What problem does it solve? Analyzing viral short web fiction (Tomato Novel shorts, Zhihu Yanxuan, Story Club pieces) by hand is slow and inconsistent, and insights get lost in chat context. This Skill runs a rigorous five-stage deconstruction pipeline that turns a short story into a persistent, structured analysis library that downstream writing workflows can directly consume. ## Core Features & Use Cases - Five-Stage Deconstruction Pipeline: Extracts story core, plot nodes, emotional curves, twist mechanics, writing techniques, character functions, and multi-layer resonance analysis, with strict stage ordering and per-stage quality gates. - Crash-Safe Resumable Output: Writes results to deconstruction-library/{title}/ with a _meta.json tracking completed stages, enabling resume, overwrite, or archive on rerun. - Genre-Calibrated Analysis: Loads genre-specific rubrics (chase-wife, rebirth-revenge, family drama, etc.) as comparison benchmarks, plus AI-tone scanning of the report itself against banned-word lists. - Use Case: Paste an 8,000-word Tomato short story; the Skill backs up the original, produces a deconstruction report, plot node list, and technique breakdown, then hands the full package to the story-short-write workflow for drafting the next piece. ## Quick Start Tell the AI "拆这篇短篇" and provide the story text or file path to start the full deconstruction pipeline.
