Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use `wiz-inspector` when Wiz CNAPP is the source of cloud posture, vulnerability, toxic-combination, or inventory evidence.
.claude/skills/grcengclub-wiz-inspector-expert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -18% | 0% |
Use wiz-inspector when Wiz CNAPP is the source of cloud posture, vulnerability, toxic-combination, or inventory evidence.
Inputs:
WIZ_CLIENT_IDWIZ_CLIENT_SECRETWIZ_API_URL, for example https://api.<region>.app.wiz.io/graphqlWIZ_AUTH_URL, defaulting to https://auth.app.wiz.io/oauth/tokenWIZ_PROJECT_ID to constrain collection to a Wiz projectread:projects, read:issues, read:vulnerabilities, read:inventoryRun /wiz-inspector:setup, then /wiz-inspector:collect. Findings are emitted using schemas/finding.schema.json and can feed GRC Engineering gap assessments, OSCAL exports, workpaper generation, and remediation workflows.
The collector handles cursor pagination, project scoping, Wiz severity/status normalization, and partial results. If a GraphQL resource query fails, it emits an inconclusive Finding for that endpoint and continues collecting the remaining resource types.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→pass | 14,006 | 5,236 | -63% | 1 | 1 | 0% | 2,759 | 1,132 | -59% | 0 | 0 | — |
case-04 | fail→pass | 10,020 | 3,496 | -65% | 1 | 1 | 0% | 1,597 | 804 | -50% | 0 | 0 | — |
case-01 | fail→pass | 12,901 | 36,631 | +184% | 1 | 1 | 0% | 2,668 | 1,426 | -47% | 0 | 0 | — |
case-02 | fail→pass | 16,960 | 11,739 | -31% | 1 | 1 | 0% | 3,363 | 2,418 | -28% | 0 | 0 | — |
case-05 | fail→pass | 3,234 | 1,443 | -55% | 1 | 1 | 0% | 572 | 469 | -18% | 0 | 0 | — |
case-06 | fail→pass | 7,831 | 2,825 | -64% | 1 | 1 | 0% | 1,244 | 699 | -44% | 0 | 0 | — |
case-07 | pass→pass | 11,567 | 6,396 | -45% | 1 | 1 | 0% | 1,799 | 1,169 | -35% | 0 | 0 | — |
case-08 | fail→pass | 16,430 | 2,739 | -83% | 1 | 1 | 0% | 938 | 672 | -28% | 0 | 0 | — |
case-18 | pass→pass | 4,693 | 2,414 | -49% | 1 | 1 | 0% | 752 | 543 | -28% | 0 | 0 | — |
case-09 | pass→pass | 6,882 | 1,446 | -79% | 1 | 1 | 0% | 1,143 | 506 | -56% | 0 | 0 | — |
case-10 | fail→pass | 16,337 | 8,771 | -46% | 1 | 1 | 0% | 2,374 | 1,533 | -35% | 0 | 0 | — |
case-11 | fail→pass | 6,058 | 2,017 | -67% | 1 | 1 | 0% | 905 | 492 | -46% | 0 | 0 | — |
case-12 | fail→pass | 13,495 | 4,244 | -69% | 1 | 1 | 0% | 1,151 | 947 | -18% | 0 | 0 | — |
case-13 | pass→pass | 10,135 | 1,355 | -87% | 1 | 1 | 0% | 1,540 | 411 | -73% | 0 | 0 | — |
case-14 | pass→pass | 19,730 | 12,489 | -37% | 1 | 1 | 0% | 3,059 | 2,162 | -29% | 0 | 0 | — |
case-15 | fail→pass | 16,430 | 9,063 | -45% | 1 | 1 | 0% | 2,453 | 1,619 | -34% | 0 | 0 | — |
case-16 | pass→pass | 5,502 | 3,055 | -44% | 1 | 1 | 0% | 788 | 607 | -23% | 0 | 0 | — |
case-17 | pass→pass | 6,465 | 1,451 | -78% | 1 | 1 | 0% | 972 | 433 | -55% | 0 | 0 | — |
case-19 | pass→pass | 5,312 | 1,227 | -77% | 1 | 1 | 0% | 748 | 396 | -47% | 0 | 0 | — |
case-20 | fail→pass | 13,984 | 2,767 | -80% | 1 | 1 | 0% | 2,254 | 755 | -67% | 0 | 0 | — |
case-21 | pass→pass | 16,404 | 10,664 | -35% | 1 | 1 | 0% | 2,606 | 1,893 | -27% | 0 | 0 | — |
case-22 | pass→pass | 14,569 | 11,736 | -19% | 1 | 1 | 0% | 2,491 | 2,362 | -5% | 0 | 0 | — |
case-23 | pass→pass | 9,467 | 7,862 | -17% | 1 | 1 | 0% | 1,714 | 1,566 | -9% | 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, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +52 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.