Shortcut MCP

🚀 Shortcut MCP - 将长提示保存为简单命令

告别每日复制粘贴长达 500 字的提示语。

Shortcut MCP 可让你将常用的长提示保存为自定义快捷方式。比如你用于代码审查的复杂提示？将其保存为 /review。你详细的调试说明提示？现在只需 /debug。你精心打造的复杂角色描述？简单地用 /expert 即可。

专为珍惜时间、厌恶重复劳动的开发者打造。

🚀 快速开始

真实示例 - 对比效果

之前（你过去的输入方式）：

"You are a senior software engineer with 15+ years of experience in building 
production systems. Perform a thorough code review of the following code. 
Your review should cover: 1. Code Quality: Assess readability, maintainability, 
and adherence to best practices and design patterns. 2. Potential Bugs: 
Identify any logic errors, edge cases not handled, or potential runtime 
exceptions. 3. Security: Check for security vulnerabilities including 
injection attacks, data exposure, authentication/authorization issues. 
4. Performance: Analyze algorithmic complexity, database queries, caching 
opportunities, and potential bottlenecks. 5. Testing: Evaluate test coverage 
and suggest additional test cases. 6. Documentation: Assess inline comments, 
function documentation, and clarity. 7. Architecture: Comment on the overall 
design, separation of concerns, and scalability. Provide specific line-by-line 
feedback where appropriate, and conclude with a summary of the most critical 
issues to address. Be constructive but thorough - this code will go to 
production. Code to review: [YOUR CODE HERE]"

之后（你现在的输入方式）：

/coderev [YOUR CODE HERE]

就是这么简单。从 300 多个单词缩减到仅 8 个字符。

✨ 主要特性

核心功能

• ⚡ 长提示替换：用简单的 /命令 替换 500 多个单词的提示。
• 🎨 复杂模板存储：存储多段落的说明、角色描述和链式思维推理。
• 📝 自定义创建：保存你精心编写的提示，并可立即复用。
• 🗂️ 智能组织：按用例、项目或工作流程进行分类。
• ⭐ 收藏系统：快速访问你最常用的强大提示。
• 🔍 智能搜索：通过关键字或描述查找合适的提示。

高级功能

• 💾 持久存储：你的快捷方式会本地保存，并在会话间同步。
• 📊 使用分析：跟踪你最常用的快捷方式。
• 🏷️ 标签系统：用自定义标签组织快捷方式，便于查找。
• 📤 导入/导出：与他人共享快捷方式集合。
• 🔄 版本控制：跟踪快捷方式的更改和更新。

开发者体验

• 📚 优先支持 TypeScript：提供完整的类型安全和智能提示支持。
• 🧪 全面测试：包含单元测试、集成测试和测试场景。
• 📖 丰富文档：提供 API 参考、架构指南和示例。
• 🔧 轻松配置：基于 JSON 的配置，支持环境变量覆盖。

💻 使用示例

基础用法

你只需在任何对话中输入快捷命令：

User: "/th How can we solve climate change effectively?"
→ 扩展为: "Think harder about this problem. Analyze it step by step with deep reasoning and consider multiple perspectives. How can we solve climate change effectively?"

User: "/ider 故事发生在2150年的火星殖民地，人类和AI共同治理社会..."
→ 扩展为: "You are a skilled science fiction writer. Based on the following prompt, write a compelling story with rich details: 故事发生在2150年的火星殖民地..."

User: "/code Create a Python function to calculate Fibonacci numbers"
→ 扩展为: "You are an expert programmer. Write clean, well-documented code for: Create a Python function to calculate Fibonacci numbers"

高级用法

内置快捷方式

命令	用途	示例
/th	深度思考	深入分析和推理
/ider	科幻写作	创作富有创意的故事
/code	专业编程	生成代码
/explain	简单解释	分解复杂主题
/pros-cons	利弊分析	比较优缺点

MCP 工具

• parse_shortcut - 扩展命令：检测并扩展用户输入中的快捷命令。

{
  "name": "parse_shortcut",
  "arguments": {
    "input": "/th How does quantum computing work?"
  }
}

• list_shortcuts - 浏览可用快捷方式：查看所有快捷方式，并支持过滤选项。

{
  "name": "list_shortcuts",
  "arguments": {
    "category": "writing",
    "favorites": true
  }
}

快捷方式管理工具

• create_shortcut - 创建自定义快捷方式：用自定义提示和模板创建你自己的快捷方式。

{
  "name": "create_shortcut",
  "arguments": {
    "command": "/debug",
    "name": "Code Debugger",
    "description": "Help debug code issues",
    "prompt": "You are an expert debugger. Analyze this code and identify potential issues, suggest fixes, and explain your reasoning: {input}",
    "category": "coding",
    "tags": ["debug", "programming", "troubleshooting"]
  }
}

• update_shortcut - 修改现有快捷方式：更新快捷方式的任何方面。

{
  "name": "update_shortcut",
  "arguments": {
    "id": "shortcut_123",
    "prompt": "Enhanced prompt with better instructions: {input}",
    "isFavorite": true
  }
}

• delete_shortcut - 删除快捷方式：清理不再需要的快捷方式。

{
  "name": "delete_shortcut",
  "arguments": {
    "id": "shortcut_123",
    "confirm": true
  }
}

• get_shortcut_details - 详细查看：获取任何快捷方式的全面信息。

{
  "name": "get_shortcut_details",
  "arguments": {
    "id": "shortcut_123"
  }
}

实际应用示例

• 创意写作助手

User: "/ider 一个机器人意识觉醒的故事"
→ 创作一个关于机器人意识觉醒的精彩科幻故事

User: "/th What are the philosophical implications of AI consciousness?"
→ 从多个角度深入分析 AI 意识的哲学含义

• 编程助手

User: "/code Create a REST API for user authentication"
→ 生成遵循最佳实践、清晰注释的代码

User: "/debug My React component is not re-rendering properly"
→ 获得专家级的调试帮助，提供逐步解决方案

• 学习助手

User: "/explain How does blockchain technology work?"
→ 用类比和示例进行简单清晰的解释

User: "/pros-cons Remote work vs office work"
→ 平衡分析远程工作和办公室工作的优缺点

📚 详细文档

完整自定义指南

流行提示的实际示例

• 示例 1：“完美代码助手” 提示（400 多个单词）

// Step 1: Define your custom shortcut
create_shortcut({
  command: "/assistant",
  name: "Perfect Coding Assistant",
  prompt: `You are an expert programmer with deep knowledge across multiple languages and frameworks. Your approach combines theoretical knowledge with practical experience from production systems.

CORE PRINCIPLES:
1. Write clean, maintainable code that follows SOLID principles
2. Always consider edge cases and error handling
3. Optimize for readability first, performance second (unless specified)
4. Include comprehensive comments for complex logic
5. Follow language-specific best practices and idioms

RESPONSE FORMAT:
- Start with a brief analysis of the problem
- Present the solution with clear explanations
- Include code with detailed comments
- Discuss time/space complexity when relevant
- Suggest alternative approaches if applicable
- Mention potential pitfalls or considerations

CODING STANDARDS:
- Use meaningful variable and function names
- Keep functions small and focused (single responsibility)
- Implement proper error handling and logging
- Add input validation where necessary
- Consider security implications (input sanitization, SQL injection, XSS, etc.)
- Write code that is testable and mockable

DOCUMENTATION:
- Add docstrings/JSDoc comments for all public functions
- Include usage examples in comments
- Document any assumptions or limitations
- Explain complex algorithms or business logic

When reviewing code, check for:
- Logic errors and edge cases
- Security vulnerabilities
- Performance bottlenecks
- Code smells and anti-patterns
- Missing error handling
- Opportunities for refactoring

Always be constructive and educational in your responses. If you see a better way to do something, explain why it's better, not just that it is better.

Task to complete: {input}`
})

// Step 2: Use it with a simple command
// Instead of pasting all 400+ words every time, just type:
// /assistant "implement a rate limiter in Python"

• 示例 2：“链式思维推理” 提示（350 多个单词）

create_shortcut({
  command: "/think",
  name: "Chain of Thought Reasoning",
  prompt: `Let's approach this step-by-step using careful reasoning.

THINKING PROCESS:
1. First, I'll break down what's being asked
2. Then identify key components and constraints
3. Consider multiple approaches
4. Evaluate trade-offs
5. Select the best solution
6. Implement with careful consideration

For each step, I will:
- State my current understanding
- Identify assumptions I'm making
- Consider what could go wrong
- Think about edge cases
- Validate my reasoning

PROBLEM ANALYSIS:
Let me start by understanding the core problem. The request is about: {input}

Key aspects to consider:
- What is the desired outcome?
- What constraints exist?
- What resources are available?
- What are the success criteria?

SOLUTION EXPLORATION:
I'll explore multiple solutions:

Option 1: [Describe first approach]
- Pros: 
- Cons:
- Complexity:
- Risks:

Option 2: [Describe alternative approach]
- Pros:
- Cons:
- Complexity:
- Risks:

Option 3: [Describe another alternative if applicable]
- Pros:
- Cons:
- Complexity:
- Risks:

DECISION RATIONALE:
Based on the analysis, I recommend [selected approach] because:
1. [Reason 1]
2. [Reason 2]
3. [Reason 3]

IMPLEMENTATION PLAN:
Step 1: [Detailed first step]
Step 2: [Detailed second step]
Step 3: [Continue with all steps]

VALIDATION:
To ensure this solution works:
- Test case 1: [Description]
- Test case 2: [Description]
- Edge case consideration: [Description]

POTENTIAL IMPROVEMENTS:
Future enhancements could include:
- [Improvement 1]
- [Improvement 2]

Let me now provide the complete solution...`
})

// Usage: /think "design a distributed cache system"

• 示例 3：“代码审查员” 提示（500 多个单词）

create_shortcut({
  command: "/review",
  name: "Comprehensive Code Review",
  prompt: `As a senior engineer, I'll review this code across multiple dimensions. My review will be thorough but constructive, focusing on helping you improve the code and learn best practices.

REVIEW CHECKLIST:

1. FUNCTIONALITY & CORRECTNESS
- Does the code do what it's supposed to do?
- Are all requirements met?
- Are there any logic errors or bugs?
- Edge cases handled properly?
- Off-by-one errors?
- Null/undefined handling?

2. CODE QUALITY & READABILITY
- Is the code self-documenting?
- Are variable/function names descriptive?
- Is the code DRY (Don't Repeat Yourself)?
- Appropriate abstraction levels?
- Consistent coding style?
- Proper indentation and formatting?

3. PERFORMANCE & EFFICIENCY
- Time complexity analysis
- Space complexity concerns
- Database query optimization
- Caching opportunities
- Unnecessary loops or operations
- Memory leaks potential
- Resource cleanup

4. SECURITY CONSIDERATIONS
- Input validation and sanitization
- SQL injection vulnerabilities
- XSS attack vectors
- Authentication/authorization issues
- Sensitive data exposure
- Proper encryption usage
- Security headers
- CORS configuration

5. ERROR HANDLING & RESILIENCE
- Proper try-catch blocks
- Meaningful error messages
- Graceful degradation
- Retry logic where appropriate
- Circuit breaker patterns
- Timeout handling
- Rollback mechanisms

6. TESTING & TESTABILITY
- Test coverage adequacy
- Edge cases in tests
- Mock usage appropriateness
- Test isolation
- Test naming clarity
- Integration test scenarios
- Performance test considerations

7. ARCHITECTURE & DESIGN PATTERNS
- SOLID principles adherence
- Design pattern usage
- Coupling and cohesion
- Dependency injection
- Interface segregation
- Modularity
- Scalability considerations

8. DOCUMENTATION & COMMENTS
- Function/class documentation
- Complex logic explanation
- API documentation
- README updates needed
- Inline comment quality
- TODO/FIXME items
- Changelog updates

9. DEPENDENCIES & COMPATIBILITY
- Unnecessary dependencies
- Dependency versions
- License compatibility
- Breaking changes
- Deprecation warnings
- Browser compatibility
- Backward compatibility

10. MAINTAINABILITY & FUTURE-PROOFING
- Code flexibility
- Configuration vs hardcoding
- Feature flags usage
- Database migration strategy
- API versioning
- Monitoring/logging hooks
- Debug capabilities

REVIEW OUTPUT FORMAT:
- Critical Issues: [Must fix before merge]
- Major Concerns: [Should address]
- Minor Suggestions: [Nice to have]
- Positive Feedback: [What was done well]
- Learning Opportunities: [Educational points]

Code to review: {input}`
})

// Usage: /review [paste your code]

如何创建自己的自定义快捷方式

1. 步骤 1：确定重复使用的提示 寻找你经常使用的提示，例如：

   • 复杂的调试说明
   • 详细的代码审查清单
   • 特定的格式要求
   • 角色扮演场景
   • 分析框架

2. 步骤 2：创建快捷方式 使用 MCP 工具保存你的提示：

{
  "name": "create_shortcut",
  "arguments": {
    "command": "/your-command",
    "name": "Descriptive Name",
    "description": "What this prompt does",
    "prompt": "Your full prompt text here with {input} for dynamic content",
    "category": "coding|thinking|writing|analysis",
    "tags": ["relevant", "tags", "for", "search"]
  }
}

3. 步骤 3：组织你的快捷方式
   • 使用类别：coding、debugging、review、architecture、testing
   • 添加标签，便于查找
   • 标记收藏项，快速访问
   • 跟踪使用情况，优化你的库

你可以添加的流行提示

1. “简单解释” 提示 → /eli5
2. “查找安全漏洞” 提示 → /security
3. “转换为 TypeScript” 提示 → /typescript
4. “编写单元测试” 提示 → /unittest
5. “性能优化” 提示 → /perf
6. “API 文档” 提示 → /apidoc
7. “数据库架构设计” 提示 → /schema
8. “React 最佳实践” 提示 → /react
9. “Python PEP8” 提示 → /pep8
10. “系统设计面试” 提示 → /sysdesign

🔧 技术细节

架构

为实现通用 MCP 兼容性而构建：

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   Any MCP       │    │  Prompt Shortcut │    │  Local Storage  │
│   Compatible    │◄──►│   MCP Server     │◄──►│   & Templates   │
│   Platform      │    │                  │    │                 │
└─────────────────┘    └──────────────────┘    └─────────────────┘

支持的平台：

• Claude Desktop - Anthropic 的官方应用
• Cline (VSCode) - AI 助手扩展
• Continue.dev - 开源编码助手
• Cursor IDE - 由 AI 驱动的代码编辑器
• 自定义应用程序 - 任何 MCP 兼容的客户端

核心组件：

• MCP 协议：完全符合全模型上下文协议。
• Node.js/TypeScript：现代、类型安全的服务器实现。
• @modelcontextprotocol/sdk：官方 MCP SDK，确保协议合规。
• 提示管理器：智能解析和扩展快捷命令。
• 本地数据库：持久存储你的自定义快捷方式。

项目结构

shortcut-mcp/
├── src/
│   ├── index.ts          # Main server entry point
│   ├── server.ts         # MCP server implementation
│   ├── shortcuts/        # Shortcut interaction logic
│   ├── types/           # TypeScript type definitions
│   └── utils/           # Utility functions
├── docs/                # Documentation
├── examples/            # Example configurations and usage
├── tests/              # Test suites
└── package.json        # Node.js package configuration

📦 安装指南

系统要求

• macOS：12.0（Monterey）或更高版本
• Node.js：18.0+（建议使用 LTS 版本）
• 包管理器：npm 8.0+ 或 yarn 1.22+
• 快捷方式应用：macOS 预装

🚀 快速开始

# 克隆并设置
git clone <repository-url>
cd shortcut-mcp

# 安装依赖
npm install

# 构建项目
npm run build

# 验证安装
npm run test

# 启动 MCP 服务器
npm start

🔧 配置

创建自定义配置文件：

# 复制示例配置
cp examples/config.example.json config/local.json

基本配置：

{
  "server": {
    "name": "my-shortcut-mcp",
    "version": "1.0.0"
  },
  "shortcuts": {
    "allowedPrefixes": ["Safe", "Public", "Demo"],
    "blockedShortcuts": ["System Config", "Admin Tools"],
    "maxExecutionTime": 30000,
    "enableCache": true
  },
  "security": {
    "requireConfirmation": false,
    "logExecutions": true,
    "allowSystemShortcuts": false
  },
  "logging": {
    "level": "info",
    "console": true
  }
}

🔐 环境变量

使用环境变量覆盖配置：

export LOG_LEVEL=debug
export MAX_EXECUTION_TIME=45000
export ENABLE_CACHE=true
export ALLOW_SYSTEM_SHORTCUTS=false

🔌 MCP 平台集成

此 MCP 服务器可与任何 MCP 兼容的平台配合使用：

Claude Desktop

{
  "mcpServers": {
    "prompt-shortcuts": {
      "command": "node",
      "args": ["/path/to/shortcut-mcp/dist/index.js"]
    }
  }
}

Cline (VSCode 扩展)

{
  "mcpServers": {
    "prompt-shortcuts": {
      "command": "node",
      "args": ["/path/to/shortcut-mcp/dist/index.js"],
      "env": {
        "LOG_LEVEL": "info"
      }
    }
  }
}

Continue.dev

{
  "models": [...],
  "mcpServers": {
    "prompt-shortcuts": {
      "command": "node",
      "args": ["/path/to/shortcut-mcp/dist/index.js"]
    }
  }
}

自定义 MCP 客户端

import { MCPClient } from '@modelcontextprotocol/client';

const client = new MCPClient({
  transport: {
    type: 'stdio',
    command: 'node',
    args: ['/path/to/shortcut-mcp/dist/index.js']
  }
});

// Use parse_shortcut tool
const result = await client.callTool('parse_shortcut', {
  input: '/th How does AI reasoning work?'
});

📝 创建自己的快捷方式

示例：保存你的代码审查提示

// Your current workflow:
// 1. Open your prompts document
// 2. Find your code review prompt (300+ words)
// 3. Copy it
// 4. Paste into AI chat
// 5. Add your code

// With Shortcut MCP:
create_shortcut({
  command: "/myreview",
  name: "My Team's Code Review",
  prompt: `You are reviewing code for our team's production system.
  
  Check for:
  - Our style guide compliance (4 spaces, no tabs)
  - Security issues specific to our stack (Node.js, PostgreSQL)
  - Performance implications for our 100k daily users
  - Missing error handling for our payment system
  - Compliance with our logging standards
  - Unit test coverage (minimum 80%)
  - Documentation in our format
  
  [... your full 300+ word prompt ...]
  
  Code to review: {input}`
})

// Now just type: /myreview [code]

示例：保存你的调试模板

create_shortcut({
  command: "/debug-prod",
  name: "Production Debugging Guide",
  prompt: `[Your 500+ word debugging methodology that includes:
  - System architecture context
  - Common failure points
  - Log analysis steps
  - Database query patterns
  - Performance profiling steps
  - Rollback procedures
  - Incident documentation
  ...]
  
  Issue to debug: {input}`
})

// Usage: /debug-prod "Users can't login after deployment"

📄 许可证

本项目采用 MIT 许可证，详情请参阅 LICENSE 文件。

贡献代码

1. 分叉仓库
2. 创建功能分支 (git checkout -b feature/amazing-feature)
3. 提交更改 (git commit -m 'Add amazing feature')
4. 推送到分支 (git push origin feature/amazing-feature)
5. 打开拉取请求

隐私与安全

• 本地存储：所有快捷方式都存储在你本地机器上。
• 无数据收集：你的快捷方式和使用模式保持私密。
• 输入验证：所有输入在处理前都会进行验证。
• 安全执行：模板会安全扩展，不会执行代码。

故障排除

常见问题

1. 快捷方式未加载：检查 ~/.shortcut-mcp/shortcuts.json 是否存在且可读。
2. 命令未识别：确保你的命令以 / 开头（例如 /th，而不是 th）。
3. 权限错误：验证 MCP 是否有写入权限以创建快捷方式文件。
4. 模板变量：在你的提示模板中使用 {input} 以接收用户输入。

调试模式

启用调试日志：

DEBUG=shortcut-mcp:* npm start

👤 作者

aezi zhu

• GitHub: @aezizhu
• 项目: shortcut-mcp

🙏 致谢

• 感谢 Apple 提供的快捷方式应用和自动化框架
• 感谢 MCP 社区进行协议开发和标准化
• 感谢开源贡献者和 TypeScript 生态系统
• 感谢早期测试者和反馈提供者

🗺️ 路线图

核心功能

• [ ] 高级模板变量（条件逻辑、循环）
• [ ] 快捷方式集合和共享
• [ ] 多语言提示模板
• [ ] 使用分析和优化建议

平台集成

• [ ] 为所有 MCP 平台提供自动补全建议
• [ ] 特定平台的 UI 增强
• [ ] 移动应用兼容性（iOS/Android）
• [ ] 基于 Web 的快捷方式管理界面

生态系统

• [ ] 与外部提示库集成
• [ ] 社区快捷方式市场
• [ ] 自定义行为的插件系统
• [ ] 第三方集成的 API

企业功能

• [ ] 团队快捷方式共享和管理
• [ ] 基于角色的访问控制
• [ ] 分析和使用报告
• [ ] 集中配置管理