test-execution

name	Test Execution
description	Strategies for running tests including inline, pytest, and Playwright

Test Execution Skill

This skill provides test execution strategies for different project types.

When to Use

• QA verification of completed tasks
• Running unit/integration/e2e tests
• Debugging test failures

Strategy 1: Inline Test (Script/Library Projects)

Best for simple validation without import complexity.

# Create test in SAME directory as code
write_file("project_tracking/test_inline.py", """
from dummy_math import add  # No import issues - same directory!

result = add(2, 3)
expected = 5

if result == expected:
    print(f"✅ PASS: add(2, 3) = {result}")
    exit(0)
else:
    print(f"❌ FAIL: add(2, 3) = {result}, expected {expected}")
    exit(1)
""")

# Run from correct directory
run_command("cd project_tracking && python test_inline.py")

Strategy 2: Pytest (Standard Projects)

# Run specific test file
pytest tests/test_feature.py -v

# Run with pattern matching
pytest -k "test_user" -v

# Run with coverage
pytest --cov=src tests/

Strategy 3: Playwright (Web UI Testing)

CRITICAL: Always use headless mode for autonomous execution.

// playwright.config.ts
export default defineConfig({
  use: {
    headless: true,  // MANDATORY - never false
    baseURL: 'http://localhost:5173',
  },
});

# Run tests
npx playwright test

# NEVER use these flags:
# --headed (opens browser)
# --debug (opens inspector)

Circuit Breaker Pattern

Max 3 retries for same error:

1. Attempt 1: pytest → ModuleNotFoundError
2. Attempt 2: sys.path fix → ModuleNotFoundError
3. Attempt 3: importlib → ModuleNotFoundError
4. STOP → Switch to inline strategy

Failure Investigation

When test fails, categorize:

Category	Action
Environment Issue	Mark BLOCKED, document
Test Code Issue	Fix test, re-run
Application Defect	Create DEFECT task

Evidence Logging

Always include test output in completion:

VERIFIED:
- pytest tests/test_auth.py: 5 passed, 0 failed
- Coverage: 87%