autogenlisted

# AutoGen (AG2) — Conversational Multi-Agent Framework **Source:** ag2ai/ag2 (Apache 2.0) — formerly microsoft/autogen; conversational agent orchestration ## Core Concepts | Component | Description | |-----------|-------------| | `ConversableAgent` | Base class — can send/receive messages, use LLM, execute code | | `AssistantAgent` | LLM-powered agent; generates replies using the model | | `UserProxyAgent` | Proxy for human input; can auto-execute code | | `GroupChat` | Manages conversation between multiple agents | | `GroupChatManager` | Routes messages in a GroupChat (round-robin or auto) | ## Install ```bash pip install ag2 # Optional: code execution support pip install ag2[jupyter-executor] ``` ## Two-Agent Conversation ```python import autogen config_list = [ { "model": "claude-sonnet-4-5", "api_key": "your-anthropic-key", "api_type": "anthropic", } ] llm_config = {"config_list": config_list, "cache_seed": 42} assistant = autogen.AssistantAgent( name="assistant", llm_config=llm_config, system_message="You are a helpful AI assistant.", ) user_proxy = autogen.UserProxyAgent( name="user_proxy", human_input_mode="NEVER", # NEVER | ALWAYS | TERMINATE max_consecutive_auto_reply=10, is_termination_msg=lambda x: x.get("content", "").rstrip().endswith("TERMINATE"), code_execution_config=False, # no code execution ) # Start conversation user_proxy.initiate_chat( assistant, message="Write a Pytho