LLMAgentPapers

Why we included this project

Anyone researching how large language models behave as autonomous agents will find this list one of the more practical starting points available. The ZJUNLP group maintains it, and the papers are grouped by the sub-problems people actually work on: memory, planning, tool use, personality, and reinforcement-learning training, so you can jump straight to the question you care about instead of grinding through surveys end to end. Multi-agent systems get their own section that separates collaborative exchanges from adversarial and open-ended conversations, which is useful when you are trying to pick a coordination pattern to study. The benchmark and tool-list resources give newcomers a concrete idea of what to measure and what primitives exist to build on, and the maintainers keep the list current. Treat it as a reading guide for your own research or for getting a team up to speed on the field, not as runnable software.