autogen

AutoGen is a programming framework for creating multi-agent AI applications that can act autonomously or work alongside humans. Pioneered in Microsoft Research, it uses a layered and extensible design with clearly divided responsibilities: - Core API: Implements message passing, event-driven agents, and local and distributed runtime. Supports cross-language development for .NET and Python. - AgentChat API: A simpler, opinionated API built on top of Core for rapid prototyping, supporting common multi-agent patterns like two-agent chat or group chats. - Extensions API: Enables first- and third-party extensions, including LLM clients (e.g., OpenAI, AzureOpenAI) and code execution capabilities. The ecosystem includes two developer tools: AutoGen Studio, a no-code GUI for building and prototyping multi-agent applications, and AutoGen Bench, a benchmarking suite for evaluating agent performance. The framework supports MCP (Model Context Protocol) servers for tool integration, such as web browsing via Playwright. AutoGen requires Python 3.10 or later. It is now in maintenance mode and will not receive new features or enhancements, being community-managed going forward. New users are directed to start with Microsoft Agent Framework, the enterprise-ready successor offering stable APIs and long-term support. Existing users are provided with a migration guide. Contributions to AutoGen are limited to bug fixes, security patches, and documentation improvements. The project is licensed under MIT for code and Creative Commons Attribution 4.0 for documentation.