: pick your Shopify store and approve access. If the window does not open, type /mcp auth storeseo inside Gemini CLI. Then gemini mcp list should show storeseo as connected. ```text /mcp auth storeseo ``` ## Other Ways to Connect ### Option: Edit settings.json Add this to ~/.gemini/settings.json for every project, or to .gemini/settings.json in one project. It is the entry gemini mcp add writes: url plus type set to http. The docs page still shows the older httpUrl key for streamable HTTP. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "type": "http" } } } ``` ### Option: Connect with an access token On a headless machine the browser sign-in cannot run. Generate a token in StoreSEO under Settings, AI Toolkit and pass it as a header. Keep the token out of version control. ```bash gemini mcp add --scope user --transport http \ --header "Authorization: Bearer YOUR_TOKEN" \ storeseo https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new session and send this prompt. Gemini CLI should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Gemini CLI - Use StoreSEO to list every product scoring under 50 and preview new meta for the first five. - Which product images have no alt text? Show me the ten heaviest first. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. - Which schema types are enabled for my store, and what would turning on Product schema add? ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Gemini CLI - Google's site says Gemini CLI was replaced by Antigravity CLI on June 18th, 2026 for unpaid tier and Google One users. If that is you, check which command-line agent your account can still use before you set this up. - gemini mcp add defaults to project scope, so a server added without --scope user is only available in that project. - Do not set --trust or trust: true on the storeseo server. Trust skips every tool confirmation, and then the confirmation StoreSEO asks for in chat is the only check. - The browser sign-in needs a browser on the same machine. In a headless session or over SSH without a browser, use the access token header instead. ## Disconnect Remove the StoreSEO server in Gemini CLI, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Gemini CLI: jobs, prompts and FAQs](/agent/gemini-cli) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Gemini CLI: MCP servers](https://geminicli.com/docs/tools/mcp-server/) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to GitHub Copilot Source: https://storeseo.com/docs/connect-storeseo-to-github-copilot GitHub Copilot is GitHub's AI coding assistant. In Visual Studio Code, Copilot Chat has an Agent mode that can plan multi-step work and call tools, including tools from Model Context Protocol servers. This guide connects GitHub Copilot to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, GitHub Copilot can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for GitHub Copilot](/agent/github-copilot). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** StoreSEO supports OAuth with dynamic client registration, so VS Code can sign you in through the browser: pick your Shopify store and approve access. No token has to sit in mcp.json. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to GitHub Copilot ### Step 1: Add StoreSEO to .vscode/mcp.json Create or open .vscode/mcp.json in your workspace and paste this. VS Code uses a top-level servers key, not mcpServers, and a remote server needs type http. ```json { "servers": { "storeseo": { "type": "http", "url": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Start the server and sign in Run MCP: List Servers from the Command Palette, pick storeseo and start it. If it asks you to sign in, pick your Shopify store in the browser and approve access. Then open Copilot Chat in Agent mode, check the tools icon lists StoreSEO and ask this. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Add it from the Command Palette Run this command, choose a remote HTTP server, paste https://mcp.storeseo.com/mcp and name it storeseo. VS Code asks where to save it: the workspace shares the entry with your team, your user configuration keeps it to yourself. ```text MCP: Add Server ``` ## Check the Connection Start a new session and send this prompt. GitHub Copilot should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask GitHub Copilot - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Preview alt text for my top collection. - Check my schema settings and tell me which theme file renders the product JSON-LD. - Is my llms.txt live? Tell me what generating one will change before you run it. - Show Search Console queries from the last 28 days where we rank between 8 and 20. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About GitHub Copilot - VS Code uses the servers key in .vscode/mcp.json. The mcpServers key belongs to the portable .mcp.json file at the project root, so copying a Cursor or Claude snippet unchanged will not work here. - On Copilot Business and Enterprise plans an organization admin must enable the MCP servers in Copilot policy, otherwise the server will not be usable. - A .vscode/mcp.json committed to the repository is shared with your team. Keep any token out of it and use input variables if you ever add one. - Keep the apply, optimize, generate and delete tools on VS Code's confirmation prompt. StoreSEO asks for your yes in the chat, and the editor's tool confirmation is the hard stop. ## Disconnect Remove the StoreSEO server in GitHub Copilot, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for GitHub Copilot: jobs, prompts and FAQs](/agent/github-copilot) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [VS Code: add and manage MCP servers](https://code.visualstudio.com/docs/copilot/customization/mcp-servers) (vendor documentation these steps were checked against) - [GitHub Docs: extend Copilot Chat with MCP](https://docs.github.com/en/copilot/how-tos/provide-context/use-mcp/extend-copilot-chat-with-mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Windsurf Source: https://storeseo.com/docs/connect-storeseo-to-windsurf Windsurf is an AI code editor that began at Codeium and is now part of Cognition. Its Cascade agent reads your project, edits files and calls tools, and Windsurf's documentation now lives under Devin Desktop. This guide connects Windsurf to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Windsurf can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Windsurf](/agent/windsurf). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Windsurf's docs say OAuth works for remote MCP servers, so you sign in by picking your Shopify store in the browser and approving. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and add it as an Authorization header in mcp_config.json. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Windsurf ### Step 1: Add StoreSEO to mcp_config.json In the Cascade panel, open the ... menu and click the Open MCP config file icon in the MCPs section. Add the entry under mcpServers and save. Remote servers use serverUrl. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Sign in to your store Windsurf supports OAuth for remote servers. When it asks you to sign in to storeseo, a browser window opens: pick your Shopify store and approve access. Then ask this to confirm the connection. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Connect with an access token Generate a token in StoreSEO under Settings, AI Toolkit and send it as a header. mcp_config.json reads environment variables with the env syntax, so the token can stay out of the file. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer ${env:STORESEO_TOKEN}" } } } } ``` ## Check the Connection Start a new session and send this prompt. Windsurf should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Windsurf - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest first. - Check my schema settings and tell me which types are off and what each would add. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Windsurf - Windsurf's MCP page says its configuration applies to the legacy Cascade agent. The Devin Local agent, the default for new tabs, reads Devin CLI configuration files instead, so check which agent you are using. - Cascade allows 100 tools in total. StoreSEO uses 44, so other MCP servers you have enabled count against the same 100. - On Teams and Enterprise plans an admin can turn MCP off or allowlist servers by name. If storeseo does not load, ask your admin to allow the server name you used in mcp_config.json. - The Cascade tool-approval behavior is not described on the MCP page. Treat StoreSEO's confirmation as an instruction the agent follows and review each write in the chat before you accept it. ## Disconnect Remove the StoreSEO server in Windsurf, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Windsurf: jobs, prompts and FAQs](/agent/windsurf) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Windsurf: Cascade MCP integration](https://docs.windsurf.com/windsurf/cascade/mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Grok Source: https://storeseo.com/docs/connect-storeseo-to-grok Grok is xAI's AI assistant. You use it at grok.com and in xAI's apps, and it can answer questions, search and analyze, and take actions in external tools you connect. This guide connects Grok to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Grok can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Grok](/agent/grok). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Grok connectors authenticate with OAuth: when you add StoreSEO you pick your Shopify store in the browser and approve access. There is no token to paste. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Grok ### Step 1: Add a custom connector Open the connectors page on grok.com, choose New Connector and then Custom. Enter this as the MCP server URL and complete the authentication when it opens. ```text https://mcp.storeseo.com/mcp ``` ### Step 2: Approve your store and check it Pick your Shopify store in the browser and approve access. Then start a chat with the connector available and ask this to confirm the connection works. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Have a team admin provision it Only on Grok Business and Enterprise: xAI's documentation says a team admin must first provision the connector in the cloud console before members can use it. Give the admin the same URL. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new chat and send this prompt. Grok should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Grok - Use StoreSEO to find my lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for five of them. - Is my store ready for AI answer engines? Check llms.txt and my AI optimization settings. - Show my Search Console queries from the last 28 days with high impressions and few clicks. - Which structured data types are switched on for my store? ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Grok - xAI's connector documentation does not describe a per-tool approval setting. StoreSEO's instruction to preview and wait for your yes is the main check in the conversation, so read what Grok proposes before you say yes. - On Grok Business and Enterprise a team admin has to provision the connector in the cloud console first, so you may need your admin to add it. - This page covers the Grok chat connector on grok.com. xAI's terminal agent is a separate client with its own setup. - Because Grok connects over the public internet, nothing needs to be opened on your side, but the StoreSEO sign-in must complete in a browser. ## Disconnect Remove the StoreSEO server in Grok, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Grok: jobs, prompts and FAQs](/agent/grok) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [xAI: Grok connectors](https://docs.x.ai/grok/connectors) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to ChatGPT dots Source: https://storeseo.com/docs/connect-storeseo-to-chatgpt-dots A ChatGPT dot is one of OpenAI's always-on agents inside ChatGPT. According to OpenAI's help center, a dot has its own cloud computer, memory, and the apps you choose to connect. You give it a goal, set what it may do on its own, and it keeps making progress between conversations and brings results back for your review. This guide connects ChatGPT dots to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, ChatGPT dots can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for ChatGPT dots](/agent/chatgpt-dots). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** You sign in with StoreSEO once, in the browser: pick your Shopify store and approve access. Your dot uses the same ChatGPT plugin, and ChatGPT can register itself through dynamic client registration, so there is no token to paste. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to ChatGPT dots ### Step 1: Create the StoreSEO app in ChatGPT OpenAI's help center says plugins are shared across dots and ChatGPT, so you set StoreSEO up once. In ChatGPT on the web, turn on Developer mode under Settings, Security and login, open Plugins, select the plus button and create an app named StoreSEO with this MCP server URL and OAuth. Sign in when the browser asks: pick your Shopify store and approve access. If ChatGPT already has the StoreSEO app, skip this step. ```text https://mcp.storeseo.com/mcp ``` ### Step 2: Make sure your dot can see it Open your dot's profile, then Customize and Plugins, and check StoreSEO is connected. Then give your dot a first task that names StoreSEO. ```text Using StoreSEO, show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Ask a workspace admin to publish it On a Business workspace only an admin or owner can turn on developer mode and publish the app, and on Enterprise and Edu an admin has to grant you access first. The controls sit under Settings, Apps. Give the admin the same URL, then start at the previous step once the app is published. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new session and send this prompt. ChatGPT dots should call StoreSEO and answer with your store's real SEO score. ```text Using StoreSEO, show me my store's SEO score breakdown. ``` ## What to Ask ChatGPT dots - Using StoreSEO, what is my store's SEO score and what are the three biggest problems? - Every Monday, use StoreSEO to list products scoring under 50 and send me the list. Report only and change nothing. - Check whether my llms.txt is live and up to date, and tell me what generating a new one would change before you do anything. - Using StoreSEO, which images have no alt text? Preview alt text for five of them. - Once a month, pull my Search Console queries for the last 28 days and summarize where we rank between 8 and 20. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About ChatGPT dots - Developer mode and write actions depend on your plan and workspace. OpenAI's developer guide lists developer mode for Pro, Plus, Business, Enterprise and Education accounts on the web. OpenAI's help center, as we read it, puts full MCP support with write actions in beta on Business, Enterprise and Edu and gives Pro accounts read access only. - On a Business workspace only an admin or owner can turn on developer mode and publish the app, and on Enterprise and Edu you need access granted by an admin. If you do not see the setting under Settings, Security and login, ask your admin. - A scheduled task has no one watching it. Write it as report only: StoreSEO tells the agent to preview and wait for your yes, but that yes is an instruction, and a task paired with a rule that takes actions without asking could apply changes unattended. - Generated previews spend AI Content Optimizer credit, so a recurring task that previews content every week uses credit every week. Keep scheduled work to reads unless you want that spend. ## Run SEO checks on a schedule with your dot OpenAI's help center describes Scheduled Tasks that run once, repeat or monitor for changes. Ask your dot in plain words what to check, when, and where to report, and keep the task to reads: scores, image status, setup and Google reports. A scheduled run is unattended, so there is nobody to say yes to a change. StoreSEO tells the agent to preview and wait for your yes before it writes, spends credit or publishes, but that is an instruction, so word the task as report only and keep your custom rules set to ask before any action that changes your store. ```text Every Monday at 8:00, use StoreSEO to list my lowest-scoring products, any images without alt text and whether my llms.txt is live, and send me a short report. Report only and change nothing. If StoreSEO is unreachable, tell me instead of reusing last week's data. ``` ## Disconnect Remove the StoreSEO server in ChatGPT dots, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for ChatGPT dots: jobs, prompts and FAQs](/agent/chatgpt-dots) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [OpenAI: ChatGPT developer mode](https://developers.openai.com/api/docs/guides/developer-mode) (vendor documentation these steps were checked against) - [OpenAI help center: Getting started with your dot](https://help.openai.com/en/articles/20001530-getting-started-with-your-dot) (vendor documentation these steps were checked against) - [OpenAI help center: Developer mode and MCP apps](https://help.openai.com/en/articles/12584461) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Grok Bot Source: https://storeseo.com/docs/connect-storeseo-to-grok-bot Grok Bot is xAI's product for Bots you keep around: AI teammates with names, jobs and context that builds up over time. According to xAI's documentation, all your Bots share one persistent cloud computer with a browser, a filesystem and a terminal, which runs in Cursor's cloud, so work can continue while your device is closed. It runs in desktop and mobile apps and is included with paid individual Cursor plans, Cursor Teams, or a linked SuperGrok subscription. This guide connects Grok Bot to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Grok Bot can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Grok Bot](/agent/grok-bot). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** xAI's documentation says a custom remote HTTPS server uses the Bot's own credential, or each person's sign-in when the server uses OAuth. StoreSEO uses OAuth: you pick your Shopify store in the browser and approve access. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Grok Bot ### Step 1: Ask your Bot to add StoreSEO In a chat with your Bot, ask it to add a custom remote HTTPS MCP server with this URL. xAI's documentation says a plugin added from your own chat is personal and account-wide, so every Bot on your account can use it. Complete the StoreSEO sign-in when the browser opens: pick your Shopify store and approve access. ```text Add a custom remote HTTPS MCP server called StoreSEO with the URL https://mcp.storeseo.com/mcp and sign me in. ``` ### Step 2: Attach it and check it works In a chat, type @ to attach the StoreSEO connector to the task, then ask this to confirm the connection. ```text Use @StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Add it to a Team Bot from the Setup panel For a Team Bot, only its owner can change the setup, and the change reaches every teammate. The owner adds the plugin from the Plugins card in the Bot's Setup panel, choosing a custom MCP server over remote HTTPS, with this URL. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new session and send this prompt. Grok Bot should call StoreSEO and answer with your store's real SEO score. ```text Use @StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Grok Bot - Use @StoreSEO to find my lowest-scoring products and tell me what is wrong with each. - Which images in my store have no alt text? Preview alt text for five of them with @StoreSEO. - Is my llms.txt live and current? Tell me what generating a new one would change before you do anything. - Use @StoreSEO to show Search Console queries from the last 28 days where we rank between 8 and 20. - Which of my collections have no meta description? Preview one for the weakest and wait for my yes. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Grok Bot - A plugin added from your own chat is personal and account-wide. A Team Bot's owner adds it from the Plugins card in the Setup panel, and that reaches only that Team Bot's teammates. - Be careful with Always allow. It saves a rule for matching actions, so a StoreSEO write tool you always allow no longer stops for your yes in the client, and StoreSEO's preview-first instruction becomes the only check. Use Ask first rules in Auto-review for apply, generate and publish tools. - A routine's Test run performs real work, including calls to connected tools, so run it with safe inputs. Write each routine as report only. - xAI's documentation says Grok Bot inherits your team's Cursor connector policy, a blocked connector shows as Disabled by team admin, and the MCP allowlist is Enterprise only. On a team plan an admin may have to allow the StoreSEO server first. ## Run SEO checks on a schedule with a routine A routine tells one Bot when to run a workflow. Ask the Bot in plain words when to run, what to check, where to report and what it must not do, and it creates the routine and shows its next run. Background routines can run while your laptop is closed, and Routines under the conversation details let you pause one or read its recent runs. Keep a routine to reads and reports. Nobody is there to say yes to a change, and a Test run performs real work, so word it as report only and keep Ask first rules on StoreSEO's write tools. ```text Every Monday at 8:00 AM, use @StoreSEO to list my lowest-scoring products, any images without alt text and whether my llms.txt is live. Post a short report in this conversation. Report only and change nothing. If StoreSEO is unreachable, say so instead of reusing last week's data. ``` ## Disconnect Remove the StoreSEO server in Grok Bot, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Grok Bot: jobs, prompts and FAQs](/agent/grok-bot) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [xAI: Grok Bot overview](https://docs.x.ai/grok-bot/overview) (vendor documentation these steps were checked against) - [xAI: Connectors and plugins](https://docs.x.ai/grok-bot/computer-and-apps) (vendor documentation these steps were checked against) - [xAI: Team Bots](https://docs.x.ai/grok-bot/team-bots) (vendor documentation these steps were checked against) - [xAI: Routines](https://docs.x.ai/grok-bot/skills-routines-and-automations) (vendor documentation these steps were checked against) - [xAI: Approvals and security](https://docs.x.ai/grok-bot/approvals-security-and-privacy) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Muse Source: https://storeseo.com/docs/connect-storeseo-to-muse Muse is a personal AI agent from Meta. It is meant to complete tasks for you rather than only answer questions, and you talk to it like a message thread in the Muse app, on muse.ai or in WhatsApp. Meta says it is rolling out in the US first, and that it is free for most of what people need, with subscription plans for people who want to do more. This guide connects Muse to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Muse can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Muse](/agent/muse). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** StoreSEO uses OAuth: the sign-in opens a browser page where you pick your Shopify store and approve access. Muse has no connector form you fill in for this, so you start it by sending Muse a message that contains the StoreSEO MCP URL. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Muse ### Step 1: Send this message to Muse Paste it into a new chat with Muse in the app, on muse.ai or in WhatsApp. Muse is rolling out in the US first, so you need access to it before this works. ```text Add a custom connector for my Shopify SEO app StoreSEO: https://mcp.storeseo.com/mcp - It is an MCP server over HTTP and it signs in with OAuth. Send me the sign-in link and I will approve it. - Once it is connected, show me my store's SEO score breakdown. ``` ### Step 2: Approve the sign-in for your store Open the sign-in link Muse sends back, pick your Shopify store in the browser and approve access. Meta does not review custom connectors, so only approve this one if you started it. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new chat and send this prompt. Muse should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Muse - Use StoreSEO to show me which of my products have the lowest SEO scores. - Preview a better meta title and description for my best seller, and wait for my yes before you save it. - Which of my product images are missing alt text? Tell me how many before you change anything. - Check whether my store has an llms.txt file and tell me what generating one would change. - How did my store's organic search traffic change over the last 28 days? ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Muse - Muse is rolling out in the US first, so you may not have access yet. Check Meta's rollout before planning around it. - StoreSEO is a custom connector in Muse, not a listed one. Meta does not review custom connectors, so approve the sign-in only for the StoreSEO link you sent yourself. - Muse decides when to ask your permission. Do not rely on that alone: tell it to preview and ask before it applies, publishes or deletes anything, and read each question before you answer yes. - Pasted messages are not a settings screen, so there is no list of connected tools to switch off in Muse. To cut off access, revoke the connection in StoreSEO under Settings, AI Toolkit. ## Disconnect Remove the StoreSEO server in Muse, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Muse: jobs, prompts and FAQs](/agent/muse) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Meta: Introducing Muse](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) (vendor documentation these steps were checked against) - [Meta Help Center: How Muse works with Connectors](https://www.meta.com/help/artificial-intelligence/1687253048996149/) (vendor documentation these steps were checked against) - [Muse Connector Platform](https://muse.ai/platform) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Manus Source: https://storeseo.com/docs/connect-storeseo-to-manus Manus is a general AI agent. You describe a task in plain words and it works through the steps, such as research, writing and acting in other apps, until it has a result for you. This guide connects Manus to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Manus can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Manus](/agent/manus). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Manus documents custom MCP servers that authenticate with an API key, OAuth token or bearer token. Generate a personal access token in StoreSEO under Settings, AI Toolkit and enter it as the server's authentication. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Manus ### Step 1: Create an access token in StoreSEO In the StoreSEO app, open Settings, then AI Toolkit, and generate a named personal access token. Copy it now and keep it out of any chat. ```text StoreSEO → Settings → AI Toolkit → generate a token ``` ### Step 2: Add StoreSEO as a custom MCP server In Manus, open Settings, then Integrations, then Custom MCP Servers, and choose Add Server. Give it a name, paste this URL as the server URL and enter your token as the authentication, as a Bearer token. ```text https://mcp.storeseo.com/mcp ``` ### Step 3: Check it and use it Manus verifies it can reach the server and read its tools. Then start a task that names StoreSEO, so Manus uses the connector. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Check the Connection Start a new session and send this prompt. Manus should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Manus - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Use StoreSEO to find product images with no alt text and show me what you would write for my best-selling collection. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Use StoreSEO to pull my Search Console queries from the last 28 days where we rank between 8 and 20, and write me a short brief. - Check my store's SEO setup with StoreSEO and give me the three fixes that matter most. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Manus - StoreSEO is not in the Manus connector directory. You add it yourself as a custom MCP server, and you name it in your prompt so Manus uses it. - Manus documents API keys, OAuth tokens and bearer tokens for custom servers but does not describe a browser sign-in for them, so connect with a StoreSEO access token. Anyone with that token can use your StoreSEO tools, so revoke it in Settings, AI Toolkit if it leaks. - Scheduled tasks run unattended, and Manus offers a setting to skip confirmations. Do not turn that on for a StoreSEO task that writes. Keep scheduled tasks report-only. - Rank tracking is a Pro feature, Google reports need a paid plan with Google connected, and generated content spends AI credit. ## Schedule a weekly SEO check in Manus Manus documents scheduled tasks that run on a timetable, can use your connected apps and run unattended. For StoreSEO, make the task report-only: it should read scores and reports and write you a summary. Do not ask a scheduled task to apply content, generate files or delete anything, because you are not there to say yes, and leave the skip confirmations setting off. ```text Every Monday morning, use StoreSEO to read my store's SEO score breakdown, my lowest-scoring products and last week's Search Console insights. Summarize what changed. Do not apply, generate, publish or delete anything. ``` ## Disconnect Remove the StoreSEO server in Manus, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Manus: jobs, prompts and FAQs](/agent/manus) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Manus: custom MCP servers](https://manus.im/docs/integrations/custom-mcp) (vendor documentation these steps were checked against) - [Manus: MCP connectors](https://manus.im/docs/integrations/mcp-connectors) (vendor documentation these steps were checked against) - [Manus: automations](https://manus.im/docs/automations) (vendor documentation these steps were checked against) - [Manus: Scheduled Tasks 2.0](https://manus.im/blog/manus-schedules) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to OpenClaw Source: https://storeseo.com/docs/connect-storeseo-to-openclaw OpenClaw is an open-source AI agent runtime that you run on your own machine or server. You talk to it through chat apps such as WhatsApp, Telegram, Slack and Discord, or from its Control UI and CLI. It connects to MCP servers, and their tools pass through the same tool policy as everything else the agent can do. This guide connects OpenClaw to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, OpenClaw can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for OpenClaw](/agent/openclaw). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** openclaw mcp login prints an authorization URL. Open it, pick your Shopify store and approve. OpenClaw normally captures the redirect on a loopback address and saves the credentials on the installation that owns the connector, so scheduled automations reuse them. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to OpenClaw ### Step 1: Add StoreSEO to OpenClaw Run this on the installation that owns the connector. The --auth oauth flag marks the server as an OAuth server, which is what openclaw mcp login needs. ```bash openclaw mcp add storeseo \ --url https://mcp.storeseo.com/mcp \ --transport streamable-http \ --auth oauth ``` ### Step 2: Sign in to your store The login prints an authorization URL. Open it in a browser, pick your Shopify store and approve access. OpenClaw captures the redirect and saves the credentials. ```bash openclaw mcp login storeseo ``` ### Step 3: Check the connection The probe opens a live connection and reports the tools the server offers. A saved server entry proves nothing until the probe passes. ```bash openclaw mcp doctor storeseo --probe ``` ## Check the Connection Start a new session and send this prompt. OpenClaw should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask OpenClaw - Use StoreSEO to list my ten lowest-scoring products and the top issue on each. - Which of my products have no AI snippet or FAQ yet? Preview them for the first three and wait for my yes. - Pull my Search Console queries from the last 28 days where we rank between 8 and 20. - Is my agents.md live? If not, tell me what generating one will publish and what it costs before you run it. - Every Monday at 8am, send me my store's SEO score breakdown and the five weakest products. Change nothing. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About OpenClaw - Run the mcp commands on the installation that owns the connector. A Gateway or agent on another host needs its own reload, config publish or restart before it sees the storeseo server. - StoreSEO has 44 tools. OpenClaw lets you narrow them with openclaw mcp tools and its include and exclude options if you want a smaller set in front of the agent. - OpenClaw has its own per-server approval setting, openclaw mcp configure storeseo --approval followed by approve, prompt or auto. If you want the client to stop before StoreSEO writes, do not leave it on a mode that skips the check. - A scheduled run has nobody to say yes. Keep automations to reads and reports, and ask for changes in a message you can answer. ## Send a weekly SEO report with OpenClaw automations Automations are OpenClaw's built-in scheduler, and openclaw cron is an alias for the same commands. This job runs every Monday in an isolated session, reads your store's scores and announces the result in a Telegram chat. List past runs with openclaw automations runs followed by the job ID. A scheduled run should be report-only: the prompt asks for reads, because nobody is there to say yes to a write. ```bash openclaw automations create "0 8 * * 1" \ "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products. Read only: do not preview, apply or generate anything." \ --name "Monday SEO report" \ --tz "Europe/Berlin" \ --session isolated \ --announce --channel telegram --to "-1001234567890" ``` ## Disconnect Remove the StoreSEO server in OpenClaw, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for OpenClaw: jobs, prompts and FAQs](/agent/openclaw) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [OpenClaw: connect MCP servers](https://docs.openclaw.ai/tools/mcp) (vendor documentation these steps were checked against) - [OpenClaw: MCP registry commands](https://docs.openclaw.ai/cli/mcp/registry) (vendor documentation these steps were checked against) - [OpenClaw: automations](https://docs.openclaw.ai/automation/cron-jobs) (vendor documentation these steps were checked against) - [OpenClaw: manage automations](https://docs.openclaw.ai/automation/cron-jobs/managing-jobs) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Hermes Agent Source: https://storeseo.com/docs/connect-storeseo-to-hermes-agent Hermes Agent is an open-source AI agent from Nous Research. It keeps memory across sessions, runs from a terminal or from chat apps such as Telegram, Discord and Slack, and loads skills from its own folder under ~/.hermes. This guide connects Hermes Agent to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Hermes Agent can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Hermes Agent](/agent/hermes-agent). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Set auth: oauth and run hermes mcp login storeseo: pick your Shopify store in the browser and approve. Hermes caches the tokens locally and refreshes them. Without a browser, send a StoreSEO personal access token from Settings, AI Toolkit as an Authorization header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Hermes Agent ### Step 1: Add StoreSEO to config.yaml Add this under mcp_servers in ~/.hermes/config.yaml. If the file already has an mcp_servers block, add only the storeseo entry. Then start a new Hermes session, or run /reload-mcp in the one that is open. ```yaml mcp_servers: storeseo: url: "https://mcp.storeseo.com/mcp" auth: oauth ``` ### Step 2: Sign in to your store On first connect Hermes prints an authorize URL and waits for the sign-in. Run this command to start it yourself or to re-authorize later. Pick your Shopify store and approve. Hermes caches the tokens in ~/.hermes/mcp-tokens/storeseo.json. ```bash hermes mcp login storeseo ``` ## Other Ways to Connect ### Option: Connect with an access token On a machine with no browser, generate a token in StoreSEO under Settings, AI Toolkit and send it as a header in place of auth: oauth. Keep config.yaml out of version control when it holds a token. ```yaml mcp_servers: storeseo: url: "https://mcp.storeseo.com/mcp" headers: Authorization: "Bearer YOUR_TOKEN" ``` ## Check the Connection Start a new session and send this prompt. Hermes Agent should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Hermes Agent - Use StoreSEO to list my products scoring under 50 and preview new meta for the first five. - Which products have no alt text on their images? Show me the list before writing anything. - Pull my Search Console queries from the last 28 days and tell me which ones sit between positions 8 and 20. - Is my llms.txt current? If not, say what generating a new one will publish and cost before you run it. - Summarize my store's SEO score breakdown in five lines. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Hermes Agent - Start a new session or run /reload-mcp after editing config.yaml, otherwise the open session does not see the storeseo server. - A token in config.yaml is a secret. Keep the file out of version control, and revoke the token in StoreSEO under Settings, AI Toolkit if it leaks. - StoreSEO has 44 tools. If you want a read-only agent, use Hermes' tools include list to expose only the StoreSEO list and get tools. - A scheduled run has nobody to say yes. Keep cron jobs to reads and reports, and ask for changes in a chat you can answer. ## Send a weekly SEO report with hermes cron hermes cron create takes a schedule and a prompt, and --deliver sends the result to a chat target. Inside a chat you can use /cron add with the same arguments. A scheduled run should be report-only: ask for reads, because nobody is there to say yes to a write, and do not tell the job to apply or generate anything. ```bash hermes cron create "every monday 9am" \ "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products. Read only: do not preview, apply or generate anything." \ --name "Monday SEO report" \ --deliver telegram ``` ## Disconnect Remove the StoreSEO server in Hermes Agent, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Hermes Agent: jobs, prompts and FAQs](/agent/hermes-agent) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Hermes Agent: MCP](https://hermes-agent.nousresearch.com/docs/user-guide/features/mcp) (vendor documentation these steps were checked against) - [Hermes Agent: scheduled jobs](https://hermes-agent.nousresearch.com/docs/user-guide/features/cron) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Paperclip Source: https://storeseo.com/docs/connect-storeseo-to-paperclip Paperclip is an open-source, MIT-licensed platform for managing teams of AI agents at work. You define a goal, hire agents such as Claude Code, Codex or Hermes into an org chart, set budgets and approvals, and follow every task and tool call from one dashboard. It is self-hosted, needs no Paperclip account and starts with npx paperclipai onboard. This guide connects Paperclip to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Paperclip can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Paperclip](/agent/paperclip). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Paperclip runs its agents headless, so nobody is there to click a browser prompt. Sign in once with OAuth in an interactive session as the same OS user that runs the agent, or skip the browser and send a personal access token from StoreSEO under Settings, AI Toolkit as a Bearer header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Paperclip ### Step 1: Add StoreSEO to the agent Paperclip runs Run this on the machine that hosts the agent, as the same OS user the Paperclip heartbeat runs as, from the agent's working directory. It is the claude_local adapter path, and user scope makes StoreSEO available to every Claude Code agent that user runs. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp ``` ### Step 2: Sign in once, then let Paperclip run headless Start claude interactively as that same user, run /mcp, choose storeseo and finish the browser sign-in: pick your Shopify store and approve. Claude Code keeps the credentials, so later Paperclip runs need no browser. Check with claude mcp get storeseo. ```text /mcp ``` ## Other Ways to Connect ### Option: Connect with an access token If the host has no browser, generate a token in StoreSEO under Settings, AI Toolkit and add the server with the token as a header instead of signing in. Keep the token in a Paperclip secret rather than in a file you commit. ```bash claude mcp add --transport http --scope user storeseo https://mcp.storeseo.com/mcp \ --header "Authorization: Bearer YOUR_TOKEN" ``` ### Option: Connect it as a Paperclip connector On the Connectors page choose Connect your own MCP server and paste the URL. Paperclip walks you through the sign-in and lists StoreSEO's tools so you can set each one to Allowed, Ask first or Off. It shows as an Unverified server, which describes Paperclip's review, not the connection. ```text https://mcp.storeseo.com/mcp ``` ## Check the Connection Start a new session and send this prompt. Paperclip should call StoreSEO and answer with your store's real SEO score. ```text List the MCP tools you have access to without calling any of them, then use StoreSEO to show my store's SEO score breakdown. ``` ## What to Ask Paperclip - Check my StoreSEO scores and report the five lowest-scoring products on this ticket. Do not change anything. - Weekly review: list products with no meta description and images with no alt text, then summarize what a fix would involve. - Preview better meta for the lowest-scoring product, show me old and new, and wait for my reply before applying. - Pull my Search Console queries for the last 28 days and note the ones ranking between 8 and 20. - Is my llms.txt current? Report its status and tell me what generating a new one would change before running it. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Paperclip - The Claude Code adapter runs headless with dangerouslySkipPermissions on by default, which lets the agent use connected MCP tools without asking. Set it to false so a StoreSEO tool that needs approval does not run unattended, or use the Paperclip connector and keep write tools on Ask first. - Paperclip's action permissions and review queue apply to connectors only. A server added with claude mcp add on the adapter does not pass through them, so the agent's own settings and StoreSEO's confirmation are your checks there. - Remote sign-in cannot finish headless. Complete OAuth once in an interactive claude session as the same OS user that runs the heartbeat, or use a token. A 401 after some hours means the refresh failed and you should sign in again. - User scope in ~/.claude.json is shared by every agent that OS user runs, so every Claude Code agent on that host sees StoreSEO. Use a dedicated OS user or a project scope if only one agent should. ## Schedule a report-only store review Paperclip routines fire on a cron schedule or a webhook, create a task and wake the agent you assign. Keep any scheduled StoreSEO run to reads and a written report: nobody is there to say yes, and the Claude Code adapter does not stop to ask by default. Leave applying, generating and publishing for a task you start and approve yourself. ```text Weekly store review. Use StoreSEO to read my SEO scores, list the five lowest-scoring products and any Search Console drops, and post the findings on this task. Read only: do not preview, apply, generate or publish anything. ``` ## Disconnect Remove the StoreSEO server in Paperclip, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Paperclip: jobs, prompts and FAQs](/agent/paperclip) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Paperclip: add an MCP server to an agent](https://docs.paperclip.ing/how-to/add-mcp-server-to-agent/) (vendor documentation these steps were checked against) - [Paperclip: Claude Code adapter](https://docs.paperclip.ing/reference/adapters/claude-code/) (vendor documentation these steps were checked against) - [Paperclip: connect a custom MCP server](https://docs.paperclip.ing/connectors/custom-mcp-servers/) (vendor documentation these steps were checked against) - [Paperclip: heartbeats and routines](https://docs.paperclip.ing/guides/projects-workflow/routines/) (vendor documentation these steps were checked against) - [Paperclip](https://paperclip.ing) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to NanoClaw Source: https://storeseo.com/docs/connect-storeseo-to-nanoclaw NanoClaw is an open-source AI assistant that runs agents in their own containers. It is built on Anthropic's Claude Agent SDK and is meant to be small enough to read, so you own a fork of the code and change it to fit what you need rather than editing a settings file. This guide connects NanoClaw to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, NanoClaw can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for NanoClaw](/agent/nanoclaw). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** NanoClaw's documentation describes no browser sign-in for an MCP server inside the agent's container, so connect with a personal access token from StoreSEO under Settings, AI Toolkit, sent as an Authorization: Bearer header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to NanoClaw ### Step 1: Create a StoreSEO access token In the StoreSEO app open Settings, AI Toolkit and generate a personal access token. Keep it out of chat messages and out of Git. ```text Settings → AI Toolkit → create a token ``` ### Step 2: Add StoreSEO to your agent's container Run this on the machine that hosts NanoClaw, using your agent group's ID from ncl groups list. The header goes in as a JSON object. NanoClaw may ask an admin to approve the change. ```bash ncl groups config add-mcp-server --id --name storeseo \ --url https://mcp.storeseo.com/mcp \ --headers '{"Authorization":"Bearer YOUR_TOKEN"}' ``` ### Step 3: Restart the agent's container The new server takes effect after a restart. Then message your agent with its trigger word, @Andy by default, and ask for your store's SEO score. ```bash ncl groups restart --id ``` ## Other Ways to Connect ### Option: Ask Claude Code to wire it into your fork NanoClaw is customized by changing your own copy, and its README says to tell Claude Code what you want. Run this inside your NanoClaw folder if you would rather not type the command yourself. ```text Add the remote MCP server https://mcp.storeseo.com/mcp to my agent as storeseo, authenticated with this Authorization: Bearer token, then restart the agent. ``` ## Check the Connection Start a new session and send this prompt. NanoClaw should call StoreSEO and answer with your store's real SEO score. ```text @Andy use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask NanoClaw - @Andy use StoreSEO to list my five lowest-scoring products and what is wrong with each. - @Andy which of my product images have no alt text? Show me the first ten. - @Andy preview better meta for my best-selling collection and wait for my yes before saving anything. - @Andy is my llms.txt live? Tell me when it was last generated. - @Andy show Search Console queries from the last 28 days where we rank between 8 and 20. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About NanoClaw - NanoClaw's documentation describes no browser OAuth sign-in for an MCP server inside the container, so this setup uses a StoreSEO access token. Treat the token like a password: the header is stored in the agent group's container configuration on your host, so keep that machine and its backups private. - NanoClaw says secrets belong with its credential gateway rather than in the agent. If you run a gateway, you can hold the StoreSEO token there for mcp.storeseo.com instead of in the header, following NanoClaw's gateway documentation. - The server is added per agent group. Add it to each group that should reach your store, and restart that group's container afterwards. - NanoClaw needs Docker on your own machine and the setup is yours to maintain. Whether it asks you before a tool runs depends on how your fork and your admins are set up, so the StoreSEO confirmation in the chat may be the only check. ## Send a weekly SEO report with a NanoClaw scheduled task NanoClaw runs a prompt on a cron schedule in the agent's own session. A scheduled task has no chat attached, so the prompt has to say where to send the result. A scheduled run should be report-only: the prompt asks for reads, because nobody is there to say yes to a write. ```bash ncl tasks create \ --group \ --name "Monday SEO report" \ --recurrence "0 8 * * 1" \ --prompt "Use StoreSEO to report my SEO score breakdown and the five lowest-scoring products, and send it to telegram. Read only: do not preview, apply, generate or delete anything." ``` ## Disconnect Remove the StoreSEO server in NanoClaw, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for NanoClaw: jobs, prompts and FAQs](/agent/nanoclaw) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [NanoClaw on GitHub](https://github.com/nanocoai/nanoclaw) (vendor documentation these steps were checked against) - [NanoClaw scheduled tasks](https://github.com/nanocoai/nanoclaw/blob/main/docs/scheduled-tasks.md) (vendor documentation these steps were checked against) - [NanoClaw agent templates and MCP servers](https://github.com/nanocoai/nanoclaw/blob/main/docs/templates.md) (vendor documentation these steps were checked against) - [NanoClaw ncl groups commands](https://github.com/nanocoai/nanoclaw/blob/main/src/cli/resources/groups.ts) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to OpenCode Source: https://storeseo.com/docs/connect-storeseo-to-opencode OpenCode is an open-source AI coding agent built by the team behind SST, now Anomaly. It runs in your terminal, and also as a desktop and IDE app, and works through tasks by calling its built-in tools and tools from MCP servers. This guide connects OpenCode to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, OpenCode can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for OpenCode](/agent/opencode). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** OpenCode signs in with OAuth by default: run opencode mcp auth storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit, set oauth to false and send it as a header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to OpenCode ### Step 1: Add StoreSEO to OpenCode Run this from the project that should use StoreSEO, or add --global to make it available in every project. OpenCode writes a remote server entry to its configuration; the equivalent hand-written entry is shown in the next step. ```bash opencode mcp add storeseo --url https://mcp.storeseo.com/mcp ``` ### Step 2: Sign in to your store Remote servers use OAuth by default. Run the auth command, then pick your Shopify store in the browser and approve access. Inside OpenCode, /mcps does the same: select storeseo and sign in. opencode mcp list should then show it as connected. ```bash opencode mcp auth storeseo opencode mcp list ``` ## Other Ways to Connect ### Option: Write it in the config In OpenCode V2 servers go under mcp.servers. Older versions put the server directly under mcp, so use the layout your version documents. ```json { "$schema": "https://opencode.ai/config.json", "mcp": { "servers": { "storeseo": { "type": "remote", "url": "https://mcp.storeseo.com/mcp" } } } } ``` ### Option: Connect with an access token Generate a token in StoreSEO under Settings, AI Toolkit, export it as STORESEO_TOKEN, and turn OAuth off for this server so OpenCode sends the header instead. ```json { "mcp": { "servers": { "storeseo": { "type": "remote", "url": "https://mcp.storeseo.com/mcp", "oauth": false, "headers": { "Authorization": "Bearer {env:STORESEO_TOKEN}" } } } } } ``` ## Check the Connection Start a new session and send this prompt. OpenCode should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask OpenCode - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Show me the ten heaviest first. - Is my agents.md live? If it is only Shopify's default, tell me what a custom one would publish. - Show me Search Console queries from the last 28 days where we rank between 8 and 20. - Which schema types are enabled for my store, and what would turning on Product schema add? ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About OpenCode - OpenCode V2 expects servers under mcp.servers. Older versions put the server directly under mcp, so check the layout your installed version documents before you edit the file. - opencode mcp add writes to the project configuration unless you pass --global, so a server added without it only exists in that project. - There is no remove command for MCP servers. To remove StoreSEO, delete its entry from the config, and run opencode mcp logout storeseo to clear the stored sign-in. - The OpenCode docs describe deny rules for MCP tools but no ask-first mode, so StoreSEO's own confirmation in the chat is the main check on writes. Deny the write tools by name if you want a client-side stop. ## Disconnect Remove the StoreSEO server in OpenCode, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for OpenCode: jobs, prompts and FAQs](/agent/opencode) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [OpenCode: MCP servers](https://opencode.ai/v2/docs/mcp-servers) (vendor documentation these steps were checked against) - [OpenCode: MCP servers (earlier docs)](https://opencode.ai/docs/mcp-servers/) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Zed Source: https://storeseo.com/docs/connect-storeseo-to-zed Zed is a code editor from Zed Industries. It has a built-in Agent Panel where an AI agent reads your project and works through tasks, and it supports MCP tools and prompts through context servers. This guide connects Zed to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Zed can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Zed](/agent/zed). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** With no Authorization header in the entry, Zed prompts you to sign in through the standard MCP OAuth flow: pick your Shopify store and approve. To use a StoreSEO personal access token instead, add it as a Bearer header in the same entry. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Zed ### Step 1: Add StoreSEO as a remote server Open Settings, then AI, then MCP Servers and add a remote server, or add this to settings.json yourself. Leave out headers so Zed can sign you in. ```json { "context_servers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Sign in and check the dot Because the entry has no Authorization header, Zed prompts you to authenticate. Pick your Shopify store and approve in the browser. In Settings, AI, MCP Servers a green dot next to storeseo means the server is active. Then ask this in the Agent Panel. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Connect with an access token If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Keep settings.json out of a public repository when it holds a token. ```json { "context_servers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` ## Check the Connection Start a new session and send this prompt. Zed should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Zed - Use StoreSEO to list every product scoring under 50 and preview new meta for the first five. - Which product images have no alt text? Show me the list before you change anything. - Is product schema turned on for my store? If not, tell me what enabling it changes. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Open the theme file that renders my FAQ section and tell me whether StoreSEO's FAQ output will show there. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Zed - Zed's default is to ask before every tool action, reads included, so a catalog audit means several prompts. Allow only the read tools with per-tool keys such as mcp:storeseo:storeseo_list_resources and leave the write tools on confirm. - Setting agent.tool_permissions.default to allow removes Zed's check. Then the confirmation StoreSEO asks the agent for in the chat is the only one, so use allow sparingly. - A token in settings.json is a secret. Revoke it in StoreSEO under Settings, AI Toolkit if the file is exposed. ## Disconnect Remove the StoreSEO server in Zed, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Zed: jobs, prompts and FAQs](/agent/zed) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Zed: Model Context Protocol](https://zed.dev/docs/ai/mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Kiro Source: https://storeseo.com/docs/connect-storeseo-to-kiro Kiro is an agentic IDE built and operated by AWS. An AI agent works on your project from a chat panel, plans multi-step changes and can call tools exposed through the Model Context Protocol, including remote HTTP servers. This guide connects Kiro to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Kiro can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Kiro](/agent/kiro). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Kiro handles the browser OAuth flow for remote servers and supports Dynamic Client Registration: pick your Shopify store and approve. To use a StoreSEO personal access token instead, add an Authorization Bearer value under headers in the same entry. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Kiro ### Step 1: Add StoreSEO to your Kiro MCP config Open the Command Palette (Cmd+Shift+P on Mac, Ctrl+Shift+P on Windows and Linux), search for MCP and open the workspace config (.kiro/settings/mcp.json) or the user config (~/.kiro/settings/mcp.json). Paste this and save. Kiro applies the change on save, and the first connection opens a browser sign-in: pick your Shopify store and approve. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ## Other Ways to Connect ### Option: Choose which StoreSEO tools Kiro may run Optional. autoApprove lists tools Kiro runs without asking, and disabledTools hides tools from the agent. The names here are real StoreSEO read tools. Keep write tools such as apply, generate and update out of autoApprove so Kiro's own prompt stays as a second check. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "autoApprove": ["storeseo_start", "storeseo_list_resources", "storeseo_get_resource_seo"] } } } ``` ### Option: Connect with an access token If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and send it in a headers entry. Keep the file out of a public repository when it holds a token. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` ## Check the Connection Start a new session and send this prompt. Kiro should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Kiro - Use StoreSEO to list my five lowest-scoring products and preview new meta for each. - Which product images have no alt text? List them before changing anything. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Turn on product schema and tell me what that changes before you do it. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Kiro - A workspace .kiro/settings/mcp.json overrides a user-level entry with the same name, so if the sign-in seems to point at the wrong store, check which file defines storeseo. - autoApprove with a wildcard would run every StoreSEO tool without Kiro asking, writes included. List individual read tools instead, so Kiro's prompt stays a second check. - A token in mcp.json is a secret. Do not commit a workspace file that holds one, and revoke the token in StoreSEO under Settings, AI Toolkit if it leaks. ## Disconnect Remove the StoreSEO server in Kiro, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Kiro: jobs, prompts and FAQs](/agent/kiro) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Kiro: MCP configuration](https://kiro.dev/docs/mcp/configuration/) (vendor documentation these steps were checked against) - [Kiro: MCP overview](https://kiro.dev/docs/mcp/) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Antigravity Source: https://storeseo.com/docs/connect-storeseo-to-antigravity Antigravity is Google's agentic coding product. Its agent plans multi-step work in your project and can call tools exposed through the Model Context Protocol, from the IDE's agent side panel. This guide connects Antigravity to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Antigravity can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Antigravity](/agent/antigravity). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Antigravity supports OAuth through dynamic client registration, which StoreSEO's server offers. Click Authenticate next to the server, approve your Shopify store in the browser, then paste the authorization code back. Or send a StoreSEO personal access token as an Authorization Bearer header. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Antigravity ### Step 1: Add StoreSEO in View raw config In the Antigravity IDE open the menu in the agent side panel, choose MCP Servers, then Manage MCP Servers, then View raw config. Paste this into mcp_config.json and save. The field is serverUrl: Antigravity does not support url or httpUrl for remote servers. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Sign in, then check the connection Open Agent settings (Cmd+, on Mac, Ctrl+, on Windows and Linux), go to Customizations and click Authenticate next to storeseo. Pick your Shopify store and approve in the browser, copy the authorization code it shows, paste it into the settings panel and click Submit. Then ask this in the agent panel. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## Other Ways to Connect ### Option: Connect with an access token If you cannot use the browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header. Keep mcp_config.json out of a public repository when it holds a token. ```json { "mcpServers": { "storeseo": { "serverUrl": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` ## Check the Connection Start a new session and send this prompt. Antigravity should call StoreSEO and answer with your store's real SEO score. ```text Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Antigravity - Use StoreSEO to list my five lowest-scoring products and preview new meta for each. - Which product images are not optimized? Show me the list before you compress anything. - Is my llms.txt up to date? If not, say what generating a new one will publish and cost. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Which products have no FAQ or AI snippet yet? Preview them for the first three. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Antigravity - The field is serverUrl. Antigravity's documentation says legacy fields such as url or httpUrl are not supported, so an entry copied from another client will not connect until you rename it. - The sign-in ends with a code you paste back into Agent settings, under Customizations, so do it in the same window where you started it. - If you add a policy that allows mcp(storeseo/*), Antigravity stops asking and the confirmation StoreSEO asks the agent for is the only check. Allow individual read tools instead. ## Disconnect Remove the StoreSEO server in Antigravity, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Antigravity: jobs, prompts and FAQs](/agent/antigravity) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Antigravity: MCP](https://antigravity.google/docs/mcp) (vendor documentation these steps were checked against) --- # How to Connect StoreSEO MCP to Kimi Code Source: https://storeseo.com/docs/connect-storeseo-to-kimi Kimi Code is Moonshot AI's terminal coding agent. You start it with the kimi command in a project folder, and it reads and edits files, runs commands, searches the code and fetches web pages while you watch. It runs the Kimi model, and you sign in with /login using either a Kimi Code OAuth account or a Moonshot platform API key. This guide connects Kimi Code to the StoreSEO MCP server at `https://mcp.storeseo.com/mcp`. Once connected, Kimi Code can read your Shopify store's SEO scores and settings and, when you agree, apply optimized meta, alt text, schema and other SEO changes. For what to ask once you are set up, see [StoreSEO for Kimi Code](/agent/kimi). ## Before You Start - **Server URL:** `https://mcp.storeseo.com/mcp` - **Sign-in:** Kimi Code signs you in with OAuth: add the server, run /mcp-config login storeseo, pick your Shopify store in the browser and approve. If you prefer a token, generate one in StoreSEO under Settings, AI Toolkit and put it in the server's headers. - **Plan:** The MCP server works on any active StoreSEO plan, including the free plan. A store with no active plan gets a message with a link to choose one in the app. - **Access token (if you need one):** create it in the StoreSEO app under Settings, AI Toolkit. Treat it like a password. ## Connect StoreSEO to Kimi Code ### Step 1: Add StoreSEO to Kimi Code Create or edit ~/.kimi-code/mcp.json so every project on your machine can use StoreSEO. For one repository only, use .kimi-code/mcp.json inside it. You can also run /mcp-config in Kimi Code and have it add the server for you. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp" } } } ``` ### Step 2: Sign in to your store Start a new Kimi Code session (a server added mid-session registers in the next one), run this command, then follow the browser window: pick your Shopify store and approve access. ```text /mcp-config login storeseo ``` ## Other Ways to Connect ### Option: Connect with an access token If you cannot use a browser sign-in, generate a token in StoreSEO under Settings, AI Toolkit and add it as a header in the same server entry. ```json { "mcpServers": { "storeseo": { "url": "https://mcp.storeseo.com/mcp", "headers": { "Authorization": "Bearer YOUR_TOKEN" } } } } ``` ## Check the Connection Start a new session and send this prompt. Kimi Code should call StoreSEO and answer with your store's real SEO score. ```text Run /mcp to confirm storeseo is connected, then ask: Use StoreSEO to show me my store's SEO score breakdown. ``` ## What to Ask Kimi Code - Use StoreSEO to list every product scoring under 50 and preview new meta for the first ten. - Which product images have no alt text? Write it for my top collection, one image at a time. - Is my llms.txt up to date? If not, tell me what generating a new one will change before you run it. - Show me Search Console queries for the last 28 days where we rank between 8 and 20. - Check the product schema on my store and tell me which products are missing it. ## How Approvals Work Reads such as SEO scores, store listings, settings and Google reports run straight away. Generating a preview spends AI credit but writes nothing, and StoreSEO tells the agent to say what a generation will cost before it runs one. Applying content, optimizing images, changing settings, publishing llms.txt or agents.md, writing a product description and deleting anything wait for your yes in the conversation. That yes is an instruction the agent follows, not a lock on the server, so your client's own tool-approval setting is the hard stop. ## Things to Know About Kimi Code - A server added while Kimi Code is open does not register until the next session starts, so restart before you run /mcp-config login. - Kimi Code reads read-only tool calls without asking, but asks for the rest by default. The /yolo and /auto modes and Approve for this session reduce those prompts; keep apply, write and generate tools on ask so the confirmation StoreSEO asks for is not your only check. - A project file at .kimi-code/mcp.json overrides a user entry with the same name, so a stale storeseo entry in a repository can shadow the one in your home folder. - Older Kimi CLI builds keep their settings in a different folder than ~/.kimi-code. These steps are for Kimi Code CLI. ## Disconnect Remove the StoreSEO server in Kimi Code, then revoke the connection or the access token in the StoreSEO app under Settings, AI Toolkit, where every connected client is listed. ## Related - [StoreSEO for Kimi Code: jobs, prompts and FAQs](/agent/kimi) - [Connect StoreSEO MCP to any AI assistant](/docs/connect-storeseo-mcp-to-your-ai-assistant) - [Every AI agent that works with StoreSEO](/agent) - [Kimi Code CLI: MCP configuration](https://moonshotai.github.io/kimi-code/en/customization/mcp.html) (vendor documentation these steps were checked against) - [Kimi Code CLI: slash commands](https://moonshotai.github.io/kimi-code/en/reference/slash-commands) (vendor documentation these steps were checked against) - [Kimi Code CLI: interaction and approvals](https://moonshotai.github.io/kimi-code/en/guides/interaction) (vendor documentation these steps were checked against) - [Kimi Code CLI on GitHub](https://github.com/MoonshotAI/kimi-code) (vendor documentation these steps were checked against)