AutoGen — Animate | AI Wisdom

← Back to topic
🎞 ANIMATE · Mode 3 of 5 · +100 XP— AutoGen
🤖Agents
👥Team
💬Converse
🏁Terminate
🎯 Next: Test →
Step 1 / 4
Speed
Creating Agents
Define specialized AssistantAgents with role prompts💡 Analogy
Two engineers with different code review checklists — same background, specialized focus.
Each AssistantAgent has a name and a system_message that defines its role. They share the same model but behave differently due to their instructions.
Console Output
> Coder agent created: "You write Python code..."
> Reviewer agent created: "You review Python code..."
●autogen_chat.py
Creating AgentsL8–20
1import asyncio
2from autogen_agentchat.agents import AssistantAgent, UserProxyAgent
3from autogen_agentchat.teams import RoundRobinGroupChat
4from autogen_agentchat.conditions import TextMentionTermination
5from autogen_ext.models.openai import OpenAIChatCompletionClient
6 
7model_client = OpenAIChatCompletionClient(model="gpt-4o")
8 
9coder = AssistantAgent(
10    name="Coder",
11    model_client=model_client,
12    system_message="You write Python code. Always end with APPROVE if the solution is complete.",
13)
14 
15reviewer = AssistantAgent(
16    name="Reviewer",
17    model_client=model_client,
18    system_message="You review Python code for correctness and best practices. Say APPROVE if the code is good.",
19)
20 
21user_proxy = UserProxyAgent(name="User")
22 
23termination = TextMentionTermination("APPROVE")
24 
25team = RoundRobinGroupChat(
26    [coder, reviewer, user_proxy],
27    termination_condition=termination,
28    max_turns=10,
29)
30 
31async def main():
32    result = await team.run(
33        task="Write a Python function to check if a string is a palindrome. Include tests."
34    )
35    for msg in result.messages:
36        print(f"[{msg.source}]: {msg.content[:200]}...")
37    print(f"Reason: {result.stop_reason}")
38 
39asyncio.run(main())
39 lines
Timeline — 4 Steps
Creating Agents
Define specialized AssistantAgents with role prompts💡 Analogy
Two engineers with different code review checklists — same background, specialized focus.
Each AssistantAgent has a name and a system_message that defines its role. They share the same model but behave differently due to their instructions.
Console Output
> Coder agent created: "You write Python code..."
> Reviewer agent created: "You review Python code..."

Was this page helpful?

Discussion

Define specialized AssistantAgents with role prompts

💡 Analogy

Two engineers with different code review checklists — same background, specialized focus.

Each AssistantAgent has a name and a system_message that defines its role. They share the same model but behave differently due to their instructions.

Console Output

> Coder agent created: "You write Python code..."
> Reviewer agent created: "You review Python code..."

Define specialized AssistantAgents with role prompts

💡 Analogy

Two engineers with different code review checklists — same background, specialized focus.

Each AssistantAgent has a name and a system_message that defines its role. They share the same model but behave differently due to their instructions.

Console Output

> Coder agent created: "You write Python code..."
> Reviewer agent created: "You review Python code..."