Intlayer MCP Server

The Intlayer MCP (Model Context Protocol) Server provides AI-powered IDE assistance tailored for the Intlayer ecosystem.

Where can I use it?

• On modern developer environments like Cursor, VS Code, and any IDE supporting the MCP protocol.
• On your favorite AI assistant like Claude Desktop, Gemini, ChatGPT, etc.

Why Use the Intlayer MCP Server?

By enabling the Intlayer MCP Server in your IDE, you unlock:

• Context-Aware Documentation The MCP server loads and exposes the documentation of Intlayer. To speed up your set up, your migrations, etc. This ensures that code suggestions, command options, and explanations are always up to date and relevant.

• Smart CLI Integration Access and run Intlayer CLI commands directly from your IDE interface. Using the MCP server, you can let your AI assistant run commands like intlayer dictionaries build to update your dictionaries, or intlayer dictionaries fill to fill your missing translations.

  View the full list of commands and options in the Intlayer CLI documentation.

• CMS and Analytics Access With an access key, your AI assistant can manage the dictionaries of your Intlayer CMS project and analyze its audience: traffic per locale and page, content exposure, conversions and A/B tests.

Local server (stdio) vs Remote server (Streamable HTTP)

The MCP server can be used in two ways:

• Local server (stdio)
• Remote server (Streamable HTTP)

Local server (stdio) (recommended)

Intlayer provides a NPM package that can be installed locally on your machine. It can be installed in your favorite IDE, as VS Code, Cursor, as well as your locale assistant application, as ChatGPT, Claude Desktop, etc.

This server is the recommended way to use the MCP server. As it integrates all the features of the MCP server, including the CLI tools.

Remote server (Streamable HTTP)

The MCP server can also be used remotely, using the Streamable HTTP transport method. This server is hosted by Intlayer, and is available at https://mcp.intlayer.org. This server can be accessed publicly, without any authentication, and is free to use.

Note that the remote server does not integrate CLI tools, AI autocompletion, etc. Without credentials, it exposes the documentation tools only. Send an access key to also use the CMS and analytics tools, see Analytics tools.

Due to server hosting costs, the availability of the remote server cannot be guaranteed. We limit the number of simultaneous connections. We recommend using the local server (stdio) transport method for the most reliable experience.

Setup via Intlayer CLI (recommended)

Intlayer provides a CLI command to automatically configure the MCP server in your project.

npx intlayer init mcp

This command will:

1. Ask you which platform you are using (Cursor, VS Code, Claude Desktop, etc.).
2. Ask you which transport method you want to use (Local server (stdio) or Remote server (Streamable HTTP)).
3. Automatically update your configuration file (e.g., .cursor/mcp.json, .vscode/mcp.json, or the global Claude Desktop config).

Setup via Intlayer VS Code extension

1. Open the Command Palette (Ctrl+Shift+P or Cmd+Shift+P).
2. Type Intlayer: Setup AI Agent Skills
3. Pick the platform you use (e.g. VS Code, Cursor, Windsurf, OpenCode, Claude Code, GitHub Copilot Workspace, etc.).
4. Pick the MCP to install (stdio, Streamable HTTP)
5. Press Enter.

Manual setup

{
  "mcpServers": {
    "intlayer": {
      "command": "npx",
      "args": ["-y", "@intlayer/mcp"]
    }
  }
}

{
  "mcpServers": {
    "intlayer-sse": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://mcp.intlayer.org"]
    }
  }
}

{
  "servers": {
    "intlayer": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@intlayer/mcp"]
    }
  }
}

{
  "servers": {
    "intlayer-sse": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://mcp.intlayer.org"]
    }
  }
}

{
  "mcpServers": {
    "intlayer": {
      "command": "npx",
      "args": ["-y", "@intlayer/mcp"]
    }
  }
}

{
  "mcpServers": {
    "intlayer-sse": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://mcp.intlayer.org"]
    }
  }
}

claude mcp add intlayer npx -y @intlayer/mcp

claude mcp add intlayer https://mcp.intlayer.org -t http

Using the MCP Server via CLI

You can also run the Intlayer MCP server directly from the command line for testing, debugging, or integration with other tools.

# Install globally
npm install -g @intlayer/mcp

# Or use directly with npx (recommended)
npx @intlayer/mcp

Analytics tools

The MCP server can answer questions such as "which locales grew this month?", "which pages do French visitors land on?" or "why did conversions drop last week?" using the analytics collected by your Intlayer CMS project.

Reports only contain aggregated, anonymous data. All dates are in UTC.

Authentication

Create an access key in the dashboard of your project (Project settings > Access keys) and enable the ANALYTICS : READ right. Keep only the rights your assistant needs.

{
  "mcpServers": {
    "intlayer": {
      "command": "npx",
      "args": ["-y", "@intlayer/mcp"],
      "env": {
        "INTLAYER_CLIENT_ID": "<your-client-id>",
        "INTLAYER_CLIENT_SECRET": "<your-client-secret>"
      }
    }
  }
}

{
  "mcpServers": {
    "intlayer": {
      "url": "https://mcp.intlayer.org",
      "headers": {
        "x-intlayer-client-id": "<your-client-id>",
        "x-intlayer-client-secret": "<your-client-secret>"
      }
    }
  }
}

Avoid pasting your client secret in the chat: it would be stored in the conversation history.