Microsoft Agent Framework Documentation
Microsoft Agent Framework provides libraries for building AI agents in Python and .NET. It supports multiple LLM providers (Azure OpenAI, OpenAI, Anthropic, Ollama, Bedrock), durable execution with Azure Functions, MCP server integration, and agent-to-agent communication.
Quick Reference
| Topic | Entry Point |
|---|---|
| Python getting started | contents/python/README.md |
| .NET getting started | contents/dotnet/README.md |
| Core package | contents/python/packages/core/ |
| Agent samples | contents/python/samples/getting_started/agents/ |
| Azure Functions | contents/python/samples/getting_started/azure_functions/ |
| MCP integration | contents/python/samples/getting_started/mcp/ |
| Workflows | contents/python/samples/getting_started/workflows/ |
| Agent-to-agent | contents/python/packages/a2a/ |
When to use
Use this skill when the user asks about:
- •Microsoft Agent Framework setup and usage
- •Building agents in Python or .NET
- •Azure OpenAI, Anthropic, or Ollama providers
- •MCP server/client integration
- •Durable agents with Azure Functions
- •Agent tools and middleware
- •Agent-to-agent communication
- •Workflow orchestration patterns
How to find information
- •First, read
TOC.mdto see all available documentation files - •Identify the relevant section/files based on the user's question
- •Read specific files for detailed information
TOC.md contains a complete file listing organized by directory - always check it first before searching.