AutoGen

AutoGen is a framework for cooperating AI agents, with local Playwright browser automation, code execution and an experimental OllamaChatCompletionClient for local models. Its orchestration can run on your machine or across servers. Standard OpenAI and Azure OpenAI service connections send model requests to remote services; compatible endpoint overrides are separate configurations.

Developers can choose how much of the workflow to build themselves. AgentChat handles common conversation patterns, including two-agent exchanges and group chats. Core provides event-driven agents and message passing for custom applications, with runtime support across Python and .NET. Agents can call other agents as tools or connect to MCP servers.

AutoGen Studio runs locally and provides a graphical interface for building and trying multi-agent workflows without writing code. It's intended for prototypes, so developers need to supply authentication, security and other requirements for production applications. AutoGen Bench evaluates agent performance. The Magentic-One application combines agents for browsing, code execution and file handling.

Python applications require Python 3.10 or later. Code uses the MIT license; documentation and other repository content use Creative Commons Attribution 4.0 International.