Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generates README.md and DESIGN.md scaffolds by analyzing module structure. Use when creating documentation templates for new modules. Automatically triggered at module creation.
.claude/skills/telagod-generating-docs/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -27% | 0% |
> 生成的是骨架，不是答案。填决策理由是人的工作，工具不能代劳。
| 场景 | 跑 | 理由 | |------|------|------| | 新模块创建后第一时间 | ✅ | 趁记忆新鲜填决策 | | verifying-modules 报「缺 README/DESIGN」 | ✅ | 直接补救 | | 重构后边界变化 | ⚠ | 看是否要重生成或手动改；不要 --force 覆盖已有内容 | | 已有完整文档 | ❌ | 不要破坏人工写的内容 |
| 字段 | 自动来源 | 待人补 | |------|----------|--------| | 模块名 | 目录名 | 是否有更好的展示名 | | 描述 | 顶级 docstring（Python） | 「存在理由」必须人写 | | 特性 | TODO（占位） | 列 3–5 个核心能力 | | 依赖 | requirements.txt / package.json / Cargo.toml 等 | 是否有运行时假设 | | 使用方法 | TODO（占位） | 5 分钟跑通示例 | | API 概览 | 公开类/函数签名 | 调用顺序、生命周期 | | 目录结构 | tree 输出 | — |
| 字段 | 自动 | 待人补（最重要） | |------|------|------------------| | 设计目标 / 非目标 | TODO | 核心决策驱动力 | | 架构 | TODO | 含一张图最好 | | 核心组件 | 公开类/函数列表 | 各组件的协作关系 | | 决策表 | TODO | 方案对比 + 选择理由 | | 技术选型 | 依赖列表 | 选 X 不选 Y 的理由 | | 权衡 | TODO | 牺牲了什么 | | 安全考量 | TODO | 信任边界、威胁模型 |
| 语言 | 提取深度 | |------|----------| | Python | 类、函数、docstring、依赖（最完整） | | Go / TypeScript / Rust | 目录结构 + 依赖 | | 其他 | 基础目录结构 |
doc_generator.js 生成骨架
↓
人工填 TODO（决策理由、使用示例、非目标）
↓
verify-module 校验完整性
↓
通过则提交--force--force 会无差别覆盖。先 diff 旧版，手动合并。bashnode scripts/doc_generator.js <模块路径> node scripts/doc_generator.js <模块路径> --force # 强制覆盖（慎用） node scripts/doc_generator.js <模块路径> --json # CI 用
工具给骨架，TODO 不补完不算交付。生成后 5 分钟内人工填决策；超过 1 天会忘原因，文档质量陡降。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,583 | 8,179 | -48% | 1 | 1 | 0% | 2,876 | 1,734 | -40% | 0 | 0 | — |
case-02 | fail→pass | 10,666 | 7,868 | -26% | 1 | 1 | 0% | 1,834 | 2,056 | +12% | 0 | 0 | — |
case-03 | fail→pass | 14,069 | 7,724 | -45% | 1 | 1 | 0% | 2,148 | 2,121 | -1% | 0 | 0 | — |
case-04 | pass→fail | 13,916 | 12,390 | -11% | 1 | 1 | 0% | 2,127 | 2,561 | +20% | 0 | 0 | — |
case-05 | pass→pass | 14,774 | 7,645 | -48% | 1 | 1 | 0% | 2,528 | 2,080 | -18% | 0 | 0 | — |
case-06 | pass→pass | 15,619 | 13,590 | -13% | 1 | 1 | 0% | 2,514 | 2,985 | +19% | 0 | 0 | — |
case-07 | pass→pass | 12,946 | 9,187 | -29% | 1 | 1 | 0% | 2,026 | 2,315 | +14% | 0 | 0 | — |
case-08 | pass→pass | 7,467 | 2,227 | -70% | 1 | 1 | 0% | 1,159 | 1,140 | -2% | 0 | 0 | — |
case-09 | pass→pass | 9,660 | 6,720 | -30% | 1 | 1 | 0% | 1,251 | 1,880 | +50% | 0 | 0 | — |
case-10 | fail→pass | 9,928 | 6,468 | -35% | 1 | 1 | 0% | 1,408 | 1,710 | +21% | 0 | 0 | — |
case-11 | pass→pass | 14,745 | 5,360 | -64% | 1 | 1 | 0% | 2,127 | 1,683 | -21% | 0 | 0 | — |
case-12 | fail→pass | 13,948 | 5,648 | -60% | 1 | 1 | 0% | 2,151 | 1,560 | -27% | 0 | 0 | — |
case-13 | pass→pass | 6,466 | 2,256 | -65% | 1 | 1 | 0% | 834 | 1,151 | +38% | 0 | 0 | — |
case-14 | pass→pass | 10,536 | 2,429 | -77% | 1 | 1 | 0% | 1,657 | 1,166 | -30% | 0 | 0 | — |
case-15 | pass→pass | 8,281 | 5,123 | -38% | 1 | 1 | 0% | 1,152 | 1,630 | +41% | 0 | 0 | — |
case-16 | fail→pass | 12,204 | 3,871 | -68% | 1 | 1 | 0% | 1,567 | 1,433 | -9% | 0 | 0 | — |
case-17 | fail→pass | 9,691 | 4,027 | -58% | 1 | 1 | 0% | 1,640 | 1,498 | -9% | 0 | 0 | — |
case-18 | pass→fail | 8,914 | 5,917 | -34% | 1 | 1 | 0% | 1,409 | 1,693 | +20% | 0 | 0 | — |
case-19 | pass→pass | 8,741 | 5,915 | -32% | 1 | 1 | 0% | 1,306 | 1,737 | +33% | 0 | 0 | — |
case-20 | fail→pass | 3,799 | 2,393 | -37% | 1 | 1 | 0% | 515 | 1,178 | +129% | 0 | 0 | — |
case-21 | pass→pass | 8,537 | 2,695 | -68% | 1 | 1 | 0% | 1,099 | 1,203 | +9% | 0 | 0 | — |
case-22 | fail→pass | 9,966 | 2,615 | -74% | 1 | 1 | 0% | 1,455 | 1,310 | -10% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +32 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.