Quality Patterns (Level 1)

Add when CI matters. Fast feedback, test separation, complexity analysis, duplicate detection, test performance.

Commands

test-watch

Continuous test execution on file changes.

# Run tests in watch mode
test-watch:
    <continuous test execution on file changes>

Python:

test-watch:
    uv run pytest-watch -v -m "not integration"

JavaScript:

test-watch:
    pnpm vitest watch --project unit

integration-test

Integration tests. No coverage threshold. Never blocks merge.

# Run integration tests with coverage report (no threshold)
integration-test:
    <integration tests (marked/tagged), report only, never blocks>

Python:

integration-test:
    uv run pytest -v -m "integration" --durations=10

JavaScript:

integration-test:
    pnpm exec playwright test

Java:

integration-test:
    mvn failsafe:integration-test

Key: Marked/tagged tests. Report coverage, no threshold. Never in check-all.

complexity

Detailed complexity report. Informational, not blocking.

# Detailed complexity report for refactoring decisions
complexity:
    <per-function/class complexity, informational, does not block>

Python:

complexity:
    uv run radon cc . -a -nc

JavaScript:

complexity:
    pnpm exec eslint . --ext .ts,.tsx --format complexity

Java:

complexity:
    mvn pmd:pmd
    cat target/pmd.xml

loc

Show N largest files by lines of code.

# Show N largest files by lines of code
loc N="20":
    <show N largest files by LOC, sorted descending>

Universal (works all stacks):

loc N="20":
    find . -name "*.py" -o -name "*.ts" -o -name "*.tsx" -o -name "*.java" | \
      xargs wc -l | sort -rn | head -n {{N}}

duplicates

Find duplicate code blocks. Informational, not blocking.

# Find duplicate code blocks
duplicates:
    <detect copy-paste code, threshold 30%, informational>

Universal (requires jscpd):

duplicates:
    jscpd . --threshold 30 --min-lines 5 --reporters console

Configuration: .jscpd.json

{
  "threshold": 30,
  "minLines": 5,
  "minTokens": 50,
  "ignore": [
    "node_modules/**",
    "dist/**",
    "build/**",
    "**/*.test.js",
    "**/*.test.ts",
    "**/*_test.py",
    "**/*Test.java"
  ],
  "reporters": ["console", "json"],
  "format": ["javascript", "typescript", "python", "java"]
}

Install:

• npm install -g jscpd (global)
• Or: npx jscpd (no install)

Key: Detects copy-paste opportunities. Used by /refactor analysis.

slowtests

Show tests slower than threshold. Identifies optimization targets.

# Show tests slower than N milliseconds
slowtests N="50":
    <run tests, filter by duration > N ms>

Python:

slowtests N="50":
    pytest -v --durations=0 | python -c "import sys; [print(l) for l in sys.stdin if any(int(t.split('.')[0]) > int('{{N}}') for t in l.split() if t.endswith('s'))]"

JavaScript:

slowtests N="50":
    vitest run --reporter=verbose --reporter=json --outputFile=.test-results.json && \
    jq -r '.testResults[].assertionResults[] | select(.duration > {{N}}) | "\(.title): \(.duration)ms"' .test-results.json

Java:

slowtests N="50":
    mvn test | grep "Time elapsed:" | awk -v n={{N}} '{ split($4,a,"."); if(a[1]*1000+a[2] > n) print $0 }'

Key: Identifies slow tests for optimization. Target: unit tests <50ms.

test-profile

Run tests with detailed timing information.

# Profile all tests with detailed timing
test-profile:
    <run all tests with comprehensive timing data>

Python:

test-profile:
    pytest -v --durations=0 --cov=. --cov-report=term-missing

JavaScript:

test-profile:
    vitest run --reporter=verbose --coverage

Java:

test-profile:
    mvn test -X

Key: Full diagnostic timing. Use to analyze test suite performance.

Test Separation Philosophy

Unit tests (unmarked):

• Fast (< 5s total)
• No external dependencies
• 96% coverage threshold
• Block merge if fail
• Run in CI every commit
• Run in pre-commit hook

Integration tests (marked/tagged):

• Slower (> 10s, possibly minutes)
• Require external services
• Coverage reported, no threshold
• Never block merge
• Run in CI, don't fail build
• Optional in pre-commit

Rationale: Quality gates must be fast. Integration tests validate but shouldn't slow feedback loop.

Markers/Tags

Python (pytest.ini):

[pytest]
markers =
    integration: Integration tests requiring external services
    unit: Fast unit tests (default, unmarked)

JavaScript (file naming):

src/lib/**/*.test.ts          # Unit tests
src/components/**/*.test.tsx  # Component tests
tests/integration/**/*.spec.ts # Integration tests (Playwright)

Java (JUnit tags):

@Tag("integration")
@Test
void testDatabaseConnection() { }

Coverage Thresholds

Unit tests: 96% (blocks merge) Integration tests: Report only (no threshold)

Why 96%?

• 100% impractical (edge cases, defensive code)
• 90% too permissive
• 96% forces thinking, allows rare exceptions

When to Add Level 1

Add when:

• Setting up CI/CD
• Multiple developers
• Need fast local feedback (test-watch)
• Integration tests exist (separate from unit)
• Identifying code quality issues (complexity, duplicates)
• Optimizing test suite performance (slowtests, test-profile)

Skip when:

• Solo project, no CI
• No integration tests yet
• Test suite already fast (<5s)
• Prefer simplicity over thoroughness