llm-agent-paper-list is a curated collection of research papers supporting a comprehensive survey on large language model based agents.
The repository addresses the need to organize and track the rapidly expanding body of research on LLM-based agents, which are artificial systems that use large language models as their foundation to sense environments, make decisions, and take actions. The approach is straightforward: it maintains a structured list of papers that are considered essential reading for understanding the field, drawn from and organized around a formal survey paper that systematically examines LLM-based agents and their potential.
Developers and researchers studying LLM agents, building agent systems, or exploring reinforcement learning approaches to agent training should find this collection useful as a reference point. The repository serves researchers interested in understanding the landscape of agent-based work rather than those seeking implementation code or frameworks, though the maintainers have released complementary projects including AgentGym for developing and evaluating agents across diverse environments and AgentGym-RL for training agents through reinforcement learning.
The project shows sustained development activity with regular updates introducing new resources and capabilities. The maintainers have released multiple related research papers and frameworks alongside the paper list, including work on reverse curriculum reinforcement learning for LLM reasoning and an interactive frontend for visualizing agent trajectories. The repository has expanded beyond a static list to serve as a hub connecting to active research implementations and datasets hosted on external platforms.
