What problem does it solve? Reviewers often edit text while claiming to audit it, which masks structural problems and lets AI-generated patterns slip through. This Skill enforces a diagnosis-only review that produces an evidence-backed defect list before any rewriting begins. ## Core Features & Use Cases - Dimension-by-dimension checks: Run a 10-item professional checklist (chatbot residue, density, stance, specificity, templatedness, rhythm) plus fiction rubric groups A–E, each in a separate pass. - Evidence rule: Every finding must include a short quoted passage with location, so reports are actionable instead of vague impressions. - Cumulative verdicts: Score findings as clean, isolated hits, cluster, or structural cluster, then route to the right repair skill (refactor vs. recreate). - Use Case: Paste a draft blog post and receive a structured report quoting each AI tell, a verdict of "cluster", and a recommendation to run natural-refactor on the three deepest defects first. ## Quick Start Review this draft article for AI writing tells and give me a defect list with quoted evidence and a verdict, without editing the text.
