AgentSociety, developed by the FIB Lab at Tsinghua University, is an advanced framework designed to simulate human behavior and social interactions using LLM-driven agents. It leverages the planning, memory, and reasoning capabilities of large language models (LLMs) to generate realistic behavior patterns, and supports environment design based on datasets, text, and rules. This framework holds significant importance in social science research, urban planning, and education, enabling researchers to better understand human behavior and social dynamics.