WooooDyy/LLM-Agent-Paper-List

Curates foundational research papers organized around a three-component agent architecture (brain, perception, action) with comprehensive coverage of single-agent, multi-agent, and human-agent cooperation scenarios. Complements the survey with **AgentGym**, a practical platform for developing and benchmarking LLM agents across diverse environments, plus **AgentGym-RL** for training agents via multi-turn reinforcement learning. Structured taxonomy spans agent construction, real-world applications, emergent social phenomena in agent societies, and open research challenges.

8,072 stars. No commits in the last 6 months.