n8n-MCP lets Claude, Cursor build n8n workflows

The smartest thing n8n-MCP does is replace prompt-to-JSON guesswork with grounded schema access, examples, and validation. That turns AI workflow building from a demo trick into something a team can actually trust.

• –The repo claims coverage across 1,084 nodes, 2,709 templates, and 2,646 extracted configs, which is the kind of breadth that makes the MCP layer actually useful.
• –Validation is the real moat: catching missing params, AI-agent wiring issues, and workflow mistakes before deployment matters more than flashy natural-language generation.
• –Support for Claude Desktop, Claude Code, Cursor, Windsurf, VS Code, plus a hosted free tier, lowers the friction from first try to first workflow.
• –The open-source MIT license and self-hosting options make it easier for teams to adopt without handing sensitive automation over to a closed black box.
• –Strong GitHub traction suggests the pain point is real; developers want agents that understand n8n, not just ones that can describe it.