playwright-agent-mcp-starter: Instructions file for GitHub Copilot

GitHub Copilot Instructions

General

Read and follow the conventions defined in AGENTS.md at the repository root. That file is the single source of truth for coding standards, selector strategy, test structure, and project layout.

Shared AI Workspace

This repository uses a shared AI workspace at /ai. Use the resources there:

• /ai/skills/ — Reusable skills (e.g., MCP server selection, prompt templates)
• /ai/agents/ — Agent definitions for specialized tasks
• /ai/prompts/ — Ready-to-use prompt templates
• /ai/knowledge/ — Domain knowledge and reference material

Generating Playwright Tests

When generating or suggesting Playwright tests:

1. Use @playwright/test — import test and expect from @playwright/test
2. Follow the Page Object Model — interact through page objects in pages/, not raw selectors
3. Prefer accessibility selectors — getByRole, getByLabel, getByText over CSS or XPath
4. Save tests under tests/e2e/ — use descriptive filenames like checkout-flow.spec.ts
5. Tag tests — add @smoke, @regression, or @api annotations
6. Use fixtures — extend the base fixture in tests/fixtures/base.fixture.ts
7. Generate test data — use utils/data-factory.ts, never hard-code values
8. No waitForTimeout — rely on Playwright auto-waiting and assertion retries
9. Keep tests deterministic — no shared state, no order dependencies
10. TypeScript strict mode — proper types, no any

Example Test Skeleton

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

test.describe('Feature Name', () => {
  test('should perform expected behavior @smoke', async ({ page }) => {
    // Arrange
    await page.goto('/feature');

    // Act
    await page.getByRole('button', { name: 'Submit' }).click();

    // Assert
    await expect(page.getByText('Success')).toBeVisible();
  });
});

Code Style

• Use async/await consistently
• Prefer const over let
• Use descriptive test names starting with "should"
• Group related tests with test.describe
• One assertion focus per test (multiple related assertions are fine)