Activity

1. vladman-25

   added

   questionFurther information is requestedFurther information is requested

   on Mar 12, 2026
2. maxisbey commented on Mar 12, 2026

   

   maxisbey

   on Mar 12, 2026

   Contributor

   More actions

   Thanks for the report! A couple of things to untangle here:

   On the model_validate_json → validate_json change — that's from #1908, which dropped RootModel from JSONRPCMessage as a refactoring cleanup. Both paths use the same underlying pydantic-core JSON parser, so this change doesn't affect Unicode handling and won't change the behavior you're seeing.

   On the error itself — ó in properly UTF-8-encoded JSON parses fine (I tested this directly). The most likely explanation is that your server is sending responses encoded as latin-1/windows-1252 rather than UTF-8. In those encodings, ó is the single byte 0xF3, which isn't a valid UTF-8 lead byte — so pydantic's parser rejects it. Responses containing only ASCII would pass through fine since ASCII bytes are identical across these encodings, which matches the "only fails with Unicode" pattern you're describing.

   Note that both RFC 8259 §8.1 and the MCP spec require UTF-8 for JSON-RPC messages, so if this is the cause, the fix belongs server-side.

   To confirm: could you share the full ValidationError output? Your traceback is truncated right after 1 validation error for JSONRPCMessage — the next line should say either invalid unicode code point at line 1 column N (which would confirm the encoding mismatch) or something else entirely (which would mean we're looking at a different issue). Also helpful: which server you're connecting to, and whether you're able to capture the raw response bytes.

   _{AI Disclaimer}
3. maxisbey

   added

   needs confirmationNeeds confirmation that the PR is actually required or needed.Needs confirmation that the PR is actually required or needed.

   on Mar 12, 2026
4. vladman-25 commented on Mar 12, 2026

   

   vladman-25

   on Mar 12, 2026

   Last edited by vladman-25

   Author

   More actions

   Thanks for the clarification!

   Here is the full error message:

   pydantic_core._pydantic_core.ValidationError: 1 validation error for JSONRPCMessage
     Invalid JSON: invalid unicode code point at line 1 column 12945
   

   So it does appear to be the exact error you mentioned.

   From your explanation, my understanding is the correct fix would be to update the server to emit properly UTF-8-encoded JSON.

   However, in the hypothetical case that changing the server encoding is not possible, would there be any recommended workaround on the client side? For example, would it be possible to introduce a preprocessing step that attempts to re-decode the raw response bytes (e.g., latin-1 -> UTF-8) before passing it to the JSON parser?

   LE: This should not be possible, our server responds with � characters by default.
5. Kludex

   closed this as completedon Mar 16, 2026