Get Started
Research questionHow can scientific agents choose domain-specific procedures that make analyses defensible?Scientific agents may produce runnable analyses while overlooking accepted tests, authoritative identifier namespaces, or caveats needed to interpret results. These procedural requirements differ across scientific fields and can determine whether an analysis is defensible.
AI
AI Agents
Latest papersRecent research connected to this question, newest first.Scientific Agent Skills: A Library of Procedural Knowledge for Research AgentsThe source concerns language-model agents analyzing experiments and describes Scientific Agent Skills, an open library of 163 versioned procedures across 16 practice areas. Its evidence is limited to documentation-corpus measurements; it reports no task-level evaluation or host selection rate.research paper · Sep 2, 2026
Related questions
How can coding agents repair scientific software when domain guidance may mislead them?How should scientific agents be evaluated on underspecified, attachment-rich requests without ground truth?How do we evaluate whether scientific agents make justified discoveries from data rather than reproduce known analyses?How can coding agents reliably implement systems-level requirements and detect the defects they introduce?
Home
Topics
Search
Library