What problem does it solve? Drafted web novel chapters often contain character voice drift, plot contradictions with earlier chapters, platform-banned words, and filler content that would fail review on the Fanqie (番茄) publishing platform. This Skill automates a structured compliance and consistency audit so problems are caught before publication. ## Core Features & Use Cases - OOC Detection Algorithm: Scores each character's dialogue against their character card using weighted feature vectors (formality, aggression, verbosity, catchphrases) with tiered thresholds from warning to blocking. - Plot Contradiction Cross-Validation: Verifies timelines, relationship levels, prior events, item/ability acquisition, and foreshadowing payoff against the previous five chapters and the story state files. - Banned Word & Typo Screening: Checks against a categorized banned-word library (political, sexual, violent, platform-specific, AI-signature phrases) and distinguishes intentional dialect/colloquial typos from real errors. - Quantified Quality Scoring: Rates seven dimensions (dialogue naturalness, pacing, hook, information increment, prose, platform fit, OOC control) into a weighted 1-5 composite score with a publish-readiness rating. - Use Case: After drafting chapter 12 of a system-flow web novel, run this review to confirm no unregistered characters appear, the system panel settlement formula matches the book config, and no first-level blocking issues remain before polishing. ## Quick Start Review the draft of chapter 12 against the character cards, foreshadowing table, and previous five chapters, then output the full audit report with quality scores.
