What problem does it solve? French text generated by LLMs carries recognizable artifacts: overused vocabulary like « crucial » and « notamment », anglicisms such as « faire du sens », superficial -ant participles, em dash overuse, and formulaic openings. This Skill identifies and removes 27 documented AI-writing patterns from French content, then injects genuine voice and personality so the result reads as human-authored. ## Core Features & Use Cases - 27-pattern detection and repair: Covers AI vocabulary, anglicisms from English-first models, copula avoidance, negative parallelisms, rule of three, synonym cycling, false ranges, redundant adjective doublets, sycophantic tone, conversation artifacts, and structural uniformity. - French typography correction: Converts curly quotes to guillemets, fixes spacing before punctuation, removes Oxford commas, and corrects number formatting. - Register preservation: Maintains « langage soutenu » when the input is formal; never downgrades legal or academic prose to casual French. - Voice injection (Part 3): Adds opinions, first person, varied rhythm, and second-degree irony so cleaned text does not read as sterile. - Use Case: Paste a ChatGPT-drafted French blog post full of « dans le paysage actuel » and « il convient de noter que »; receive a draft rewrite, a self-audit of remaining AI tells, and a final humanized version with a summary of removed patterns. ## Quick Start Humanize this French text and remove all AI-writing patterns while keeping its original register.
