Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Look up vehicle information from a license plate number using the CarsXE API. Use this when a user mentions a license plate and wants to know what vehicle it belongs to.
.claude/skills/hashgraph-online-plate-decoder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 161% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -52% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 213% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -15% | 0% |
When the user provides a license plate number:
plate (required): the license plate number — if not provided, ask:> "Please provide a license plate number."
country (required): ISO 3166-1 alpha-2 country code (e.g., US, AU, GB) — if not provided, ask:> "Please provide a country code (e.g., US, AU, GB)."
state (required): 2-letter state/province code (e.g., CA, NY) — if not provided, ask:> "Please provide a state code (e.g., CA, NY, TX)."
GET https://api.carsxe.com/v2/platedecoder?key={CARSXE_API_KEY}&plate={PLATE}&country={COUNTRY}&state={STATE}&source=codex_plugin
CARSXE_API_KEY environment variable (see AGENTS.md).| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 5,187 | 13,854 | +167% | 1 | 1 | 0% | 631 | 1,646 | +161% | 0 | 0 | — |
case-02 | fail→pass | 13,476 | 2,062 | -85% | 1 | 1 | 0% | 1,320 | 633 | -52% | 0 | 0 | — |
case-03 | fail→pass | 9,194 | 13,883 | +51% | 1 | 1 | 0% | 573 | 1,796 | +213% | 0 | 0 | — |
case-04 | fail→pass | 12,594 | 12,917 | +3% | 1 | 1 | 0% | 1,271 | 1,733 | +36% | 0 | 0 | — |
case-05 | fail→pass | 18,483 | 9,365 | -49% | 1 | 1 | 0% | 1,088 | 920 | -15% | 0 | 0 | — |
case-06 | pass→pass | 15,811 | 10,184 | -36% | 1 | 1 | 0% | 1,571 | 956 | -39% | 0 | 0 | — |
case-07 | pass→pass | 15,958 | 13,884 | -13% | 1 | 1 | 0% | 1,551 | 1,567 | +1% | 0 | 0 | — |
case-08 | fail→pass | 16,387 | 14,921 | -9% | 1 | 1 | 0% | 1,849 | 2,034 | +10% | 0 | 0 | — |
case-09 | fail→pass | 11,704 | 6,065 | -48% | 1 | 1 | 0% | 1,888 | 1,358 | -28% | 0 | 0 | — |
case-10 | fail→pass | 10,906 | 2,784 | -74% | 1 | 1 | 0% | 896 | 619 | -31% | 0 | 0 | — |
case-11 | pass→pass | 13,919 | 7,176 | -48% | 1 | 1 | 0% | 1,437 | 647 | -55% | 0 | 0 | — |
case-12 | pass→pass | 16,855 | 7,244 | -57% | 1 | 1 | 0% | 1,689 | 651 | -61% | 0 | 0 | — |
case-13 | fail→pass | 9,446 | 9,789 | +4% | 1 | 1 | 0% | 1,383 | 960 | -31% | 0 | 0 | — |
case-14 | fail→pass | 12,053 | 8,457 | -30% | 1 | 1 | 0% | 1,826 | 648 | -65% | 0 | 0 | — |
case-15 | pass→pass | 16,297 | 8,742 | -46% | 1 | 1 | 0% | 1,744 | 801 | -54% | 0 | 0 | — |
case-16 | fail→pass | 13,433 | 8,960 | -33% | 1 | 1 | 0% | 1,323 | 1,809 | +37% | 0 | 0 | — |
case-17 | fail→pass | 5,330 | 8,652 | +62% | 1 | 1 | 0% | 859 | 874 | +2% | 0 | 0 | — |
case-18 | pass→pass | 15,175 | 14,174 | -7% | 1 | 1 | 0% | 1,846 | 1,974 | +7% | 0 | 0 | — |
case-19 | pass→pass | 10,528 | 4,164 | -60% | 1 | 1 | 0% | 777 | 827 | +6% | 0 | 0 | — |
case-20 | pass→pass | 10,018 | 6,587 | -34% | 1 | 1 | 0% | 1,693 | 1,310 | -23% | 0 | 0 | — |
case-21 | pass→pass | 20,131 | 16,017 | -20% | 1 | 1 | 0% | 2,139 | 1,955 | -9% | 0 | 0 | — |
case-22 | fail→pass | 7,874 | 12,485 | +59% | 1 | 1 | 0% | 1,284 | 1,360 | +6% | 0 | 0 | — |
case-23 | fail→pass | 11,510 | 8,037 | -30% | 1 | 1 | 0% | 911 | 779 | -14% | 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. 23 cases were attempted. The headline lift of +61 percentage points is the difference between those two pass rates over the 23 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.