Claude Code gets a no-ai-slop skill that rewrites drafts and flags AI-sounding prose

aigclink · x · 2026-07-25

A new Claude Code skill called no-ai-slop can both rewrite drafts and diagnose “AI-sounding” prose.

Default mode: /no-ai-slop performs minimal effective editing, preserving the writer’s voice while removing the parts that make text harder to read, such as jargon, long sentences, abstract nouns, and over-complicated structure.
Detection mode: instead of guessing whether text was AI-written, it flags specific slop patterns, quotes the offending sentence, and suggests a one-line fix.
The accompanying eval.md acts as a pass/fail self-check so the skill can be evaluated by the same agent without a second reviewer.
The post argues this is a better approach than generic “AI detectors,” and includes 20+ patterns such as binary contrasts, throat-clearing openers, faux-insight setups, colon reveals, superficial analysis, and fake-profound endings.

More from coding & agent