Playwright

Playwright is an end-to-end (e2e) testing framework to simulate real user interactions across an entire web app.

Microsoft's Playwright MCP server allows an agent to interact with a Playwright server so agents can automatically test.

Add Playwright MCP server:

claude mcp add playwright npx '@playwright/mcp@latest'

Instruct the agent to use Playwright: Use playwright mcp to open a browser to example.com

Example generated e2e test:

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

test.describe('Task app E2E', () => {
  test('user can log in, create a task, and log out', async ({ page }) => {
    // 1. Go to the app
    await page.goto('http://localhost:3000');

    // 2. Log in
    await page.getByLabel('Email').fill('andrew@example.com');
    await page.getByLabel('Password').fill('SuperSecret');
    await page.getByRole('button', { name: 'Sign in' }).click();

    // 3. Verify dashboard loaded
    await expect(page.getByRole('heading', { name: 'Dashboard' })).toBeVisible();

    // 4. Create a new task
    await page.getByRole('button', { name: 'New Task' }).click();
    await page.getByLabel('Task title').fill('Review PR');
    await page.getByLabel('Description').fill('Make sure the end-to-end test covers a real flow');
    await page.getByRole('button', { name: 'Save Task' }).click();

    // 5. Verify success message
    await expect(page.getByText('Task created successfully')).toBeVisible();
  });
});

Related: MCP, Claude Chrome Extension