Repository navigation
Reorganize AI integration-related test fixtures #5993
Description
Activity
Hi! I'd love to contribute. Could you assign this issue to me?
- moved this from Waiting for: Product Owner to No status in GitHub Issues with 👀 3
on May 12, 2026 I traced the fixture cluster and this looks like a good small first split.
The cleanest scoped pass I see is to move the MCP transport helpers and AI model response fixtures out of the global
tests/conftest.pyinto the integrations layer, where the current users already live:- MCP/FastMCP helpers:
stdio,json_rpc,json_rpc_sse, and the selector/payload helpers - AI response fixtures used by OpenAI, OpenAI Agents, LiteLLM, and LangChain tests
That would leave the default behavior unchanged for integration tests while cutting down the global conftest surface. I also noticed the contribution guide asks for prior maintainer discussion before PR work, so I am checking here before opening anything.
This would be Codex-assisted, with the final patch reviewed locally and kept to that fixture move only.
- MCP/FastMCP helpers:
Hi @omribz156 ,
Thanks for offering to take a look at this, the split that you've proposed looks good. I'll assign the issue to you, and don't hesitate to reach out if you have any questions as you dig into this.
- moved this from Waiting for: Product Owner to No status in GitHub Issues with 👀 3
on May 25, 2026 Hi @ericapisani, since there hasn't been any activity on this for a couple of months, I'd like to take over if that's okay.
My plan is to move only the AI provider response fixtures to a new tests/integrations/conftest.py and load it via pytest_plugins in the root conftest.py. Specifically, I plan to move these 9 fixtures:
get_model_response
get_rate_limit_model_response
streaming_chat_completions_model_response
nonstreaming_chat_completions_model_response
openai_embedding_model_response
nonstreaming_responses_model_response
nonstreaming_anthropic_model_response
nonstreaming_google_genai_model_response
responses_tool_call_model_responses
- moved this from Waiting for: Product Owner to No status in GitHub Issues with 👀 3
on Aug 5, 2026 - moved this from Waiting for: Product Owner to No status in GitHub Issues with 👀 3
on Sep 3, 2026
Metadata
Metadata
Assignees
Labels
Projects
- Status
Context: #5937
Right now, a large number of our pytest fixtures live in the global
conftest.pyfile. For maintainability reasons, it'd be better to move provider-specific (or just AI integration-specific) fixtures into separateconftest.pyfiles and load them into the globalconftest.pyusingpytest_plugins.