MCP: n8n as Server and Client

What Is MCP?

The Model Context Protocol (MCP) is an open standard for how AI models communicate with external tools and data sources. Instead of every AI application building its own custom integrations, MCP provides a universal interface: any MCP-compatible client (Claude Desktop, Cursor, custom agents) can connect to any MCP-compatible server and call its tools through a standardised protocol. n8n added native MCP support in v1.88.0.

Two MCP Nodes with Opposite Roles

n8n's MCP integration consists of two nodes:

• MCP Server Trigger — turns an n8n workflow into an MCP server. Any MCP-compatible client pointed at the generated endpoint URL can call the workflow's tools remotely.
• MCP Client Tool — connects an n8n AI Agent to an external MCP server, giving the agent access to that server's tools without any custom integration code.

MCP Server Trigger: Exposing n8n Workflows as Tools

When you add an MCP Server Trigger node to a workflow, n8n generates a Streamable HTTP endpoint URL. Any MCP client pointed at this URL can discover and call the tools attached to that trigger. The tools exposed are whatever sub-nodes you connect to the MCP Server Trigger.

A practical example: build a workflow that queries your internal Postgres database and returns formatted results. Add an MCP Server Trigger. Point Claude Desktop at that URL. Claude can now query your database directly as a tool — no custom API layer required, no separate backend to maintain.

MCP Client Tool: Consuming External MCP Servers

The MCP Client Tool sub-node works in the opposite direction — it connects your n8n AI Agent to an external MCP server. Add it to the Tools input of an AI Agent, configure the external MCP server's URL and authentication, and the agent gains access to all of that server's tools automatically.

External MCP servers you can connect to include:

• Filesystem MCP server — read and write files on a local or remote filesystem.
• GitHub MCP server — query repositories, open issues, create pull requests.
• Brave Search MCP server — web search via the MCP standard.
• Custom internal tools — any service your team exposes as an MCP server becomes immediately accessible to n8n agents.

Why MCP Matters for n8n Users

MCP is valuable because it removes integration silos. A workflow you build in n8n — a CRM query, a document generator, a data pipeline trigger — becomes a tool that any compatible AI client can use. You are not building integrations for one specific AI tool; you are publishing tools that work across the entire MCP ecosystem. This means n8n workflows become reusable infrastructure that AI assistants, IDE extensions, and custom agents can all call.

Getting Started

Both MCP nodes appear in the n8n canvas like any other node. The MCP Server Trigger requires no external dependencies — n8n hosts and serves the endpoint. The MCP Client Tool requires the URL of the external MCP server you are connecting to, plus any required credentials. Start by exposing one simple workflow as an MCP server and testing it from Claude Desktop to see the integration in action end-to-end.