autogen

Files (1)

• SKILL.md

SKILL.md (first lines)

# AutoGen — Multi-Agent Conversation Framework

## Overview

AutoGen (Microsoft Research) is a framework for building LLM applications using multiple AI agents that collaborate through conversations. Agents can be LLM-powered, tool-enabled, or human-in-the-loop. AutoGen v0.4+ (the new AgentChat API) provides a clean async-first API with teams of agents (RoundRobinGroupChat, SelectorGroupChat) and structured message passing. Widely used for code generation, research automation, and agentic task completion.

GitHub: https://github.com/microsoft/autogen (42k+ stars)

## When to Use

- Multi-agent task completion (researcher + critic + executor pattern)
- Automated code generation and debugging loops
- Human-in-the-loop workflows with AI assistance
- Building agent teams with specialized roles
- Complex reasoning that benefits from agent debate/critique

## Installation

```bash
pip install autogen-agentchat autogen-ext[openai]

# For code execution support
pip install autogen-ext[docker]
```

## Key Patterns / Usage

### Simple Two-Agent Conversation
```python
import asyncio
from autogen_agentchat.agents import AssistantAgent
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_agentchat.conditions import TextMentionTermination
from autogen_ext.models.openai import OpenAIChatCompletionClient

async def main():
    model_client = OpenAIChatCompletionClient(model="gpt-4o-mini")
    
    # Define agents
    writer = AssistantAgent(
        name="Writer",
        model_client=model_client,
        system_message="You write Python code solutions.",
    )
    
    reviewer = AssistantAgent(
        name="Reviewer",
        model_client=model_client,
        system_message=(
            "You review code for bugs and improvements. "
            "If the code is correct and complete, say 'APPROVED'."
        ),
    )
    
    # Team with termination condition
    termination = TextMentionTermination("APPROVED")
    team = RoundRobinGroupChat([writer, reviewer], termination_condition=termination)
    
    # Run
    result = await team.run(task="Write a Python function to check if a number is prime.")
    for msg in result.messages:
        print(f"\n[{msg.source}]: {msg.content}")

asyncio.run(main())
```

### Agent with Tools
```python
import asyncio
from autogen_agentchat.agents import AssistantAgent
from autogen_ext.models.openai import OpenAIChatCompletionClient

def get_weather(city: str) -> str:
    """Get current weather for a city."""
    # Simulate weather API
    return f"Weather in {city}: 72°F, partly cloudy"

def search_web(query: str) -> str:
    """Search the web for information."""
    return f"Search results for '{query}': [simulated results]"

async def main():
    model_client = OpenAIChatCompletionClient(model="gpt-4o")
    
    agent = AssistantAgent(
        name="ToolAgent",
        model_client=model_client,
        tools=[get_weather, search_web],
        system_message="Use available tools to answer questions accurately.",
    )
    
    # Single agent run
    response = await agent.run(task="What's the weather in Paris and Tokyo?")
    print(response.messages[-1].content)

asyncio.run(main())
```

### Code Execution Agent
```python
import asyncio
from autogen_agentchat.agents import AssistantAgent, CodeExecutorAgent
from autogen_agentchat.teams import RoundRobinGroupChat
from autogen_agentchat.conditions import TextMentionTermination
from autogen_ext.code_executors.local import LocalCommandLineCodeExecutor
from autogen_ext.models.openai import OpenAIChatCompletionClient

async def main():
    model_client = OpenAIChatCompletionClient(model="gpt-4o")
    
    # Code writer agent
    code_writer = AssistantAgent(
        name="CodeWriter",
        model_client=model_client,
        system_message=(
            "Write Python code to solve the given task. "
            "Wrap code in ```python ... ``` blocks. "
            "After execution succeeds, say TASK_COMPLETE."
        ),

From UltronCore/claude-skill-vault, MIT.

More from UltronCore/claude-skill-vault