chore(deps): weekly dependency update - #765
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 439d11156f
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "Codex (@codex) review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "Codex (@codex) address that feedback".
|
|
||
| [tool.braintrust.matrix.openai] | ||
| latest = "openai==3.8.0" | ||
| latest = "openai==3.13.0" |
There was a problem hiding this comment.
Before merging this pin, re-record and commit the affected cassettes/latest/ directories. This commit moves OpenAI from 3.8.0 to 3.13.0, along with minor upgrades for Anthropic, Pipecat, Pydantic AI, Google ADK, Mistral, and others, but changes no cassette files; consequently, the latest sessions replay interactions recorded with the previous SDKs rather than validating the newly pinned releases' actual request/response behavior, allowing integration regressions to ship despite passing playback tests.
AGENTS.md reference: AGENTS.md:L194-L199
Useful? React with 👍 / 👎.
….9 compat Re-record `latest` cassettes for the minor provider bumps in the weekly dependency update: openai 3.13.0 (incl. http2 + btx), anthropic 1.5.0, pydantic-ai 2.43.0, google-genai 2.23.0, google-adk 2.9.0, strands-agents 1.55.1, pipecat-ai 1.10.0, instructor 1.17.0, and a partial refresh of mistralai 2.10.0 (32/45; the remaining 13 kept from main because the local Mistral key was persistently rate-limited). Kept from main on purpose: the hand-authored instructor retry cassette, the harbor ATIF cassettes (they record Braintrust API traffic), the huggingface_hub cassettes (HF no longer routes Llama-3.1-8B to cerebras), and the anthropic managed-agents sessions cassette (live recording hangs). Two SDK changes were needed for the bumps: - anthropic 1.5.0 made `accumulate_event()` require a caller-owned `json_bufs` dict. TracedMessageStream now keeps a per-stream buffer and passes it only when the installed SDK accepts the keyword, so 0.48.0 and 1.4.x keep working. - google-adk 2.9.0 moved tool execution from `llm_flows.functions.__call_tool_async` to `llm_flows._tool_caller._call_tool_async`. ToolCallAsyncPatcher is now a composite over both targets, with the legacy target yielding to the new one so a single tool call never emits two tool spans. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 89e30ba311
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "Codex (@codex) review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".
|
|
||
| [tool.braintrust.matrix.openai-agents] | ||
| latest = "openai-agents==0.22.0" | ||
| latest = "openai-agents==0.22.2" |
There was a problem hiding this comment.
When CI runs the newly pinned openai-agents==0.22.2 session, it still replays openai_agents/cassettes/latest/test_auto_openai_agents.yaml, whose request identifies Agents/Python 0.22.0; the same mismatch remains for LiteLLM (1.100.0 versus 1.100.1) and unchanged Mistral chat cassettes (2.9.0 versus 2.10.0). Although the follow-up re-recorded several directories, these concrete stale headers show that the earlier issue remains for part of the matrix: because CI uses playback-only cassettes, those latest jobs do not validate responses recorded with the upgraded SDKs. Re-record every bumped provider's latest session, not only the directories currently changed.
AGENTS.md reference: AGENTS.md:L194-L205
Useful? React with 👍 / 👎.
1 participant
Automated weekly dependency update via
python scripts/update-matrix-latest.py && uv lock --upgrade.