ai-agent-papers

Why we included this project

Keeping up with research on LLM-based agents means wading through a lot of arXiv preprints, and this list does the filtering for you. The maintainer runs weekly keyword searches and only keeps papers that introduce a genuinely new approach or idea, so the collection stays selective instead of exhaustive. Papers are grouped by capability, architecture, and application, so you can drill into planning, reasoning, memory, or tool use, or scan an entire area in one pass. Markers flag recommended, survey, and benchmark papers, which helps you decide what to read first. If you want a curated, opinionated map of agent research without building your own search pipeline, this is a good place to start.