autogen

Microsoft's framework for building multi-agent AI applications now sits in maintenance mode. The README directs new users to Microsoft Agent Framework and keeps the layered API design documented.

AutoGen is a framework for creating multi-agent AI applications that can act autonomously or work alongside humans. The README opens with a caution: the project is now in maintenance mode, will not receive new features or enhancements, and is community managed going forward. New users are pointed to Microsoft Agent Framework, described as the enterprise ready successor. The repository is Python and the documentation is released under CC-BY-4.0.

The design is layered and extensible, with clearly divided responsibilities that build on top of each other. The Core API implements message passing, event driven agents, and local and distributed runtimes, with cross language support for .NET and Python. The AgentChat API is simpler and more opinionated, built for rapid prototyping and closest to what v0.2 users already know. The Extensions API lets first and third party extensions expand the framework, including specific LLM clients like OpenAI and AzureOpenAI and code execution.

AutoGen Studio is a no-code GUI for building multi-agent applications, though the README warns it is meant for rapid prototyping rather than production. AutoGen Bench is a benchmarking suite for evaluating agent performance. Magentic-One is described as a multi-agent team built with the AgentChat and Extensions APIs, handling tasks that require web browsing, code execution, and file handling.

The quickstart calls for an OpenAI API key, since the samples call the OpenAI API. A warning says to only connect to trusted MCP servers, because they may execute commands in your local environment or expose sensitive information. A migration guide covers the move from AutoGen v0.2, and contributions in maintenance mode are limited to bug fixes, security patches, and documentation improvements.

• Source 1 · Project repository
• Source 2 · README