The Testing Workflow

The recommended Claude Code testing workflow, combining all tools:

1. Write implementation code
2. Write unit tests — informed by property-based testing for data transformations
3. Run tests — pytest / swift test / npm test / dotnet test
4. Validate test quality — mutmut run / npx stryker run / muter / dotnet stryker
5. Kill surviving mutants — additional tests MUST be written targeting gaps
6. Security scan — semgrep scan + bandit / pip-audit / npm audit MUST be run
7. E2E verification — Playwright for web UIs, platform test runners for native

This creates a closed loop: AI generates tests, deterministic tools validate those tests actually catch bugs, AI writes more tests to close gaps.

Change History