MobileWright: Playwright Finally Comes to Mobile

The Mobile Automation Problem

Web test automation has a clear winner in 2026: Playwright. Developers love its auto-wait mechanism, semantic locators (getByRole, getByLabel), and rock-solid TypeScript support.

Mobile has no equivalent. Appium is powerful but verbose and inherently fragile. XCUITest and Espresso are platform-specific. Detox works for React Native but struggles with native UI layers. And all of these frameworks make AI agent integration a nightmare — they require coordinate-based clicks or opaque XPath selectors that LLMs can't reason about.

MobileWright aims to fix this.

What Is MobileWright?

MobileWright (mobilewright.dev) is an open-source mobile automation framework with a unified TypeScript API for both iOS and Android. Instead of relying on fragile coordinates or XPath, it exposes the device's accessibility tree — the same semantic structure that screen readers use.

This has two massive benefits:

1. Human-readable, resilient locators: Find elements by role, label, or text — just like Playwright for the web.
2. AI agent-friendly: AI agents (Claude, GPT-4, Gemini) can interact with your app using readable commands without needing expensive screenshot-based vision models.

MobileWright vs. Appium: A Quick Comparison

Code Example: Your First MobileWright Test

Here's how clean a MobileWright test looks compared to traditional Appium:

import { test, expect } from '@mobilewright/test';

test('login flow', async ({ device }) => {
  await device
    .getByLabel('Username')
    .fill('john@example.com');

  await device
    .getByLabel('Password')
    .fill('secret123');

  await device
    .getByRole('button', { name: 'Sign In' })
    .tap();

  await expect(
    device.getByText('Welcome back, John')
  ).toBeVisible();
});

const username = await driver.findElement(
  By.xpath('//XCUIElementTypeTextField')
);
await username.sendKeys('john@example.com');

const password = await driver.findElement(
  By.xpath('//XCUIElementTypeSecureTextField')
);
await password.sendKeys('secret123');

const loginBtn = await driver.findElement(
  By.accessibilityId('sign-in-button')
);
await loginBtn.click();

await driver.waitUntil(async () => {
  const el = await driver.findElement(
    By.xpath('//*[@label="Welcome back, John"]')
  );
  return await el.isDisplayed();
}, { timeout: 5000 });

The AI Agent Superpower

This is where MobileWright gets genuinely exciting. Because it exposes the accessibility tree rather than coordinates, you can connect AI agents directly to your mobile app.

Imagine prompting Claude: "Navigate to the checkout screen and verify the order total matches the cart" — and having the agent execute it on a real iOS simulator, using getByRole and getByLabel locators without needing to process a single screenshot.

Getting Started

1. Install the CLI:

   npm init mobilewright@latest ./

2. Auto-discover simulators: The CLI automatically finds connected devices and running simulators — no configuration needed.
3. Run your tests:

   npx mobilewright test

Full docs at mobilewright.dev and the source on GitHub.

Should You Adopt It?

MobileWright is still early (open-source, not yet v1.0), but its trajectory is exciting. As AI-agent-driven testing becomes standard, a framework that's purpose-built for that paradigm has a massive advantage.