feat[crewa]: Init crewai instrumentation

instrumentation-genai/opentelemetry-instrumentation-crewai/CHANGELOG.md

@@ -0,0 +1,35 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+## [0.1.0] - 2025-11-25
+
+### Added
+- Initial release of CrewAI instrumentation
+- Wrapper-based instrumentation for CrewAI workflows, agents, tasks, and tools
+- Support for `Crew.kickoff()` → `Workflow` spans
+- Support for `Task.execute_sync()` → `Step` spans
+- Support for `Agent.execute_task()` → `AgentInvocation` spans
+- Support for `BaseTool.run()` and `CrewStructuredTool.invoke()` → `ToolCall` spans
+- Integration with `splunk-otel-util-genai` for standardized GenAI telemetry
+- Proper trace context propagation using `contextvars`
+- Rich span attributes for all CrewAI components
+- Defensive instrumentation that doesn't break applications on errors
+
+### Documentation
+- Comprehensive README with usage examples
+- Compositional instrumentation patterns (CrewAI + OpenAI + Vector Stores)
+- Configuration and environment variable documentation
+
+### Limitations
+- Synchronous workflows only (async support planned for future release)
+- LLM calls not instrumented (use provider-specific instrumentation)
+
+[Unreleased]: https://github.com/signalfx/splunk-otel-python-contrib/compare/v0.1.0...HEAD
+[0.1.0]: https://github.com/signalfx/splunk-otel-python-contrib/releases/tag/v0.1.0
+

instrumentation-genai/opentelemetry-instrumentation-crewai/README.rst

@@ -0,0 +1,198 @@
+OpenTelemetry CrewAI Instrumentation
+=====================================
+
+|pypi|
+
+.. |pypi| image:: https://badge.fury.io/py/splunk-otel-instrumentation-crewai.svg
+   :target: https://pypi.org/project/splunk-otel-instrumentation-crewai/
+
+This library provides OpenTelemetry instrumentation for `CrewAI <https://github.com/joaomdmoura/crewAI>`_,
+a framework for orchestrating autonomous AI agents.
+
+Installation
+------------
+
+.. code-block:: bash
+
+    pip install splunk-otel-instrumentation-crewai
+
+
+Usage
+-----
+
+.. code-block:: python
+
+    from opentelemetry.instrumentation.crewai import CrewAIInstrumentor
+    from crewai import Agent, Task, Crew
+
+    # Instrument CrewAI
+    CrewAIInstrumentor().instrument()
+
+    # Create your crew
+    agent = Agent(
+        role="Research Analyst",
+        goal="Provide accurate research",
+        backstory="Expert researcher with attention to detail",
+    )
+
+    task = Task(
+        description="Research the latest AI trends",
+        expected_output="A comprehensive report on AI trends",
+        agent=agent,
+    )
+
+    crew = Crew(agents=[agent], tasks=[task])
+
+    # Run your crew - telemetry is automatically captured
+    result = crew.kickoff()
+
+
+What Gets Instrumented
+-----------------------
+
+This instrumentation captures:
+
+- **Crews** → Mapped to ``Workflow`` spans
+- **Tasks** → Mapped to ``Step`` spans
+- **Agents** → Mapped to ``AgentInvocation`` spans
+- **Tool Usage** → Mapped to ``ToolCall`` spans
+
+All spans are properly nested with correct parent-child relationships and include
+rich attributes about the operation.
+
+
+Compositional Instrumentation
+------------------------------
+
+This instrumentation focuses on CrewAI's workflow orchestration. For complete observability:
+
+**CrewAI Only**
+
+.. code-block:: python
+
+    from opentelemetry.instrumentation.crewai import CrewAIInstrumentor
+
+    CrewAIInstrumentor().instrument()
+
+Provides workflow structure but no LLM call details.
+
+**CrewAI + OpenAI**
+
+.. code-block:: python
+
+    from opentelemetry.instrumentation.crewai import CrewAIInstrumentor
+    from opentelemetry.instrumentation.openai import OpenAIInstrumentor
+
+    CrewAIInstrumentor().instrument()
+    OpenAIInstrumentor().instrument()
+
+Adds LLM call spans with token usage, model names, and latency metrics.
+
+**Full Stack (CrewAI + OpenAI + Vector Store)**
+
+.. code-block:: python
+
+    from opentelemetry.instrumentation.crewai import CrewAIInstrumentor
+    from opentelemetry.instrumentation.openai import OpenAIInstrumentor
+    from opentelemetry.instrumentation.chromadb import ChromaDBInstrumentor
+
+    CrewAIInstrumentor().instrument()
+    OpenAIInstrumentor().instrument()
+    ChromaDBInstrumentor().instrument()
+
+Complete RAG workflow visibility with vector store operations.
+
+
+Configuration
+-------------
+
+Environment Variables
+~~~~~~~~~~~~~~~~~~~~~
+
+.. code-block:: bash
+
+    # Disable CrewAI's built-in telemetry (recommended)
+    export CREWAI_DISABLE_TELEMETRY=true
+
+
+Instrumentation Options
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. code-block:: python
+
+    from opentelemetry.instrumentation.crewai import CrewAIInstrumentor
+
+    # Basic instrumentation
+    CrewAIInstrumentor().instrument()
+
+    # With custom tracer provider
+    CrewAIInstrumentor().instrument(tracer_provider=my_tracer_provider)
+
+    # Uninstrumentation
+    CrewAIInstrumentor().uninstrument()
+
+
+Requirements
+------------
+
+- Python >= 3.9
+- CrewAI >= 0.70.0
+- OpenTelemetry API >= 1.38
+- ``splunk-otel-util-genai`` >= 0.1.4
+
+
+Trace Hierarchy Example
+------------------------
+
+.. code-block::
+
+    Crew: Customer Support (Workflow)
+    ├── Task: inquiry_resolution (Step)
+    │   └── Agent: Senior Support Representative
+    │       ├── LLM: gpt-4o-mini (via openai-instrumentation)
+    │       └── Tool: docs_scrape
+    └── Task: quality_assurance (Step)
+        └── Agent: QA Specialist
+            └── LLM: gpt-4o-mini (via openai-instrumentation)
+
+
+Each span includes rich attributes:
+
+- ``gen_ai.system`` = "crewai"
+- ``gen_ai.operation.name`` = "invoke_workflow" | "invoke_agent" | "execute_tool"
+- Framework-specific attributes (agent role, task description, tool names, etc.)
+
+
+Limitations
+-----------
+
+- **Async Support**: Currently supports synchronous workflows only. Async support (``kickoff_async()``)
+  is planned for a future release.
+- **LLM Calls**: Not instrumented here. Use provider-specific instrumentation
+  (e.g., ``opentelemetry-instrumentation-openai``).
+
+
+Contributing
+------------
+
+Contributions are welcome! Please ensure:
+
+- All tests pass
+- Code follows project style guidelines
+- Instrumentation is defensive (catches exceptions)
+- Documentation is updated
+
+
+Links
+-----
+
+- `CrewAI Documentation <https://docs.crewai.com/>`_
+- `OpenTelemetry Python <https://opentelemetry.io/docs/languages/python/>`_
+- `Splunk GenAI Utilities <https://github.com/signalfx/splunk-otel-python-contrib>`_
+
+
+License
+-------
+
+Apache-2.0
+

instrumentation-genai/opentelemetry-instrumentation-crewai/examples/manual/.env

@@ -0,0 +1,7 @@
+CREWAI_DISABLE_TELEMETRY=true
+OPENAI_API_KEY=
+OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317
+PYTHONUNBUFFERED=1
+OTEL_SERVICE_NAME=crewai-examples
+DEEPEVAL_TELEMETRY_OPT_OUT="YES"
+OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=true

instrumentation-genai/opentelemetry-instrumentation-crewai/examples/manual/Dockerfile

@@ -0,0 +1,35 @@
+FROM python:3.12-slim
+
+WORKDIR /app
+
+# Install git for pip dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    git \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy only the CrewAI instrumentation package and example
+COPY instrumentation-genai/opentelemetry-instrumentation-crewai /app/opentelemetry-instrumentation-crewai
+
+# Set working directory to examples
+WORKDIR /app/opentelemetry-instrumentation-crewai/examples
+
+# Install Python dependencies (including genai utils from PyPI)
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Install local CrewAI instrumentation package
+RUN pip install --no-cache-dir /app/opentelemetry-instrumentation-crewai
+
+# Verify packages are installed correctly
+RUN python3 -c "from opentelemetry.instrumentation.crewai import CrewAIInstrumentor; print('✓ CrewAI instrumentation available')" && \
+    python3 -c "from opentelemetry.util.genai.handler import get_telemetry_handler; print('✓ GenAI handler available (from PyPI)')"
+
+# Set default environment variables
+ENV OTEL_PYTHON_LOG_CORRELATION=true \
+    OTEL_PYTHON_LOG_LEVEL=info \
+    OTEL_EXPORTER_OTLP_PROTOCOL=grpc \
+    PYTHONUNBUFFERED=1 \
+    CREWAI_DISABLE_TELEMETRY=true
+
+# Run the customer support example
+CMD ["python3", "customer_support.py"]
+