Connect to the n8n docs MCP server

The n8n documentation publishes two Model Context Protocol (MCP) servers. Connect an AI tool to a server to let it search n8n's knowledge and answer your questions as you work.

The two servers draw on different sources, so pick the one that fits how you want your AI tool to answer:

Server	Sources	Best for
GitBook docs server	The published n8n docs only	Searching and reading exact documentation pages. The docs are the single, authoritative source.
Kapa.ai server	The docs, the community forum, and the n8n blog	Broader questions that the docs alone don't cover, such as troubleshooting and real-world examples. This is the same engine that powers the AI Assistant on the docs site.

You can connect both. Use the GitBook server when you want answers grounded in the docs, and the Kapa.ai server when you want answers drawn from n8n's wider knowledge base.

Both servers use HTTP transport, so your AI tool needs to support remote MCP servers over HTTP. Neither supports stdio or SSE.

MCP server URLs

Use the endpoint for the server you want:

Server	URL
GitBook docs server	https://docs.n8n.io/~gitbook/mcp
Kapa.ai server	https://n8n.mcp.kapa.ai

Connect your AI tool

The steps differ by tool, but each one needs a server URL from above. The following examples set up both servers. To connect only one, keep the entry you want and remove the other.

The Kapa.ai server requires authentication. Your AI tool opens a browser-based sign-in flow when you first connect to it, or the first time you use it. Follow the prompts to authorize the connection.

Claude Code

Run these commands in your terminal:

claude mcp add --transport http n8n-docs https://docs.n8n.io/~gitbook/mcp
claude mcp add --transport http n8n-kapa https://n8n.mcp.kapa.ai

Cursor

VS Code

Other tools

After you connect, your tool can search n8n's knowledge as you work.

The exact configuration steps and file locations vary between tools and versions. Check your tool's documentation for how it adds a remote MCP server. For more on the Kapa.ai server, see the Kapa.ai MCP documentation.