Komut#cc-toolkit
0
E2e
Kritik kullanıcı akışları için Playwright'ı kullanarak uçtan uca testler oluşturun.
yazan kuscubasiece·8 Eylül 2026
Kaynak: rohitg00/awesome-claude-code-toolkit · Apache-2.0
Kritik kullanıcı akışları için Playwright'ı kullanarak uçtan uca testler oluşturun.
Kaynak: rohitg00/awesome-claude-code-toolkit · Apache-2.0
mkdir -p .claude/commands && curl -sL https://www.antrophiclove.com/api/kaynak/e2e/raw -o .claude/commands/e2e.mdHedef dosya: .claude/commands/e2e.md
Generate end-to-end tests using Playwright for critical user flows.
## Steps
### 1. Identify Critical Flows
- Ask which user flow to test if not specified.
- Common critical flows:
- Authentication (sign up, log in, log out, password reset)
- Core CRUD operations (create, read, update, delete)
- Payment/checkout flows
- Search and filtering
- Navigation between key pages
### 2. Set Up Test Structure
```typescript
import { test, expect } from "@playwright/test";
test.describe("Feature Name", () => {
test.beforeEach(async ({ page }) => {
// Navigate to starting point
// Set up required auth state if needed
});
test("should complete the happy path", async ({ page }) => {
// Steps matching real user behavior
});
test("should handle error state gracefully", async ({ page }) => {
// Verify error messages, recovery options
});
});
```
### 3. Write Tests Following Best Practices
- Use user-visible selectors: `getByRole`, `getByText`, `getByLabel`.
- Avoid CSS selectors and XPaths unless absolutely necessary.
- Add `data-testid` attributes only when semantic selectors are insufficient.
- Wait for network requests to complete before asserting: `page.waitForResponse`.
- Use `test.step` to document multi-step flows for readable reports.
### 4. Handle Test Data
- Use API calls in `beforeEach` to set up test data (faster than UI).
- Clean up test data in `afterEach` or use isolated test databases.
- Never depend on data created by other tests.
### 5. Run and Verify
- Run with `npx playwright test --headed` to watch execution.
- Verify tests pass in headless mode for CI.
- Check that tests are not flaky by running 3 times.
## Rules
- Test user-visible behavior, not implementation details.
- Each test should be independent and runnable in isolation.
- Keep tests under 30 seconds each.
- Use Playwright's auto-waiting instead of manual `page.waitForTimeout`.
- Configure retries in CI but fix flaky tests rather than relying on retries.
- Take screenshots on failure for debugging: `use: { screenshot: "only-on-failure" }`.