Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
The table of contents is too big for display.
Diff view
Diff view
  •  
  •  
  •  
55 changes: 28 additions & 27 deletions py/pyproject.toml
Original file line number Diff line number Diff line change
Expand Up @@ -381,54 +381,54 @@ conflicts = [
[tool.braintrust.matrix]

[tool.braintrust.matrix.ai-sdk]
latest = "ai[openai]==0.7.0"
latest = "ai[openai]==0.8.0"
"0.4.0" = "ai[openai]==0.4.0"

[tool.braintrust.matrix.openai]
latest = "openai==3.19.2"
latest = "openai==3.24.0"
"1.92.0" = "openai==1.92.0"
"1.77.0" = "openai==1.77.0"
"1.71.0" = "openai==1.71.0"

[tool.braintrust.matrix.anthropic]
latest = "anthropic==1.8.0"
latest = "anthropic==1.11.0"

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Refresh the remaining Anthropic latest cassettes

The 1.11.0 bump is only partially re-recorded: for example, test_anthropic_beta_messages_create_captures_compaction_metadata.yaml still identifies Anthropic/Python 1.8.0, even though that test explicitly runs only for the latest matrix entry; the sessions-events cassette is likewise unchanged. Because test_anthropic(latest) selects this directory and CI is replay-only, these paths continue replaying 1.8-era traffic rather than validating the newly pinned client. Re-record the remaining latest-only tests with 1.11.0.

AGENTS.md reference: AGENTS.md:L194-L204

Useful? React with 👍 / 👎.

"0.48.0" = "anthropic==0.48.0"

[tool.braintrust.matrix.cohere]
latest = "cohere==7.1.1"
latest = "cohere==7.2.0"
"5.0.0" = "cohere==5.0.0"

[tool.braintrust.matrix.openai-agents]
latest = "openai-agents==0.22.3"
latest = "openai-agents==0.23.1"
"0.13.6" = "openai-agents==0.13.6"
"0.0.19" = "openai-agents==0.0.19"

[tool.braintrust.matrix.litellm]
# 1.92.0 only publishes Linux wheels through CPython 3.13, but the test
# matrix also covers Windows and CPython 3.14.
latest = "litellm==1.103.0"
latest = "litellm==1.104.0"
"1.74.0" = "litellm==1.74.0"

[tool.braintrust.matrix.livekit-agents]
latest = "livekit-agents==1.8.3"
latest = "livekit-agents==1.8.4"
"1.3.1" = "livekit-agents==1.3.1"

[tool.braintrust.matrix.pipecat-ai]
latest = "pipecat-ai==1.12.0"
"1.3.0" = "pipecat-ai==1.3.0"

[tool.braintrust.matrix.claude-agent-sdk]
latest = "claude-agent-sdk==0.2.160"
latest = "claude-agent-sdk==0.2.163"
"0.1.10" = "claude-agent-sdk==0.1.10"

[tool.braintrust.matrix.cursor-sdk]
# 1.0.25 fixes bridge callback tokens that could be parsed as CLI flags while
# retaining per-turn usage, sync/async consumption, multimodal, and custom tools.
latest = "cursor-sdk==1.0.32"
latest = "cursor-sdk==1.0.35"
"1.0.25" = "cursor-sdk==1.0.25"

[tool.braintrust.matrix.agno]
latest = "agno==3.0.11"
latest = "agno==3.1.1"
"2.4.0" = "agno==2.4.0"
"2.1.0" = "agno==2.1.0"

Expand All @@ -445,15 +445,15 @@ latest = "autogen-ext[openai]==0.7.5"
"0.7.0" = "autogen-ext[openai]==0.7.0"

[tool.braintrust.matrix.strands-agents]
latest = "strands-agents==1.57.1"
latest = "strands-agents==1.57.2"
"1.20.0" = "strands-agents==1.20.0"

[tool.braintrust.matrix.pydantic-ai-integration]
latest = "pydantic-ai==2.51.0"
latest = "pydantic-ai==2.54.0"
"1.10.0" = "pydantic-ai==1.10.0"

[tool.braintrust.matrix.pydantic-ai-wrap-openai]
latest = "pydantic-ai==2.51.0"
latest = "pydantic-ai==2.54.0"
"1.0.1" = "pydantic-ai==1.0.1"
"0.1.9" = "pydantic-ai==0.1.9"

Expand All @@ -466,47 +466,48 @@ latest = "instructor==1.17.0"
"1.11.0" = "instructor==1.11.0"

[tool.braintrust.matrix.google-genai]
latest = "google-genai==2.25.0"
latest = "google-genai==2.28.0"
"1.75.0" = "google-genai==1.75.0"
"1.30.0" = "google-genai==1.30.0"

[tool.braintrust.matrix.google-cloud-discoveryengine]
latest = "google-cloud-discoveryengine==0.20.4"
latest = "google-cloud-discoveryengine==0.20.5"

[tool.braintrust.matrix.dspy]
latest = "dspy==3.4.0"
"2.6.0" = "dspy==2.6.0"

[tool.braintrust.matrix.deepagents]
latest = "deepagents==0.7.19"
latest = "deepagents==0.7.21"
"0.6.0" = "deepagents==0.6.0"

[tool.braintrust.matrix.crewai]
# 1.13.0 is the first release with the full causal-id surface (event_id /
# parent_event_id / started_event_id on BaseEvent) plus the ``usage`` field on
# LLMCallCompletedEvent that the Braintrust CrewAI integration depends on.
latest = "crewai==1.15.22"
latest = "crewai==1.15.23"
"1.13.0" = "crewai==1.13.0"

[tool.braintrust.matrix.google-adk]
latest = "google-adk[mcp]==2.10.0"
latest = "google-adk[mcp]==2.11.0"
"2.6.3" = "google-adk[mcp]==2.6.3"
"1.14.1" = "google-adk[mcp]==1.14.1"

[tool.braintrust.matrix.langchain-core]
latest = "langchain-core==1.6.5"
latest = "langchain-core==1.6.6"
"0.3.28" = "langchain-core==0.3.28"

[tool.braintrust.matrix.llama-index-core]
latest = "llama-index-core==0.14.25"
"0.13.0" = "llama-index-core==0.13.0"

[tool.braintrust.matrix.openrouter]
latest = "openrouter==1.2.32"
latest = "openrouter==1.3.22"
"0.6.0" = "openrouter==0.6.0"

[tool.braintrust.matrix.mistralai]
latest = "mistralai==2.10.1"
latest = "mistralai==3.0.0"

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Re-record the Mistral latest cassettes

Updating this integration across a major SDK release without refreshing its latest cassettes leaves test_mistral(latest) replaying Mistral 2.9/2.10 exchanges (for example, the request fixtures under integrations/mistral/cassettes/latest/ still identify those clients). That session selects the latest cassette directory while CI runs VCR in replay-only mode, so this change cannot validate the request/response behavior of the newly shipped v3 client. Re-record the Mistral latest fixtures and make any compatibility adjustments the v3 playback exposes.

AGENTS.md reference: AGENTS.md:L194-L205

Useful? React with 👍 / 👎.

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Finish refreshing the Mistral v3 latest cassettes

Fresh evidence after the earlier review comment is that the refresh remains incomplete: 14 files in mistral/cassettes/latest, covering chat, agents, OCR, and auto-instrumentation, are unchanged and still record mistral-client-python/2.9.0 or 2.10.0. Thus test_mistral(latest) installs v3 but replays pre-v3 exchanges for those endpoint families in replay-only CI, leaving much of the major-version upgrade unvalidated. Re-record the complete latest session rather than only the conversations/audio/FIM subset.

AGENTS.md reference: AGENTS.md:L194-L204

Useful? React with 👍 / 👎.

"2.10.1" = "mistralai==2.10.1"
"1.12.4" = "mistralai==1.12.4"

[tool.braintrust.matrix.typesafe-sdk]
Expand All @@ -517,20 +518,20 @@ latest = "typesafe-sdk==0.7.2"
# ``provider="auto"`` routing mode the integration relies on for multi-
# provider use, while still keeping the same chat_completion /
# text_generation / feature_extraction / sentence_similarity method names.
latest = "huggingface-hub==2.0.0"
latest = "huggingface-hub==2.1.1"
"0.32.0" = "huggingface-hub==0.32.0"

[tool.braintrust.matrix.transformers]
latest = "transformers==5.17.0"
latest = "transformers==5.18.0"
"4.42.0" = "transformers==4.42.0"

[tool.braintrust.matrix.temporalio]
latest = "temporalio==1.33.0"
latest = "temporalio==1.34.0"
"1.20.0" = "temporalio==1.20.0"
"1.19.0" = "temporalio==1.19.0"

[tool.braintrust.matrix.harbor]
latest = "harbor==0.23.0"
latest = "harbor==0.24.0"
# 0.16.0 is the oldest supported release: it is the first with
# TrialEvent.AGENT_END, which the lifecycle state machine subscribes to. Harbor
# gained its plugin system in 0.13.2, so 0.13.2-0.15.0 can load the plugin but
Expand All @@ -547,11 +548,11 @@ latest = "pytest==9.1.1"
latest = "braintrust-core==0.0.59"

[tool.braintrust.matrix.boto3]
latest = "boto3==1.43.103"
latest = "boto3==1.43.108"
"1.34.116" = "boto3==1.34.116"

[tool.braintrust.matrix.botocore]
latest = "botocore==1.43.103"
latest = "botocore==1.43.108"
"1.34.116" = "botocore==1.34.116"

# OpenTelemetry api/sdk/exporter release in lockstep; keep their keys in sync.
Expand Down

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

Loading
Loading