Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Grafana、SigNoz、および同様のプラットフォーム用の実際のオペレータ質問に答える監視ダッシュボードを構築します。メトリクスを虚栄ボードではなく機能するダッシュボードに変える場合に使用します。
.claude/skills/affaan-m-dashboard-builder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-21 | ✓→✗ | ▼ Worse | 18% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 9% | 0% |
当任务需要构建一个可供操作人员使用的仪表盘时使用此方案。
目标不是"展示所有指标"，而是回答以下问题：
围绕以下方面组织：
首先检查现有仪表盘：
推荐结构：
每个面板都应回答一个真实问题。如果不能，则移除。
research-opsbackend-patternsterminal-ops| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 25,902 | 27,003 | +4% | 1 | 1 | 0% | 6,198 | 6,748 | +9% | 0 | 0 | — |
case-02 | pass→pass | 16,430 | 28,236 | +72% | 1 | 1 | 0% | 2,748 | 6,750 | +146% | 0 | 0 | — |
case-03 | fail→pass | 21,391 | 22,522 | +5% | 1 | 1 | 0% | 3,403 | 4,877 | +43% | 0 | 0 | — |
case-04 | fail→pass | 15,286 | 20,985 | +37% | 1 | 1 | 0% | 2,554 | 3,196 | +25% | 0 | 0 | — |
case-05 | pass→pass | 32,153 | 17,642 | -45% | 1 | 1 | 0% | 2,995 | 3,300 | +10% | 0 | 0 | — |
case-06 | pass→pass | 14,697 | 12,833 | -13% | 1 | 1 | 0% | 2,250 | 2,518 | +12% | 0 | 0 | — |
case-07 | pass→pass | 17,760 | 14,800 | -17% | 1 | 1 | 0% | 3,346 | 2,967 | -11% | 0 | 0 | — |
case-08 | fail→pass | 19,915 | 16,293 | -18% | 1 | 1 | 0% | 3,269 | 3,433 | +5% | 0 | 0 | — |
case-09 | pass→pass | 15,365 | 15,308 | -0% | 1 | 1 | 0% | 2,403 | 2,880 | +20% | 0 | 0 | — |
case-10 | pass→pass | 10,435 | 10,973 | +5% | 1 | 1 | 0% | 1,705 | 2,089 | +23% | 0 | 0 | — |
case-11 | pass→pass | 12,711 | 9,451 | -26% | 1 | 1 | 0% | 1,950 | 2,190 | +12% | 0 | 0 | — |
case-12 | pass→pass | 3,247 | 5,887 | +81% | 1 | 1 | 0% | 579 | 1,601 | +177% | 0 | 0 | — |
case-13 | pass→pass | 8,460 | 8,442 | -0% | 1 | 1 | 0% | 1,555 | 2,217 | +43% | 0 | 0 | — |
case-14 | pass→pass | 5,114 | 7,617 | +49% | 1 | 1 | 0% | 948 | 1,437 | +52% | 0 | 0 | — |
case-15 | pass→pass | 10,410 | 8,428 | -19% | 1 | 1 | 0% | 1,692 | 1,997 | +18% | 0 | 0 | — |
case-16 | pass→pass | 17,218 | 11,467 | -33% | 1 | 1 | 0% | 2,564 | 2,349 | -8% | 0 | 0 | — |
case-17 | pass→pass | 13,532 | 10,799 | -20% | 1 | 1 | 0% | 2,041 | 2,170 | +6% | 0 | 0 | — |
case-18 | pass→pass | 19,532 | 21,568 | +10% | 1 | 1 | 0% | 3,322 | 4,498 | +35% | 0 | 0 | — |
case-19 | fail→fail | 6,451 | 3,051 | -53% | 1 | 1 | 0% | 879 | 1,002 | +14% | 0 | 0 | — |
case-20 | pass→pass | 12,884 | 12,180 | -5% | 1 | 1 | 0% | 1,935 | 2,448 | +27% | 0 | 0 | — |
case-21 | pass→fail | 16,800 | 17,461 | +4% | 1 | 1 | 0% | 2,811 | 3,328 | +18% | 0 | 0 | — |
case-22 | pass→pass | 13,265 | 11,064 | -17% | 1 | 1 | 0% | 2,191 | 2,245 | +2% | 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 +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.