Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reviewing a project's security posture, setting up CI pipelines, or responding to a reported vulnerability in a dependency.
.claude/skills/thedaviddias-dependency-audit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 95% | 0% |
| case-07 | ✓→✗ | ▼ Worse | -2% | 0% |
| case-10 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
| case-15 | ✓→✓ | ＝ Same ✓ | -12% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 24% | 0% |
Third-party packages are the most common attack surface in modern web applications. The 2021 Log4Shell incident, the 2022 node-ipc supply-chain attack, and countless npm package hijackings demonstrate that a single vulnerable transitive dependency can compromise every application that depends on it. Automated, continuous scanning drastically reduces the window between a CVE being published and your team being aware of it.
Check the project's dependencies for known security vulnerabilities using the package manager audit command.
Upgrade, patch, or replace vulnerable dependencies and configure automated scanning in the CI pipeline.
Explain how supply-chain attacks work and why dependency auditing is a critical part of modern application security.
Review the lock file and package.json for unpinned version ranges, abandoned packages, and any packages flagged in recent CVE databases.
For full implementation details, code examples, and framework-specific guidance, see references/rule.md.
Rule page: https://frontendchecklist.io/en/rules/security/dependency-audit
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 10,850 | 15,621 | +44% | 1 | 1 | 0% | 2,506 | 2,783 | +11% | 0 | 0 | — |
case-15 | pass→pass | 10,782 | 7,402 | -31% | 1 | 1 | 0% | 1,812 | 1,598 | -12% | 0 | 0 | — |
case-01 | fail→pass | 6,371 | 10,914 | +71% | 1 | 1 | 0% | 1,269 | 2,470 | +95% | 0 | 0 | — |
case-02 | pass→pass | 11,581 | 12,218 | +6% | 1 | 1 | 0% | 1,980 | 2,459 | +24% | 0 | 0 | — |
case-03 | pass→pass | 12,356 | 9,519 | -23% | 1 | 1 | 0% | 2,135 | 2,058 | -4% | 0 | 0 | — |
case-04 | pass→pass | 12,938 | 9,831 | -24% | 1 | 1 | 0% | 2,213 | 2,063 | -7% | 0 | 0 | — |
case-05 | pass→pass | 10,472 | 11,830 | +13% | 1 | 1 | 0% | 2,255 | 2,538 | +13% | 0 | 0 | — |
case-06 | pass→pass | 12,381 | 11,978 | -3% | 1 | 1 | 0% | 2,178 | 2,417 | +11% | 0 | 0 | — |
case-07 | pass→fail | 14,290 | 12,232 | -14% | 1 | 1 | 0% | 2,559 | 2,505 | -2% | 0 | 0 | — |
case-08 | pass→pass | 11,897 | 10,121 | -15% | 1 | 1 | 0% | 2,143 | 2,211 | +3% | 0 | 0 | — |
case-09 | pass→pass | 12,183 | 7,462 | -39% | 1 | 1 | 0% | 2,252 | 1,717 | -24% | 0 | 0 | — |
case-11 | pass→pass | 11,917 | 9,909 | -17% | 1 | 1 | 0% | 2,170 | 2,165 | -0% | 0 | 0 | — |
case-12 | pass→pass | 16,879 | 18,153 | +8% | 1 | 1 | 0% | 2,721 | 3,103 | +14% | 0 | 0 | — |
case-13 | pass→pass | 12,946 | 14,152 | +9% | 1 | 1 | 0% | 2,280 | 2,786 | +22% | 0 | 0 | — |
case-14 | pass→pass | 9,460 | 9,213 | -3% | 1 | 1 | 0% | 2,149 | 1,904 | -11% | 0 | 0 | — |
case-16 | pass→pass | 19,404 | 16,888 | -13% | 1 | 1 | 0% | 3,232 | 3,184 | -1% | 0 | 0 | — |
case-17 | pass→pass | 12,535 | 10,333 | -18% | 1 | 1 | 0% | 2,230 | 1,987 | -11% | 0 | 0 | — |
case-18 | pass→pass | 12,723 | 11,181 | -12% | 1 | 1 | 0% | 2,334 | 2,173 | -7% | 0 | 0 | — |
case-19 | pass→pass | 9,574 | 8,638 | -10% | 1 | 1 | 0% | 1,682 | 1,855 | +10% | 0 | 0 | — |
case-20 | pass→pass | 11,101 | 9,199 | -17% | 1 | 1 | 0% | 2,262 | 1,991 | -12% | 0 | 0 | — |
case-21 | pass→pass | 13,391 | 14,712 | +10% | 1 | 1 | 0% | 2,702 | 3,216 | +19% | 0 | 0 | — |
case-22 | pass→pass | 11,997 | 10,559 | -12% | 1 | 1 | 0% | 2,116 | 2,122 | +0% | 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 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.