What problem does it solve? Research results are often delivered as bare conclusions with no trace of how they were reached, leaving human reviewers unable to verify claims, spot hidden judgment calls, or steer the work. This Skill turns research reporting into a transparent, auditable process handoff. ## Core Features & Use Cases - Decision-trail reporting: Structures every report as question → hypotheses → experiments (failures included) → decision points with options, evidence, and recommendations. - Claim verification ladder: Tags each headline claim with its verification level, from formal verifier down to self-assessment, and marks unverified claims explicitly. - Role separation and reader analysis: Enforces renderer ≠ decider checkpoints, requires a written reader analysis, and links every headline number to its run (command, log path, seed). - Use Case: When finishing an experiment series, produce a progress report that lists each experiment with a one-line result, states the statistical verification of the main claim, flags what is not verified, and ends with an explicit decision request for the human reviewer. ## Quick Start Use the present skill to write a research progress report for my experiment results, including hypotheses, failed runs, verification levels for each claim, and the decision I need from my advisor.
