n8n‑MCP – AI‑friendly knowledge base for the n8n workflow platform

What it is

• A server that exposes the full catalog of n8n workflow nodes (≈3 000 nodes) as a Model Context Protocol (MCP) API.
• AI assistants (Claude, OpenAI, etc.) can call this API to get structured information about node properties, operations, documentation, example configurations, and ready‑made workflow templates.

Why it matters

• Building n8n automations with an LLM is hard because the model needs up‑to‑date details about every node. n8n‑MCP gives the model a reliable, searchable knowledge source, turning vague prompts into concrete, runnable workflows.
• The service also bundles validation tools, so the AI can check that a generated node configuration is complete and will not fail at runtime.

Key features (as described in the README)

Feature	What you get
Node catalog	2 951 nodes (838 core + 2 113 community), with 99 % of their property schemas covered.
Operations coverage	About two‑thirds of all node actions are described (66.5 %).
Documentation	86 % of official n8n docs are indexed, including the AI‑specific nodes.
AI‑capable tools	267 node variants flagged as usable by AI agents, with full docs.
Real‑world examples	156 ranked configurations extracted from popular community templates.
Template library	2 352 ready‑to‑use workflow templates, each annotated with AI‑friendly metadata (99.96 % coverage).
Search	Full‑text node and template search, filters for community vs. verified nodes, and ability to pull example configs.
Validation	validate_node (quick minimal or full runtime checks) and validate_workflow to catch missing parameters before deployment.
Management tools	When you provide your own n8n instance credentials, the API can create, update, delete, and test workflows directly, explore dynamic dropdown values, and manage folders, data tables, etc.
Safety guidance	Strong warnings not to edit production workflows directly with AI and to always back up and test changes.

How you use it

1. Get an API key – sign up at the hosted dashboard (dashboard.n8n-mcp.com) or self‑host via Docker, Railway, etc.
2. Ask the AI – an LLM (e.g., Claude) calls MCP tools like search_nodes, get_node, validate_node, search_templates, then assembles a workflow JSON.
3. Deploy – with n8n_create_workflow / n8n_update_partial_workflow the generated workflow can be pushed to your own n8n server.
4. Iterate – run n8n_test_workflow, fix any validation errors, and redeploy.

Who it’s for

• Developers who want to automate the creation of n8n workflows using LLMs.
• Teams building internal AI assistants that need to understand the n8n ecosystem.
• Anyone who prefers a programmatic, up‑to‑date source of node specs rather than scraping the n8n docs manually.

Getting started

• Hosted demo: Visit the dashboard link, sign up, receive a free tier API key (100 tool calls/day).
• Self‑host: Follow the SELF_HOSTING.md guide for Docker, npm (npx n8n-mcp), or Railway deployment.
• Connect to n8n: Set N8N_API_URL and N8N_API_KEY (or Cloudflare Access tokens) so the MCP can talk to your n8n instance.
• IDE integration: The repo provides quick‑start docs for Claude Code, VS Code, Cursor, etc., letting you invoke MCP tools directly from your editor.

Bottom line n8n‑MCP turns the sprawling n8n node ecosystem into a clean, searchable, AI‑ready knowledge base, enabling large language models to build, validate, and deploy real n8n workflows with far fewer mistakes.