Qt Coverage Workflow

Coverage-driven test generation is a loop: run tests with instrumentation → generate report → identify gaps → generate targeted tests → re-run to verify improvement. This skill covers the full loop for both Python and C++ Qt projects.

The Coverage Loop

run instrumented tests
        ↓
parse coverage report (gaps list)
        ↓
send gaps to Claude / test-generator agent
        ↓
generate targeted tests
        ↓
run tests again → verify delta
        ↓
repeat until threshold met

Use /qt:coverage to execute this loop. The test-generator agent activates automatically after /qt:coverage identifies gaps.

Python Projects (coverage.py)

Full Python coverage walkthrough — see references/python-coverage-workflow.md for installation, all report formats, branch coverage, CI integration, and agent-handoff parsing patterns.

Key CI step pattern:

- name: Run coverage
  run: pytest --cov=myapp --cov-report=xml --cov-fail-under=80 tests/

C++ Projects (gcov + lcov)

Full gcov/lcov walkthrough — see references/gcov-lcov-workflow.md for CMake presets, the complete lcov command sequence, Clang/LLVM alternative, gap parsing, and troubleshooting.

Coverage Thresholds

Configure thresholds in .qt-test.json:

{
  "coverage_threshold": 80,
  "coverage_exclude": ["tests/*", "*/migrations/*"]
}

| Threshold | When appropriate | |---|---| | 60–70% | Early-stage projects, rapid prototyping | | 80% | General production code (recommended default) | | 90%+ | Safety-critical components | | 100% MC/DC | Aerospace/automotive (requires Coco) |

Identifying High-Value Coverage Gaps

When analyzing gaps, prioritize:

1. Business logic classes — highest risk of regression
2. Error paths (exception handlers, validation failures) — often untested
3. Complex conditionals — branches with multiple conditions
4. Public API methods — surface area for other code to depend on
5. Skip test infrastructure, generated moc_* files, pure UI glue code

Handoff to test-generator Agent

After identifying gaps, structure the handoff:

Gaps found in calculator.py: lines 18-22 (divide by zero path), line 45 (overflow check)
Gaps found in formatter.py: lines 8-10 (empty string handling)
Current coverage: 74%. Target: 80%.
Generate tests targeting these specific lines.

The test-generator agent activates automatically after /qt:coverage completes and gaps are found.

Additional Resources

• references/gcov-lcov-workflow.md — Full gcov/lcov command reference, CMake preset patterns, troubleshooting
• references/python-coverage-workflow.md — coverage.py configuration, branch coverage, parallel test runs
• templates/qt-coverage.yml — Ready-to-use GitHub Actions workflow (Python + C++ variants)
• templates/run-coverage.sh — Portable shell script for local and generic CI use