What problem does it solve? Long-form web novel projects lose consistency across hundreds of chapters: outlines drift from the actual prose, foreshadowing gets dropped, character states contradict earlier chapters, and AI-generated text carries detectable stylistic fingerprints. This Skill structures the entire production process—topic selection, core settings, volume outlines, per-chapter blueprints, drafting, and quality review—into file-based, resumable workflows. ## Core Features & Use Cases - Five-phase writing workflow: Covers topic selection, core settings, outline building (volume-level and per-chapter blueprints with word-count budgets), chapter drafting with context loading, and multi-gate quality review including banned-word and AI-pattern scans. - Pipeline adapter mode: When a novel-studio repository is detected, it refuses to write prose directly and instead converts requests (new book, daily update, chapter rewrite) into novel-studio --pipeline invocations with prompt files. - Deterministic quality scripts: Ships Node.js scripts that detect AI-style sentence patterns, em-dashes, paragraph stutter, model degeneration, and normalize punctuation in finished chapters. - Use Case: A writer says "帮我开书" (start a new book) with a target platform and genre; the Skill guides them through benchmarking against comparable titles, builds the settings/outline/tracking file tree, then produces chapters one at a time with word-count verification and continuity tracking after each chapter. ## Quick Start Tell the assistant to start a new long-form web novel using the story-long-write skill, specifying your genre, target platform, and any benchmark titles you want to emulate.
