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
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 12 additions & 5 deletions a2a/weather_service/pyproject.toml
Original file line number Diff line number Diff line change
Expand Up @@ -24,19 +24,26 @@ dependencies = [
"opentelemetry-instrumentation-httpx>=0.49b0",
# OpenTelemetry GenAI semantic convention instrumentation
# Emits spans with gen_ai.* attributes for MLflow compatibility
"opentelemetry-instrumentation-openai>=0.34b0",
"opentelemetry-instrumentation-openai>=0.60.0",

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

suggestion: this is the biggest version jump in the PR — 0.34b00.60.0, crossing the 0.34→0.60 gap. The test-startup check is a smoke test (import + boot) and won't catch subtle API/attribute-name changes in the instrumentation layer. Since this package emits gen_ai.* attributes that MLflow downstream consumes, a breaking change in attribute names or span structure would show up later as missing traces/telemetry rather than a crash. Worth a quick manual run of the weather-service against MLflow (or a spot-check of the instrumentation changelog for attribute-name changes) before merge. Mamma mia, not blocking — just the kind of thing that burns you three weeks after a clean merge.

# OpenInference for LangChain instrumentation and AGENT span semantics
"openinference-semantic-conventions>=0.1.12",
"openinference-instrumentation-langchain>=0.1.27",
"openinference-instrumentation-langchain>=0.1.63",

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

nit: 0.1.270.1.63 — large patch span, but same 0.1.x series so semver says this should be compatible. Same observability concern as the opentelemetry-instrumentation-openai comment above but less severe (0.1.x implies no intentional breaking changes). Just flagging that these two instrumentation libraries together are the biggest behavioral delta in the PR; the direct deps (aiohttp, cryptography, python-multipart) are patch-bumps and safe.

"urllib3>=2.6.3", # Indirect; prevents CVE-2025-66418
"python-multipart>=0.0.22", # Indirect; prevents CVE-2026-24486
"cryptography>=46.0.5", # Indirect; prevents CVE-2026-26007
"python-multipart>=0.0.26", # Indirect; prevents CVE-2026-24486
"cryptography>=46.0.7", # Indirect; prevents CVE-2026-26007
"langgraph-checkpoint>=3.0.0", # Indirect; prevents CVE-2025-64439
"mcp>=1.23.0", # Indirect; prevents CVE-2025-66416
"protobuf>=6.33.5", # Indirect; prevents CVE-2026-0994
"starlette>=0.49.1", # Indirect; prevents CVE-2025-62727
"pyasn1>=0.6.3", # Indirect; prevents CVE-2026-30922
"aiohttp>=3.13.3", # Indirect; prevents CVE-2025-69223
"aiohttp>=3.13.5", # Indirect; prevents CVE-2025-69223
"gitpython>=3.1.47", # Indirect; prevents CVE-2026-42215, CVE-2026-42284

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

nit: the 7 new CVE pins all follow the repo's existing convention (# Indirect; prevents CVE-*), perfetto. Only cosmetic nit: consider interleaving them with the pre-existing indirect-CVE block higher up in the file instead of appending at the very end — future diff readers will see indirect pins split across two disjoint blocks. Purely aesthetic.

"langchain-text-splitters>=1.1.2", # Indirect; prevents CVE-2026-41481
"langsmith>=0.7.31", # Indirect; prevents CVE-2026-41182
"mako>=1.3.11", # Indirect; prevents CVE-2026-41205
"marshmallow>=3.26.2", # Indirect; prevents CVE-2025-68480
"python-dotenv>=1.2.2", # Indirect; prevents CVE-2026-28684
"requests>=2.33.0", # Indirect; prevents CVE-2026-25645
]

[project.scripts]
Expand Down
Loading
Loading