Skip to main content
Cloud / Azure / Products / Microsoft Agent Framework - Open-Source SDK for AI Agents

Microsoft Agent Framework - Open-Source SDK for AI Agents

Microsoft Agent Framework: open-source framework for AI agents and multi-agent workflows in .NET, Python, and Go. Successor to Semantic Kernel and AutoGen.

ai-machine-learning
Pricing Model Open source, free of charge; costs arise for the models and services you use
Availability SDK for .NET, Python, and Go; the Agent Framework for Go is in public preview per Microsoft
Data Sovereignty Depends on the connected models and services
Reliability No service SLA of its own since it is a framework; SLAs apply to the connected services SLA

What is the Microsoft Agent Framework?

The Microsoft Agent Framework is an open-source framework for building AI agents and multi-agent workflows in .NET, Python, and Go. It is not a billable Azure service but an SDK you embed in your applications. The source code is publicly available at github.com/microsoft/agent-framework.

Microsoft organizes the framework into four areas. Agents are individual agents that use LLMs to process inputs, call tools and MCP servers, and generate responses; supported providers include Microsoft Foundry, Anthropic, Azure OpenAI, OpenAI, and Ollama. The harness agent is an opinionated agent for long, multi-step tasks with planning and todo tracking, context compaction, file access and memory, don’t-ask-again tool approval, and observability. Workflows connect agents and functions through explicit execution paths, either functional or graph-based. Integrations bundle connections to model providers, agent services, tools, context providers, middleware, evaluation services, and UI frameworks.

The framework also provides foundational building blocks: model clients for chat completions and responses, an agent session for state management, context providers for agent memory, middleware for intercepting agent actions, and MCP clients for tool integration.

Microsoft describes Agent Framework as the direct successor to Semantic Kernel and AutoGen, created by the same teams. It combines AutoGen’s simple abstractions with Semantic Kernel’s enterprise features and adds graph-based workflows plus robust state management for long-running and human-in-the-loop scenarios.

Core Features

  • Agents with LLM integration, tool calls, and MCP server integration
  • Harness agent with planning, todo tracking, context compaction, file access, memory, and observability for multi-step tasks
  • Functional and graph-based workflows for explicit multi-agent orchestration
  • Model clients, agent session for state management, and context providers for memory
  • Middleware for intercepting and extending agent actions
  • Broad provider support including Microsoft Foundry, Anthropic, Azure OpenAI, OpenAI, and Ollama
  • SDKs for .NET, Python, and Go; the Go variant is in public preview per Microsoft
  • Migration guides from Semantic Kernel and from AutoGen

Typical Use Cases

Single agents with tool use
An agent answers open-ended or conversational requests, calls tools and MCP servers, and retains conversation context through the agent session.

Multi-agent orchestration
When several agents or functions must coordinate and execution order needs to be fixed, graph-based workflows express the flow explicitly instead of leaving it to the model.

Long-running tasks
The harness agent is built for multi-step tasks: planning and todo tracking structure the flow, context compaction keeps the context manageable, and memory plus file access preserve intermediate results.

Human in the loop
For approvals and review steps, the framework provides state management that supports long-running flows with human intervention.

Migrating existing agents
Teams already working with Semantic Kernel or AutoGen move their applications across using the migration guides Microsoft provides.

Benefits

  • Open source and free; you pay only for the models and services you use
  • Provider independence through broad model and integration support
  • Explicit control over multi-agent flows via graph-based workflows
  • Enterprise features inherited from Semantic Kernel: state management, type safety, middleware, telemetry
  • A clear migration path from Semantic Kernel and AutoGen
  • Usable in .NET, Python, and Go environments

Integration with innFactory

As a Microsoft Solutions Partner, innFactory supports you in building agent applications with the Microsoft Agent Framework: the architectural decision between agent and workflow, integration with Microsoft Foundry and other model providers, tool and MCP integration, plus observability and evaluations.

We describe how we implement agent architectures on Azure in practice in our articles CompanyGPT cloud stack on Azure and CompanyGPT with Microsoft Foundry, agents, and Bedrock. Contact us for a no-obligation consultation.

Typical Use Cases

Single agents with tool use and MCP integration
Multi-agent orchestration through graph-based workflows
Long-running tasks with planning, context compaction, and memory
Human-in-the-loop scenarios with state management
Migrating existing Semantic Kernel or AutoGen applications

Technical Specifications

0th Four primary areas: agents, harness agent, workflows, and integrations
1st Agents use LLMs, call tools and MCP servers, and generate responses
2nd Supported model providers include Microsoft Foundry, Anthropic, Azure OpenAI, OpenAI, and Ollama
3rd Harness agent for long, multi-step tasks with planning, todo tracking, context compaction, file access, memory, and observability
4th Workflows: functional and graph-based flows with explicit execution paths
5th Building blocks: model clients, agent session for state management, context providers for memory, middleware, and MCP clients
6th Packages: 'Microsoft.Agents.AI.Foundry' (.NET), 'agent-framework' (pip), 'github.com/microsoft/agent-framework-go' (Go)
7th Direct successor to Semantic Kernel and AutoGen, created by the same teams
8th Migration guides available from Semantic Kernel and from AutoGen

Frequently Asked Questions

Is the Microsoft Agent Framework a paid Azure service?

No. The Microsoft Agent Framework is an open-source SDK, not a billable Azure service. Costs arise for the models and services you call through it, for example deployments in Microsoft Foundry or Azure OpenAI. The source code is at github.com/microsoft/agent-framework.

Which languages are supported and what is their status?

Microsoft documents Agent Framework for .NET, Python, and Go. For Go, Microsoft explicitly states: 'The Agent Framework for Go is in public preview.' Declarative agents, RAG, CodeAct, and functional workflows are not yet available in the Go variant per Microsoft.

How does the framework relate to Semantic Kernel and AutoGen?

Microsoft describes Agent Framework as the direct successor to both projects, created by the same teams. It combines AutoGen's simple agent abstractions with Semantic Kernel's enterprise features such as session-based state management, type safety, filters, and telemetry, and adds graph-based workflows. Migration guides exist for both predecessors.

When should I use an agent and when a workflow?

Microsoft recommends an agent when the task is open-ended or conversational, when autonomous tool use and planning are needed, or when a single LLM call suffices. Microsoft recommends a workflow when the process has well-defined steps, when you need explicit control over execution order, or when multiple agents or functions must coordinate. If you can write a function to handle the task, Microsoft recommends doing that instead of using an AI agent.

What should I consider when connecting third-party systems?

Microsoft notes that using third-party servers, agents, code, or non-Azure Direct models is at your own risk. These are Non-Microsoft Products governed by their own license terms. Microsoft recommends reviewing all data shared with and received from such systems and managing whether data flows outside your organization's Azure compliance and geographic boundaries.

Note: All product information on this page has been compiled with care, but is provided without guarantee and may be outdated or incomplete. Cloud services evolve rapidly — features, pricing, SLAs, and availability change frequently. Authoritative and up-to-date information can only be found on the official product page of Azure (official documentation). This page does not represent an offer by Azure.

Microsoft Solutions Partner

innFactory is a Microsoft Solutions Partner. We provide expert consulting, implementation, and managed services for Azure.

Microsoft Solutions Partner Microsoft Data & AI

Ready to start with Microsoft Agent Framework - Open-Source SDK for AI Agents?

Our certified Azure experts help you with architecture, integration, and optimization.

Schedule Consultation