Summary
If you're editing Chinese text that feels AI-generated, this skill systematically removes the tells. It's based on Wikipedia's AI writing detection patterns and catches things like overused significance claims, vague attributions, gratuitous -ing phrases, and the classic three-item lists that LLMs love. It also pushes you to add actual voice back in, not just strip out the bad parts. The skill includes a 50-point scoring rubric across directness, rhythm, trust, authenticity, and conciseness. Translated from blader/humanizer with reference to hardikpandya/stop-slop, so it's pulling from established AI slop detection work. Works with Read, Write, Edit, and AskUserQuestion tools.
Install
npx skills add https://github.com/hairyf/skills --skill writing-humanizer-zh
