Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are a dependency security expert specializing in vulnerability scanning, license compliance, and supply chain security. Analyze project dependencies for known vulnerabilities, licensing issues, outdated packages, and provide actionable remediation strategies.
.claude/skills/dokhacgiakhoa-dependency-management-deps-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -54% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -59% | 0% |
You are a dependency security expert specializing in vulnerability scanning, license compliance, and supply chain security. Analyze project dependencies for known vulnerabilities, licensing issues, outdated packages, and provide actionable remediation strategies.
The user needs comprehensive dependency analysis to identify security vulnerabilities, licensing conflicts, and maintenance risks in their project dependencies. Focus on actionable insights with automated fixes where possible.
$ARGUMENTS
resources/implementation-playbook.md.resources/implementation-playbook.md for detailed tooling and templates.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 13,531 | 11,008 | -19% | 1 | 1 | 0% | 2,077 | 1,937 | -7% | 0 | 0 | — |
case-01 | fail→fail | 18,161 | 22,757 | +25% | 1 | 1 | 0% | 2,735 | 3,509 | +28% | 0 | 0 | — |
case-02 | pass→pass | 12,838 | 10,085 | -21% | 1 | 1 | 0% | 1,990 | 2,087 | +5% | 0 | 0 | — |
case-03 | pass→pass | 14,210 | 16,111 | +13% | 1 | 1 | 0% | 2,130 | 2,966 | +39% | 0 | 0 | — |
case-04 | pass→pass | 14,576 | 20,052 | +38% | 1 | 1 | 0% | 2,430 | 3,374 | +39% | 0 | 0 | — |
case-05 | pass→pass | 13,558 | 15,149 | +12% | 1 | 1 | 0% | 2,010 | 2,742 | +36% | 0 | 0 | — |
case-06 | pass→pass | 11,816 | 11,690 | -1% | 1 | 1 | 0% | 1,627 | 1,284 | -21% | 0 | 0 | — |
case-08 | fail→pass | 24,941 | 12,373 | -50% | 1 | 1 | 0% | 3,609 | 2,596 | -28% | 0 | 0 | — |
case-09 | pass→pass | 12,602 | 14,196 | +13% | 1 | 1 | 0% | 2,027 | 2,575 | +27% | 0 | 0 | — |
case-10 | fail→pass | 17,788 | 18,078 | +2% | 1 | 1 | 0% | 2,554 | 3,334 | +31% | 0 | 0 | — |
case-11 | pass→pass | 9,968 | 10,889 | +9% | 1 | 1 | 0% | 1,467 | 2,050 | +40% | 0 | 0 | — |
case-12 | pass→pass | 13,466 | 11,141 | -17% | 1 | 1 | 0% | 1,898 | 2,258 | +19% | 0 | 0 | — |
case-13 | fail→pass | 13,463 | 16,383 | +22% | 1 | 1 | 0% | 1,927 | 2,719 | +41% | 0 | 0 | — |
case-14 | pass→pass | 10,150 | 13,007 | +28% | 1 | 1 | 0% | 1,587 | 1,411 | -11% | 0 | 0 | — |
case-15 | pass→pass | 13,323 | 10,651 | -20% | 1 | 1 | 0% | 1,847 | 2,203 | +19% | 0 | 0 | — |
case-16 | fail→pass | 11,309 | 3,722 | -67% | 1 | 1 | 0% | 1,848 | 848 | -54% | 0 | 0 | — |
case-17 | pass→pass | 13,322 | 11,918 | -11% | 1 | 1 | 0% | 1,924 | 1,926 | +0% | 0 | 0 | — |
case-18 | pass→pass | 14,066 | 14,612 | +4% | 1 | 1 | 0% | 2,263 | 2,281 | +1% | 0 | 0 | — |
case-19 | pass→pass | 18,034 | 11,338 | -37% | 1 | 1 | 0% | 2,504 | 2,143 | -14% | 0 | 0 | — |
case-20 | fail→pass | 19,205 | 6,314 | -67% | 1 | 1 | 0% | 2,875 | 1,183 | -59% | 0 | 0 | — |
case-21 | pass→pass | 12,163 | 7,756 | -36% | 1 | 1 | 0% | 1,785 | 1,459 | -18% | 0 | 0 | — |
case-22 | pass→pass | 10,106 | 6,376 | -37% | 1 | 1 | 0% | 1,753 | 1,402 | -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. 22 cases were attempted. The headline lift of +23 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.