LLM Agents Papers is a curated paper list that catalogs must-read research on large language model agents.
The repository addresses the challenge of navigating the rapidly expanding research landscape around LLM-based agents by organizing papers into a structured taxonomy. It covers foundational concepts and recent advances across multiple dimensions of agent development, including how agents form personalities, maintain memory, plan actions, use tools, and learn through reinforcement training. The collection also encompasses research on multi-agent systems, covering task-oriented communication, collaborative exchanges, adversarial interactions, and open conversations between agents. Beyond core agent research, the list includes papers on practical applications, frameworks for building agents, benchmarks for evaluation, and categorized tool resources.
Developers building or researching LLM agents should use this list as a reference guide to understand the state of the field and identify relevant prior work. The structured organization by capability area—personality, memory, planning, tool use, and training approaches—makes it useful for those focusing on specific aspects of agent design. The inclusion of multi-agent interaction patterns and application domains helps contextualize how individual agent capabilities combine in real systems. The repository also maintains links to related paper collections on prompt-based reasoning and knowledge editing for language models, positioning it within a broader ecosystem of LLM research resources.
The project maintains active curation with new papers regularly added to reflect emerging research. The repository includes a contribution framework that invites community participation in expanding and refining the paper list. Development activity shows ongoing engagement with the research community through structured organization of content and coordination with related paper collection efforts.
