🔌 MCPX

One config to rule them all. Configure your MCP servers once, deploy to every AI CLI automatically.

---

😩 The Problem

Each AI CLI tool uses a different file format for configuring MCP (Model Context Protocol) servers:

If you use multiple AI tools (and you probably do), you need to manually maintain 8 different config files with different structures, field names, and quirks. For every single project.

---

✨ The Solution

MCPX maintains a single canonical config file (.mcpx.json) per project and automatically generates the correct config file for each AI CLI provider you use.

.mcpx.json  ──────►  .mcp.json                     (Claude Code)
    │       ──────►  .gemini/settings.json          (Gemini CLI)
    │       ──────►  ~/.kimi/mcp.json               (Kimi CLI)
    │       ──────►  .codex/config.toml             (OpenAI Codex)
    │       ──────►  opencode.json                  (OpenCode)
    │       ──────►  .copilot/mcp-config.json       (Copilot CLI)
    │       ──────►  .vscode/mcp.json               (VS Code)
    └─────  ──────►  .idea/mcp.json                 (IntelliJ IDEA)

---

🚀 Quick Start

📦 Installation

npm install -g mcpx-cli

⚡ First Setup

Navigate to your project directory and run:

mcpx

The interactive wizard will guide you through:

1. 🔍 Detection — Automatically detects existing MCP configs in your project
2. 📥 Import — Offers to import servers from detected configs
3. ➕ Add servers — Interactive wizard to configure new MCP servers
4. 🎯 Select providers — Choose which AI CLIs you want to generate configs for
5. ⚙️ Generate — Creates .mcpx.json and all provider config files

---

📋 Commands