Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when working with full stack orchestration full stack feature
.claude/skills/dokhacgiakhoa-full-stack-orchestration-full-stack-feature/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 6% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 3% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 2% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | -25% | 0% |
resources/implementation-playbook.md.Orchestrate full-stack feature development across backend, frontend, and infrastructure layers with modern API-first approach:
Extended thinking: This workflow coordinates multiple specialized agents to deliver a complete full-stack feature from architecture through deployment. It follows API-first development principles, ensuring contract-driven development where the API specification drives both backend implementation and frontend consumption. Each phase builds upon previous outputs, creating a cohesive system with proper separation of concerns, comprehensive testing, and production-ready deployment. The workflow emphasizes modern practices like component-driven UI development, feature flags, observability, and progressive rollout strategies.]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→pass | 18,364 | 12,516 | -32% | 1 | 1 | 0% | 2,541 | 2,692 | +6% | 0 | 0 | — |
case-05 | pass→pass | 12,499 | 7,423 | -41% | 1 | 1 | 0% | 1,800 | 1,850 | +3% | 0 | 0 | — |
case-06 | pass→pass | 12,612 | 10,590 | -16% | 1 | 1 | 0% | 2,011 | 2,046 | +2% | 0 | 0 | — |
case-01 | fail→fail | 42,036 | 30,860 | -27% | 1 | 1 | 0% | 8,238 | 5,453 | -34% | 0 | 0 | — |
case-02 | fail→fail | 34,184 | 22,945 | -33% | 1 | 1 | 0% | 6,655 | 4,797 | -28% | 0 | 0 | — |
case-03 | fail→fail | 40,528 | 30,464 | -25% | 1 | 1 | 0% | 8,229 | 7,199 | -13% | 0 | 0 | — |
case-07 | pass→pass | 24,097 | 11,023 | -54% | 1 | 1 | 0% | 2,791 | 2,087 | -25% | 0 | 0 | — |
case-08 | pass→pass | 12,819 | 11,120 | -13% | 1 | 1 | 0% | 1,976 | 2,118 | +7% | 0 | 0 | — |
case-09 | pass→pass | 7,777 | 7,741 | -0% | 1 | 1 | 0% | 1,014 | 1,875 | +85% | 0 | 0 | — |
case-10 | pass→pass | 8,506 | 6,971 | -18% | 1 | 1 | 0% | 1,138 | 1,543 | +36% | 0 | 0 | — |
case-11 | pass→pass | 9,527 | 3,770 | -60% | 1 | 1 | 0% | 1,500 | 1,127 | -25% | 0 | 0 | — |
case-12 | pass→pass | 9,009 | 5,170 | -43% | 1 | 1 | 0% | 1,382 | 1,337 | -3% | 0 | 0 | — |
case-13 | pass→pass | 7,515 | 5,870 | -22% | 1 | 1 | 0% | 1,176 | 1,512 | +29% | 0 | 0 | — |
case-14 | pass→pass | 15,781 | 12,209 | -23% | 1 | 1 | 0% | 2,329 | 2,271 | -2% | 0 | 0 | — |
case-15 | fail→fail | 21,754 | 34,619 | +59% | 1 | 1 | 0% | 3,356 | 6,317 | +88% | 0 | 0 | — |
case-16 | fail→fail | 29,290 | 34,787 | +19% | 1 | 1 | 0% | 5,924 | 6,307 | +6% | 0 | 0 | — |
case-17 | pass→pass | 15,733 | 18,719 | +19% | 1 | 1 | 0% | 2,455 | 3,575 | +46% | 0 | 0 | — |
case-18 | pass→pass | 19,155 | 15,289 | -20% | 1 | 1 | 0% | 2,637 | 3,051 | +16% | 0 | 0 | — |
case-19 | fail→pass | 10,043 | 2,078 | -79% | 1 | 1 | 0% | 1,355 | 822 | -39% | 0 | 0 | — |
case-20 | pass→pass | 10,755 | 4,534 | -58% | 1 | 1 | 0% | 1,686 | 1,333 | -21% | 0 | 0 | — |
case-21 | pass→pass | 13,364 | 9,295 | -30% | 1 | 1 | 0% | 1,765 | 1,892 | +7% | 0 | 0 | — |
case-22 | pass→pass | 14,101 | 11,966 | -15% | 1 | 1 | 0% | 2,292 | 2,706 | +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 +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.