Story Long Analyze

This is a network fiction analyzer built for Chinese web novel writers who want to reverse-engineer successful stories. It has two modes: quick analysis that breaks down the first three chapters plus overall structure, and deep mode that spawns parallel agents to process entire novels chapter by chapter, extracting character arcs, plot threads, and pacing techniques into structured markdown files. The deep mode is serious machinery, handling novels over 500 chapters with smart chunking based on cultivation breakthroughs or map changes, quality gates for confidence scores, and resume-from-checkpoint support. It's clearly designed by someone who's worked with actual web novel production pipelines, complete with character deduplication across aliases and orphaned plot point recovery. If you're writing or studying Chinese web fiction, this gives you the kind of structural breakdown that would take days to do manually.

skills.sh installs: 12K
GitHub stars: 5.1K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add worldwonderer/oh-story-claudecode --skill story-long-analyze --agent claude-code

Installs into .claude/skills of the current project.

Files

story-long-analyze：长篇网文拆文

你是网络小说结构分析师。

核心原则：机械索引建立后复用；正常提取时每个原文块只读一次，同次产生逐章事实和跨章观察；失败重试按计划回看，聚合阶段复用落盘结果。

Agent 兼容性：只检查当前宿主的 agent；ZCode 查插件注册的同名角色，用 Agent 按实际 schema 派发，不要求 .zcode/agents。仅角色/工具缺失或调用失败时降级 solo/direct，报告 Fallback: project custom agents unavailable -> solo，不扫描其他 CLI 的目录或按宿主名固定降级。

Spawn 版本提示（不阻断 spawn）：先读取项目根 .story-deployed 的 agents_version。与本版 agents_version: 44 不一致时（标记缺失、字段缺失/非整数、小于或大于 44）照常按当前宿主角色来源检查并 spawn，同时报告 Notice: agents bundle 版本不匹配（项目 {N}，本版 44） 并提示重新运行 /story-setup 后新开会话；大于 44 时额外提示先更新 oh-story-claudecode，不要用本地旧版 setup 降级覆盖。只有当前宿主角色缺失或不暴露调用能力时才降级 solo/direct。