使用说明

MCP开发提示

通过模型上下文协议为开发人员提供精心策划的AI提示

一个轻量级的MCP服务器，为人工智能驱动的开发工作流程提供久经考验的提示。通过结合模块化技能创建专门的代理，或使用专家AI配置文件进行调试、SQL优化等。

🚀 快速开始

安装

git clone https://github.com/LeonNonnast/mcpdevprompts
cd mcpdevprompts
npm install
npm run build

claude mcp add mcpdevprompts --node /path/to/mcpdevprompts/build/server.js

与Claude一起使用


# Use prompts directly
claude prompt debug-andy "My API returns 500 errors randomly"
claude prompt sql-expert "Optimize this slow query"
claude prompt clean-code-clarity-readability "Create a user service with clean code"

快速工作流示例

🔍 发现可用的个人资料和技能

# See all available AI profiles
claude search_profiles
# Returns: debug-andy, sql-expert, performance-kai, lovable-ai-editor-base, etc.

# List all development skills
claude list_skills
# Returns: clean-code-clarity-readability, testing-strategies, error-handling-best-practices, etc.

👥 使用AI配置文件

# Use a specialized debugging expert
claude prompt debug-andy "My React app crashes randomly on mobile devices"

# Get SQL optimization help
claude prompt sql-expert "This query takes 30 seconds, how can I optimize it?"

# Performance analysis
claude prompt performance-kai "My Node.js API response time increased by 200%"

🧠 将技能加载到代理中

# Load clean code skills
claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions"],
  "agent_name": "Clean Code Assistant"
}'

# Load testing expertise
claude load_skills '{
  "skill_ids": ["testing-strategies", "error-handling-best-practices"],
  "agent_name": "Testing Expert"
}'

# Create a full-stack quality agent
claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "testing-strategies"],
  "agent_name": "Full-Stack Expert"
}'

🎯 现实世界场景

场景1：调试会话

# 1. Start with debugging expert
claude prompt debug-andy "API endpoint returns 500 error intermittently"

# 2. Load additional skills for comprehensive solution
claude load_skills '{
  "skill_ids": ["error-handling-best-practices", "testing-strategies"],
  "agent_name": "Debug & Test Expert"
}'

# 3. Now ask for complete solution
"Create robust error handling and tests for this API endpoint"

场景2：代码质量审查

# 1. Load clean code skills
claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "clean-code-commenting"],
  "agent_name": "Code Quality Reviewer"
}'

# 2. Review and improve code
"Review this function and suggest improvements following clean code principles"

场景3：机载新AI代理

# 1. Onboard a specialized editor
claude prompt project-onboarding "Introduce lovable-ai-editor-base for code refactoring tasks"

# 2. Combine with additional skills
claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "testing-strategies"],
  "agent_name": "Enhanced Code Editor"
}'

本地开发测试

# Clone and test locally
git clone https://github.com/mcpdevprompts.git
cd server
npm install
npm run build
npm run inspector  # Opens MCP Inspector for testing

📚 可用提示

👥 AI配置文件

专门任务的专家AI助手：

调试andy：系统调试和根本原因分析
sql专家：高级SQL查询优化和数据库设计
表演凯：性能分析和优化策略
可爱的人工智能编辑库：AI驱动的代码编辑和重构
困惑搜索助手：研究和信息收集
replit专家软件开发人员编辑器：全栈开发指南

🛠️ 技能与技巧

发展技能和方法：

项目入职培训：将专门的人工智能代理引入您的工作流程
清晰的代码可读性：生成清晰、不言自明的代码
简洁的代码小函数：编写专注、单一目的的函数
干净的代码注释：添加有意义的评论和文件
错误处理最佳实践：实施稳健的错误处理
测试策略：创建全面的测试套件

🎯 专用工具

增强功能的内置工具：

搜索_提示：按关键字或类别查找提示
搜索文件：按专业查找AI配置文件
get_prompt_stats：查看提示收集统计信息
get_tool_stats：查看可用工具和用法
list_skills：列出所有可用的开发技能
load_skills：将多种技能加载到代理的知识库中

工具示例：

# Search for profiles
claude search_profiles
# Returns: debug-andy, sql-expert, performance-kai, etc.

# List all available skills
claude list_skills
# Returns: clean-code-clarity-readability, testing-strategies, etc.

# Load multiple skills into an agent
claude load_skills '{"skill_ids": ["clean-code-clarity-readability", "testing-strategies"], "agent_name": "Clean Code Expert"}'
# Returns: Agent loaded with specified skills and ready to use them

# Get collection statistics
claude get_prompt_stats
# Returns: total prompts, categories, effectiveness ratings

💡 用例

面向前端开发人员

# Get clean code generation
claude prompt clean-code-clarity-readability "Create a React component for user profile"

# Small, focused functions
claude prompt clean-code-small-functions "Refactor this large function into smaller parts"

# Add proper documentation
claude prompt clean-code-commenting "Add documentation to this API endpoint"

面向后端开发人员

# Database optimization
claude prompt sql-expert "Optimize this N+1 query problem"

# Performance debugging
claude prompt performance-kai "My Node.js API is slow under load"

# Systematic debugging
claude prompt debug-andy "Random 500 errors in production"

对于团队领导

# Agent onboarding
claude prompt project-onboarding "Introduce lovable-ai-editor profile for code review"

# Error handling standards
claude prompt error-handling-best-practices "Establish error handling guidelines"

# Testing strategies
claude prompt testing-strategies "Create testing plan for new microservice"

用于代码质量和技能管理

# Clean code generation
claude prompt clean-code-clarity-readability "Create a user authentication service"

# Create a specialized agent with multiple skills
claude list_skills  # First see all available skills
claude load_skills '{"skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "testing-strategies"], "agent_name": "Code Quality Expert"}'
# Now the agent has all three skills loaded and ready to use

🎪 技能管理系统

可用技能

我们的技能体系提供模块化开发专业知识，可以组合成专业代理：

整洁代码技能：

clean-code-clarity-readability：生成具有有意义名称的自解释代码
clean-code-small-functions：编写专注、单一职责的职能
clean-code-commenting：添加有意义的评论和文件

发展技能：

testing-strategies：创建全面的测试套件和测试计划
error-handling-best-practices：实施稳健的错误处理模式
project-onboarding：引导专业AI代理进行工作流集成

技能工作流程

1.发现可用技能

claude list_skills

退货：

[
  {
    "id": "clean-code-clarity-readability",
    "title": "Generate Clear and Readable Code",
    "description": "Generate self-explanatory code with meaningful names",
    "tags": ["clean code", "readability", "maintainability"],
    "effectiveness": 5
  },
  {
    "id": "clean-code-small-functions",
    "title": "Generate Small, Single-Responsibility Functions",
    "description": "Write focused, single-purpose functions under 20 lines",
    "tags": ["clean code", "functions", "modularity"],
    "effectiveness": 5
  }
]

2.将技能加载到代理中

claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "testing-strategies"],
  "agent_name": "Clean Code Expert"
}'

退货：

You are now a Clean Code Expert with the following specialized skills:

[Combined skill prompts...]

You have been equipped with these 3 specialized skills:
- Generate Clear and Readable Code: Generate self-explanatory code with meaningful names
- Generate Small, Single-Responsibility Functions: Write focused, single-purpose functions
- Create Comprehensive Test Suites: Design testing strategies and test plans

Please acknowledge that you have integrated these skills and are ready to apply them.

3.使用您的专业代理

加载技能后，您的代理会自动将其应用于相关请求：

# Agent now combines all loaded skills
"Create a user authentication service with clean code and tests"
# → Uses clean code + testing skills together

预先构建的技能组合

🧹 干净代码专家

claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "clean-code-commenting"],
  "agent_name": "Clean Code Expert"
}'

🧪 测试专家

claude load_skills '{
  "skill_ids": ["testing-strategies", "error-handling-best-practices"],
  "agent_name": "Testing Specialist"
}'

🎯 全栈质量代理

claude load_skills '{
  "skill_ids": ["clean-code-clarity-readability", "clean-code-small-functions", "testing-strategies", "error-handling-best-practices"],
  "agent_name": "Full-Stack Quality Agent"
}'

技能管理的好处

🔧 模块化：根据特定需求混合搭配技能
📈 可扩展的：新技能无缝集成
👥 协作的：与团队分享技能组合
🎯 专注的：为特定任务创建高度专业化的代理
💡 智能：技能在情境中协同工作

🏗️ 建筑

mcpdevprompts/
├── src/                    # TypeScript source code
│   ├── server.ts          # Main MCP server
│   ├── prompt-manager.ts  # Prompt loading and management
│   ├── tool-manager.ts    # Tool management system
│   └── utils/
├── public/
│   ├── prompts/
│   │   ├── profiles/      # AI assistant profiles
│   │   ├── skills/        # Development skills & techniques
│   │   └── onboarding/    # Project & team setup
│   ├── tools/             # MCP tools definitions
│   └── schema/            # JSON schemas for validation
├── build/                 # Compiled JavaScript
└── docs/                  # Documentation

🔧 发展

先决条件

Node.js 18.0.0或更高版本
npm或纱线

设置

# Clone the repository
git clone https://github.com/mcpdevprompts/server.git
cd server

# Install dependencies
npm install

# Build the project
npm run build

# Run in development mode
npm run dev

# Test with MCP Inspector
npm run inspector

添加新提示

选择正确的类别:
- profiles/ -AI助理个性
- skills/ -开发技术
- onboarding/ -设置和集成

遵循模式 （参见 public/schema/prompt-schema.json):

{
  "id": "unique-prompt-id",
  "title": "Human-readable title",
  "description": "What this prompt does",
  "category": "profiles|skills|onboarding",
  "tags": ["relevant", "searchable", "tags"],
  "prompt": "Your detailed prompt text here...",
  "examples": [
    {
      "input": "Example input",
      "output": "Expected output description"
    }
  ],
  "effectiveness": 4.5,
  "author": "Your Name",
  "version": "1.0.0",
  "created_at": "2024-01-01T00:00:00Z",
  "updated_at": "2024-01-01T00:00:00Z"
}

彻底测试 具有各种输入
提交拉取请求 描述清晰

添加新工具

工具通过自定义操作扩展了MCP服务器的功能。请按照以下步骤操作：

选择一个有意义的工具名称 （例如。， analyze_code, generate_tests, check_dependencies)

创建工具定义 在……里面 public/tools/your-tool.json:

{
  "id": "analyze_code_complexity",
  "name": "analyze_code_complexity",
  "description": "Analyze code complexity and suggest improvements",
  "input_schema": {
    "type": "object",
    "properties": {
      "code": {
        "type": "string",
        "description": "The code to analyze"
      },
      "language": {
        "type": "string",
        "description": "Programming language (js, ts, py, etc.)"
      },
      "metrics": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Metrics to calculate (cyclomatic, cognitive, etc.)"
      }
    },
    "required": ["code", "language"]
  }
}

执行工具逻辑 在……里面 src/server.ts CallToolRequestSchema处理程序：

case "analyze_code_complexity":
  if (!args || typeof args.code !== "string") {
    throw new McpError(ErrorCode.InvalidRequest, "Code parameter is required");
  }

  const analysis = await this.analyzeCodeComplexity(args.code, args.language);
  return {
    content: [
      {
        type: "text",
        text: JSON.stringify(analysis, null, 2)
      }
    ]
  };

测试工具 配备MCP检查器和各种输入

工具创意：

validate_env_vars：检查环境变量的完整性
generate_tests：为给定的代码创建单元测试
check_dependencies：分析package.json的漏洞
format_sql：格式化和验证SQL查询
analyze_performance：检测性能瓶颈
generate_docs：根据代码注释创建文档

📊 质量标准

所有提示必须符合以下标准：

有效性：社区测试的平均评分为4.0+
清晰度：清晰、可操作的指示
完整性：任务的全面覆盖
最佳实践：遵循当前行业标准
无障碍：在适用的情况下考虑无障碍要求

🤝 贡献

我们欢迎捐款！请查看我们的贡献指南了解详情。

贡献类型

新提示：为常见开发任务添加高质量提示
迅速改进：根据用户反馈增强现有提示
文档：改进设置指南和使用示例
漏洞修补：报告并修复服务器问题
工具：添加新的MCP工具以增强功能

🔒 安全

提示中没有敏感数据
所有用户输入的输入验证
API端点的速率限制
定期安全审计
所有提示和工具的模式验证

🎯 路线图

[x] 第一阶段：带有核心提示的TypeScript MCP服务器
[x] 第2阶段：人工智能配置文件和专用工具
[x] 第三期：具有可组合专业知识的技能管理系统
[x] 阶段4：社区捐款和评级系统
[x] 阶段5：IDE集成和高级分析
[x] 第6阶段：自定义提示集合和企业功能

📝 许可证

MIT许可证-请参阅许可证了解详情。

🙏 致谢

灵感来自 Context7 MCP服务器
建立在模型上下文协议
社区快速贡献者

📞 支持

由以下材料制成❤️ 由MCP DevPrompts社区发布