Template Skill 完整参考

✅ pdf-editor
✅ bigquery-analytics
✅ brand-guidelines

❌ PDF_Editor
❌ pdfEditor
❌ pdf editor

[做什么]. [何时使用]: (1) [场景1], (2) [场景2], (3) [场景3].
[可选: 不做什么].

---

### 示例 2：领域知识型 Skill

```markdown
---
name: bigquery-sales
description: >
  Query and analyze sales data in BigQuery including opportunities, 
  pipeline metrics, and conversion rates. Use when analyzing sales 
  performance, forecasting, or generating sales reports.
---

# BigQuery Sales Analytics

## Quick Reference

### Table Schema

Main tables:
- `sales.opportunities` - Sales opportunities and deals
- `sales.pipeline` - Pipeline stages and transitions
- `sales.conversions` - Conversion events and rates

Detailed schema: [SCHEMA.md](references/SCHEMA.md)

## Common Queries

### Today's Closed Deals

```sql
SELECT 
  opportunity_id,
  deal_value,
  closed_by
FROM sales.opportunities
WHERE 
  status = 'closed_won'
  AND DATE(closed_date) = CURRENT_DATE()

---

### 示例 3：工作流程型 Skill

```markdown
---
name: data-pipeline
description: >
  Execute ETL data pipeline with extract, clean, transform, validate, 
  and load stages. Use when processing raw data into analytics-ready format.
---

# Data Pipeline

## Pipeline Stages

The pipeline consists of 5 sequential stages:

1. **Extract** - Fetch data from sources
2. **Clean** - Remove duplicates and handle nulls
3. **Transform** - Apply business logic
4. **Validate** - Check data quality
5. **Load** - Write to destination

## Running the Pipeline

### Full Pipeline

```bash
python scripts/run_pipeline.py --config pipeline.yaml

.pipeline/
├── extract.done
├── clean.done
├── transform.done
├── validate.done
└── load.done

---

## 目录结构模板

### 基础结构

**何时使用**：
- 纯文本指令
- 无需脚本或资源
- 简单工作流程

**示例**：撰写内部沟通、设计指南

---

### 标准结构

**何时使用**：
- 需要可执行脚本
- 需要参考文档
- 需要模板或资产

**示例**：PDF 编辑、前端构建、品牌应用

---

### 复杂结构

**何时使用**：
- 支持多个变体/框架
- 复杂的目录结构
- 共享资源 + 变体特定资源

**示例**：云部署、多框架前端构建

---

## 资源文件模板

### scripts/ 模板

#### Python 脚本模板

```python
#!/usr/bin/env python3
"""
[脚本功能简短描述]

Usage:
    script_name.py <arg1> <arg2> [options]

Examples:
    script_name.py input.txt output.txt
    script_name.py input.txt output.txt --verbose
"""

import argparse
import sys
import os

def main():
    """主函数"""
    parser = argparse.ArgumentParser(
        description='[详细描述]',
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog='''
Examples:
  %(prog)s input.txt output.txt
  %(prog)s input.txt output.txt -v
        '''
    )
    
    # 必需参数
    parser.add_argument('input', help='输入文件路径')
    parser.add_argument('output', help='输出文件路径')
    
    # 可选参数
    parser.add_argument('-v', '--verbose', action='store_true',
                       help='显示详细输出')
    parser.add_argument('--format', choices=['json', 'csv', 'xml'],
                       default='json', help='输出格式')
    
    args = parser.parse_args()
    
    # 验证输入
    if not os.path.exists(args.input):
        print(f"✗ 错误: 文件不存在: {args.input}", file=sys.stderr)
        sys.exit(1)
    
    # 执行主逻辑
    try:
        process(args.input, args.output, args.format, args.verbose)
        print(f"✓ 成功: 已保存到 {args.output}")
    except Exception as e:
        print(f"✗ 错误: {str(e)}", file=sys.stderr)
        sys.exit(1)

def process(input_path, output_path, format, verbose):
    """核心处理逻辑"""
    if verbose:
        print(f"处理 {input_path}...")
    
    # TODO: 实现处理逻辑
    
    if verbose:
        print("处理完成")

if __name__ == '__main__':
    main()

X-RateLimit-Limit: 100
X-RateLimit-Remaining: 87
X-RateLimit-Reset: 1642248600

---

## 常见场景模板

### 场景 1：简单工具 Skill

**特点**：
- 单一功能
- 无复杂依赖
- 直接操作

**模板**：

**SKILL.md**：

```markdown
---
name: simple-tool
description: [工具功能]. Use when [场景].
---

# Simple Tool

## Usage

```bash
python scripts/tool.py <input> <output>

---

### 场景 2：多步骤工作流 Skill

**特点**：
- 顺序执行
- 状态管理
- 错误恢复

**模板**：

**SKILL.md**：

```markdown
---
name: workflow
description: [工作流程描述]. Use for [场景].
---

# Workflow

## Stages

1. Stage 1 - [说明]
2. Stage 2 - [说明]
3. Stage 3 - [说明]

## Running

Full workflow:
```bash
python scripts/run_pipeline.py

---

## 验证和打包

### 打包前检查

```bash
# 1. 验证 YAML 语法
python -c "import yaml; yaml.safe_load(open('my-skill/SKILL.md').read().split('---')[1])"

# 2. 检查文件存在
ls my-skill/SKILL.md
ls my-skill/scripts/*.py
ls my-skill/references/*.md

# 3. 测试脚本
python my-skill/scripts/main_script.py --help

# 4. 打包
scripts/package_skill.py my-skill/