Back to MCP Servers

n8n MCP Server

Community MCP server that gives AI agents structured access to n8n nodes, workflow templates, and the n8n management API for building and operating workflows.

Automation by czlonkowski (community) API Key active
Overview

n8n-MCP is a community-maintained Model Context Protocol server that bridges n8n's workflow automation platform with AI assistants like Claude, Cursor, and Windsurf. It exposes structured documentation for the full n8n node catalog (the README claims 1,650+ nodes across core and community), along with workflow templates and node property metadata, so AI agents can generate, validate, and deploy n8n workflows with high fidelity instead of guessing at JSON shapes.

The server provides two layers of tooling. The core layer works without any credentials and offers node search, node inspection, workflow validation, and template discovery. The management layer requires an n8n API URL and API key, and adds CRUD operations on workflows, execution control, credential management, and instance health/audit tools. This means an agent can both research the right nodes and then create or update real workflows on a running n8n instance.

This is not an official n8n project. It is built and maintained by czlonkowski and is the most popular community implementation, with a hosted dashboard option at dashboard.n8n-mcp.com for users who do not want to self-host. n8n itself ships separate first-party building blocks (the "MCP Server Trigger" and "MCP Client Tool" nodes inside n8n) which solve a different problem: turning n8n workflows into MCP tools and consuming external MCP servers from within a workflow.

Tools

Tool Description
tools_documentation Reference guide describing the available tools and how to use them.
search_nodes Full-text search across n8n nodes with filtering for community or verified nodes.
get_node Retrieve node details in minimal, standard, or full mode including docs, properties, and version history.
validate_node Validate a node configuration with minimal or full profiles.
validate_workflow Validate a complete workflow including AI Agent nodes and connections.
search_templates Find workflow templates by keyword, node type, task, or metadata.
get_template Fetch full workflow JSON for a template.
n8n_create_workflow Create a new workflow on the connected n8n instance.
n8n_get_workflow Retrieve an existing workflow by ID.
n8n_update_full_workflow Replace an existing workflow with a new definition.
n8n_update_partial_workflow Apply targeted updates to a workflow without full replacement.
n8n_delete_workflow Delete a workflow from the n8n instance.
n8n_list_workflows List workflows on the connected instance.
n8n_validate_workflow Validate a workflow against the live n8n instance.
n8n_autofix_workflow Automatically apply fixes to a workflow definition.
n8n_workflow_versions List previous versions of a workflow.
n8n_deploy_template Deploy a template as a new workflow.
n8n_test_workflow Trigger a test run of a workflow.
n8n_executions List or inspect workflow executions.
n8n_manage_credentials Manage credentials on the n8n instance.
n8n_audit_instance Run an audit against the connected n8n instance.
n8n_health_check Check the health/status of the n8n instance.
Setup Guide

Prerequisites

  • Node.js 18+ (for the npx install path) or Docker
  • An MCP-compatible client (Claude Desktop, Claude Code, Cursor, Windsurf, VS Code)
  • Optional but recommended: a running n8n instance with an API key (Settings, n8n API) for management tools

Install via npx (recommended)

Add the following to your claude_desktop_config.json (or .mcp.json for Claude Code / Cursor):

{
  "mcpServers": {
    "n8n-mcp": {
      "command": "npx",
      "args": ["n8n-mcp"],
      "env": {
        "MCP_MODE": "stdio",
        "LOG_LEVEL": "error",
        "DISABLE_CONSOLE_OUTPUT": "true",
        "N8N_API_URL": "https://your-n8n-instance.com",
        "N8N_API_KEY": "your-api-key"
      }
    }
  }
}

If you omit N8N_API_URL and N8N_API_KEY, you still get the documentation, search, validation, and template tools. Adding them unlocks the workflow and execution management tools.

Install via Docker

A prebuilt image is published at ghcr.io/czlonkowski/n8n-mcp. The repo also includes a docker-compose.yml for self-hosting. Pass the same N8N_API_URL and N8N_API_KEY env vars to the container.

Hosted option

For quick evaluation, the maintainer hosts a dashboard at dashboard.n8n-mcp.com with a free tier (100 tool calls per day).

Use Cases
  • Build a new n8n workflow from a natural language description, with the agent looking up correct node types and properties before generating JSON.
  • Validate and autofix an existing workflow before promoting it to production, catching missing required fields or invalid connections.
  • Search the template library for a starting point (for example "Slack to HubSpot lead routing") and deploy it to your n8n instance.
  • Inspect recent executions of a workflow to debug a failure, then patch the workflow with n8n_update_partial_workflow.
  • Audit an n8n instance for misconfigured credentials, outdated nodes, or security issues.
Example Prompts
  • "Search n8n nodes for ones that can read Gmail and show me the required parameters."
  • "Build a workflow that triggers on a new Stripe charge, looks up the customer in HubSpot, and posts a Slack message. Validate it before saving."
  • "List my last 20 workflow executions and find the ones that failed in the last 24 hours."
  • "Find a template for syncing Airtable rows to Google Sheets and deploy it to my n8n instance."
  • "Autofix the workflow with ID 142 and run a test execution against this sample payload."
Pros
  • Very comprehensive: claims 99% node property coverage and 2,300+ templates, far beyond a simple API wrapper.
  • Two-tier design works with or without n8n API credentials, so you can use it for workflow research even without a server.
  • Multiple deploy paths: npx, Docker image, Railway one-click, or a hosted dashboard with a free tier.
  • Actively maintained with frequent releases and a large community following.
Limitations
  • Community-maintained, not an official n8n project, so support and roadmap depend on a single maintainer.
  • Management tools require an n8n API key, and the maintainer explicitly warns against pointing the agent at production workflows.
  • Surface area is large (20+ tools), which can confuse smaller models and inflate context usage.
Alternatives