Package Overview
AutoGen for .NET provides multiple NuGet packages to suit different needs. You can install the all-in-one package or choose specific packages based on your requirements.All packages require .NET 6.0 or later.
Quick Install
For most users, install the all-in-one package:AutoGen.Core- Core abstractions and typesAutoGen.OpenAI- OpenAI integrationAutoGen.LMStudio- LM Studio supportAutoGen.SemanticKernel- Semantic Kernel integrationAutoGen.SourceGenerator- Type-safe function generation
Package Versions
Individual Packages
Install only the packages you need for a minimal footprint:AutoGen.Core
AutoGen.Core
Core abstractions for agents, messages, and group chat. Use this if you want to avoid dependencies on specific LLM providers.Includes:
- Agent abstractions (
IAgent,IStreamingAgent) - Message types (
TextMessage,ImageMessage, etc.) - Group chat and orchestration
- Middleware infrastructure
AutoGen.OpenAI
AutoGen.OpenAI
OpenAI model integration including GPT-4, GPT-3.5, and Azure OpenAI.Features:
OpenAIChatAgentfor OpenAI models- Streaming support
- Function calling
- Vision capabilities
AutoGen.Anthropic
AutoGen.Anthropic
Claude model integration from Anthropic.Features:
AnthropicClientAgentfor Claude models- Prompt caching support
- Tool/function calling
AutoGen.AzureAIInference
AutoGen.AzureAIInference
Azure AI Inference integration for models deployed on Azure.Features:
ChatCompletionsClientAgentfor Azure AI models- Support for GitHub Models, Azure AI Studio
- Compatible with various model providers
AutoGen.SemanticKernel
AutoGen.SemanticKernel
Integration with Microsoft Semantic Kernel.Features:
SemanticKernelAgentwrapper- Access to Semantic Kernel plugins
- Unified agent interface
AutoGen.SourceGenerator
AutoGen.SourceGenerator
Source generator for type-safe function definitions.Features:
[Function]attribute for methods- Automatic schema generation from XML docs
- Compile-time type safety
- Zero runtime reflection
AutoGen.DotnetInteractive
AutoGen.DotnetInteractive
Code execution support using dotnet-interactive.Supports:
- C# code execution
- F# code execution
- PowerShell scripts
- Python scripts (with Python kernel)
AutoGen.Mistral
AutoGen.Mistral
Mistral AI model integration.
AutoGen.Gemini
AutoGen.Gemini
Google Gemini model integration.
AutoGen.Ollama
AutoGen.Ollama
Local model support via Ollama.
AutoGen.LMStudio
AutoGen.LMStudio
LM Studio local model support.
Choosing the Right Package
All-in-One
Install
AutoGen if you want to get started quickly with common providers like OpenAI and don’t mind the additional dependencies.Core Only
Install
AutoGen.Core if you:- Want minimal dependencies
- Plan to implement custom agents
- Only need the abstraction layer
Verification
Verify your installation by checking the installed packages:Example Project File
Here’s a complete.csproj example with common packages:
YourProject.csproj
Replace version numbers with the latest available versions from NuGet.
Next Steps
Quick Start
Build your first AutoGen application
Core Concepts
Learn about agents and messaging