Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Look up CVE records via x cve — cached, zero-API-key, daily xz TSV. Load for cve, vulnerability id, kev, epss, nvd, cvelist, or security advisory.
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -75% | 0% |
Per-year xz-compressed TSV; current and last year auto-refresh (1 h TTL), older years stay frozen 180 d. After the first online day, x cve works offline indefinitely.
x cvebashx cve info CVE-2024-0001 # id lookup (shorthand: x cve 2024-0001) x cve year 2024 # per-year stream, pipe-friendly TSV x cve detail CVE-2024-0001 # full upstream record (products, CWE, refs, ADP)
x cve -h for all flags, subcommands, and shorthand forms.
Daily xz TSV published at github.com/x-cmd/cve/releases/download/data/ (schema documented at the release). The cached record is a thin projection — for the questions that go one hop beyond it:
x shodan cve CVE-X — EPSS score, KEV membership, exploit articles, vendor advisoriesx kev ls — CISA KEV catalog (known-exploited list, full table)x osv — per-package version-level query ("is MY version vulnerable"); accepts osv-* and cve-* idsx cve detail — raw CVEProject/cvelistV5 JSON (canonical upstream, fetched on demand)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 6,826 | 3,433 | -50% | 1 | 1 | 0% | 1,061 | 841 | -21% | 0 | 0 | — |
case-02 | fail→pass | 15,338 | 7,759 | -49% | 1 | 1 | 0% | 2,730 | 805 | -71% | 0 | 0 | — |
case-03 | fail→pass | 19,595 | 3,670 | -81% | 1 | 1 | 0% | 2,024 | 917 | -55% | 0 | 0 | — |
case-04 | fail→pass | 4,681 | 3,638 | -22% | 1 | 1 | 0% | 611 | 860 | +41% | 0 | 0 | — |
case-05 | fail→pass | 15,308 | 8,072 | -47% | 1 | 1 | 0% | 2,791 | 709 | -75% | 0 | 0 | — |
case-06 | fail→pass | 6,783 | 4,155 | -39% | 1 | 1 | 0% | 1,292 | 822 | -36% | 0 | 0 | — |
case-07 | fail→pass | 9,500 | 2,724 | -71% | 1 | 1 | 0% | 1,467 | 612 | -58% | 0 | 0 | — |
case-08 | fail→pass | 13,916 | 3,748 | -73% | 1 | 1 | 0% | 1,915 | 883 | -54% | 0 | 0 | — |
case-09 | pass→pass | 12,406 | 12,980 | +5% | 1 | 1 | 0% | 1,967 | 876 | -55% | 0 | 0 | — |
case-10 | fail→pass | 50,992 | 2,614 | -95% | 1 | 1 | 0% | 2,596 | 622 | -76% | 0 | 0 | — |
case-11 | pass→pass | 9,218 | 3,145 | -66% | 1 | 1 | 0% | 1,472 | 609 | -59% | 0 | 0 | — |
case-12 | fail→pass | 10,670 | 5,232 | -51% | 1 | 1 | 0% | 1,764 | 1,138 | -35% | 0 | 0 | — |
case-13 | fail→pass | 13,896 | 5,271 | -62% | 1 | 1 | 0% | 2,326 | 815 | -65% | 0 | 0 | — |
case-14 | fail→pass | 6,082 | 5,761 | -5% | 1 | 1 | 0% | 957 | 650 | -32% | 0 | 0 | — |
case-15 | pass→pass | 6,894 | 3,569 | -48% | 1 | 1 | 0% | 1,103 | 804 | -27% | 0 | 0 | — |
case-16 | pass→pass | 11,236 | 2,961 | -74% | 1 | 1 | 0% | 1,694 | 757 | -55% | 0 | 0 | — |
case-17 | pass→pass | 7,562 | 2,705 | -64% | 1 | 1 | 0% | 765 | 700 | -8% | 0 | 0 | — |
case-18 | pass→pass | 16,111 | 2,365 | -85% | 1 | 1 | 0% | 2,669 | 568 | -79% | 0 | 0 | — |
case-19 | pass→pass | 22,370 | 2,160 | -90% | 1 | 1 | 0% | 2,289 | 547 | -76% | 0 | 0 | — |
case-20 | fail→pass | 9,553 | 2,955 | -69% | 1 | 1 | 0% | 961 | 683 | -29% | 0 | 0 | — |
case-21 | fail→pass | 8,301 | 4,494 | -46% | 1 | 1 | 0% | 1,457 | 578 | -60% | 0 | 0 | — |
case-22 | fail→pass | 11,222 | 2,425 | -78% | 1 | 1 | 0% | 1,458 | 594 | -59% | 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 +68 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.