Scientific Agent Skills packages research procedures that runnable code alone does not supply, such as choosing an appropriate statistical test or checking a data identifier. Its 163 skills contain versioned instructions and, where applicable, reference material and scripts.
The library applies the on-demand skills pattern: an AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → reads a procedure when it judges that procedure relevant. This makes methodological guidance available during analysis. The paper measures properties of the documentation, however, and does not establish that agents choose the right skills or produce better scientific results.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
agent skill — A reusable instruction file that teaches an agent how to do one job well — the procedure, the tools, and what counts as done.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
articleLabAgent: Customize Any Research Hubs for Scientific Discoveries Using AI AgentsLei Liu, Yikun Zhang, Jialin Chen, Wanjia Zhao, Rex Ying, Wengong Jin, Hua Xu, James Zou, Tianyu Liu, Hongyu Zhao
articleSubagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic TasksWasu Top Piriyakulkij, Rachel Lawrence, Alicia Curth, Sushrut Karmalkar, Niranjani Prasad
articleAgentLogs: A Dataset for Opening the Black Box of GitHub's Cloud AgentJonan Richards, Kosei Horikawa, Youmei Fan, Yutaro Kashiwa, Mairieli Wessel