Connect with MCP

The Conduit docs are served as an MCP (Model Context Protocol) endpoint, so any AI assistant that speaks MCP can search and read them directly. It exposes two tools, search and fetch, backed by the same content this site renders, so it never goes stale.

Endpoint: https://conduit-docs.vercel.app/mcp

It is read-only, public, and stateless. No API key, no login.

Claude Code

claude mcp add --transport http conduit-docs https://conduit-docs.vercel.app/mcp

Cursor

Add to .cursor/mcp.json in your project (or the global one):

{
  "mcpServers": {
    "conduit-docs": {
      "url": "https://conduit-docs.vercel.app/mcp"
    }
  }
}

VS Code

Add to .vscode/mcp.json (note the top-level servers key):

{
  "servers": {
    "conduit-docs": {
      "type": "http",
      "url": "https://conduit-docs.vercel.app/mcp"
    }
  }
}

Claude Desktop and Windsurf

These use a local bridge to reach a remote endpoint. Add to the MCP config:

{
  "mcpServers": {
    "conduit-docs": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://conduit-docs.vercel.app/mcp"]
    }
  }
}

ChatGPT

In a Deep Research or connector-enabled workspace, add a custom connector and paste the endpoint URL:

https://conduit-docs.vercel.app/mcp

The search and fetch tools follow the connector contract ChatGPT expects.

No MCP client?

Every chatbot can still read the plain-text exports:

• /llms.txt is a compact index of the whole site.
• /llms-full.txt is the full docs in one file.

Paste either into any chat, or point a tool that accepts a URL at them.