Tauri

You can automate and test your Tauri desktop app using the MCP (Model Context Protocol) server. This setup lets AI assistants interact with your app for end-to-end testing and automation across platforms, with a direct bridge to your app’s UI and actions.

How to use

You run an MCP server locally and connect an MCP client to it. The server exposes a set of testing tools that let you launch your Tauri app, inspect the UI, simulate clicks and text input, take screenshots, and retrieve logs. Use the standard MCP bridge to initialize the connection from your frontend during development. The workflow typically looks like: initialize the MCP bridge, launch the app, perform actions like clicking and filling fields, capture screenshots for verification, and stop the app when testing is complete.

How to install

Prerequisites you need before installing the MCP server: Node.js >= 18, Tauri v2.x, pnpm (recommended) or npm, and Rust with cargo.

Step-by-step commands you should run in your project directory:

Install the Rust plugin dependency in your Tauri project by editing the Rust manifest and adding the plugin dependency.

Additional setup and configuration

Configure the MCP server by adding a local stdio MCP entry that points to the runtime script provided with the plugin. This is a local server configuration that runs directly in your development environment.

{
  "mcpServers": {
    "tauri_mcp": {
      "command": "node",
      "args": ["node_modules/tauri-plugin-mcp/packages/tauri-mcp/dist/index.js"],
      "env": {
        "TAURI_PROJECT_ROOT": "."
      }
    }
  }
}

Troubleshooting and notes

If the MCP bridge does not initialize, verify that you call initMcpBridge() in your frontend during development, and ensure your app runs in dev mode so the bridge can attach properly.

Available tools

app_status

Check if the Tauri app is currently running.

launch_app

Launch the Tauri app, with optional timeout and features to enable specific configurations.

stop_app

Stop the running Tauri app.

snapshot

Capture the accessibility tree and return element references.

click

Click an element by a reference number or a CSS selector.

fill

Fill an input field identified by a ref or selector with a value.

press_key

Press a specific keyboard key.

navigate

Navigate the app or an embedded web view to a given URL.

screenshot

Take a screenshot of the current UI state.

evaluate_script

Execute a custom JavaScript snippet within the app context.

get_console_logs

Retrieve console logs from the app for debugging.

get_network_logs

Retrieve network logs for debugging.