augmanitai

AutoGen RoundRobinGroupChat AUGMANITAI Example

A Microsoft AutoGen example demonstrating collaborative terminology research using RoundRobinGroupChat.

Architecture

Three agents collaborate in a round-robin discussion to answer research questions:

1. Query Analyzer — Breaks down questions into key concepts and identifies relevant domains
2. Terminology Expert — Searches and retrieves AUGMANITAI definitions and examples
3. Synthesis Officer — Integrates findings into coherent, structured responses

Communication Pattern

The RoundRobinGroupChat cycles through agents in sequence:

• Agent 1 responds to the initial message
• Agent 2 responds to Agent 1
• Agent 3 responds to Agent 2
• Cycle repeats up to max_turns (continues until conversation terminates)

Each agent sees the full conversation history and can reference previous contributions.

Files

• agents.py — AssistantAgent definitions with system prompts
• tools.py — AUGMANITAI lookup tools for agents
• mock_client.py — Mock ChatCompletionClient for offline execution
• groupchat_example.py — RoundRobinGroupChat orchestration
• test_example.py — Pytest tests
• run_example.py — CLI runner
• requirements.txt — Dependencies

Running

Install dependencies:

pip install -r requirements.txt

Run example:

python run_example.py

Run tests:

pytest test_example.py -v

Extending

To extend this example:

• Replace MockChatCompletionClient with real client (e.g., Azure OpenAI, Claude)
• Add function calling so agents can invoke AUGMANITAI tool methods directly
• Implement custom termination conditions (e.g., consensus reached, max turns)
• Add agent reflection: agents review peer responses and refine understanding
• Create specialized group chats for different task types
• Integrate human-in-the-loop review between agent turns

Key Learning Points

• Round-robin conversation: How agents with distinct expertise can collaborate iteratively
• Group chat state management: Conversation history tracking and context propagation
• System prompts for specialization: Role-specific instructions guide agent behavior
• Async/await patterns: AutoGen’s async architecture for scalable multi-agent workflows
• Mock clients for development: Testing group chat logic without API costs or latency

---

Bound by the Ethical Disclaimer §1–§20 (DISCLAIMER.md) and the repository licenses (LICENSE). Contact and provider: IMPRESSUM.md.

This site is open source. Improve this page.