What problem does it solve? Japanese text drafted with LLMs often reads smoothly but carries recognizable AI tells: missing human agency, inflated abstractions, uniform paragraph rhythm, full-width dashes, and overused stock phrases. This Skill provides a systematic rule set to find and fix those patterns before publishing. ## Core Features & Use Cases - Structural pattern detection: Identifies false agency, proposition-style H2 headings, hook-and-reversal openings, uniform rhythm, and both-sides-ism across six structural categories. - Vocabulary blacklist: Flags AI-favored words, translationese verbs, IT-loanword metaphors, hedging stacks, and decorative emoji across eight vocabulary categories. - Scoring and prioritized revision: Rates text on five axes (stance, rhythm, agency, specificity, reduction) out of 50 points and applies fixes in priority order from stance down to symbols. - Use Case: Paste an AI-drafted blog post and ask for a review; the Skill lists issues by severity across stance, agency, structure, vocabulary, and symbols, then rewrites while preserving meaning. ## Quick Start Ask the AI to review your Japanese draft using stop-ai-slop-jp and list the issues in severity order before rewriting.
