NPMs.io MCP Server

Overview

MCP (Model Context Protocol) is a framework that allows you to integrate custom tools into AI-assisted development environments—such as Cursor AI. MCP servers expose functionality (like data retrieval or code analysis) so that an LLM-based IDE can call these tools on demand. Learn more about MCP in the Model Context Protocol Introduction.

This project demonstrates an MCP server built in Node.js that provides powerful NPM package search and information retrieval capabilities through the npms.io API. It offers tools for searching packages, getting package suggestions, and retrieving detailed package information.

Requirements

• Node.js: Version 20 or higher is required.

Features

• MCP Integration: Exposes NPM package search and information tools to LLM-based IDEs.
• Advanced Package Search: Search for NPM packages with support for various filters and modifiers:
  • Scope filtering (e.g., scope:types)
  • Author and maintainer filtering
  • Keyword-based filtering
  • Package status filtering (deprecated, unstable, insecure)
  • Search score customization
• Search Suggestions: Get package name suggestions with highlighted matches
• Package Information: Retrieve detailed information about single or multiple packages
• Input Validation: Uses Zod for schema validation
• Standard I/O Transport: Connects via StdioServerTransport for integration with development environments

Installation

1. Clone the Repository

   git clone <repository_url>
   cd <repository_directory>

2. Install Dependencies

   npm install

Integrating with Cursor AI

This project includes a ./cursor subdirectory that contains an mcp.json file for configuring the MCP server. Cursor AI uses this file to automatically discover and launch your MCP server.

The ./cursor/mcp.json Structure

{
  "mcpServers": {
    "npms": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-server-npms"
      ]
    }
  }
}

You can verify the absolute path to your Node.js executable by running which node in your terminal.

Optional: Configuration Automation Scripts

Easily configure your local environment by automatically updating the mcp.json file with the correct absolute paths:

chmod +x ./scripts/update_config.sh
./scripts/update_config.sh

Optional: Global Cursor settings

You can also move the mcp.json file to your global Cursor AI configuration directory located at ~/.cursor to make the configuration available globally.

Using the MCP Tools in Cursor Composer (Agent Mode)

With the MCP server integrated into Cursor AI and with Agent mode enabled in Cursor Composer, you can use natural language prompts like:

search for react packages that are not deprecated

get information about the express package

get suggestions for "react-router"

The AI agent will infer the appropriate tool from your MCP server and execute it accordingly.

API

Resources

• https://api.npms.io/v2: NPM package search and information interface

Tools

• search

  • Search for npm packages with support for advanced filters and modifiers
  • Inputs: q (String): Search query with support for filters and modifiers:
    • scope:types - Show/filter results that belong to the @types scope
    • author:username - Show/filter results by author
    • maintainer:username - Show/filter results by maintainer
    • keywords:keyword1,keyword2 - Show/filter results by keywords (use -keyword to exclude)
    • not:deprecated - Exclude deprecated packages
    • not:unstable - Exclude packages with version < 1.0.0
    • not:insecure - Exclude insecure packages
    • is:deprecated - Show only deprecated packages
    • is:unstable - Show only unstable packages
    • is:insecure - Show only insecure packages
    • boost-exact:false - Disable exact match boosting
    • score-effect:14 - Set score effect (default: 15.3)
    • quality-weight:1 - Set quality weight (default: 1.95)
    • popularity-weight:1 - Set popularity weight (default: 3.3)
    • maintenance-weight:1 - Set maintenance weight (default: 2.05)
  • Returns: JSON object containing:
    • package: Package data (name, version, etc.)
    • flags: Package flags (deprecated, unstable, insecure)
    • score: Package score details
    • searchScore: Computed search score

• search_suggestions

  • Get package name suggestions with highlighted matches
  • Inputs: q (String): Search query (qualifiers will be ignored)
  • Returns: JSON object containing:
    • package: Package data (name, version, etc.)
    • flags: Package flags (deprecated, unstable, insecure)
    • score: Package score details
    • searchScore: Computed search score
    • highlight: Highlighted matched text

• get_package_info

  • Get detailed information about a package
  • Inputs: name (String): Package name to get information for
  • Returns: JSON object containing:
    • score: Package score information
    • analyzedAt: Date of last package analysis
    • collected: Information from all sources
    • evaluation: Package evaluation details
    • error: Any error from last analysis attempt

• get_multi_package_info

  • Get detailed information about multiple packages in a single request
  • Inputs: names (String[]): Array of package names to get information for
  • Returns: JSON object containing:
    • score: Package score information
    • analyzedAt: Date of last package analysis
    • collected: Information from all sources
    • evaluation: Package evaluation details
    • error: Any error from last analysis attempt

What is MCP?

Model Context Protocol (MCP) provides a standardized approach to integrate custom tools into AI-assisted development environments. With MCP, you can define tools that perform specific tasks—such as retrieving external data, validating code, or enforcing coding standards—and the AI assistant in your IDE can call these tools automatically based on context. This helps improve developer productivity, ensures consistent quality, and streamlines workflows.

References & Resources

• Model Context Protocol: typescript-sdk
• Use Your Own MCP on Cursor in 5 Minutes
• Model Context Protocol Introduction
• npms.io API Documentation

License

This project is licensed under the MIT License.