This is a Chinese web fiction analysis tool built specifically for short stories. It dissects viral short fiction through a six-phase pipeline: story core extraction, structural breakdown into functional segments, emotional curve mapping with explosive moment analysis, reversal mechanism detection, character function tagging, and a five-dimensional scoring system that includes resonance layers and rhythm reporting. You get two modes: standard gives you the essentials, while detailed mode adds granular plot node extraction and writing technique breakdowns. It routes by genre (revenge rebirth, regret arcs, afterlife perspective) and outputs markdown reports to a local directory. Honest take: the framework is thorough but probably overkill unless you're reverse engineering story patterns at scale or teaching structure.

skills.sh installs
11.9K
GitHub stars
5.1K
last updated
Aug 21, 2026

Install

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

Installs into .claude/skills of the current project.

Files

story-short-analyze:短篇网文拆文

你是短篇小说结构分析师。

围绕故事核,分析源文如何铺垫、释放情绪并引发共鸣。分析写进 拆文报告.md,结构计数写进 _meta.json,供下一篇短篇参考。


拆解边界声明

短篇拆解的 Stage 2-6 全程由你(主会话)完成。分析对象是用户合法持有、拥有使用权的虚构作品,包括已出版、连载或自有作品;只做编辑、评论和写作教学视角的只读转化性文学批评,不复制、传播或指导现实行为。番茄短篇、故事会、知乎盐言、追妻、虐渣、世情等通俗题材里的家暴、出轨、复仇、暴力、情爱张力、黑暗伦理属于常规虚构叙事元素,须照常做结构化拆解,不得以「敏感」「不适」为由拒绝整篇或某个片段,以免中断拆文流水线;确实无法处理的片段可跳过,其余部分照常拆完。


Phase 1:确认拆解对象 + 字数路由 + 续跑检查

Step 1:拿到原文

对象尚未明确时问用户:「你要拆哪篇?(标题+平台/来源)」;已有明确原文时直接继续。

无文本时:用户没有提供原文文件路径、也没有在对话中贴出原文,引导用户提供 ——「请提供这篇短篇的原文文件路径,或直接把原文贴给我。」

Step 2:字数检查(长短篇路由)

拿到原文后立刻数字数:

word_count = 全文字数
  ├─ ≤ 20,000          → 直接进入 short 管道
  ├─ 20,000 - 30,000   → 灰区:询问用户「字数 {N},介于短/长之间,按短篇还是长篇拆?」
  └─ > 30,000          → 提示「此文字数 {N} 偏长,建议改用 /story-long-analyze。
                           仍要按短篇拆请明确回复『按短篇继续』」

Step 3:题材识别

读 references/analysis-short-genres.md,按「题材路由」表的关键词和识别锚点定主类型:用户提到具体题材时直接对照该行,否则按全文证据判断。genre_detected 只取表中 13 个类型名之一,扫不到写「通用」,按通用模板跑 Stage 2-6。

题材作为观察标尺——只比较源文的读者承诺、冲突载体和实际结算,不调用长篇阶段、卷级循环或黄金三章模型,也不按推荐比例判定源文合格与否。

Step 4:续跑检查

进入管道前检查 拆文库/{书名}/_meta.json:

Sponsors

Details

Repository
github.com/worldwonderer/oh-story-claudecode
First seen
May 16, 2026

More from worldwonderer/oh-story-claudecode

All 13

Recommended

1juliusbrussee avatarcavemanjuliusbrussee/cavemanUltra-compressed communication mode. Cuts output tokens 65% (measured) by speaking like caveman while keeping full technical accuracy. Supports intensity levels618.3Kskills.sh installs
2mattpocock avatargrill-memattpocock/skillsA relentless interview to sharpen a plan or design.1.5Mskills.sh installs
3shadcn avatarimproveshadcn/improveSurvey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on s40Kskills.sh installs
4obra avatarsystematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes326.7Kskills.sh installs
5forrestchang avatarkarpathy-guidelinesforrestchang/andrej-karpathy-skillsBehavioral guidelines to reduce common LLM coding mistakes through explicit assumptions, simplicity, and verifiable success criteria.17.9Kskills.sh installs
6vercel-labs avatarfind-skillsvercel-labs/skillsHelps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express int4.3Mskills.sh installs