Audits and trust-ranks existing research assets to select the next workflow anchor.
npx skills add https://github.com/OpenLAIR/dr-claw --skill ds-intake-audit
Please help me install this Agent Skill. Skill: ds-intake-audit Source: https://github.com/OpenLAIR/dr-claw/tree/main/skills/ds-intake-audit Command: npx skills add https://github.com/OpenLAIR/dr-claw --skill ds-intake-audit
Research quests often do not start from a blank state: baselines, experiment results, drafts, or reviewer comments may already exist. This Skill audits that pre-existing state, ranks each asset by trustworthiness, reconciles it with durable records, and decides which research stage should take over next instead of blindly restarting from zero.
Audit the current quest's existing baselines, results, and drafts, then tell me which research stage should continue next.
High-intent search queries and answers about installing and using this skill.
Run an intake audit that inventories existing baseline assets, verifies the metric contract and provenance, and confirms the baseline through durable artifact records. It then routes the quest to the experiment stage without rerunning completed work.
Classify each result by trust level such as trusted, usable_with_verification, or stale_or_conflicting, then reconcile verified results into durable records. The audit recommends whether analysis, writing, or rebuttal should run next.
Skip the intake audit when the quest is genuinely blank and should start with normal scouting or baseline stages, when the state is already normalized with an obvious next step, or when the task is not research-related.
Yes. When review materials or revision requests define the real task, the audit classifies the state as review_package_ready and routes the quest to the rebuttal stage instead of treating it as a fresh paper-writing pass.
A baseline is only marked trusted when its metric contract, source, and comparability are clear. If provenance or verification evidence is missing, the audit records the gap explicitly rather than fabricating a clean history.
WILLOSCARCopy standard artifact templates into a new research workspace directory.
hungthinh04Draft and refine ML and AI research papers for NeurIPS, ICML, ICLR, ACL, AAAI, and COLM.
AndreJorgeLopesClassify new learnings and store them in Hindsight's retain tool.
leoplastureAudit and normalize existing research state, ranking asset trustworthiness and reconciling conflicts.
