Bart

You can draft, edit, or review prose to remove AI-generated writing patterns, and audit an entire docs folder for AI slop. It detects 41 patterns across content, language, style, filler, and rhetoric, then supports a draft, audit, final rewrite loop.

It targets writers, editors, and developers maintaining documentation who want natural, human-sounding text. It suits anyone reviewing drafts or auditing repositories for statistical AI tells, chatbot artifacts, and promotional or inflated language before publication.

It detects 41 patterns including significance inflation, promotional language, -ing tails, vague attributions, AI vocabulary, copula avoidance, negative parallelisms, false agency, em dash overuse, chatbot artifacts, hedging stacks, staccato drama, and aphorism formulas.

The skill is released under the MIT license, making it free to use, modify, and redistribute. It is version 1.0.0 and descends from the blader/humanizer, hardikpandya/stop-slop, and brandonwise/humanizer lineage.

Yes. It includes a no-fabrication rule so rewrites preserve factual content while improving naturalness. Voice calibration keeps the author's tone consistent, and statistical tells guide detection rather than inventing new claims during the rewrite loop.