Skip to main content

​
What is AutoGen for .NET?

AutoGen for .NET is a powerful framework for building multi-agent conversational AI applications using C#. It provides a comprehensive set of tools and abstractions for creating intelligent agents that can communicate, collaborate, and solve complex problems together.

​
Key Features

Conversable Agents

Create AI agents with customizable behaviors, including AssistantAgent and UserProxyAgent for human-in-the-loop workflows.

Group Chat

Orchestrate multiple agents in dynamic conversations with flexible workflows and role-based coordination.

Type-Safe Functions

Define functions with source generators for compile-time type safety and automatic schema generation.

Code Execution

Execute C#, F#, Python, and PowerShell code snippets using dotnet-interactive integration.

​
Architecture

AutoGen for .NET consists of two package families:

​
AutoGen.* Packages (Stable)

The mature, stable packages derived from AutoGen 0.2:
  • AutoGen.Core: Core abstractions for agents, messages, and group chat
  • AutoGen.OpenAI: OpenAI model integration
  • AutoGen.Anthropic: Claude model integration
  • AutoGen.AzureAIInference: Azure AI Inference integration
  • AutoGen.SemanticKernel: Semantic Kernel integration
  • AutoGen.SourceGenerator: Type-safe function generation
  • AutoGen.DotnetInteractive: Code execution capabilities

​
Microsoft.AutoGen.* Packages (Preview)

New event-driven packages with modern architecture (APIs subject to change):
  • Microsoft.AutoGen.Core: Event-driven core runtime
  • Microsoft.AutoGen.Agents: Agent hosting infrastructure
  • Microsoft.AutoGen.AgentChat: Conversational agents
This documentation focuses on the stable AutoGen.* packages, which are recommended for production use.

​
Core Concepts

​
Agents

Agents are the building blocks of AutoGen applications. Each agent has:
  • Name: Unique identifier for the agent
  • System Message: Instructions that define the agent’s role and behavior
  • LLM Configuration: Model settings and API credentials
  • Human Input Mode: Control when human intervention is required

​
Messages

AutoGen supports multiple message types:
  • TextMessage: Plain text content
  • ImageMessage: Visual content
  • ToolCallMessage: Function invocation requests
  • ToolCallResultMessage: Function execution results
  • ToolCallAggregateMessage: Combined tool call and result

​
Middleware

Middleware components intercept and process messages, enabling:
  • Function calling
  • Message transformation
  • Logging and debugging
  • Custom workflows

​
Quick Example

Here’s a minimal example to get you started: 
using AutoGen;
using AutoGen.OpenAI;

var openAIKey = Environment.GetEnvironmentVariable("OPENAI_API_KEY");
var gpt35Config = new OpenAIConfig(openAIKey, "gpt-3.5-turbo");

var assistant = new AssistantAgent(
    name: "assistant",
    systemMessage: "You are a helpful AI assistant.",
    llmConfig: new ConversableAgentConfig
    {
        Temperature = 0,
        ConfigList = [gpt35Config],
    })
    .RegisterPrintMessage();

var user = new UserProxyAgent(
    name: "user",
    humanInputMode: HumanInputMode.ALWAYS)
    .RegisterPrintMessage();

await user.InitiateChatAsync(
    receiver: assistant,
    message: "Hello, can you help me?",
    maxRound: 10);

​
Next Steps

Quick Start

Build your first AutoGen application in minutes

Installation

Install AutoGen packages via NuGet

Core Concepts

Learn about agents and messaging

Examples

Explore sample applications

​
Resources

Build docs developers (and LLMs) love

Get started for freeTalk to us