AutoGen

Trace Microsoft AutoGen multi-agent applications in Muster via OpenLit instrumentation.

AutoGen is an open-source framework from Microsoft for building LLM applications with conversational agents. Muster traces AutoGen runs through the OpenLit instrumentation, capturing every agent message and LLM call as OpenTelemetry spans.

Setup

1. Install dependencies

pip install langfuse openlit autogen-agentchat autogen-ext

2. Configure credentials

import os

os.environ["LANGFUSE_PUBLIC_KEY"] = "pk-lf-..."
os.environ["LANGFUSE_SECRET_KEY"] = "sk-lf-..."
os.environ["LANGFUSE_BASE_URL"] = "https://app.getmuster.io"
os.environ["OPENAI_API_KEY"] = "sk-proj-..."

3. Initialize OpenLit instrumentation

import openlit

openlit.init()

Once initialised, every AutoGen agent invocation is captured automatically as an OTel span and forwarded to Muster.

4. Run an AutoGen agent

from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient
import asyncio

async def main():
    model_client = OpenAIChatCompletionClient(model="gpt-4o-mini")
    assistant = AssistantAgent("assistant", model_client=model_client)
    response = await assistant.on_messages_stream(
        [{"role": "user", "content": "Tell me about Muster"}],
        cancellation_token=None,
    )
    async for msg in response:
        print(msg)

asyncio.run(main())

Add user/session attributes

from langfuse import propagate_attributes

with propagate_attributes(user_id="user_123", session_id="sess_abc"):
    # run your AutoGen workflow
    ...

Troubleshooting

Enable debug logging: export LANGFUSE_DEBUG="True".
Call langfuse.flush() before exit in short-lived scripts.
Filter unrelated OTel spans to reduce billable units.