The Open-Source AI Stack
RSS

The Stack · Agents · Open source

Microsoft Agent Framework

Microsoft's open agent framework consolidating AutoGen and Semantic Kernel; MIT, Python and .NET, model-agnostic via connectors, native MCP and A2A, graph-based orchestration. 1.0 GA April 2026.

The Microsoft Agent Framework is Microsoft's MIT-licensed agent framework for Python and .NET. It is the consolidation of two earlier Microsoft projects: AutoGen (multi-agent conversations, from Microsoft Research) and Semantic Kernel (the enterprise orchestration SDK). Both remain supported with bug and security fixes, but Microsoft has said most new investment now goes to the Agent Framework; Semantic Kernel 1.x carries a support commitment of at least one year past the framework's general availability. It reached 1.0 in April 2026. Multi-agent work uses graph-based workflows that compose agents into sequential, concurrent, handoff, and group-chat patterns. It is model-agnostic through first-party service connectors (Microsoft Foundry, Azure OpenAI, OpenAI, Anthropic Claude, Amazon Bedrock, Google Gemini, Ollama), with native MCP and A2A support. For open-source AI, the Agent Framework is the Microsoft entry in the fully-open vendor-SDK pattern, and it resolves a real source of confusion: AutoGen is no longer the current Microsoft agent framework. Teams starting fresh are pointed here rather than at AutoGen or Semantic Kernel.

Sources

Want a follow-up? Ask the chat about Microsoft Agent Framework in context. It will compare to siblings at the same layer and ground every claim in the wiki.

Other projects at the Agents layer

19 siblings · ordered open first