Open Source · MIT · v1.0.0
1000+
verified MCP servers

Install any of them in one command.

mcpm manages MCP servers across Claude, Cursor, Windsurf, Zed, and 5 more AI clients — writing the right config for each one, automatically.

9 AI clients
5 runtimes
11 curated bundles
npm · PyPI · Docker · Go · Deno
$ npx mcp-fleet install github
0
verified MCP servers
0
AI clients supported
0
runtimes supported
0
curated bundles
How it works

Stop editing JSON configs by hand

One command installs a server into every AI client you have — using the right config format for each.

✗ Before — manual config editing
~/.claude.json
# Edit claude config
vim ~/.claude.json
# Then edit cursor...
vim ~/.cursor/mcp.json
# Then zed...
vim ~/.config/zed/settings.json
# ... (4 more clients)
# Hope you got the format right 🤞
✓ After — one command
terminal
$ mcpm install github
Detected 4 clients…
Claude Code
Cursor
Zed
Windsurf

Done in 0.3s
🔍

Search 1000+ servers

Browse by name, tag, or description. Find exactly what you need across 5 runtimes.

🩺

Doctor checks

Verify all installed servers are reachable and correctly configured.

🔄

Auto-update

Keep servers current across npm, PyPI, Docker Hub, and the Go module proxy.

📋

Project config

Pin servers in .mcpmrc. Teammates sync with one command.

📤

Import / Export

Back up your setup or migrate to a new machine in seconds.

🚀

Publish

Submit your MCP server to the registry. Opens a PR automatically.

Clients

Works with every major AI client

mcpm detects installed clients automatically and writes the correct config format — even Zed's unique context_servers structure.

Claude Code
~/.claude.json
Cursor
~/.cursor/mcp.json
VS Code Copilot
~/.vscode/mcp.json
Windsurf
~/.codeium/windsurf/…
Zed
~/.config/zed/settings
Gemini CLI
~/.gemini/settings.json
Cline
VS Code extension
Continue
~/.continue/config.json
OpenAI Codex
~/.codex/config.json
Runtimes

Not just npm

mcpm manages servers across all major runtimes — install, update, and health-check them all with the same commands.

🟢
Node.js
npx
🐍
Python
uvx
🐳
Docker
docker run
🦫
Go
go run
🦕
Deno
deno run
mcpm doctor
Runtimes
Node.js (npx)
Python (uvx)
Docker
○ Go → https://go.dev/dl
○ Deno → curl -fsSL https://deno.land/install.sh | sh

github
postgres
puppeteer
fetch (uvx)

✓ All servers healthy
Bundles

An entire workflow in one command

Bundles are curated packs of the best servers for a use case. Install all of them at once.

terminal
$ mcpm install @bundle/webdev
Resolving bundle… 8 servers
github   filesystem   puppeteer   fetch
sqlite   postgres   sentry   vercel

8 servers installed across 3 clients
@bundle/webdev
@bundle/ai
@bundle/data
@bundle/devops
@bundle/cloud
@bundle/productivity
@bundle/finance
@bundle/social
@bundle/media
@bundle/startup
@bundle/ai-tools
Reference

All commands

mcpm install <server>Install a server (or @bundle/name) into all detected clients
mcpm uninstall <server>Remove a server from all clients
mcpm updateUpdate all installed servers (npm, uvx, docker, go)
mcpm search [query]Browse or filter the 1000-server registry
mcpm info <server>Show details, env vars, and package link
mcpm listList all installed servers per client
mcpm outdatedCheck which servers are out of date
mcpm doctorVerify runtimes and server health
mcpm syncSync servers from .mcpmrc in current directory
mcpm export [file]Export your full config to JSON
mcpm import <file>Import a config from a file
mcpm publishSubmit a server to the public registry via PR
mcpm run <server>Run a server directly in stdio mode

Start in 10 seconds.

No account. No config. Just install and go.

View on GitHub Full docs
$ npx mcp-fleet install github