What problem does it solve? Russian text written or rewritten by LLMs carries detectable machine patterns: officialese, calques from English, uniform sentence rhythm, and formulaic phrases like «важно отметить» or «не просто X, а Y». This Skill diagnoses those markers and rewrites the text so it reads as natural human-authored Russian, without flattening the author's voice. ## Core Features & Use Cases - Pattern-based diagnosis: A catalog of 52 AI-generation patterns in Russian, prioritized from critical HARD BANS to stylistic issues, with a regex scanner script that counts marker density per 100 words. - Voice calibration and contrastive rewriting: Builds a voice passport from the author's samples, then replaces the most predictable word choices with less probable, characterful ones while preserving personal anecdotes and colloquialisms. - Audit and verification loop: Quad-pass self-audit plus a before/after re-scan that catches new clichés introduced during the rewrite itself. - Use Case: A blogger pastes a Russian article draft that sounds robotic and asks to make it lively. The Skill scans it, flags officialese and empty openings, rewrites red-flag paragraphs in the author's register, and verifies no new AI markers slipped in. ## Quick Start Humanize this Russian text and remove all traces of AI generation while keeping my personal style.
