fix(anthropic): preserve caller errors in streaming responses

[alexliluz]

Fixes #389

Anthropic's with_streaming_response context manager closes the raw response without forwarding caller exceptions, so an error raised while consuming a stream is recorded as a successful span. Wrap the dynamically-created sync and async streaming methods and pass the exception to the active raw-response stream wrapper before SDK cleanup.

Tests: targeted sync/async regression tests pass on the latest and oldest supported environments; pyright and ruff pass. The full suite still has five pre-existing timeout-classification failures in this environment.

AI disclosure: OpenAI Codex (GPT-5) assisted with repository inspection, implementation, and test execution; the changes and results were manually reviewed.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: alexliluz / name: alexliluz (59fd839, 5dacce8, 6361e42, c229f4e)

[opentelemetry-pr-dashboard]

Pull request dashboard status

Waiting on the author · refreshed 2026-08-22 16:12 UTC

Resolve merge conflicts.

Investigate required status check failures.

Status above doesn't look right?

• Just replied or pushed? Anything around or after the refresh time above may not be picked up yet — give it a few minutes.
• Should this be with reviewers? Comment /dashboard route:reviewers to route it to them.
• Anything wrong — including the routing? Report it with what you expected; it helps us improve the dashboard.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

This PR addresses a previously known instrumentation gap where caller-side exceptions raised inside Anthropic streaming raw-response context managers were not reaching the raw-response proxy, causing spans to be incorrectly finalized as successful.

Changes:

• Adds sync/async context-manager wrappers that forward caller exceptions to the raw-response proxy before delegating to the Anthropic SDK context manager.
• Instruments MessagesWithStreamingResponse / AsyncMessagesWithStreamingResponse to dynamically wrap their create methods at construction time.
• Unskips existing sync/async tests covering user exceptions in streaming raw-response contexts.

Reviewed changes

Copilot reviewed 7 out of 7 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
instrumentation/opentelemetry-instrumentation-genai-anthropic/tests/test_sync_messages.py	Unskips the sync test that asserts spans fail when users raise during streaming raw-response usage.
instrumentation/opentelemetry-instrumentation-genai-anthropic/tests/test_async_messages.py	Unskips the async test that asserts spans fail when users raise during streaming raw-response usage.
instrumentation/opentelemetry-instrumentation-genai-anthropic/src/opentelemetry/instrumentation/genai/anthropic/wrappers.py	Introduces sync/async response context-manager wrappers to preserve caller exceptions.
instrumentation/opentelemetry-instrumentation-genai-anthropic/src/opentelemetry/instrumentation/genai/anthropic/patch.py	Adds wrapping logic to patch dynamically-created with_streaming_response.create methods.
instrumentation/opentelemetry-instrumentation-genai-anthropic/src/opentelemetry/instrumentation/genai/anthropic/_raw_response.py	Adds a _fail() hook to finalize spans on caller-side failures.
instrumentation/opentelemetry-instrumentation-genai-anthropic/src/opentelemetry/instrumentation/genai/anthropic/init.py	Wires the new patch points into instrument/uninstrument flows.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.