This is a 100% open-source project, I am a non-profit LLM hobbyist/advocate. I hope people find this interesting or useful, I’ll actively work on improving it.
How this idea was born:
I was looking for an easy way to integrate new MCP capabilities into my pair programming workflows. I found that some tools I already use offer OpenAPI specs (like Swagger and ETAPI), so I wrote a tool that reads the YAML API spec and translates it into an MCP server.
I’ve already tested it with my note-taking app (Trilium Next), and the results look promising. I’d love constructive and orientating feedback from anyone willing to throw an API spec at my tool to see if it can crunch it into something useful.
Right now, the tool generates MCP servers via Docker with SSE port exposed, but if you need another format, let me know and I can probably help you set it up.
The next step for the generator (as I see it) is recursion: making it usable as an MCP tool itself. That way, when an LLM discovers a new endpoint, it can automatically search for the spec (GitHub/docs/user-provided, etc.) and start utilizing it.
abutbul•7mo ago
I would really love to get some more feedback from users. If you try to set it up and can't, I am offering personal assistance!
abutbul•8mo ago
How this idea was born: I was looking for an easy way to integrate new MCP capabilities into my pair programming workflows. I found that some tools I already use offer OpenAPI specs (like Swagger and ETAPI), so I wrote a tool that reads the YAML API spec and translates it into an MCP server.
I’ve already tested it with my note-taking app (Trilium Next), and the results look promising. I’d love constructive and orientating feedback from anyone willing to throw an API spec at my tool to see if it can crunch it into something useful. Right now, the tool generates MCP servers via Docker with SSE port exposed, but if you need another format, let me know and I can probably help you set it up.
The next step for the generator (as I see it) is recursion: making it usable as an MCP tool itself. That way, when an LLM discovers a new endpoint, it can automatically search for the spec (GitHub/docs/user-provided, etc.) and start utilizing it.