Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Every product will be AI-powered. The question is whether you'll build it right or ship a demo that falls apart in production. This skill covers LLM integration patterns, RAG architecture, prompt engineering that scales, AI UX that users trust, and cost optimization that doesn't bankrupt you. Use when: keywords, file_patterns, code_patterns.
.claude/skills/dokhacgiakhoa-ai-product/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -8% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 17% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 14% | 0% |
You are an AI product engineer who has shipped LLM features to millions of users. You've debugged hallucinations at 3am, optimized prompts to reduce costs by 80%, and built safety systems that caught thousands of harmful outputs. You know that demos are easy and production is hard. You treat prompts as code, validate all outputs, and never trust an LLM blindly.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 19,195 | 40,593 | +111% | 1 | 1 | 0% | 2,777 | 2,550 | -8% | 0 | 0 | — |
case-02 | fail→pass | 15,253 | 18,112 | +19% | 1 | 1 | 0% | 2,597 | 2,985 | +15% | 0 | 0 | — |
case-03 | pass→pass | 15,822 | 13,953 | -12% | 1 | 1 | 0% | 2,166 | 2,537 | +17% | 0 | 0 | — |
case-04 | pass→pass | 19,517 | 16,809 | -14% | 1 | 1 | 0% | 2,727 | 3,119 | +14% | 0 | 0 | — |
case-05 | pass→pass | 16,599 | 21,252 | +28% | 1 | 1 | 0% | 2,745 | 3,447 | +26% | 0 | 0 | — |
case-06 | pass→pass | 19,570 | 20,200 | +3% | 1 | 1 | 0% | 2,709 | 3,150 | +16% | 0 | 0 | — |
case-07 | pass→pass | 13,681 | 17,219 | +26% | 1 | 1 | 0% | 2,027 | 2,772 | +37% | 0 | 0 | — |
case-08 | pass→pass | 12,072 | 14,462 | +20% | 1 | 1 | 0% | 2,082 | 2,664 | +28% | 0 | 0 | — |
case-09 | fail→fail | 18,836 | 21,791 | +16% | 1 | 1 | 0% | 2,966 | 3,895 | +31% | 0 | 0 | — |
case-10 | pass→pass | 18,648 | 19,074 | +2% | 1 | 1 | 0% | 2,715 | 3,293 | +21% | 0 | 0 | — |
case-11 | pass→pass | 18,439 | 16,283 | -12% | 1 | 1 | 0% | 2,934 | 2,960 | +1% | 0 | 0 | — |
case-12 | pass→pass | 13,425 | 13,899 | +4% | 1 | 1 | 0% | 2,384 | 2,561 | +7% | 0 | 0 | — |
case-13 | pass→pass | 13,073 | 13,505 | +3% | 1 | 1 | 0% | 2,593 | 2,771 | +7% | 0 | 0 | — |
case-14 | pass→pass | 13,281 | 17,781 | +34% | 1 | 1 | 0% | 2,330 | 2,957 | +27% | 0 | 0 | — |
case-15 | pass→pass | 17,036 | 14,349 | -16% | 1 | 1 | 0% | 2,257 | 2,624 | +16% | 0 | 0 | — |
case-16 | fail→pass | 21,945 | 18,552 | -15% | 1 | 1 | 0% | 3,067 | 3,235 | +5% | 0 | 0 | — |
case-17 | pass→pass | 15,594 | 14,829 | -5% | 1 | 1 | 0% | 2,376 | 2,327 | -2% | 0 | 0 | — |
case-18 | pass→pass | 15,546 | 16,289 | +5% | 1 | 1 | 0% | 2,561 | 2,763 | +8% | 0 | 0 | — |
case-19 | pass→pass | 25,843 | 24,005 | -7% | 1 | 1 | 0% | 3,769 | 3,818 | +1% | 0 | 0 | — |
case-20 | pass→pass | 14,876 | 15,277 | +3% | 1 | 1 | 0% | 2,526 | 2,536 | +0% | 0 | 0 | — |
case-21 | pass→pass | 9,485 | 6,384 | -33% | 1 | 1 | 0% | 1,759 | 1,521 | -14% | 0 | 0 | — |
case-22 | pass→pass | 12,886 | 14,577 | +13% | 1 | 1 | 0% | 2,499 | 2,863 | +15% | 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.
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.