Strips out the telltale signs of AI-generated text using 30 rules pulled from Wikipedia's AI writing cleanup guide. Catches the usual suspects: "delve," "leverage," dual adjectives everywhere, fake emotional ranges, and those formulaic "It's not just X, it's Y" constructions. Scores your text before and after so you can see what actually changed. Works best when you've got a draft that reads like it came from a chatbot and you need it to sound like a human wrote it. Won't add facts or change your meaning, just removes the robotic cadence and corporate padding. The self-check step matters because it's easy to swap one AI pattern for another without realizing it.
npx skills add https://github.com/richtabor/agent-skills --skill humanize