Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Custom MCP Server & many servers #9

Open
lnvestor opened this issue Jan 9, 2025 · 3 comments
Open

Custom MCP Server & many servers #9

lnvestor opened this issue Jan 9, 2025 · 3 comments
Assignees

Comments

@lnvestor
Copy link

lnvestor commented Jan 9, 2025

Hi

is there a way anything on the config file all MCP tools should be bridged in one sse, also when i try custom MCP i get this error

Thread Run failed: 400 {"type":"error","error":{"type":"invalid_request_error","message":"tools.0.input_schema: input_schema does not support oneOf, allOf, or anyOf at the top level"}}

Thanks

@Nedomas
Copy link
Member

Nedomas commented Jan 9, 2025

Hey @lnvestor , oops - yeah, you shouldn’t be getting this error.

  1. Just to try to recreate - I understand you are running a local SSE server over something like ngrok and added it to the assistant MCPs list?
  2. Are you running a native SSE server or is it STDIO server wrapper with Supergateway?
  3. Are you running some community MCP? Which then - maybe I could run it and see if it responds with the right tools input schema.
  4. Regarding all MCP tools bridged in one SSE - maybe you can just add all the MCPs as separate servers? Superinterface supports adding multiple MCPs. I’m not sure if you can run multiple MCPs over a single SSE transport connection.

Thanks!

@Nedomas Nedomas self-assigned this Jan 9, 2025
@Nedomas
Copy link
Member

Nedomas commented Jan 9, 2025

by the way, maybe you could try using another model from another provider? I think the error ("input_schema does not support oneOf, allOf, or anyOf at the top level") might be actually valid and OpenAI or Anthropic might not necessary support this feature of OpenAPI spec. Here’s some context: https://community.openai.com/t/oneof-allof-usage-has-problems-with-strict-mode/966047/5 https://community.openai.com/t/custom-gpts-action-openapi-3-0-specification-limited-support-oneof-anyof-allof/581015/8

If your server is custom MCP, you might need to transform your OpenAPI schema response into something simpler - I know oneOf can be simplified with some parsers/transformers as we do something like that when importing custom OpenAPI into Superinteface as function import.

@lnvestor
Copy link
Author

lnvestor commented Jan 9, 2025

@Nedomas Thanks for your answer, yeah i have pre difinded servers and i build custom ones as well,
i think that can be achieved over SSE i found this method worked on this repo by ConnectX.io
https://github.com/EvalsOne/MCP-connect

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants