Software Quality Workflows
Use this umbrella for implementation-quality loops: test-first work, systematic debugging, short spikes, simplification passes, debugger sessions, and pre-commit review.
Pick the mode
| Situation |
Mode |
| New behavior or bug fix with clear expected output |
TDD: RED → GREEN → REFACTOR. |
| Unknown root cause |
Systematic debugging: reproduce, localize, hypothesize, fix, verify. |
| Uncertain approach or API |
Spike: time-boxed throwaway experiment before production changes. |
| Recent change is too complex |
Simplify-code: parallel or focused cleanup with behavior preserved. |
| Ready to hand off/commit |
Requesting code review: security scan, quality gates, auto-fix only when safe. |
| Runtime behavior is opaque |
Debugger-assisted investigation via pdb/debugpy or Node inspector. |
TDD loop
- Write or identify a failing test that captures the requirement.
- Run only the relevant test to see RED.
- Implement the smallest fix.
- Run focused then broader tests.
- Refactor while tests stay GREEN.
Systematic debugging loop
- Reproduce with exact command/input.
- Observe logs, state, failing assertions, and environment.
- Narrow the fault with instrumentation or debugger probes.
- Fix the root cause, not the symptom.
- Verify the original failure and a regression test.
Spikes
Create disposable experiments under a clearly temporary path. Do not mix spike code into production without reimplementation or review. End with a decision: adopt, reject, or investigate further.
Simplification and review
Before committing, inspect diff shape, remove incidental complexity, check security-sensitive paths, and run project gates. If using subagents for cleanup/review, independently verify their claims.
Debugger notes
- Python: use
pdb for local stepping and debugpy when a DAP/remote attach flow is needed.
- Node: start with
node --inspect/--inspect-brk and inspect through Chrome DevTools Protocol when logs are insufficient.
1---2name: software-quality-workflows3description: Umbrella workflow for TDD, debugging, code simplification, spikes, pre-commit review, and debugger-assisted investigation.4---56# Software Quality Workflows78Use this umbrella for implementation-quality loops: test-first work, systematic debugging, short spikes, simplification passes, debugger sessions, and pre-commit review.910## Pick the mode1112| Situation | Mode |13|---|---|14| New behavior or bug fix with clear expected output | TDD: RED → GREEN → REFACTOR. |15| Unknown root cause | Systematic debugging: reproduce, localize, hypothesize, fix, verify. |16| Uncertain approach or API | Spike: time-boxed throwaway experiment before production changes. |17| Recent change is too complex | Simplify-code: parallel or focused cleanup with behavior preserved. |18| Ready to hand off/commit | Requesting code review: security scan, quality gates, auto-fix only when safe. |19| Runtime behavior is opaque | Debugger-assisted investigation via `pdb`/`debugpy` or Node inspector. |2021## TDD loop22231. Write or identify a failing test that captures the requirement.242. Run only the relevant test to see RED.253. Implement the smallest fix.264. Run focused then broader tests.275. Refactor while tests stay GREEN.2829## Systematic debugging loop30311. Reproduce with exact command/input.322. Observe logs, state, failing assertions, and environment.333. Narrow the fault with instrumentation or debugger probes.344. Fix the root cause, not the symptom.355. Verify the original failure and a regression test.3637## Spikes3839Create disposable experiments under a clearly temporary path. Do not mix spike code into production without reimplementation or review. End with a decision: adopt, reject, or investigate further.4041## Simplification and review4243Before committing, inspect diff shape, remove incidental complexity, check security-sensitive paths, and run project gates. If using subagents for cleanup/review, independently verify their claims.4445## Debugger notes4647- Python: use `pdb` for local stepping and `debugpy` when a DAP/remote attach flow is needed.48- Node: start with `node --inspect`/`--inspect-brk` and inspect through Chrome DevTools Protocol when logs are insufficient.