---

The Problem

Managing MCP servers across multiple AI clients (Claude Desktop, Cursor, Windsurf, VS Code, Zed) is fragmented and tedious:

• 🔧 Scattered configs - Each client has its own config file in different locations
• 🔄 Manual sync - Adding a server means manually editing multiple JSON files
• 👁️ No visibility - Hard to know which servers are working or failing
• 🧪 No testing - Can't easily verify servers before using them in your workflow

This friction slows down development and makes server management error-prone.

The Solution

MCP Server Manager is a unified CLI + TUI tool that centralizes MCP server management with a single gateway pattern. Add servers once, connect your clients, and everything stays in sync automatically.

Key Features:

• 🎯 Gateway Pattern - One gateway, all clients
• 🎨 Interactive TUI - Beautiful terminal UI with keyboard shortcuts and customizable themes
• ⚡ Automatic Sync - Changes propagate to all clients instantly
• 🧪 Built-in Testing - Verify servers before using them
• 📊 Token Tracking - Monitor context usage per server
• 📦 Profile Management - Group servers for different contexts with cloning support
• 🎨 Theme Customization - Choose from three color palettes (default, minimal, colorful)

---

Installation

NPM

npm install -g mcp-server-manager

Bash (curl)

curl -fsSL https://raw.githubusercontent.com/sardine-ai/mcp-server-manager/main/scripts/install.sh | bash

PowerShell

irm https://raw.githubusercontent.com/sardine-ai/mcp-server-manager/main/scripts/install.ps1 | iex

---

Quick Start

Launch the TUI

mcpsm

Or use CLI commands

# List all servers
mcpsm list

# Add a server
mcpsm add myserver --type stdio --command "npx" --args "-y @modelcontextprotocol/server-filesystem /tmp"

# Test all servers
mcpsm test

# Connect a client
mcpsm clients connect claude

For complete command reference and guides, see the documentation.

---

Contributing

Contributions are welcome! Please see our contributing guidelines for details.

---

License

MIT © Sardine AI