Peter Yang's no-ai-slop skill nears 1,000 GitHub stars

Peter Yang released no-ai-slop on July 22 as a free skill for Claude Code. It picked up close to 1,000 GitHub stars in a day, and the launch post pulled 4,388 likes and 454,000 views. Most AI cleanup tools rewrite a draft after the fact, and that's the assumption this one breaks.

The tool catches more than 20 AI prose patterns while you write, not after. Binary contrasts like "It's not X. It's Y." Throat-clearing openers like "Here's what nobody tells you." Fake-deep closers that land on a moral nobody asked for. It flags each one, explains the edit, and keeps the writer's own voice instead of swapping in a generic house style.

Yang shipped it for three surfaces: Claude Code, Cursor via .cursorrules, and Codex CLI.

A rewrite still carries the shape of the thing it's fixing. The sentence rhythm, the paragraph lengths, the places where a human would have stopped and didn't, stay put underneath the fix. Catching the pattern while the words are still being generated skips that step. The draft comes out clean instead of needing a second pass to hide the first one.

I run something similar on my own newsletter, and the gap between the two approaches is real, not cosmetic.

The GitHub repo doesn't say how it holds up on longer writing, where the tells spread across paragraphs instead of sitting in one sentence.