GH Repository · tsinghua-fib-lab
AgentSociety
AgentSociety 2 is a modern, LLM-native agent simulation platform designed for social science research and experimental design. It provides a flexible framework for creating and managing intelligent agents in simulated environments.
- stars
- 1,289
- 30-day movement
- +145/day
- Related entries
- 60
- Connections
- 3
python/uvpythonagentsociety2ai-social-scientistagentagent-based-modelingmulti-agentssocial-simulationsimulationharness-engineeringPythonai-assistantsai-for-scienceagent-appmakedocker
AgentSociety 2 is an LLM-native agent simulation platform from Tsinghua FIB Lab, aimed at social science research and experimental design. It provides a framework for creating and managing intelligent agents within simulated environments.
You want to run multi-agent social simulations driven by LLMs rather than hand-coded agent rules.
Use it to
- Build LLM-driven multi-agent simulations
- Design social science experiments with simulated agents
- Prototype agent-based models in Python
- Set up reproducible simulation environments via Docker and uv
For Social science researchers and agent-based modeling practitioners
- Role
- agent-app
- Language
- Python
- Licence
- Apache-2.0
- Forks
- 213
- Open issues
- 26
- Last push
- 2026-09-15
- Latest release
- agentsociety2-v2.0.0 · 2026-03-05
topicssocial-simulationmulti-agentsagent-based-modelingllm-agentsresearchpython