Arch-Master MCP
Multi-stack scaffolding engine for generating production-ready module structures (NestJS, Java Spring, Python FastAPI) directly from an MCP-compatible AI client, reducing token usage for boilerplate code.
README
Arch-Master MCP
Multi-stack scaffolding engine exposed as a Model Context Protocol (MCP) server.
Generates production-ready module structures for NestJS, Java Spring, and Python FastAPI — directly from any MCP-compatible AI client.
Zero-token boilerplate. Designed for projects that are just getting started — instead of asking the AI to write every file (which burns thousands of tokens), the AI simply calls a tool and the entire module structure is generated locally on your machine in milliseconds.
What it does
Arch-Master MCP exposes two tools over stdio:
| Tool | Description |
|---|---|
scaffold_module |
Generates all architecture files for a named module in the chosen stack/pattern |
list_patterns |
Returns a markdown table of every supported stack and its available patterns |
No files are uploaded to the AI. All generation runs locally — only the file list is returned to the client.
Why this matters for new projects
When starting a project from scratch, the AI would normally write every file from scratch — entity, repository, service, controller, module, interfaces, mappers — burning thousands of output tokens per module. Arch-Master flips this: the AI decides what to generate, the MCP generates how.
| Approach | Tokens per module (approx.) |
|---|---|
| AI writes each file manually | ~2 000 – 5 000 tokens |
scaffold_module via Arch-Master |
~50 – 100 tokens |
Supported stacks & patterns
| Stack | Patterns | Base path |
|---|---|---|
nestjs |
hexagonal, layers, clean |
src/modules |
java_spring |
hexagonal, layers, clean |
src/main/java/com/company/modules |
python_fastapi |
hexagonal, layers, clean |
src/modules |
Quick start
1. Install dependencies
pnpm install
2. Build
pnpm build
# output → dist/
3. Register in your MCP client (e.g. Claude Desktop)
{
"mcpServers": {
"arch-master": {
"command": "node",
"args": ["dist/index.js"],
"cwd": "/absolute/path/to/arch-master-mcp"
}
}
}
Development
pnpm test # unit tests (Jest)
pnpm test:cov # coverage report
pnpm build # compile TypeScript → dist/
Project structure
src/
├── index.ts # MCP server entry point (tool registration)
├── core/
│ ├── generator.ts # File creation engine
│ └── template-engine.ts # {{Variable}} template renderer
├── handlers/
│ └── scaffold.handler.ts # Tool request handlers
├── strategies/
│ ├── index.ts # Strategy registry (single source of truth)
│ ├── hexagonal/ # Hexagonal pattern templates per stack
│ ├── layers/ # Layered pattern templates per stack
│ └── clean/ # Clean architecture templates per stack
├── types/
│ └── mcp-types.ts # Shared TypeScript interfaces
└── utils/
└── file-system.ts # fs helpers
Extending
- New stack: add an entry in
src/strategies/index.tsand its pattern files undersrc/strategies/<pattern>/. - New pattern for an existing stack: add a pattern file and wire it in
src/strategies/index.ts. - The
src/index.tsentry point is generic — it never needs to change.
Technical stack
- Runtime: Node.js 20+ / TypeScript (ESM,
NodeNextmodule resolution) - MCP SDK:
@modelcontextprotocol/sdk - Validation: Zod
- Build: NestJS CLI →
dist/ - Package manager: pnpm
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。