project-scaffold

project-scaffold

An MCP server that scaffolds full-stack projects with consistent structure, Docker setup, CI/CD pipelines, and database configuration.

Category
访问服务器

README

project-scaffold

An MCP server that scaffolds full-stack projects with consistent structure, Docker setup, CI/CD pipelines, and database configuration. Built with FastMCP.

Alpha — This project is under active development. APIs and generated output may change between versions. Contributions are welcome! See CONTRIBUTING.md.

What it generates

Every scaffolded project comes with:

  • FastAPI backend with async SQLAlchemy and Pydantic v2
  • PostgreSQL database with Docker Compose (includes pgAdmin)
  • Frontend — React (Vite + TypeScript) or HTMX (Tailwind CSS)
  • Alembic database migrations (optional)
  • JWT authentication boilerplate (optional)
  • CI/CD — GitHub Actions for Azure, or Render.com blueprint
  • Makefile with common dev commands

Template combinations

Project type Frontend Deploy target Repo structure
work react Azure (App Service + Static Web Apps) Multi-repo (API + frontend)
work htmx Azure (App Service) Single repo
personal react Render.com Multi-repo (API + frontend)
personal htmx Render.com Single repo

Requirements

  • Python 3.12+
  • uv (recommended) or pip

Installation

From GitHub (pip)

pip install git+https://github.com/dawiegriesel/mcp-dev.git

From GitHub (uv)

uv pip install git+https://github.com/dawiegriesel/mcp-dev.git

Local development

git clone https://github.com/dawiegriesel/mcp-dev.git
cd mcp-dev
uv sync

For development dependencies (pytest, ruff):

uv sync --extra dev

Installing the MCP server

Claude Code

claude mcp add project-scaffold -- uvx --from "git+https://github.com/dawiegriesel/mcp-dev.git" project-scaffold

Or if installed locally:

claude mcp add project-scaffold -- uv run --directory /path/to/mcp-dev project-scaffold

Project-scoped config (local development)

For local development, create a .mcp.json in the repo root to register the server automatically when you open the repo in Claude Code — no manual claude mcp add needed:

{
  "mcpServers": {
    "project-scaffold": {
      "command": "uv",
      "args": ["run", "--directory", "/your/local/path/mcp-dev", "project-scaffold"]
    }
  }
}

Replace the --directory value with your actual clone path. This file is gitignored to avoid committing local paths.

Claude Desktop

Add this to your Claude Desktop config file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows):

{
  "mcpServers": {
    "project-scaffold": {
      "command": "uvx",
      "args": [
        "--from", "git+https://github.com/dawiegriesel/mcp-dev.git",
        "project-scaffold"
      ]
    }
  }
}

Other MCP clients

Run the server with stdio transport (the default):

project-scaffold

Or without installing:

uvx --from "git+https://github.com/dawiegriesel/mcp-dev.git" project-scaffold

MCP tools

The server exposes five tools via MCP:

list_templates

Returns all available project template combinations and their default stack.

create_project

Scaffolds a complete project. Configuration options:

Parameter Required Default Description
name yes — Project name (lowercase, hyphens allowed)
project_type yes — "work" (Azure) or "personal" (Render.com)
frontend_type yes — "react" (SPA) or "htmx" (server-rendered)
output_dir yes — Parent directory for the project folder
description no "" Short project description
db_name no derived from name Database name
include_auth no true Include JWT auth boilerplate
include_alembic no true Include Alembic migrations
api_port no 8000 Local dev API port
frontend_port no 3000 Local dev frontend port (React only)

add_component

Adds a component to an existing scaffolded project. Supported components:

  • api_router — FastAPI router with full CRUD endpoints, plus auto-generated model and schema if fields are provided
  • db_model — SQLAlchemy model with matching Pydantic schemas

Components not yet implemented: frontend_page, github_action, docker_service.

validate_project

Checks that a scaffolded project has all expected files. Reports any missing files.

get_project_info

Reads the .scaffold.json metadata from an existing project and returns its configuration.

Example prompts

Once the MCP server is installed, you can use natural language to scaffold and extend projects. Here are some example prompts to get started:

Scaffold a new project

Create a new personal project called "recipe-box" with an HTMX frontend. Put it in ~/projects. Include auth and migrations.

Scaffold a work project named "inventory-tracker" with a React frontend in ~/work. No auth needed but include Alembic.

I want to start a new side project called "budget-app". It should be a simple server-rendered app I can deploy to Render. Create it in ~/dev.

What project templates are available?

Add components to an existing project

Add a "products" API router to my project at ~/projects/recipe-box with fields: title (str), prep_time (int), servings (int), instructions (text).

Add a database model called "category" with fields name (str) and description (text) to ~/projects/recipe-box.

I need a new "orders" endpoint in my inventory-tracker project. It should have fields for customer_name (str), total (float), and fulfilled (bool).

Inspect and validate projects

Check if my project at ~/projects/recipe-box has all the expected files.

What's the configuration for the project in ~/work/inventory-tracker?

Validate the project structure at ~/dev/budget-app and tell me if anything is missing.

Project structure

React projects (multi-repo)

my-app/
├── Makefile
├── README.md
├── .env.example
├── docker-compose.yml
├── docker-compose.override.yml
├── .scaffold.json
├── my-app-api/
│   ├── pyproject.toml
│   ├── Dockerfile
│   ├── app/
│   │   ├── main.py
│   │   ├── config.py
│   │   ├── database.py
│   │   ├── models/
│   │   ├── schemas/
│   │   ├── routers/
│   │   └── auth/          (optional)
│   ├── alembic/            (optional)
│   ├── tests/
│   └── .github/workflows/
└── my-app-frontend/
    ├── package.json
    ├── vite.config.ts
    ├── Dockerfile
    ├── nginx.conf
    ├── src/
    │   ├── App.tsx
    │   ├── api/
    │   ├── pages/
    │   ├── components/
    │   └── styles/
    └── .github/workflows/

HTMX projects (single repo)

my-app/
├── Makefile
├── README.md
├── .env.example
├── docker-compose.yml
├── pyproject.toml
├── Dockerfile
├── .scaffold.json
├── app/
│   ├── main.py
│   ├── config.py
│   ├── database.py
│   ├── models/
│   ├── schemas/
│   ├── routers/
│   ├── templates/
│   │   ├── base.html
│   │   ├── index.html
│   │   └── partials/
│   ├── static/css/
│   └── auth/              (optional)
├── alembic/                (optional)
└── tests/

Development

# Lint
uv run ruff check src/

# Format
uv run ruff format src/

# Test
uv run pytest

Automated documentation updates

This repo includes a docs-pre-commit Claude agent (.claude/agents/docs-pre-commit.md). When working in Claude Code, it triggers automatically before commits to update CHANGELOG.md, README.md, and CLAUDE.md to reflect the changes being committed.

To trigger it manually, tell Claude: "Update the documentation before I commit."

Architecture

src/project_scaffold/
├── __main__.py          # Entry point — runs mcp.run()
├── server.py            # FastMCP server with 5 tool definitions + main()
├── models.py            # Pydantic models (ProjectConfig, ComponentConfig)
├── config.py            # Constants, type maps, template paths
├── generator.py         # Core engine — renders Jinja2 templates to disk
├── renderers/
│   └── api.py           # Component generators (add_router, add_model)
└── templates/           # Jinja2 templates (included in package)
    ├── api/             # FastAPI backend
    ├── frontend/react/  # React + Vite + TypeScript
    ├── frontend/htmx/   # HTMX + Tailwind
    ├── docker/          # Docker Compose
    ├── cicd/azure/      # GitHub Actions for Azure
    ├── cicd/render/     # Render.com blueprint
    ├── alembic/         # Database migrations
    └── common/          # README, Makefile, .env, .gitignore

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选