registry

The problem: Every AI app uses a different config format for MCP servers. Claude Desktop uses mcpServers, VS Code uses servers, Goose uses YAML with cmd/envs, Codex uses TOML with mcp_servers... there are 19 apps, 6 root keys, and 4 formats.

The solution: One canonical format, config generators for every app, a registry of 105+ servers, and a CLI that auto-detects your apps and writes the correct config.

Browse the full server catalog at getmcp.es

Quick Start

# Install the GitHub MCP server into all your detected AI apps
npx @getmcp/cli add github

# Browse available servers
npx @getmcp/cli list

# Search for database servers
npx @getmcp/cli list --search=database

# Remove a server
npx @getmcp/cli remove github

Note: If you run into issues with npx, try npx @getmcp/cli@latest — npx can cache stale versions, and v0.7.0 had a broken release.

Why getmcp?

How It Works

                      Canonical Format
                           |
    +----------+-----------+-----------+----------+
    |          |           |           |          |
 Claude     VS Code     Goose      Codex   + 15 more
 Desktop   (servers)   (YAML)     (TOML)     apps
(mcpServers)          (extensions)(mcp_servers)

1. Server definitions are stored in a canonical format (aligned with the official MCP registry schema)
2. Config generators transform the canonical format into each app's specific format
3. The CLI auto-detects installed apps, prompts for env vars, and merges configs into existing files (never overwrites)

Supported Apps

Registry

105+ MCP servers included out of the box. Here are some highlights:

Browse the full catalog at getmcp.es or from the CLI:

npx @getmcp/cli list

Packages

Package	Description	npm
@getmcp/cli	CLI tool for installing MCP servers
@getmcp/core	Zod schemas, types, and utilities
@getmcp/generators	Config generators for 19 apps
@getmcp/registry	Registry of 105+ MCP server definitions

Library Usage

Generate config for any app

import { generateConfig, generateAllConfigs } from "@getmcp/generators";

// Generate for a specific app
const config = generateConfig("goose", "github", {
  command: "npx",
  args: ["-y", "@modelcontextprotocol/server-github"],
  env: { GITHUB_TOKEN: "ghp_xxx" },
});
// => { extensions: { github: { cmd: "npx", args: [...], envs: {...} } } }

// Generate for ALL apps at once
const all = generateAllConfigs("github", {
  command: "npx",
  args: ["-y", "@modelcontextprotocol/server-github"],
  env: { GITHUB_TOKEN: "ghp_xxx" },
});
// all["claude-desktop"] => JSON string
// all["goose"]          => YAML string
// all["codex"]          => TOML string

Validate configs with Zod schemas

import { StdioServerConfig, CanonicalMCPConfig } from "@getmcp/core";

StdioServerConfig.parse({ command: "npx", args: ["server"] });
// throws ZodError if invalid

Search the registry

import { searchServers, getServersByCategory } from "@getmcp/registry";

searchServers("database"); // [{ id: "postgres", ... }]
getServersByCategory("web"); // [{ id: "brave-search", ... }, { id: "fetch", ... }]

Development

# Install dependencies
npm install

# Build all packages
npm run build

# Run all tests (515 tests)
npm run test

# Build and test a specific package
npm run build --workspace=@getmcp/core
npm run test --workspace=@getmcp/core

# Run the CLI locally (no build needed)
npx tsx packages/cli/src/bin.ts add
npx tsx packages/cli/src/bin.ts list --search=database

# Run the web directory locally
npm run website

Architecture

getmcp/
  packages/
    core/          # Zod schemas, TS types, utility functions
    generators/    # 19 config generators (one per app)
    registry/      # 105+ MCP server definitions
    cli/           # add/remove/list/doctor/import commands, app detection
    web/           # Next.js web directory at getmcp.es

The CLI supports JSON, JSONC, YAML, and TOML config files natively. Format is auto-detected from the file extension, so Goose configs are read/written as YAML and Codex configs as TOML.

Contributing

See ROADMAP.md for planned improvements and open tasks. Contributions are welcome.

Credits

This CLI is inspired by vercel-labs/skills.

License

MIT