If you've ever read something and thought "an AI definitely wrote this," you know the patterns: everything is pivotal or crucial, sentences end with those dangling -ing phrases, and there's way too much talk about broader implications. This skill catches that stuff based on Wikipedia's AI writing guide. It flags the obvious tells (serves as a testament to, nestled in the heart of, fostering innovation) and rewrites them into normal sentences. The interesting part is the second half, which pushes you past just removing slop into actually adding voice back. It won't magically make boring content fascinating, but it's solid for cleanup passes on marketing copy or articles that came out of ChatGPT a little too polished.
npx -y skills add countbot-ai/countbot --skill humanizer --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills