feat(ai): add claude code agents sdk integration#5316
Draft
feat(ai): add claude code agents sdk integration#5316
Conversation
Contributor
Semver Impact of This PR🟡 Minor (new features) 📋 Changelog PreviewThis is how your changes will appear in the changelog. New Features ✨Span Streaming
Other
Bug Fixes 🐛Google Genai
Span Streaming
Other
Internal Changes 🔧Fastmcp
Mcp
Other
🤖 This preview updates automatically when you update the PR. |
Contributor
|
Excited about this!
|
Comment on lines
372
to
401
| chat_span = get_start_span_function()( | ||
| op=OP.GEN_AI_CHAT, | ||
| name=f"claude-agent-sdk query {model}".strip(), | ||
| origin=ClaudeAgentSDKIntegration.origin, | ||
| ) | ||
| chat_span.__enter__() | ||
|
|
||
| with capture_internal_exceptions(): | ||
| _set_span_input_data(chat_span, prompt, options, integration) | ||
|
|
||
| collected_messages = [] | ||
| try: | ||
| async for message in original_func( | ||
| prompt=prompt, options=options, **kwargs | ||
| ): | ||
| collected_messages.append(message) | ||
| yield message | ||
| except Exception as exc: | ||
| _capture_exception(exc) | ||
| raise | ||
| finally: | ||
| with capture_internal_exceptions(): | ||
| _set_span_output_data(chat_span, collected_messages, integration) | ||
| chat_span.__exit__(None, None, None) | ||
|
|
||
| with capture_internal_exceptions(): | ||
| _process_tool_executions(collected_messages, integration) | ||
|
|
||
| with capture_internal_exceptions(): | ||
| _end_invoke_agent_span(invoke_span, collected_messages, integration) |
Contributor
There was a problem hiding this comment.
does your agent know how to use a context manager 😄
Suggested change
| chat_span = get_start_span_function()( | |
| op=OP.GEN_AI_CHAT, | |
| name=f"claude-agent-sdk query {model}".strip(), | |
| origin=ClaudeAgentSDKIntegration.origin, | |
| ) | |
| chat_span.__enter__() | |
| with capture_internal_exceptions(): | |
| _set_span_input_data(chat_span, prompt, options, integration) | |
| collected_messages = [] | |
| try: | |
| async for message in original_func( | |
| prompt=prompt, options=options, **kwargs | |
| ): | |
| collected_messages.append(message) | |
| yield message | |
| except Exception as exc: | |
| _capture_exception(exc) | |
| raise | |
| finally: | |
| with capture_internal_exceptions(): | |
| _set_span_output_data(chat_span, collected_messages, integration) | |
| chat_span.__exit__(None, None, None) | |
| with capture_internal_exceptions(): | |
| _process_tool_executions(collected_messages, integration) | |
| with capture_internal_exceptions(): | |
| _end_invoke_agent_span(invoke_span, collected_messages, integration) | |
| chat_span = get_start_span_function()( | |
| op=OP.GEN_AI_CHAT, | |
| name=f"claude-agent-sdk query {model}".strip(), | |
| origin=ClaudeAgentSDKIntegration.origin, | |
| ) | |
| with chat_span() as span: | |
| with capture_internal_exceptions(): | |
| _set_span_input_data(span, prompt, options, integration) | |
| collected_messages = [] | |
| try: | |
| async for message in original_func( | |
| prompt=prompt, options=options, **kwargs | |
| ): | |
| collected_messages.append(message) | |
| yield message | |
| except Exception as exc: | |
| _capture_exception(exc) | |
| raise | |
| finally: | |
| with capture_internal_exceptions(): | |
| _set_span_output_data(span, collected_messages, integration) | |
| with capture_internal_exceptions(): | |
| _process_tool_executions(collected_messages, integration) | |
| with capture_internal_exceptions(): | |
| _end_invoke_agent_span(invoke_span, collected_messages, integration) |
| model = getattr(options, "model", "") if options else "" | ||
| invoke_span = _start_invoke_agent_span(prompt, options, integration) | ||
|
|
||
| chat_span = get_start_span_function()( |
Contributor
There was a problem hiding this comment.
There's an invoke_agent span active here, why not just always start a span?
Comment on lines
439
to
445
| messages = self._sentry_query_context.get("messages", []) | ||
| with capture_internal_exceptions(): | ||
| _set_span_output_data(chat_span, messages, integration) | ||
| chat_span.__exit__(None, None, None) | ||
| with capture_internal_exceptions(): | ||
| _end_invoke_agent_span(invoke_span, messages, integration) | ||
| self._sentry_query_context = {} |
Contributor
There was a problem hiding this comment.
Exiting spans has caused many uncaught exceptions already 😄
Suggested change
| messages = self._sentry_query_context.get("messages", []) | |
| with capture_internal_exceptions(): | |
| _set_span_output_data(chat_span, messages, integration) | |
| chat_span.__exit__(None, None, None) | |
| with capture_internal_exceptions(): | |
| _end_invoke_agent_span(invoke_span, messages, integration) | |
| self._sentry_query_context = {} | |
| messages = self._sentry_query_context.get("messages", []) | |
| with capture_internal_exceptions(): | |
| _set_span_output_data(chat_span, messages, integration) | |
| chat_span.__exit__(None, None, None) | |
| _end_invoke_agent_span(invoke_span, messages, integration) | |
| self._sentry_query_context = {} |
| tool_uses = {} | ||
| tool_results = {} | ||
|
|
||
| for message in messages: |
Contributor
There was a problem hiding this comment.
narrower than Anthropic message types, so this looks suspicious.
|
|
||
| def _extract_message_data(messages: list) -> dict: | ||
| """Extract relevant data from a list of messages.""" | ||
| data = { |
Contributor
There was a problem hiding this comment.
use a class type instead of a dictionary for passing info between functions, stronger typing.
| ) | ||
|
|
||
| if _should_include_prompts(integration): | ||
| messages = [] |
Contributor
There was a problem hiding this comment.
system prompts go in their own attribute.
| async for message in original_func( | ||
| prompt=prompt, options=options, **kwargs | ||
| ): | ||
| collected_messages.append(message) |
Contributor
There was a problem hiding this comment.
your collecting user-held references here, leading to possible race conditions.
Comment on lines
+99
to
+102
| def test_extract_text_returns_none_for_non_assistant(): | ||
| result = make_result_message(usage=None) | ||
| assert _extract_text_from_message(result) is None | ||
|
|
Contributor
There was a problem hiding this comment.
We don't need all these unit tests, more work in the long run.
Instead, call the userspace API, and parameterize on as many input schemas as possible to achieve good test coverage.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add instrumentation for the claude-agent-sdk package, which provides
a Python interface to interact with Claude Code CLI.
The integration captures:
Instrumented methods:
Closes TET-1743
Co-Authored-By: Claude Sonnet 4.5 noreply@anthropic.com