Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →FAQ identification from support tickets, step-by-step tutorial creation, screenshot/video script guidance, search optimization, and self-service deflection tracking.
.claude/skills/onewave-ai-knowledge-base-builder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 11% | 0% |
FAQ identification from support tickets, step-by-step tutorial creation, screenshot/video script guidance, search optimization, and self-service deflection tracking.
You are an expert knowledge management specialist. Analyze support tickets to identify recurring questions and create comprehensive knowledge base articles. Generate: FAQ sections, step-by-step tutorials with screenshots placement, video script guidance, search-optimized content, and self-service deflection metrics. Focus on reducing ticket volume through excellent documentation.
markdown# Knowledge Base Builder Output **Generated**: {timestamp} --- ## Results [Your formatted output here] --- ## Recommendations [Actionable next steps]
Trigger Phrases:
Example Request: > "Sample user request here]"
Response Approach:
Remember: Focus on delivering value quickly and clearly!
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 20,390 | 21,700 | +6% | 1 | 1 | 0% | 3,308 | 2,968 | -10% | 0 | 0 | — |
case-02 | fail→pass | 16,844 | 15,038 | -11% | 1 | 1 | 0% | 2,748 | 2,800 | +2% | 0 | 0 | — |
case-03 | fail→pass | 20,011 | 18,040 | -10% | 1 | 1 | 0% | 2,787 | 3,037 | +9% | 0 | 0 | — |
case-04 | fail→pass | 32,563 | 24,239 | -26% | 1 | 1 | 0% | 4,943 | 3,943 | -20% | 0 | 0 | — |
case-05 | fail→pass | 18,672 | 19,046 | +2% | 1 | 1 | 0% | 3,097 | 3,446 | +11% | 0 | 0 | — |
case-06 | fail→pass | 11,366 | 15,295 | +35% | 1 | 1 | 0% | 2,036 | 2,891 | +42% | 0 | 0 | — |
case-07 | fail→pass | 17,589 | 18,651 | +6% | 1 | 1 | 0% | 2,924 | 3,736 | +28% | 0 | 0 | — |
case-08 | fail→pass | 7,977 | 12,295 | +54% | 1 | 1 | 0% | 1,566 | 2,609 | +67% | 0 | 0 | — |
case-09 | fail→pass | 24,552 | 18,596 | -24% | 1 | 1 | 0% | 4,272 | 3,655 | -14% | 0 | 0 | — |
case-10 | fail→pass | 18,845 | 17,110 | -9% | 1 | 1 | 0% | 2,958 | 2,977 | +1% | 0 | 0 | — |
case-11 | fail→pass | 16,617 | 16,537 | -0% | 1 | 1 | 0% | 2,861 | 3,031 | +6% | 0 | 0 | — |
case-12 | pass→pass | 18,641 | 18,511 | -1% | 1 | 1 | 0% | 3,261 | 3,572 | +10% | 0 | 0 | — |
case-13 | fail→pass | 13,527 | 21,668 | +60% | 1 | 1 | 0% | 2,367 | 4,493 | +90% | 0 | 0 | — |
case-14 | pass→pass | 12,254 | 15,648 | +28% | 1 | 1 | 0% | 2,171 | 3,054 | +41% | 0 | 0 | — |
case-15 | pass→pass | 13,886 | 14,708 | +6% | 1 | 1 | 0% | 2,485 | 3,009 | +21% | 0 | 0 | — |
case-16 | fail→pass | 15,488 | 14,066 | -9% | 1 | 1 | 0% | 2,728 | 2,906 | +7% | 0 | 0 | — |
case-17 | fail→pass | 17,962 | 20,895 | +16% | 1 | 1 | 0% | 3,415 | 4,301 | +26% | 0 | 0 | — |
case-18 | fail→pass | 12,098 | 15,667 | +30% | 1 | 1 | 0% | 2,148 | 3,045 | +42% | 0 | 0 | — |
case-19 | fail→fail | 18,803 | 20,897 | +11% | 1 | 1 | 0% | 2,999 | 4,195 | +40% | 0 | 0 | — |
case-20 | pass→pass | 17,596 | 26,974 | +53% | 1 | 1 | 0% | 3,230 | 4,385 | +36% | 0 | 0 | — |
case-21 | pass→fail | 9,654 | 12,739 | +32% | 1 | 1 | 0% | 1,986 | 2,968 | +49% | 0 | 0 | — |
case-22 | pass→pass | 10,528 | 13,054 | +24% | 1 | 1 | 0% | 1,870 | 2,454 | +31% | 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 +64 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.