MCPX solves the fragmentation problem of MCP (Model Context Protocol) server configuration. Instead of maintaining separate config files for Claude Code, Copilot, VS Code, IntelliJ, and five other AI tools — each with a different format — you write one file and MCPX takes care of the rest.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/davidpastorvicente/mcpx-cli/llms.txt
Use this file to discover all available pages before exploring further.
Installation
Install MCPX globally and get ready to configure your first project.
Quickstart
Run the interactive wizard and have all your AI tools configured in minutes.
Command Reference
Explore every command, flag, and argument MCPX supports.
Supported Providers
Claude Code, Copilot, VS Code, IntelliJ, OpenAI Codex, and more.
How it works
MCPX maintains a single source of truth —.agents/mcp.json for your project or ~/.agents/mcp.json globally — and generates the correct provider-specific config files whenever you add, remove, or update a server.
Run the interactive wizard
Navigate to your project directory and run
mcpx. The wizard detects existing configs, lets you add servers, and asks which AI providers to target.Add MCP servers
Use
mcpx add to configure additional servers at any time. Each server can use stdio (local command) or http (remote URL) transport.Key features
Single source of truth
One canonical
.agents/mcp.json file drives every provider config. Edit once, deploy everywhere.Smart merge
MCPX only rewrites the MCP section of provider files, preserving unrelated settings like OpenAI Codex’s
model and approval_mode.Import existing configs
Already using Claude Code or Copilot? Import your existing servers directly into the canonical config with
mcpx import.Enable / disable servers
Temporarily disable a server without removing it. The
enabled flag is reflected across all provider configs on the next sync.