Install any skill in seconds. Free to start, no credit card required.
Get Started Free →生成面向敏捷开发团队的精炼设计文档。MVP导向,避免过度设计。使用场景:(1) 需要为新功能或系统模块生成设计文档 (2) 需要明确功能边界和交互流程 (3) 需要提供实现思路和关键方法 (4) 需要阐述技术难点和解决方案。该skill会先分析项目技术栈和现有组件,然后生成精炼、重点突出的设计文档。
.claude/skills/aiskillstore-agile-design-doc/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-18 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-06 | ✓→✗ | ▼ Worse | 53% | 0% |
| case-05 | ✓→✓ | = Same ✓ | 45% | 0% |
| case-15 | ✓→✓ | = Same ✓ | 45% | 0% |
在生成设计文档之前,必须先完成以下分析:
询问用户或分析项目:
- 项目背景?现在是什么系统,要做什么功能/模块?必须读取的配置文件:
- Python项目:pyproject.toml, requirements.txt, setup.py
- Node.js项目:package.json, yarn.lock, pnpm-lock.yaml
- Java项目:pom.xml, build.gradle
- Go项目:go.mod, go.sum
- 其他:根据项目类型识别
分析内容:
- 主要框架和库
- 数据库和存储
- 消息队列和中间件
- 部署和运维工具通过以下方式了解现有组件:
1. 询问用户:有哪些现有组件可以复用?
2. 读取项目结构:分析src/、lib/、components/等目录
3. 查看文档:README.md、docs/等
记录可复用的组件:
- 基础服务类
- 工具函数
- 中间件
- 数据模型按照以下结构生成设计文档:
明确说明:
- 本次设计要解决什么问题
- 提供什么功能简要列出本次设计的功能点:
- 功能1:一句话描述
- 功能2:一句话描述
- 功能3:一句话描述
不要展开详细说明,保持简洁为每个主要功能绘制时序图:sequenceDiagram participant User participant API participant Service participant DB
User->>API: 请求 API->>Service: 调用 Service->>DB: 查询 DB-->>Service: 返回 Service-->>API: 结果 API-->>User: 响应
时序图目的:
- 展示组件间的交互顺序
- 明确系统边界
- 识别外部依赖针对每个功能点,按以下结构描述:
功能点名称
实现思路(2-3句话)
关键方法(代码示例)
python# 示例:用户认证 def authenticate_user(token: str) -> User: """验证用户token并返回用户信息""" # 1. 验证token格式 # 2. 从缓存或数据库查询 # 3. 返回用户信息 pass
技术难点(如有)
仅列出新增或修改的数据模型:
- User: {id, name, email}
- Order: {id, userId, amount, status}
使用简洁的表格或JSON格式仅列出新增或修改的API:
POST /api/users
- 请求:{name, email}
- 响应:{id, name, email, createdAt}
保持简洁,不要展开所有字段生成文档后,进行以下检查:
优先级顺序:
1. 读取项目文件(pyproject.toml、package.json等)
2. 分析项目结构
3. 询问用户具体问题
4. 基于常见模式做合理假设(并在文档中说明)原则:
- 如果是核心功能,必须包含
- 如果是辅助功能,可以后续迭代
- 如果不确定,询问用户
- 在文档中明确标注"可选"或"后续迭代"处理方式:
1. 读取项目配置文件了解技术栈
2. 搜索相关文档和最佳实践
3. 参考项目现有代码的实现方式
4. 在文档中说明技术选型的理由| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | pass→fail | 19,823 | 19,175 | -3% | 1 | 1 | 0% | 2,698 | 4,122 | +53% | 0 | 0 | — |
case-16 | fail→fail | 37,341 | 37,334 | -0% | 1 | 1 | 0% | 6,350 | 7,575 | +19% | 0 | 0 | — |
case-01 | fail→fail | 21,343 | 24,808 | +16% | 1 | 1 | 0% | 3,784 | 4,737 | +25% | 0 | 0 | — |
case-02 | fail→fail | 23,423 | 21,877 | -7% | 1 | 1 | 0% | 4,077 | 5,339 | +31% | 0 | 0 | — |
case-03 | fail→fail | 29,201 | 15,911 | -46% | 1 | 1 | 0% | 4,257 | 4,809 | +13% | 0 | 0 | — |
case-04 | fail→fail | 31,014 | 38,278 | +23% | 1 | 1 | 0% | 6,349 | 9,403 | +48% | 0 | 0 | — |
case-05 | pass→pass | 21,264 | 18,916 | -11% | 1 | 1 | 0% | 2,433 | 3,532 | +45% | 0 | 0 | — |
case-07 | fail→fail | 32,214 | 24,448 | -24% | 1 | 1 | 0% | 4,846 | 4,900 | +1% | 0 | 0 | — |
case-08 | fail→fail | 30,831 | 23,906 | -22% | 1 | 1 | 0% | 4,866 | 4,534 | -7% | 0 | 0 | — |
case-09 | fail→fail | 31,456 | 23,231 | -26% | 1 | 1 | 0% | 4,893 | 4,826 | -1% | 0 | 0 | — |
case-10 | fail→fail | 47,242 | 40,891 | -13% | 1 | 1 | 0% | 7,470 | 7,123 | -5% | 0 | 0 | — |
case-21 | fail→pass | 31,265 | 21,549 | -31% | 1 | 1 | 0% | 3,913 | 4,998 | +28% | 0 | 0 | — |
case-11 | fail→fail | 28,637 | 21,424 | -25% | 1 | 1 | 0% | 4,365 | 4,530 | +4% | 0 | 0 | — |
case-12 | fail→fail | 29,058 | 19,847 | -32% | 1 | 1 | 0% | 4,238 | 4,206 | -1% | 0 | 0 | — |
case-13 | fail→fail | 44,902 | 27,887 | -38% | 1 | 1 | 0% | 3,717 | 5,641 | +52% | 0 | 0 | — |
case-14 | fail→fail | 35,087 | 31,786 | -9% | 1 | 1 | 0% | 4,920 | 5,116 | +4% | 0 | 0 | — |
case-15 | pass→pass | 26,851 | 20,565 | -23% | 1 | 1 | 0% | 3,931 | 5,688 | +45% | 0 | 0 | — |
case-17 | fail→fail | 35,621 | 28,536 | -20% | 1 | 1 | 0% | 4,961 | 4,277 | -14% | 0 | 0 | — |
case-18 | fail→pass | 41,571 | 22,177 | -47% | 1 | 1 | 0% | 5,512 | 4,214 | -24% | 0 | 0 | — |
case-19 | fail→fail | 22,810 | 22,294 | -2% | 1 | 1 | 0% | 3,080 | 4,485 | +46% | 0 | 0 | — |
case-20 | fail→fail | 25,218 | 27,414 | +9% | 1 | 1 | 0% | 3,686 | 5,478 | +49% | 0 | 0 | — |
case-22 | pass→pass | 22,807 | 15,485 | -32% | 1 | 1 | 0% | 2,629 | 3,234 | +23% | 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 +5 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.