Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Run the full Kubernetes Operator audit (CRD + reconcile + capability) on the current repo
.claude/skills/alirezarezvani-operator-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 19% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -36% | 0% |
Run the full audit on a Kubernetes Operator repository:
/operator-audit
/operator-audit --operator-dir ./my-operator
/operator-audit --crd-dir ./config/crd --controller-dir ./controllersbashSKILL=engineering/kubernetes-operator/skills/kubernetes-operator DIR="${OPERATOR_DIR:-.}" echo "## CRD validation" python "$SKILL/scripts/crd_validator.py" --crd "$DIR/config/crd" || true echo "" echo "## Reconcile lint" python "$SKILL/scripts/reconcile_lint.py" --controller "$DIR/controllers" || python "$SKILL/scripts/reconcile_lint.py" --controller "$DIR/internal/controller" || true echo "" echo "## Capability audit" python "$SKILL/scripts/operator_capability_audit.py" --operator-dir "$DIR"
A markdown report with:
controllers/ or internal/controller/config/crd/ (kubebuilder layout)kubernetes-operator skill installed| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | 23,251 | 6,351 | -73% | 1 | 1 | 0% | 3,996 | 1,245 | -69% | 0 | 0 | — |
case-04 | fail→fail | 20,340 | 3,488 | -83% | 1 | 1 | 0% | 3,547 | 761 | -79% | 0 | 0 | — |
case-01 | fail→pass | 17,991 | 13,071 | -27% | 1 | 1 | 0% | 3,089 | 2,756 | -11% | 0 | 0 | — |
case-02 | fail→fail | 19,567 | 2,134 | -89% | 1 | 1 | 0% | 3,673 | 660 | -82% | 0 | 0 | — |
case-05 | fail→pass | 9,060 | 5,609 | -38% | 1 | 1 | 0% | 1,530 | 921 | -40% | 0 | 0 | — |
case-06 | pass→pass | 10,652 | 24,159 | +127% | 1 | 1 | 0% | 1,857 | 4,655 | +151% | 0 | 0 | — |
case-07 | pass→pass | 10,205 | 6,300 | -38% | 1 | 1 | 0% | 1,836 | 1,513 | -18% | 0 | 0 | — |
case-08 | fail→fail | 11,904 | 8,439 | -29% | 1 | 1 | 0% | 2,092 | 1,722 | -18% | 0 | 0 | — |
case-09 | fail→fail | 18,361 | 7,873 | -57% | 1 | 1 | 0% | 3,218 | 1,841 | -43% | 0 | 0 | — |
case-10 | fail→pass | 12,879 | 13,614 | +6% | 1 | 1 | 0% | 2,367 | 2,823 | +19% | 0 | 0 | — |
case-11 | fail→pass | 15,731 | 10,111 | -36% | 1 | 1 | 0% | 3,087 | 2,389 | -23% | 0 | 0 | — |
case-12 | pass→pass | 11,551 | 8,192 | -29% | 1 | 1 | 0% | 2,011 | 1,819 | -10% | 0 | 0 | — |
case-13 | pass→pass | 8,367 | 2,152 | -74% | 1 | 1 | 0% | 1,334 | 811 | -39% | 0 | 0 | — |
case-14 | pass→pass | 4,071 | 1,442 | -65% | 1 | 1 | 0% | 661 | 606 | -8% | 0 | 0 | — |
case-15 | fail→pass | 10,356 | 3,249 | -69% | 1 | 1 | 0% | 1,758 | 1,126 | -36% | 0 | 0 | — |
case-16 | fail→pass | 13,038 | 9,307 | -29% | 1 | 1 | 0% | 2,112 | 1,980 | -6% | 0 | 0 | — |
case-17 | pass→pass | 11,237 | 1,403 | -88% | 1 | 1 | 0% | 2,023 | 625 | -69% | 0 | 0 | — |
case-18 | fail→fail | 14,317 | 4,770 | -67% | 1 | 1 | 0% | 2,950 | 774 | -74% | 0 | 0 | — |
case-19 | fail→fail | 5,934 | 1,860 | -69% | 1 | 1 | 0% | 1,117 | 746 | -33% | 0 | 0 | — |
case-20 | pass→pass | 3,568 | 2,119 | -41% | 1 | 1 | 0% | 634 | 795 | +25% | 0 | 0 | — |
case-21 | pass→pass | 3,665 | 1,627 | -56% | 1 | 1 | 0% | 639 | 726 | +14% | 0 | 0 | — |
case-22 | pass→pass | 14,387 | 9,460 | -34% | 1 | 1 | 0% | 2,668 | 2,192 | -18% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.