Multi-Agent Frameworks¶

TraceVerde instruments 8 multi-agent and LLM frameworks with complete trace hierarchy.

CrewAI¶

Complete span hierarchy: Crew -> Task -> Agent -> LLM calls.

pip install genai-otel-instrument[crewai]

import genai_otel

genai_otel.instrument(
    service_name="my-crewai-app",
    enabled_instrumentors=["crewai", "openai"],
    enable_cost_tracking=True,
)

from crewai import Crew, Agent, Task

researcher = Agent(
    role="Senior Researcher",
    goal="Research topics thoroughly",
    backstory="Expert researcher with 10 years experience",
    llm="gpt-4"
)

task = Task(
    description="Research OpenTelemetry best practices",
    expected_output="Comprehensive report",
    agent=researcher
)

crew = Crew(agents=[researcher], tasks=[task], process="sequential")
result = crew.kickoff()
# Complete traces: Crew -> Agent -> Task -> LLM calls

Instrumented methods:

Crew.kickoff(), kickoff_async(), akickoff(), kickoff_for_each() and async batch variants
Task.execute_sync(), Task.execute_async()
Agent.execute_task()

Span types:

crewai.crew.execution - crew-level attributes (process type, agent/task counts)
crewai.task.execution - task attributes (description, expected output)
crewai.agent.execution - agent attributes (role, goal, backstory, LLM model)

Includes automatic ThreadPoolExecutor context propagation for worker threads.

See CrewAI example.

LangGraph¶

Stateful workflows with graph-based orchestration.

pip install genai-otel-instrument[langgraph]

Instruments invoke(), stream(), ainvoke(), astream() with automatic session ID propagation.

See LangGraph example.

Google ADK¶

Google Agent Development Kit instrumentation.

pip install genai-otel-instrument[google-adk]

Instruments Runner.run_async() and InMemoryRunner.run_debug(). Captures agent name, model, tools, sub-agents, and session info.

See Google ADK example.

AutoGen (Legacy)¶

Microsoft multi-agent conversations with group chats.

pip install genai-otel-instrument[autogen]

import genai_otel
genai_otel.instrument(service_name="autogen-example")

import autogen

assistant = autogen.AssistantAgent(
    name="assistant",
    llm_config={"config_list": [{"model": "gpt-4", "api_key": "..."}]},
)

user_proxy = autogen.UserProxyAgent(
    name="user",
    human_input_mode="NEVER",
    max_consecutive_auto_reply=2,
    code_execution_config=False,
)

user_proxy.initiate_chat(
    assistant,
    message="What is the capital of France?",
    max_turns=2,
)
# Traces: initiate_chat -> OpenAI calls with cost tracking

Instrumented methods:

ConversableAgent.initiate_chat() - agent-to-agent conversations
GroupChat.select_speaker() - speaker selection in group chats
GroupChatManager.run() - group chat orchestration

See AutoGen example.

AutoGen AgentChat (v0.4+)¶

pip install genai-otel-instrument[autogen-agentchat]

Instruments agent run()/run_stream() and team execution (RoundRobinGroupChat, SelectorGroupChat, Swarm). Captures participants, task content, and termination conditions.

OpenAI Agents SDK¶

Agent orchestration with handoffs, sessions, and guardrails.

Automatically instrumented when OpenAI SDK is installed.

See OpenAI Agents example.

AWS Bedrock Agents¶

Managed agent runtime with knowledge bases and RAG.

pip install genai-otel-instrument[aws]

See Bedrock Agents example.

Pydantic AI¶

Type-safe agents with Pydantic validation and multi-provider support.

pip install genai-otel-instrument[pydantic-ai]

See Pydantic AI example.

Other Frameworks¶

Framework	Description	Install Extra	Example
LangChain	Chains, agents, and tools	`[langchain]`	example
LlamaIndex	Query engines and indices	`[llamaindex]`	example
Haystack	Modular NLP pipelines with RAG	`[haystack]`	example
DSPy	Declarative LM programming	`[dspy]`	example
Instructor	Structured output extraction	`[instructor]`	example

All Examples¶

Browse all framework examples in the examples/ directory.