Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Build features guided by data insights, A/B testing, and continuous measurement using specialized agents for analysis, implementation, and experimentation.
.claude/skills/dokhacgiakhoa-data-engineering-data-driven-feature/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 54% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -1% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
Build features guided by data insights, A/B testing, and continuous measurement using specialized agents for analysis, implementation, and experimentation.
Extended thinking: This workflow orchestrates a comprehensive data-driven development process from initial data analysis and hypothesis formulation through feature implementation with integrated analytics, A/B testing infrastructure, and post-launch analysis. Each phase leverages specialized agents to ensure features are built based on data insights, properly instrumented for measurement, and validated through controlled experiments. The workflow emphasizes modern product analytics practices, statistical rigor in testing, and continuous learning from user behavior.]
resources/implementation-playbook.md.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 17,108 | 16,915 | -1% | 1 | 1 | 0% | 2,400 | 3,684 | +54% | 0 | 0 | — |
case-03 | pass→pass | 41,983 | 35,445 | -16% | 1 | 1 | 0% | 6,041 | 7,170 | +19% | 0 | 0 | — |
case-01 | fail→fail | 49,484 | 49,290 | -0% | 1 | 1 | 0% | 8,261 | 7,003 | -15% | 0 | 0 | — |
case-02 | pass→pass | 36,862 | 32,648 | -11% | 1 | 1 | 0% | 6,490 | 6,420 | -1% | 0 | 0 | — |
case-04 | pass→pass | 41,232 | 38,210 | -7% | 1 | 1 | 0% | 6,184 | 6,870 | +11% | 0 | 0 | — |
case-05 | pass→pass | 18,234 | 16,171 | -11% | 1 | 1 | 0% | 2,759 | 3,379 | +22% | 0 | 0 | — |
case-06 | pass→pass | 18,147 | 22,127 | +22% | 1 | 1 | 0% | 2,904 | 3,617 | +25% | 0 | 0 | — |
case-07 | fail→pass | 25,837 | 4,799 | -81% | 1 | 1 | 0% | 4,650 | 1,544 | -67% | 0 | 0 | — |
case-08 | pass→pass | 15,526 | 9,955 | -36% | 1 | 1 | 0% | 2,002 | 2,030 | +1% | 0 | 0 | — |
case-10 | pass→pass | 13,517 | 20,854 | +54% | 1 | 1 | 0% | 2,133 | 3,609 | +69% | 0 | 0 | — |
case-11 | pass→pass | 15,388 | 16,034 | +4% | 1 | 1 | 0% | 2,678 | 2,879 | +8% | 0 | 0 | — |
case-12 | pass→pass | 14,126 | 12,558 | -11% | 1 | 1 | 0% | 1,856 | 2,501 | +35% | 0 | 0 | — |
case-13 | pass→pass | 12,481 | 10,255 | -18% | 1 | 1 | 0% | 1,767 | 2,471 | +40% | 0 | 0 | — |
case-14 | pass→pass | 14,245 | 11,501 | -19% | 1 | 1 | 0% | 2,083 | 2,544 | +22% | 0 | 0 | — |
case-15 | pass→pass | 18,293 | 14,080 | -23% | 1 | 1 | 0% | 2,250 | 2,826 | +26% | 0 | 0 | — |
case-16 | pass→pass | 17,447 | 17,036 | -2% | 1 | 1 | 0% | 2,446 | 3,064 | +25% | 0 | 0 | — |
case-17 | fail→fail | 9,042 | 7,759 | -14% | 1 | 1 | 0% | 1,302 | 1,730 | +33% | 0 | 0 | — |
case-18 | pass→pass | 14,590 | 13,122 | -10% | 1 | 1 | 0% | 2,255 | 2,491 | +10% | 0 | 0 | — |
case-19 | fail→fail | 20,432 | 22,602 | +11% | 1 | 1 | 0% | 3,021 | 3,799 | +26% | 0 | 0 | — |
case-20 | pass→pass | 16,651 | 45,137 | +171% | 1 | 1 | 0% | 2,338 | 2,763 | +18% | 0 | 0 | — |
case-21 | pass→pass | 14,939 | 19,294 | +29% | 1 | 1 | 0% | 2,487 | 3,864 | +55% | 0 | 0 | — |
case-22 | pass→pass | 14,755 | 13,181 | -11% | 1 | 1 | 0% | 2,458 | 3,465 | +41% | 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 +5 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.