feat: add spec-compliant streaming error event for Responses API

[gyliu513]

Summary

Replace the ad-hoc OpenAIErrorResponse dict-based error event in SSE streaming with a spec-compliant OpenAIResponseObjectStreamError Pydantic model.

Problem: When errors occur mid-stream (e.g., ValueError from a provider), the SSE generator previously emitted an OpenAIErrorResponse — which is the OpenAI HTTP error response format ({"error": {"message": "...", "code": "..."}}). This is the wrong format for streaming. OpenAI's streaming API uses a distinct ResponseErrorEvent with type: "error" as a top-level SSE event (openai-python SDK reference).

Fix: Introduce OpenAIResponseObjectStreamError matching OpenAI's ResponseErrorEvent schema:

{"type": "error", "code": "400", "message": "not found", "sequence_number": 5}

This allows OpenAI-compatible client SDKs to correctly deserialize streaming errors.

[github-actions]

✱ Stainless preview builds

This PR will update the llama-stack-client SDKs with the following commit message.

feat: add spec-compliant streaming error event for Responses API

Edit this comment to update it. It will appear in the SDK's changelogs.

✅ llama-stack-client-go studio · conflict

Your SDK build had at least one new note diagnostic, which is a regression from the base state.

New diagnostics (1 note)

💡 Schema/EnumHasOneMember: This enum schema has just one member, so it could be defined using [`const`](https://json-schema.org/understanding-json-schema/reference/const).

✅ llama-stack-client-openapi studio · code · diff

Your SDK build had at least one "warning" diagnostic, but this did not represent a regression.
generate ⚠️

✅ llama-stack-client-python studio · conflict

Your SDK build resulted in a merge conflict between your custom code and the newly generated changes, but this did not represent a regression.

✅ llama-stack-client-node studio · conflict

Your SDK build resulted in a merge conflict between your custom code and the newly generated changes, but this did not represent a regression.

---

This comment is auto-generated by GitHub Actions and is automatically kept up to date as you push.
If you push custom code to the preview branch, re-run this workflow to update the comment.
Last updated: 2026-03-30 14:47:28 UTC

[cdoern]

can you please add a description to this PR

[gyliu513]

@cdoern Done, this is a draft PR, so I did not add description when submit it, it is still draft, I may need more test and the description may be updated later.

[leseb]

@gyliu513 review from AI:

High: the PR adds a new streaming error model and even emits it at runtime, but the declared stream
union still does not include it. src/llama_stack_api/openai_responses.py adds
OpenAIResponseObjectStreamError, while src/llama_stack_api/responses/fastapi_routes.py now yields
that event from sse_generator(). But OpenAIResponseObjectStream is still the old discriminated
union, so the generated OpenAPI stream schema does not actually admit type: "error". In the
generated spec this shows up as an orphan OpenAIResponseObjectStreamError schema that is never
referenced from OpenAIResponseObjectStream. That undercuts the PR’s stated goal: SDKs generated
from the spec still won’t know that error is a valid stream event.

[gyliu513]

Done, thanks @leseb