Humanizer RU is an editorial assistant and readability linter designed for LLM agents processing Russian text. Rooted in the editing traditions of Kornei Chukovsky and Maxim Ilyakhov's 'Write and Cut', it systematically strips away bureaucratic clichés (канцелярит), passive voice without agents, empty introductions, and AI-typical filler words across genres such as technical specifications, Habr articles, documentation, and client emails. With a strict 'fact lock' that prohibits fabricating details and three operational modes (careful, deep, and audit), this skill ensures generated Russian copy reads naturally to IT professionals and human audiences without sterilizing personal style or altering technical terminology.
主要功能
01Integrates with an optional command-line Python readability linter to calculate metric densities and formatting defects
02Supports genre-specific rules for technical documentation, Habr/Infostart longreads, chat posts, formal letters, and specs
03Preserves technical terminology, 1C/IT domain jargon, numbers, dates, and authorial tone under a strict fact-lock guarantee
04Eliminates AI slop, bureaucratic jargon, and generic chatbot introductions without rewriting clean paragraphs
05Provides multi-mode editing from minimal non-invasive touch-ups to full readability audits with categorized findings
0617 GitHub stars
使用场景
01Running automated text audits to flag AI-generated boilerplate and readability issues before publication
02Clearing bureaucratic fluff and passive voice from software documentation, user manuals, and client communications
03Polishing AI-drafted Russian technical articles and blog posts for platforms like Habr, Infostart, or Telegram channels