Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a cloud cost optimization expert specializing in reducing infrastructure expenses while maintaining performance and reliability. Analyze cloud spending, identify savings opportunities, and implement cost-effective architectures across AWS, Azure, and GCP.
.claude/skills/dokhacgiakhoa-database-cloud-optimization-cost-optimize/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 38% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 40% | 0% |
You are a cloud cost optimization expert specializing in reducing infrastructure expenses while maintaining performance and reliability. Analyze cloud spending, identify savings opportunities, and implement cost-effective architectures across AWS, Azure, and GCP.
The user needs to optimize cloud infrastructure costs without compromising performance or reliability. Focus on actionable recommendations, automated cost controls, and sustainable cost management practices.
$ARGUMENTS
resources/implementation-playbook.md.resources/implementation-playbook.md for detailed cost analysis and tooling.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→fail | 20,320 | 22,534 | +11% | 1 | 1 | 0% | 3,871 | 4,662 | +20% | 0 | 0 | — |
case-01 | fail→fail | 23,065 | 27,913 | +21% | 1 | 1 | 0% | 3,871 | 3,700 | -4% | 0 | 0 | — |
case-02 | pass→pass | 17,837 | 9,829 | -45% | 1 | 1 | 0% | 2,000 | 1,754 | -12% | 0 | 0 | — |
case-03 | pass→pass | 16,324 | 8,156 | -50% | 1 | 1 | 0% | 2,444 | 1,549 | -37% | 0 | 0 | — |
case-05 | fail→fail | 16,124 | 22,406 | +39% | 1 | 1 | 0% | 2,389 | 3,962 | +66% | 0 | 0 | — |
case-06 | fail→pass | 17,676 | 15,712 | -11% | 1 | 1 | 0% | 2,480 | 3,050 | +23% | 0 | 0 | — |
case-07 | pass→pass | 15,883 | 11,946 | -25% | 1 | 1 | 0% | 2,322 | 2,104 | -9% | 0 | 0 | — |
case-08 | pass→pass | 18,861 | 23,160 | +23% | 1 | 1 | 0% | 3,299 | 3,890 | +18% | 0 | 0 | — |
case-09 | fail→fail | 19,282 | 21,178 | +10% | 1 | 1 | 0% | 2,880 | 3,441 | +19% | 0 | 0 | — |
case-10 | pass→pass | 17,854 | 25,171 | +41% | 1 | 1 | 0% | 3,535 | 3,995 | +13% | 0 | 0 | — |
case-11 | fail→fail | 39,128 | 25,268 | -35% | 1 | 1 | 0% | 5,621 | 4,615 | -18% | 0 | 0 | — |
case-12 | fail→pass | 19,998 | 22,277 | +11% | 1 | 1 | 0% | 2,960 | 3,431 | +16% | 0 | 0 | — |
case-13 | fail→fail | 17,740 | 21,387 | +21% | 1 | 1 | 0% | 2,800 | 3,547 | +27% | 0 | 0 | — |
case-14 | pass→fail | 18,173 | 20,743 | +14% | 1 | 1 | 0% | 2,930 | 3,730 | +27% | 0 | 0 | — |
case-15 | fail→pass | 24,226 | 33,584 | +39% | 1 | 1 | 0% | 3,926 | 5,431 | +38% | 0 | 0 | — |
case-16 | fail→fail | 18,239 | 15,079 | -17% | 1 | 1 | 0% | 2,719 | 2,807 | +3% | 0 | 0 | — |
case-17 | fail→fail | 17,496 | 17,998 | +3% | 1 | 1 | 0% | 3,128 | 3,745 | +20% | 0 | 0 | — |
case-18 | fail→fail | 46,958 | 38,259 | -19% | 1 | 1 | 0% | 8,168 | 5,969 | -27% | 0 | 0 | — |
case-19 | fail→pass | 16,820 | 20,410 | +21% | 1 | 1 | 0% | 2,446 | 3,112 | +27% | 0 | 0 | — |
case-20 | fail→pass | 17,648 | 21,231 | +20% | 1 | 1 | 0% | 3,244 | 4,545 | +40% | 0 | 0 | — |
case-21 | fail→fail | 16,410 | 18,834 | +15% | 1 | 1 | 0% | 2,546 | 3,577 | +40% | 0 | 0 | — |
case-22 | fail→fail | 14,266 | 13,000 | -9% | 1 | 1 | 0% | 2,438 | 2,821 | +16% | 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 +14 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
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.