Skip to main content
Glama
yuyao1999

RT-Prompt-MCP

by yuyao1999
README.md
<!-- @format -->

[![smithery badge](https://smithery.ai/badge/@yuyao1999/rt-prompt-mcp-server)](https://smithery.ai/server/@yuyao1999/rt-prompt-mcp-server)

# RT-Prompt-MCP

RT-Prompt-MCP 是一个基于 Model Context Protocol (MCP) 的服务器,专注于提供开发和设计相关的提示词补充建议。

## 功能特点

- 提供特定领域的提示词补充,帮助 LLM 生成更符合要求的内容
- 支持后端开发、前端开发和通用场景的提示词
- 易于集成到支持 MCP 协议的客户端
- 使用 TypeScript 开发,类型安全

## 安装

### Installing via Smithery

To install rt-prompt-mcp-server for Claude Desktop automatically via [Smithery](https://smithery.ai/server/@yuyao1999/rt-prompt-mcp-server):

```bash
npx -y @smithery/cli install @yuyao1999/rt-prompt-mcp-server --client claude
```

### Manual Installation
全局安装:

```bash
npm install -g rt-prompt-mcp
```

## 使用方法

### 作为命令行工具运行

安装后,直接运行:

```bash
rt-prompt-mcp
```

### 作为 MCP Server 与 MCP 客户端集成

在支持 MCP 协议的应用中(如 Claude Desktop)配置:

```json
{
  "mcpServers": {
    "rt-prompt-mcp": {
      "command": "rt-prompt-mcp",
      "args": []
    }
  }
}
```

## 工具说明

该 MCP Server 提供五个主要工具:

1. **get_backend_suggestions**: 获取后端开发相关的提示词补充

   - `context`: 当前上下文或任务描述
   - `databaseType`: 数据库类型(如 MySQL、PostgreSQL 等)
   - `language`: 编程语言(如 Java、Python 等)

2. **get_frontend_suggestions**: 获取前端开发相关的提示词补充

   - `context`: 当前上下文或任务描述
   - `framework`: 前端框架(如 React、Vue 等)
   - `deviceType`: 设备类型(如移动端、桌面端等)

3. **get_general_suggestions**: 获取通用场景的提示词补充

   - `context`: 当前上下文或任务描述
   - `taskType`: 任务类型(如代码生成、文档生成等)

4. **get_ui_design_suggestions**: 获取 UI 设计图转化相关的提示词补充

   - `context`: 当前上下文或任务描述
   - `designType`: 设计类型(如线框图、高保真原型图等)
   - `platform`: 平台类型(如 Web、iOS、Android 等)

5. **get_rt_crud_suggestions**: 获取荣通后端标准 CRUD 开发规范提示词。

   - `base_path` (可选): Java/Kotlin 根包路径,例如 'com.example.myapp' 或 'cn.teamy'。如果提供,将替换提示中默认的 'cn.teamy'。请使用点分隔路径。

6. **get_feishu_prompt**: 获取飞书相关的提示词
   - `prompt_name`: 提示词名称,如'UI 转化提示词'、'AI 生成 UI-3D 风格'等

## 示例

例如,要获取 MySQL 数据库设计的建议:

```
使用 get-backend-suggestions 工具,并提供以下参数:
- context: "创建用户和订单的数据库表结构"
- databaseType: "MySQL"
- language: "SQL"
```

## 开发

### 前提条件

- Node.js 16+
- npm 或 yarn

### 本地开发

1. 克隆仓库:

   ```bash
   git clone https://github.com/yourusername/rt-prompt-mcp.git
   cd rt-prompt-mcp
   ```

2. 安装依赖:

   ```bash
   npm install
   ```

3. 构建项目:

   ```bash
   npm run build
   ```

4. 本地测试:
   ```bash
   npm start
   ```

## 许可证

MIT

TDQS

D1.8/5.0

Scored across 6 tools

Disambiguation4/5

The tools are clearly distinguished by their target domains (backend, frontend, general, RT-CRUD, UI design, Feishu), making misselection unlikely. However, the lack of descriptions means the exact boundaries between 'general' and domain-specific suggestions are unclear, which could cause minor confusion.

Naming Consistency5/5

All tools follow a perfect 'get_[domain]_suggestions' pattern, with consistent snake_case and verb-noun structure. This predictability makes it easy for agents to understand and use the toolset without naming-related errors.

Tool Count4/5

Six tools is a reasonable number for a prompt suggestion server, covering multiple domains without being overwhelming. It could be slightly thin if more granular domains are needed, but the scope appears well-defined for common development areas.

Completeness2/5

The toolset is severely incomplete as it only provides 'get' operations with no ability to create, update, delete, or manage prompts. For a prompt management server, this represents significant gaps that will limit agent workflows and cause dead ends in multi-step tasks.

Maintenance

ActivityInactive
ResponsivenessNo issues