Connect AutoGen to Anoman AI

Step 1: Install AutoGen

pip install pyautogen

Step 2: Configure Anoman as the LLM

Pass an llm_config dict with Anoman's base URL. AutoGen uses this config for all agent completions.

import autogen

config_list = [
    {
        "model": "gpt-4o-mini",
        "api_key": "anm-sk-your-key-here",
        "base_url": "https://api.anoman.io/v1",
    }
]

llm_config = {
    "config_list": config_list,
    "temperature": 0.7,
}

Step 3: Define agents with Anoman config

assistant = autogen.AssistantAgent(
    name="assistant",
    llm_config=llm_config,
)

user_proxy = autogen.UserProxyAgent(
    name="user_proxy",
    human_input_mode="NEVER",
    max_consecutive_auto_reply=3,
    code_execution_config={"work_dir": "workspace"},
)

user_proxy.initiate_chat(
    assistant,
    message="Analyse the attached CSV and summarise the key findings.",
)

Group chat with Anoman

Multi-agent group chats work identically — every LLM call in the conversation routes through Anoman.

researcher = autogen.AssistantAgent("researcher", llm_config=llm_config)
writer = autogen.AssistantAgent("writer", llm_config=llm_config)
critic = autogen.AssistantAgent("critic", llm_config=llm_config)

groupchat = autogen.GroupChat(
    agents=[user_proxy, researcher, writer, critic],
    messages=[],
    max_round=6,
)
manager = autogen.GroupChatManager(groupchat=groupchat, llm_config=llm_config)

user_proxy.initiate_chat(manager, message="Write a research report on AI safety.")

Guardrail transparency headers

X-Anoman-Guardrail-Injection: pass score=0.08
X-Anoman-Guardrail-Pii: pass
X-Anoman-Guardrail-Content: pass
X-Anoman-Cache: none