Novel Deconstruct

拆书——对网络小说进行深度结构分析。输入番茄小说等平台的链接，自动抓取、解码、分析，输出七步骨架/冲突三层/人物弧光/对话技法/节奏控制等维度的拆书报告。触发词：拆书、拆解、分析这本小说、帮我看看这本书。

Install

skills CLI (Claude Code, Cursor, Codex and more)

npx skills add QQ-L-XX/novel-deconstruct --skill novel-deconstruct

Manual install

git clone --depth 1 https://github.com/QQ-L-XX/novel-deconstruct.git /tmp/novel-deconstruct
mkdir -p ~/.claude/skills
cp -r /tmp/novel-deconstruct/. ~/.claude/skills/novel-deconstruct

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Novel Deconstruct — 拆书技能

对网络小说进行系统性的写作技法分析。自动处理字体反爬、章节抓取、定量扫描、场景级深度拆解全流程。

## 检测流程

```
用户提供小说链接
├─ 番茄小说 (fanqienovel.com) → 字体解码 + API抓取
├─ 其他平台 → 通用网页抓取
└─ 用户直接粘贴文本 → 跳过抓取，直接分析
```

---

## Step 1: 获取元数据

读取小说主页，提取：
- 书名、作者、简介、题材标签、总字数、更新状态
- 完整章节目录（含每章标题）
- 章节总数 N

---

## Step 2: 抓取正文

### 番茄小说

1. 从主页 HTML (`__INITIAL_STATE__` JSON) 提取 `chapterListWithVolume`
2. 从 CSS 提取自定义字体 URL (woff2) → `tools/decode_font.py` 用 fontTools + ddddocr 建 PUA→汉字映射表
3. 按采样策略批量请求 reader API (`fanqienovel.com/reader/{itemId}`)，用映射表解码正文

### 采样策略

| 范围 | 策略 | 用途 |
|------|------|------|
| Ch 1-10 | **全文** | 开局分析（激励事件/人物引入/世界观建立） |
| Ch 11-N | 每 N/8 章取1章 | 结构采样（幕转折/冲突升级/人物弧光） |
| 最后10章 | **全文** | 结局分析（高潮/收束/主题落点） |

总计约 20-30 章。番茄免费章节通常仅前10章，后续章节可能需要登录 Cookie 或第三方镜像站。

---

## Step 3: 定量扫描（必做，不可跳过）

对每章正文执行以下自动化扫描，输出数据表：

### 3.1 基础指标

| 指标 | 方法 |
|------|------|
| 字数 | `len(re.sub(r'\s', '', body))` |
| 段落数 | 按换行分割 |
| 场景数 | 按空行/`-`/`—`/视角切换分割 |
| 句均长 | 以 `。！？；…` 断句，取均值 |
| 短句% | ≤15字的句子占比 |
| 长句% | >25字的句子占比 |
| 对话% | 含对话标记的段落占比 |
| 了/段 | 「了」字总数 ÷ 段落数 |

### 3.2 疲劳词扫描（逐词计数，输出单章+总计）

按以下分类，使用 `body.count(word)` 逐词扫描：

| 类别 | 词表 | 阈值 |
|------|------|------|
| 副词 | 突然/忽然/猛然/竟然/居然/渐渐/逐渐/依旧/仍然/默默/微微 | 各≤3次/章 |
| 动词模板 | 露出/陷入/涌起/泛起/浮现/充斥 | 各≤3次/章 |
| 连接词 | 然而/不过/于是/随即/紧接着/与此同时 | 各≤4次/章 |
| 身体模板 | 瞳孔一缩/身体一僵/倒吸一口凉气/呼吸一滞 | 各≤2次/章 |
| 含混词 | 仿佛/宛如/好像/似乎 | 合计≤2次/500字 |
| 「让」字 | 让/令/使 | **专项统计**（指南禁止项） |
| 「不是…而是…」 | 正则 `不是.{0,20}而是` | ≤3次/章 |

### 3.3 情绪标签词扫描

Read the full skill on GitHub

Details

Skill name: novel-deconstruct
Repository: QQ-L-XX/novel-deconstruct
License: MIT
Repo updated: 9 June 2026

FAQ

What does the Novel Deconstruct skill do?

How do I install the Novel Deconstruct skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add QQ-L-XX/novel-deconstruct --skill novel-deconstruct. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Novel Deconstruct popular?

It lives in QQ-L-XX/novel-deconstruct, which has 17 GitHub stars and 3 forks.

Related skills

Simple Man

Maksim-Burtsev/simple-man

High-compression professional communication mode. Use when the user wants fewer tokens, less reading, no filler, compact coding-agent status, terse technical answers, or low-cognitive-load collaboration without reducing effort, validation, proactivity, or accuracy. Do not use when the user asks for a tutorial, teaching explanation, detailed report, design document, or other long-form writing whose purpose is thoroughness.

Writing & ContentIn 2 lists

Magicmd

didilili/MagicMD

Use when the user wants public article URLs from WeChat, Juejin, CSDN, or technical blogs converted into Markdown packages, including batch conversion, local media, metadata, and extraction reports.

Writing & Content

SEO Geo Blog Writer

weipanux/seo-geo-blog-writer

Create blog posts optimized for traditional search engines (SEO) and generative AI citations (GEO). Supports two modes - keyword-driven (user provides target keyword) or topic expansion (automated keyword research). Complete workflow from keyword research through schema markup generation.

Writing & Content

Diataxis Docs Framework

anivar/developer-docs-framework

Enterprise technical documentation best practices, patterns, and frameworks for developer and partner adoption. Covers content architecture (Diataxis four quadrants), 14 content types (tutorials, how-to guides, API reference, SDK docs, migration guides, changelogs, runbooks, integration guides, troubleshooting, architecture docs), pluggable writing styles (Diataxis, Google, Microsoft, Stripe, Canonical, Minimal), information architecture, docs-as-code workflows, documentation audit, anti-patterns checklist, and developer experience (DX) strategy. 27 rules, 5 references, 6 style guides. Baselin

Writing & Content

Chinese Grammar Proofreader

orange4664/chinese-grammar-proofreader

纯中文病句辨析与修改 Skill。严格面向现代汉语句法、语义、逻辑和标点规范，检查成分残缺或赘余、搭配不当、语序不当、结构混乱/句式杂糅、指代不明、歧义、不合逻辑、关联词错误、两面对一面、否定失当、数量范围和标点问题。适用于“找语病”“检查病句”“辨析并修改病句”“按高考语文标准检查” “这句话有没有问题”等任务。默认只纠正明确问题，采用最小修改，不做泛润色、风格改写、段落美化或事实核查。段落和长文必须逐句通读、逐句诊断，不允许粗略扫读或只挑可疑句。

Writing & Content

Voltagent Docs Bundle

VoltAgent/skills

Look up VoltAgent documentation embedded in node_modules/@voltagent/core/docs for version-matched docs. Use for API signatures, guides, and examples.

Writing & ContentIn 1 list10

Browse the most popular Claude skills, all writing & content skills, or write your own with the skills generator.