Microsoft has released version 1.0 (GA) of the Microsoft Agent Framework, its consolidated platform for building production-ready multi-agent AI systems. The release includes a VS Code extension with two built-in Copilot skills — one for agent framework guidance and one for Microsoft Foundry deployment — enabling AI-assisted code generation, debugging, and migration from other frameworks like Semantic Kernel and AutoGen. Key new features include a visual Agent Inspector for debugging multi-agent workflows, fan-in/fan-out and map-reduce parallelization patterns, support for long-running resumable/checkpointed workflows, and chat history management strategies for memory control. The framework officially supports Python and .NET/C#, with parity as a stated goal. Community engagement is encouraged via GitHub, Discord, and weekly office hours.
Sort: