humanizer

What problem does it solve? AI-generated text carries recognizable tells — inflated symbolism, promotional language, negative parallelisms, em dash overuse, and templated phrases — that make writing feel robotic and erode reader trust. This Skill systematically finds and rewrites those patterns so text reads as naturally human. ## Core Features & Use Cases - 29-pattern detection framework: Scans text against a structured index of AI writing tells drawn from Wikipedia's "Signs of AI writing" guide and Blake Stockton's breakdowns, each with words-to-watch and before/after examples. - Mandatory negation sweep: Runs a Grep-based search for negative parallelism constructions ("not just X, it's Y") that survive casual read-throughs, then re-verifies after editing. - Voice restoration: Goes beyond removing bad patterns by adding rhythm variation, first-person perspective, and specific detail so the result has genuine personality rather than sterile neutrality. - Use Case: Paste a draft blog post or product announcement that reads as AI-generated; the Skill flags each pattern instance, rewrites the affected sentences, and returns a humanized version with a summary of changes. ## Quick Start Ask the AI to humanize a piece of text, for example: "Humanize this product announcement draft and remove any AI writing patterns."