Rewrite AI-sounding prose into natural, human writing - detect and fix em-dash overuse, the rule of three, AI vocabulary, vague attributions, negative parallelisms, and filler, while preserving meaning and matching voice.
---
name: humanizer
description: Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound natural and human-written - detects inflated symbolism, promotional language, vague attributions, em-dash overuse, the rule of three, AI vocabulary, passive voice, negative parallelisms, and filler. Based on Wikipedia's "Signs of AI writing" guide.
---
An editor that strips telltale AI patterns while keeping the message and the author's voice intact.
## How to use
1. Scan for AI patterns - em-dash overuse, rule-of-three lists, vague attributions ("experts say"), AI vocabulary, negative parallelisms ("not X, but Y"), promotional language, and filler phrases.
2. Rewrite rather than delete - replace AI-isms with natural alternatives and cover everything the original covered (five paragraphs in, five out).
3. Calibrate voice - if the user supplies a writing sample, match its sentence length, word level, and punctuation habits; otherwise use a natural, varied, opinionated default.
4. Add personality (the "soul" pass) only for blog/essay/opinion writing; keep technical, legal, and reference text plain.
5. Run the draft -> audit -> final loop, re-scanning the rewrite for any patterns that crept back in.
Full skill & source: https://github.com/blader/humanizer/tree/9600f2b7241cb4eed6ad803abee5ea01d67fe8e4