Microsoft AutoGen
AutoGen is Microsoft Research's framework for building multi-agent AI applications in which agents can act autonomously or work alongside humans. Important: AutoGen is now in maintenance mode. It will not receive new features, it is community managed, and Microsoft recommends that new projects start with Microsoft Agent Framework, its enterprise-ready successor, using the official migration guide.
It remains relevant for teams with existing AutoGen code and for learning multi-agent patterns. Its layered design offers a Core API for event-driven, message-passing agents with local and distributed runtimes (including .NET and Python), an AgentChat API for common patterns such as two-agent chats and group chats, and an Extensions API for model clients, code execution and MCP tools. AutoGen Studio provides a no-code GUI for prototyping, and AutoGen Bench helps evaluate agents. Magentic-One, a multi-agent team for web and file tasks, is built on it.
Getting started: requires Python 3.10+. Install with `pip install -U "autogen-agentchat" "autogen-ext[openai]"`, and `pip install -U autogenstudio` for the GUI. Users upgrading from v0.2 should follow the migration guide, as the APIs changed significantly.
Features
Why choose it
Trade-offs
Maintenance check
- Last commit
- 2026-10-16
- Commits, last 90 days
- 1
- Releases, last 90 days
- 0
- Latest release
- python-v0.7.5 (2025-09-30)
- Open issues
- 1124
- Recent issues with at least one comment
- 2/8
Oldest week on the left; the latest four weeks are orange. Source: GitHub commit activity.
Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.
Most discussed open issues
- #7353 Feature: Cryptographic action receipts for enterprise agent governance (AAR)417 comments · 2026-03-06
- #7405 Proposal: GuardrailProvider protocol for tool call interception147 comments · 2026-03-14
- #7487 Multi-agent systems need a 'mission keeper' role — not a Boss Agent, but a dedicated goal integrity node74 comments · 2026-03-29
- #7492 Payment primitive for multi-agent systems - how are teams handling this?64 comments · 2026-03-30
- #7372 Cryptographic governance layer for AutoGen distributed agent runtime62 comments · 2026-03-11
Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.
Who it fits
- You maintain an existing AutoGen v0.2 or v0.4 system, or want to run research artefacts built on it such as Magentic-One and AutoGen Bench.
- You want to reference its layered event-driven architecture while planning a migration to Microsoft Agent Framework.
- You are starting a brand-new project; Microsoft states AutoGen is in maintenance mode with no new features and recommends Microsoft Agent Framework.
- You expect AutoGen Studio to be a turnkey production web app (the README notes Studio is for prototyping only).
How it differs
vs agent-framework★ 14.3kAutoGen is in maintenance mode; Microsoft Agent Framework is its active successor with built-in checkpointing, OpenTelemetry, and unified Python/.NET APIs.
vs CrewAI★ 60.2kAutoGen is a maintenance-mode Microsoft research framework; CrewAI is an actively developed Python framework for role-based Crews and deterministic Flows.
Compatibility
Quick start
Use cases
Alternatives
Featured in these lists
Comments
No comments yet. Be the first!