About MCPverse
MCPverse is a registry for finding setup details for MCP servers. It stores server entries in JSON, offers a local browser-based catalog, and provides a JavaScript/Node.js package for accessing the data.
Key features
- 01JSON-based server registry
- 02Local web catalog
- 03JavaScript and Node.js package
- 04Setup instructions and required variables
- 05Links to official documentation
Use cases
- 01Finding MCP server setup details
- 02Checking required environment variables
- 03Browsing server entries locally
- 04Using registry data in Node.js projects
MCP support
MCPverse catalogs MCP servers and provides their setup details; it does not describe an MCP server interface for clients.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMCPverse
What is this?
Finding setup instructions and configuration details (like environment variables) for different Model Context Protocol (MCP) servers can be difficult. This project solves that.
It provides a central, simple list (mcpverse.json) of quality MCP servers with:
- Easy-to-find setup instructions.
- Clear information on required variables.
- Links to official documentation.
Features
- Simple Data: Server info is stored in
mcpverse.json. - Web Page: Open
index.htmllocally to browse servers. - NPM Package: Use the data in your JavaScript/Node.js projects.
Install via npm
npm install mcpverse
# or
yarn add mcpverse
How to Use (JavaScript / Node.js)
Once installed via npm, you can require the registry in your code:
// Import the registry functions
const mcpverse = require('mcpverse');
// --- Example 1: List all available MCP server names ---
const allMcps = mcpverse.listMcps();
console.log("Available MCPs:");
allMcps.forEach(mcp => console.log(`- ${mcp.name} (ID: ${mcp.id})`));
// Output:
// Available MCPs:
// - Prisma MCP Server (ID: prisma)
// - Puppeteer MCP Server (ID: puppeteer)
// - Slack MCP Server (ID: slack)
// - Supabase MCP Server (ID: supabase)
// --- Example 2: Get full details for a specific MCP ---
const puppeteerDetails = mcpverse.getMcp('puppeteer');
if (puppeteerDetails) {
console.log("\nPuppeteer Details:", JSON.stringify(puppeteerDetails, null, 2));
}
// --- Example 3: Get required variables for Slack ---
const slackInfo = mcpverse.getInstallationInfo('slack');
if (slackInfo && slackInfo.variables) {
console.log("\nSlack requires the following variables:");
for (const [varName, details] of Object.entries(slackInfo.variables)) {
console.log(`- ${varName}: ${details.description}`);
}
}
// Output:
// Slack requires the following variables:
// - SLACK_BOT_TOKEN: Your Slack bot token (starts with xoxb-).
// - SLACK_TEAM_ID: Your Slack workspace Team ID (starts with T).
// - SLACK_CHANNEL_IDS: Comma-separated list of Slack channel IDs to monitor.
// --- Example 4: Generate the full install command for Supabase ---
// Imagine you have your token in an environment variable
const myToken = process.env.SUPABASE_ACCESS_TOKEN || 'your-token-placeholder';
const supabaseCommand = mcpverse.generateFullCommand('supabase', {
SUPABASE_ACCESS_TOKEN: myToken
});
if (supabaseCommand) {
console.log("\nTo run Supabase MCP Server:");
console.log(supabaseCommand);
}
// Example Output:
// To run Supabase MCP Server:
// npx -y @supabase/mcp-server-supabase@latest --access-token "your-token-placeholder"
See index.js for all available functions.
Web Interface
If you clone this repository, you can open the index.html file directly in your web browser. It loads data from mcpverse.json and shows a card for each server with its details, install command, and required variables.
Data Format (mcpverse.json)
The registry is a JSON object. Each key is a server ID (like "prisma"). The value is an object describing the server:
{
"id": "string", // Unique ID (same as key)
"name": "string", // Display name
"description": "string", // What it does
"official": "boolean", // Is it official?
"category": "string", // e.g., "Database"
"readmeUrl": "string | null", // Link to docs
"brand": {
"logoUrl": "string | null" // Logo image URL
},
"installation": {
"type": "string", // How to install (e.g., "npx")
"command": "string", // Base command (e.g., "npx")
"args": ["string"], // Command arguments (use <VAR_NAME> for placeholders)
"variables": { // Optional: Required variables
"VAR_NAME": {
"description": "string",
"secret": "boolean", // Is it sensitive?
"placeholder": "string | null" // Example format
}
}
}
}
Contributing
Want to add or update a server?
- Fork this repository.
- Edit
mcpverse.json(add/update an entry, follow the format). - Create a Pull Request.
Please ensure added servers are stable and have good documentation.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.