How to Connect StoreSEO MCP 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.

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.

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.

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.

[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.

[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.

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
• Connect StoreSEO MCP to any AI assistant
• Every AI agent that works with StoreSEO
• OpenAI Codex: Model Context Protocol (vendor documentation these steps were checked against)