thkt/use-workflow-code

Workflow: /code

Workflows

| Command | Workflow Reference | | ------- | ----------------------------------------------- | | /code | ${CLAUDE_SKILL_DIR}/references/code-workflow.md |

Patterns

| Pattern | Reference | | ------- | ------------------------------------------------------ | | TDD | ${CLAUDE_SKILL_DIR}/../use-workflow-tdd-cycle/SKILL.md |

<!-- canonical: rules/development/THRESHOLDS.md (coverage targets) -->

Quality Gates

| Gate | Target | Verification | | ------------ | --------------------- | ---------------------------------- | | Tests | All passing | npm test exit code 0 | | Lint | 0 errors | npm run lint exit code 0 | | Types | No errors | tsc --noEmit exit code 0 | | Coverage | C0 ≥90%, C1 ≥80% | Coverage report | | Test Quality | per-metric thresholds | evaluator-test (skip if no Spec) |

Test Quality Gate

When a Spec with Test Scenarios exists, spawn evaluator-test as a background agent using this invocation.

Agent(subagent_type: "evaluator-test",
      prompt: "spec_path: <path>\ntest_paths: <paths>",
      run_in_background: true)

Pass when all 5 metrics meet thresholds. On any fail, report findings (uncovered T-NNN, excess tests, duplicates, granularity issues, intent issues) and fix before proceeding. Skip when no Spec exists (e.g., /fix, ad-hoc changes).

| Metric | Threshold | | ----------- | --------- | | coverage | ≥0.8 | | excess | ≤0.1 | | duplication | ≤0.2 | | granularity | ≥0.7 | | intent | ≥0.7 |

Review Gate

After RGRC cycles, spawn reviewer-readability as a background agent:

Agent(subagent_type: "reviewer-readability",
      prompt: "Review files changed in this session: <paths>",
      run_in_background: true)

High severity → fix before Quality Gates. Medium/low → advisory. Skip for /fix and single-file changes.

Gate Result Output

Tests:        pass | fail (detail)
Lint:         pass | fail (detail)
Types:        pass | fail (detail)
Coverage:     C0 XX% / C1 XX% - pass | fail
Test Quality: cov=X.X exc=X.X dup=X.X gran=X.X int=X.X | pass | fail | skip (no Spec)

All 5 lines required. Empty lines indicate a skipped gate - investigate before proceeding.

Rationalization Counters

| Excuse | Counter | | ---------------------------------------- | -------------------------------------------------------------- | | "Tests pass, lint can wait" | Lint errors are tech debt. Zero errors before commit | | "Type errors are just warnings" | tsc --noEmit exit 0 or no ship. Type warnings are errors | | "Coverage is close enough" | "Close enough" is failure with extra steps. Meet the threshold | | "This gate doesn't apply to this change" | All 4 gates apply to every change. No exceptions |