An open-source Python framework for simulating autonomous agents in configurable environments. Inspired by complex adaptive systems research, Generative Agents (Park et al. 2023), and LLMs as Economic Agents (Horton 2023).
What It Does
- Configurable environments where agents interact according to defined rules
- Real-time graphical interface for monitoring and controlling agent behavior during simulation
- Modular architecture supporting various agent types and interaction protocols
Motivation
Exploring how agents behave in simulated economic and political environments — what happens when you give LLM-based agents resources, goals, and the ability to trade, negotiate, or compete.