electron-playwright-cli

A fork of playwright-cli that adds Electron desktop app support. All standard Playwright CLI commands work, plus Electron-specific ones for interacting with the main process and listing windows.

Install

npm install -g electron-playwright-cli playwright

playwright is a peer dependency and must be installed alongside.

Configuration

Create .playwright/cli.config.json in your project root:

{
  "browser": {
    "launchOptions": {
      "args": ["path/to/your/main.js"]
    }
  }
}

The args array is passed to Electron. Point it at your app's main process entry point.

The daemon shuts down automatically when the Electron app closes (via close or the app exiting on its own). The next command spawns a fresh daemon that reads the current config from disk, so config changes take effect immediately — no need to manually kill the daemon.

Optional launch options:

• executablePath: path to the Electron binary (defaults to the one in node_modules)
• cwd: working directory for the Electron process
• env: additional environment variables
• timeout: launch timeout in milliseconds

App readiness

By default, the CLI waits for the load event and a paint frame before accepting commands. For apps with async rendering (React, etc.), the first screenshot may still capture a blank page. Use readyCondition to tell the CLI when your app is actually ready:

{
  "browser": {
    "launchOptions": {
      "args": ["main.js"]
    },
    "readyCondition": {
      "waitForSelector": "[data-testid='app-ready']",
      "timeout": 10000
    }
  }
}

Available options (all optional, can be combined):

Option	Type	Description
waitForSelector	CSS selector	Waits for the element to be visible
waitForFunction	JS expression	Waits for the expression to return truthy (e.g. "document.fonts.ready")
waitForLoadState	"load" | "domcontentloaded" | "networkidle"	Waits for the specified load state
timeout	number (ms)	Timeout for all conditions (default: 10000)

A double requestAnimationFrame paint wait always runs after any conditions, ensuring pixels are rendered before the first command.

Usage

# take a snapshot of the running app
electron-playwright-cli snapshot

# interact with elements using refs from the snapshot
electron-playwright-cli click e15
electron-playwright-cli fill e3 "hello world"
electron-playwright-cli type "search query"
electron-playwright-cli press Enter

# take a screenshot
electron-playwright-cli screenshot --filename=app.png

# evaluate in the electron main process
electron-playwright-cli electron_evaluate "electron => electron.app.getName()"

# list all electron windows
electron-playwright-cli electron_windows

All standard commands from playwright-cli are available: goto, snapshot, click, fill, type, press, screenshot, tab-list, tab-select, cookie-list, localstorage-get, route, console, network, tracing-start, tracing-stop, run-code, and more. See the skill reference for the full command list.

How it works

electron-playwright-cli replaces the standard browser launch flow with a custom daemon that composes Playwright's internal BrowserServerBackend with an ElectronContextFactory. When you run a command, the CLI client spawns (or connects to) a background daemon process that launches your Electron app via playwright._electron.launch(), registers Electron-specific tools (like electron_evaluate and electron_windows) into Playwright's tool registry, and exposes the full Playwright CLI command surface over a Unix socket using newline-delimited JSON.

Install as Agent Skill

npx skills add spacecake-labs/electron-playwright-cli

Works with Claude Code, Cursor, Codex, and 40+ other agents.

try spacecake — the best interface for claude code

Upstream

This project is a fork of microsoft/playwright-cli, licensed under Apache-2.0.

License

Apache-2.0