fix(crewai): stop the __dict__ walk leaking credentials onto spans

[IdoGol24]

CrewAISpanAttributes stringified every attribute of the Crew, Agent and Task objects it was handed, so anything those objects held -- an LLM client, an embedder config -- reached the span through its repr, api keys included.

Read a fixed set of fields instead. Agents already had a hand-written allowlist in _extract_agent_data; the standalone-agent path just wasn't using it. Tasks reuse the dict _parse_tasks already built. Only Crew needed a new list. Tools go through the existing _serialize_tools on every path rather than being repr'd, and the dead _populate_llm_attributes is gone.

• I have added tests that cover my changes.

• If adding a new instrumentation or changing an existing one, I've added screenshots from some observability platform showing the change.

• PR name follows conventional commits format: feat(instrumentation): ... or fix(instrumentation): ....

• (If applicable) I have updated the documentation accordingly.

Summary by CodeRabbit

• Bug Fixes
  • Improved telemetry safety by preventing credentials and other sensitive configuration values from appearing in span attributes.
  • Standardized captured task and agent details, including task identifiers and serialized tool information.
  • Continued emitting supported Crew, task, and agent identifiers while excluding unsupported configuration data.
  • Improved consistency of telemetry attributes across CrewAI components.

[CLAassistant]

All committers have signed the CLA.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 4c04cd01-fc6b-41ca-bfd1-12caa740d75b

📥 Commits

Reviewing files that changed from the base of the PR and between 9ce6b03 and 75d80e3.

📒 Files selected for processing (2)

• packages/opentelemetry-instrumentation-crewai/opentelemetry/instrumentation/crewai/crewai_span_attributes.py
• packages/opentelemetry-instrumentation-crewai/tests/test_span_attribute_allowlist.py

🚧 Files skipped from review as they are similar to previous changes (2)

• packages/opentelemetry-instrumentation-crewai/opentelemetry/instrumentation/crewai/crewai_span_attributes.py
• packages/opentelemetry-instrumentation-crewai/tests/test_span_attribute_allowlist.py

Included review availability: Your plan provides up to 8 included reviews per hour; 6 remain after this review.

---

📝 Walkthrough

Walkthrough

CrewAI span extraction now uses an explicit safe-field allowlist and shared serialization for agents, tasks, and tools. New tests verify that sentinel credentials do not enter span attributes while identifiers remain available.

Changes

CrewAI attribute safety

Layer / File(s)	Summary
Allowlisted CrewAI extraction packages/opentelemetry-instrumentation-crewai/opentelemetry/instrumentation/crewai/crewai_span_attributes.py	Crew extraction uses CREW_FIELDS. Agent and task extraction reuse stringification helpers. Tool data is serialized, and agent configuration is excluded.
Credential leakage validation packages/opentelemetry-instrumentation-crewai/tests/test_span_attribute_allowlist.py	Tests construct agents, tasks, and crews with sentinel credentials. They verify that span attributes exclude the sentinel while retaining allowlisted identifiers.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Merge Risk: ⚪ Minimal · up to 75d80

CrewAI span extraction now limits serialized fields to safe allowlists to prevent credential leakage while retaining supported identifiers. No concrete merge-blocking risk remains.

Sequence Diagram(s)

sequenceDiagram
  participant CrewInstance
  participant CrewAISpanAttributes
  participant InMemorySpanExporter
  CrewInstance->>CrewAISpanAttributes: provide crew, agent, and task data
  CrewAISpanAttributes->>CrewAISpanAttributes: select allowlisted fields and serialize tools
  CrewAISpanAttributes->>InMemorySpanExporter: emit filtered span attributes

Loading

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 76.47% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 17 functions across 2 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: preventing credential leakage caused by walking CrewAI object dict values into spans.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

• Fix all pre-merge checks with AI

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}