VoltAgent/awesome-ai-agent-papers

VoltAgent/awesome-ai-agent-papers is a curated, hand-picked list of research papers about AI agents, all published in 2026 and sourced from arXiv. It's like a weekly digest that filters the flood of new papers down to the ones most relevant to building and understanding AI agent systems.

What it covers:

• Core topics in the AI agent ecosystem: multi-agent coordination, memory & retrieval-augmented generation (RAG), tooling, evaluation & observability, and security.
• Each paper is categorized into one of these sections, with a short summary and a link to the arXiv abstract.

Who it's for:

• AI engineers building agent systems
• Researchers exploring new architectures
• Developers integrating LLM agents into products

How it's maintained:

• Updated weekly from arXiv, so it stays current.
• The maintainers manually review hundreds of papers each week, filter out irrelevant ones, and organize the rest—so you don't have to.
• Only papers published from January 2026 onward are included.

Current size:

• Over 364 research papers listed, with counts per category (e.g., Multi-Agent 54, Memory & RAG 58, Eval & Observability 86, Agent Tooling 95, AI Agent Security 83).

Extras:

• A Discord community link for discussion.
• Sponsors (Crawlbase, SerpApi) that support the project.
• Some promotional content for third-party tools (EveryFeed, LaunchKit) appears in the README, but the core value is the paper list itself.

In short: it's a high-quality, frequently updated reading list that saves you time by curating the most relevant AI agent research from arXiv.