AutoGen

By Microsoft

Microsoft's multi-agent conversation framework (maintenance-only; new work moved to Microsoft Agent Framework)

Website ↗ Repository ↗ Docs ↗

Maintenance onlyNo updates in 6+ monthspython-v0.7.52025-09-30PythonMIT

#multi-agent #conversation #microsoft #python

59.9k Completely free Open source

Doc version 1.0 · Updated June 28, 2026

Related tools

CrewAI

Multi-Agent Orchestration

LangGraph

Multi-Agent Orchestration

Overview

AutoGen is Microsoft’s framework for building multi-agent systems as conversations between agents — an event-driven core with a higher-level AgentChat API.
As of 2025 it is in maintenance mode: existing projects keep working, but Microsoft directs new development to the consolidated Microsoft Agent Framework (Semantic Kernel + AutoGen).

flowchart LR
  task[task] --> a[AssistantAgent]
  a --> model[(model client)]
  model --> a
  a --> out[reply]

The Code samples tab shows a single-agent AgentChat run.

When to use it

Reasonable for existing AutoGen codebases and for its conversation-centric multi-agent patterns. For new projects, weigh the Microsoft Agent Framework or an actively developed alternative such as CrewAI or LangGraph.

Example

Python · AgentChat

When to use

Use the AgentChat API for conversational multi-agent apps — agents that message one another to solve a task.

How it flows

An AssistantAgent wraps a model client; run drives the conversation and returns the message list.

Structure

flowchart LR
  task[task] --> a[AssistantAgent]
  a --> model[(model client)]
  model --> a
  a --> out[reply]

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

async def main():
    model = OpenAIChatCompletionClient(model="gpt-4o-mini")
    agent = AssistantAgent("assistant", model_client=model)
    result = await agent.run(task="Say hello in one short sentence.")
    print(result.messages[-1].content)

asyncio.run(main())

Example

AutoGen — implementation sample

A real, runnable mini-project. Download it and run with Docker.

View samples/autogen_1 on GitHub ↗

June 28, 2026

AutoGen — implementation sample

A tiny AutoGen (AgentChat) sample: a single assistant agent runs one task.

Note: AutoGen is in maintenance mode; Microsoft directs new work to the Microsoft Agent Framework.

Configure

cd samples/autogen_1
cp .env.sample .env
# edit .env: set OPENAI_API_KEY (and optionally MODEL)

Run with Docker

cd samples/autogen_1
docker build -t aas-autogen .
docker run --rm --env-file .env aas-autogen "Say hello in one short sentence."

Run with Docker (in a devcontainer with DooD)

The foreground docker run may print nothing under Docker-outside-of-Docker — run detached and follow the logs:

docker logs -f "$(docker run -d --env-file .env aas-autogen "Say hello in one short sentence.")"

Run locally

cd samples/autogen_1
pip install -r requirements.txt
python app.py "Say hello in one short sentence."

Example run

Output varies by model and run — LLMs are non-deterministic. One run with gpt-4o-mini:

Hello!

Files

.env.sample

# Copy to `.env` and add your OpenAI API key.
#   cp .env.sample .env
OPENAI_API_KEY=

MODEL=gpt-4o-mini

app.py

"""AutoGen (AgentChat): a single assistant agent runs one task.

Run:
    docker build -t aas-autogen .
    docker run --rm --env-file .env aas-autogen "Say hello in one short sentence."
"""

import asyncio
import os
import sys

from dotenv import load_dotenv
from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient

load_dotenv()

MODEL = os.environ.get("MODEL", "gpt-4o-mini")


async def main() -> None:
    prompt = " ".join(sys.argv[1:]) or "Say hello in one short sentence."
    agent = AssistantAgent("assistant", model_client=OpenAIChatCompletionClient(model=MODEL))
    result = await agent.run(task=prompt)
    print(result.messages[-1].content)


if __name__ == "__main__":
    asyncio.run(main())

Dockerfile

FROM python:3.13-slim

ENV PYTHONUNBUFFERED=1

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY app.py .

ENTRYPOINT ["python", "app.py"]