One Page Love for Codex

Bring website and section references into the same task where Codex designs and builds your page.

Free during beta. No One Page Love account or API key.

Connect in three steps

  1. Add the MCP server

    With the Codex CLI installed, run this command in your terminal.

    codex mcp add onepagelove --url https://api.onepagelove.com/mcp
  2. Or edit your configuration

    Alternatively, add this entry to ~/.codex/config.toml. Keep the other settings and servers in that file. You only need one of these setup methods.

    [mcp_servers.onepagelove]
    url = "https://api.onepagelove.com/mcp"
  3. Verify and start a new task

    Check the configured servers with the command below, then start a fresh Codex session. Ask it to use One Page Love before building.

    codex mcp list

Check it works

Ask Codex to search One Page Love for three websites. Confirm it uses the MCP tools and returns source links, rather than relying on general suggestions.

Start with your next project

Ask Codex to inspect the result at desktop and mobile sizes, then use the references to refine the composition. It should interpret the ideas, not copy another site’s content or assets.

Improve the homepage in this project. Use One Page Love to study six relevant websites and their hero sections. Choose the strongest ideas for our content, explain the direction, then implement it with our existing design system.

A few things to know

Not seeing One Page Love?

If no tools appear, restart the session and check the server URL in your configuration. A managed environment may restrict MCP servers. If Codex skips research, explicitly ask it to use One Page Love before editing.

What does the connection do?

It makes One Page Love’s website and section searches available to your AI tool. Your agent can inspect references and apply the ideas using its own design and coding capabilities. The MCP itself does not build or host your website.

Will the results look like the demo?

Image presentation depends on your client. Some show screenshots inline, some show links, and others let the model inspect an image without displaying it. The demo illustrates the workflow, not an exact client interface.

What data is shared?

Your AI tool sends search arguments to One Page Love. Queries, filters, result counts, request metadata and IP addresses are logged to operate the service. Avoid putting sensitive project information in search requests. Review your client’s tool permissions before connecting.