What problem does it solve? Long-form posts often grow longer without getting smarter: they restate one idea several times, announce their own insights, end on tidy maxims, and carry AI writing tells like em-dashes and rule-of-three rhythms. This Skill is an adversarial editing pass that makes a post earn every word and read like a blunt human engineering manager wrote it. ## Core Features & Use Cases - Padding taxonomy hunt: Finds and quotes thesis restatements, insight signposts, tidy-maxim endings, hedge-balance paragraphs, empty generalities, and throat-clearing openers, with a cut or tighten instruction for each. - AI-tell scrub: Enforces zero em-dashes, flags rule-of-three rhythms, stacked "not X, it's Y" pivots, thesaurus-prose, and banned words, while protecting the post's edge so edits never sand off the voice. - Cross-post repetition check: When reviewing multiple posts, detects reused metaphors, biographical wells, sentence shapes, and closing beats across the set. - Use Case: After expanding a 500-word essay into 1000 words, run a diagnose pass to verify the added length bought new insight, get a prioritized list of cuts with exact quotes, and confirm the takeaway is non-obvious before publishing. ## Quick Start Ask the AI to run a diagnose pass on your draft post, for example: review my essay draft.md with stark-blog-sharpen and tell me what to cut.
