Hello guys
I’ve created MCP server which allows LLM agent in such IDE like Cursor to build n8n workflows with nodes. It’s an early stage, but it confirms this idea can work.
I’m worried about licence of this MCP server.
I would like to give it away to you (n8n team) or continue development if you approve it after looking at repo.
Basically, it was developed using Cursor with Claude 3.7 sonnet.
This is really nice, thank you for sharing!
I think the results will be excellent, especially since the node schema files in /workflow_nodes are being used, That’s the key for the AI to understand and generate a perfect workflow.
Just curious, where do you extract them from?
I’m guessing from /types/nodes.json?
Thanks for reply!
Just curious, where do you extract them from?
I went to n8n repo and asked agent to create me python script. First I asked it to investigate the structure of /packages/nodes-base/nodes so it would be able to determine syntax. Then came across some missing parameters in nodes while debugging. The script was improved with time. I don’t keep it in repo right now cuz don’t know if it’s legal in terms n8n product. But this script will be really helpful for further updates in n8n, especially if n8n team will be to introduce new nodes. That’s how this MCP server can be keep up-to-date.