Open-source Microsoft Agent Framework allows developers to build complex multi-agent workflows using .NET or Python, according to the company. Credit: StockStudio / Shutterstock Microsoft has introduced the Microsoft Agent Framework, an open-source SDK and runtime for building, orchestrating, and deploying AI agents and multi-agent workflows, with full framework support for .NET and Python. Introduced October 1 and available on GitHub, the Agent Framework is designed to let developers build everything from simple chat agents to complex multi-agent workflows with graph-based orchestration, Microsoft said. Developers can experiment locally with the framework and then deploy to the Azure AI Foundry for AI apps, with observability, durability, and compliance built-in. Multi-agent systems can be built connecting Azure AI Foundry, Microsoft 365 Copilot, and other agent platforms. [ Related: Agentic AI – Ongoing news and insights ] Microsoft said that the Microsoft Agent Framework brings together and extends the ideas from the Semantic Kernel and AutoGen projects, and that developers currently using these projects will find the transition to the Agent Framework straightforward. The Agent Framework supports the Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol and features an OpenAPI-first design and cloud-agnostic runtime. As a result, the Agent Framework enables the following capabilities, according to Microsoft: Agents can dynamically discover and invoke external tools or data servers exposed over MCP. Agents can collaborate across runtimes using structured, A2A protocol-driven messaging. Any REST API with an OpenAPI specification can be imported as a callable tool instantly. Agents can run in containers, on-premises, or across multiple clouds, making them portable across environments. The Agent Framework supports a wide range of orchestration patterns, provides an extension package for experimental features, and pluggable memory modules that allows developers to choose from Redis, Pinecone, Qdrant, Weaviate, Elasticsearch, Postgres, or their own store for conversational memory. The framework also includes built-in connectors to Azure AI Foundry, Microsoft Graph, Microsoft Fabric, SharePoint, Oracle, Amazon Bedrock, MongoDB, and SaaS systems through Azure Logic Apps, Microsoft said.