Get the FREE Ultimate OpenClaw Setup Guide →

mcp -starter

MCP (model context protocol) starter server. For UI libraries, APIs, open-source projects and more

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio mnove-mcp-server-starter node /path/to/mcp-starter/dist/server.js

How to use

This MCP server template provides a ready-to-use MCP server focused on exposing UI component registries to AI assistants. It offers a structured approach for organizing components into categories, validating data with Zod schemas, and providing dynamic tools per category (e.g., getButtons, getForms) so clients can query components, usage examples, installation instructions, and related components. Start the server in development or production modes, and use the generated tools to retrieve registry data for your UI library or design system. The server is designed to integrate with registry-based libraries (such as shadcn/ui) and can be extended with custom component types and additional categories as your project grows.

How to install

Prerequisites:

  • Node.js 18 or higher
  • pnpm (recommended) or npm
  • Git (to clone templates or repos)

Installation steps:

  1. Clone the template repository:
git clone https://github.com/mnove/mcp-starter.git
cd mcp-starter
  1. Install dependencies:
pnpm install
# or npm install
  1. Build the project:
pnpm run build
# or npm run build
  1. Run in development mode:
pnpm run dev
# or npm run dev
  1. Inspect the MCP server (optional):
pnpm run inspect

Configuration notes:

  • Update src/lib/config.ts to point to your actual component registry URLs and files.
  • If you customize categories, ensure you rebuild before running in production.

Additional notes

Tips:

  • When adding new categories, the server will automatically create corresponding tools. Ensure your registry data aligns with the category keys you define.
  • Replace placeholder URLs (e.g., https://ui.stackzero.co) with your actual project URLs for production use.
  • Environment variables like API keys or tokens may be required for private registries; document and load them via a .env file or your deployment environment.
  • Use the Inspector tool to test tools interactively; ensure authentication tokens are present if required by the inspector URL.
  • If you encounter tool limits in AI clients, consider consolidating or nesting tools by category to manage the tool count effectively.

Related MCP Servers

Sponsor this space

Reach thousands of developers