A minimal starter for Microsoft AutoGen — a framework for building multi-agent AI applications. This starter creates an AssistantAgent with a custom tool, powered by Nebius Token Factory.

Features

AssistantAgent from autogen-agentchat
Custom Python tool (get_current_time) auto-called by the model
Streaming responses rendered via Console
Nebius Token Factory via OpenAIChatCompletionClient (OpenAI-compatible)

Prerequisites

Python 3.10+
Nebius API key — Nebius Token Factory

Installation

git clone https://github.com/Arindam200/awesome-ai-apps.git
cd awesome-ai-apps/starter_ai_agents/autogen_starter

pip install -r requirements.txt
# or: uv sync

Create .env:

cp .env.example .env
# set NEBIUS_API_KEY

Usage

python main.py

Example Queries

"What time is it right now?" (triggers the get_current_time tool)
"Write a haiku about distributed systems."
"Explain multi-agent orchestration in one paragraph."

Technical Details

Framework: autogen-agentchat + autogen-ext[openai] (v0.4+)
Agent: AssistantAgent with reflect_on_tool_use=True
Model: Qwen/Qwen3-30B-A3B via Nebius (OpenAIChatCompletionClient with custom base_url)
Tool: get_current_time (plain Python function)

README.md

AutoGen Starter