A skill that cleans up 26 AI writing patterns while trying to preserve the author’s voice
Based on Wikipedia’s “Signs of AI writing” guide, it scans for 26 patterns: inflated importance, false contrasts, ornate headings, chatbot leftovers, and more. It rewrites the text first, provides a brief critique, then produces a final version that preserves sourced claims while avoiding changes to code, data, and link destinations. Its stance is refreshingly honest: “Beating detectors is not the goal.”
Polishing formulaic passages in user-facing writing such as blogs, product announcements, emails, and reports; matching your voice using your own writing samples.
It may mistake deliberate rhetoric or specialist language for “AI tells” and flatten the prose, so the final editorial call is yours. It does not replace factual verification.
npx skills add blader/humanizer --global --agent '*'