This repository provides
- Notebooks demonstrating a from-scratch implementation of four common design patterns: reflection, planning, tool use, and multi-agent collaboration.
- Code defining an agent class for each of these design patterns.
- Ready-to-use implementations of simple agent use cases across four different agent frameworks (AutoGen, CrewAI, Swarm, and LangGraph). These implementations can serve as a baseline for developing custom use cases.
Follow these steps to get started:
- create virtual environment
python3 -m venv venv
- activate the virtual environment
source venv/bin/activate
(venv\Scripts\activate
on windows) - install the requirements
pip install -r requirements.txt
- create a
.env
files and add an openai key to itOPENAI_API_KEY="sk-proj ...."
Using on of the agent frameworks:
- execute the
main.py
of the one of the frameworks, e.g.python agent_frameworks/openai_swarm_agent
- For using Autogen you need to have docker running. I you do not have it installed on your system, download it from the website (https://docs.docker.com/get-started/get-docker/). Then install and open it.