---
description: Claude Office Skills project coding standards and conventions
globs: ["**/*.md", "**/*.ts", "**/*.js", "**/*.json", "**/*.py"]
alwaysApply: true
---

# Claude Office Skills - Project Rules

## Language Requirements

### Code and Documentation Language: English Only

All project artifacts MUST be written in **English**:

1. **SKILL.md files** - All skill names, descriptions, instructions, and examples must be in English
2. **Code comments** - All comments in TypeScript, JavaScript, Python must be in English
3. **Variable/function names** - Use descriptive English names (camelCase for JS/TS, snake_case for Python)
4. **README and documentation** - All markdown documentation in English
5. **Git commit messages** - Write commits in English
6. **JSON/YAML keys** - Use English keys and values

### Example - Correct SKILL.md Header

```yaml
---
name: contract-review
description: "Analyze contracts for risks, check completeness, and provide recommendations."
category: legal
tags:
  - contract
  - review
  - risk-analysis
---
```

### Example - Incorrect (Do NOT do this)

```yaml
---
name: 合同审查  # ❌ Chinese name
description: "分析合同风险"  # ❌ Chinese description
---
```

## Skill Design Standards

### Frontmatter Structure (YAML)

Every SKILL.md must include:

```yaml
---
# Basic Information (Required)
name: skill-name-in-kebab-case
description: "Clear English description of what this skill does"
version: "1.0.0"
author: claude-office-skills
license: MIT

# Categorization (Required)
category: legal|hr|finance|sales|marketing|productivity|pdf|document|spreadsheet|presentation|workflow|research|visualization
tags:
  - relevant
  - english
  - tags
department: Target Department

# AI Model Compatibility
models:
  recommended:
    - claude-sonnet-4
    - claude-opus-4
  compatible:
    - claude-3-5-sonnet
    - gpt-4

# MCP Tools Integration (if applicable)
mcp:
  server: office-mcp
  tools:
    - tool_name_1
    - tool_name_2

# Capabilities
capabilities:
  - capability_1
  - capability_2

# Language Support
languages:
  - en
  - zh
---
```

### Content Structure

Skills should follow this structure:

1. **Overview** - What the skill does, capabilities and limitations
2. **How to Use** - Step-by-step instructions
3. **Domain Knowledge** - Embedded expertise, patterns, best practices
4. **Output Format** - Expected output structure
5. **Examples** - Real-world usage examples
6. **Tips** - Best practices for better results
7. **Limitations** - Clear boundaries

## File Naming Conventions

- Skill directories: `kebab-case` (e.g., `contract-review`, `stock-analysis`)
- TypeScript/JavaScript files: `camelCase.ts` or `kebab-case.ts`
- Python files: `snake_case.py`

## MCP Server Standards

### Tool Naming

- Use `snake_case` for tool names: `extract_text_from_pdf`, `create_docx`
- Be descriptive: `analyze_spreadsheet` not `analyze`

### Tool Description

- Write clear English descriptions
- Include parameter types and examples
- Document return values

## Git Workflow

- Branch naming: `feature/skill-name` or `fix/issue-description`
- Commit messages: Use conventional commits in English
  - `feat: add stock-analysis skill`
  - `fix: correct calculation in dcf-valuation`
  - `docs: update README with new skills`