Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
.claude/skills/dokhacgiakhoa-skill-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -54% | 0% |
This skill provides guidance for creating effective skills.
Skills are modular, self-contained packages that extend Claude's capabilities by providing specialized knowledge, workflows, and tools. Think of them as "onboarding guides" for specific domains or tasks—they transform Claude from a general-purpose agent into a specialized agent equipped with procedural knowledge that no model can fully possess.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 26,606 | 19,799 | -26% | 1 | 1 | 0% | 4,278 | 2,904 | -32% | 0 | 0 | — |
case-02 | fail→fail | 18,800 | 9,960 | -47% | 1 | 1 | 0% | 2,234 | 1,218 | -45% | 0 | 0 | — |
case-03 | pass→pass | 20,068 | 24,062 | +20% | 1 | 1 | 0% | 3,415 | 3,831 | +12% | 0 | 0 | — |
case-04 | pass→pass | 18,947 | 20,880 | +10% | 1 | 1 | 0% | 2,971 | 3,820 | +29% | 0 | 0 | — |
case-05 | pass→pass | 25,025 | 18,997 | -24% | 1 | 1 | 0% | 4,295 | 3,526 | -18% | 0 | 0 | — |
case-06 | pass→pass | 16,514 | 16,353 | -1% | 1 | 1 | 0% | 2,062 | 2,383 | +16% | 0 | 0 | — |
case-07 | pass→pass | 18,793 | 16,743 | -11% | 1 | 1 | 0% | 2,767 | 2,371 | -14% | 0 | 0 | — |
case-08 | pass→pass | 19,063 | 9,681 | -49% | 1 | 1 | 0% | 2,092 | 1,859 | -11% | 0 | 0 | — |
case-09 | pass→pass | 17,674 | 11,221 | -37% | 1 | 1 | 0% | 2,184 | 1,460 | -33% | 0 | 0 | — |
case-10 | fail→fail | 18,253 | 9,904 | -46% | 1 | 1 | 0% | 2,116 | 1,321 | -38% | 0 | 0 | — |
case-11 | fail→pass | 9,995 | 7,543 | -25% | 1 | 1 | 0% | 812 | 844 | +4% | 0 | 0 | — |
case-12 | pass→pass | 7,498 | 3,271 | -56% | 1 | 1 | 0% | 1,202 | 934 | -22% | 0 | 0 | — |
case-13 | fail→pass | 14,150 | 7,629 | -46% | 1 | 1 | 0% | 1,604 | 966 | -40% | 0 | 0 | — |
case-14 | fail→pass | 19,713 | 2,327 | -88% | 1 | 1 | 0% | 2,486 | 806 | -68% | 0 | 0 | — |
case-15 | fail→pass | 15,632 | 2,251 | -86% | 1 | 1 | 0% | 1,744 | 834 | -52% | 0 | 0 | — |
case-16 | fail→pass | 15,298 | 2,149 | -86% | 1 | 1 | 0% | 1,732 | 795 | -54% | 0 | 0 | — |
case-17 | pass→pass | 14,086 | 7,071 | -50% | 1 | 1 | 0% | 1,499 | 721 | -52% | 0 | 0 | — |
case-18 | pass→pass | 12,846 | 12,597 | -2% | 1 | 1 | 0% | 2,124 | 1,716 | -19% | 0 | 0 | — |
case-19 | pass→pass | 18,273 | 13,512 | -26% | 1 | 1 | 0% | 2,133 | 1,753 | -18% | 0 | 0 | — |
case-20 | pass→pass | 12,624 | 8,624 | -32% | 1 | 1 | 0% | 2,104 | 1,816 | -14% | 0 | 0 | — |
case-21 | pass→pass | 17,599 | 12,289 | -30% | 1 | 1 | 0% | 2,115 | 1,605 | -24% | 0 | 0 | — |
case-22 | pass→pass | 17,537 | 10,826 | -38% | 1 | 1 | 0% | 1,985 | 1,280 | -36% | 0 | 0 | — |
case-23 | fail→pass | 11,530 | 8,421 | -27% | 1 | 1 | 0% | 1,142 | 1,095 | -4% | 0 | 0 | — |
case-24 | pass→pass | 10,445 | 11,012 | +5% | 1 | 1 | 0% | 1,651 | 1,318 | -20% | 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. 24 cases were attempted. The headline lift of +25 percentage points is the difference between those two pass rates over the 24 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.