Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Fast, continuous DevSecOps pipeline for Pull Requests and active branches. Runs SAST, SCA, and secrets detection to catch vulnerabilities before they merge.
.claude/skills/hoangnguyen0403-security-test/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -51% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 12% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -33% | 0% |
> !IMPORTANT] > Fast, continuous DevSecOps pipeline for Pull Requests and active branches. Runs SAST, SCA, and secrets detection to catch vulnerabilities before they merge.
Optional args: slug=<feature>, ticket=<id/url>, mode=interactive|autonomous|channel, channel=<id>, auto_continue=true|false, profile=business|hybrid|technical.
When the user asks to perform this workflow, execute the following steps:
> Goal: Execute a high-speed security audit on a branch or PR delta and stop obvious security regressions before merge. > > Policy: Fast execution (< 2 mins). Focus on SAST, SCA, secrets, and trust-boundary regressions. No dynamic exploitation required.
git diff <base>...HEAD.<SKILLS>/common/common-security-audit/references/trust-review-policy.md.reviewContext.promptInjectionRisk to high unless host controls clearly reduce that risk.specialist-aspm-correlator.specialist-security-reviewer in fast mode for normal diffs and deep mode for auth, secrets, agent tools, or external integration changes.design-solution evidence or return BLOCKED with the missing design questions.artifacts/security-review.md with trust class, review context, scope, source provenance, runtime contract, blockers, warnings, finding confidence, exploit path, evidence gaps, and handoff notes.artifacts/security-review.dev.md, artifacts/security-review.appsec.md, or artifacts/security-review.exec.md only when a separate audience needs it.markdown### 🛡️ Security Check: [PASS / FAIL] **Scan Scope**: [branch/diff size] **Trust Class**: [trusted|semi-trusted|untrusted] #### 🔴 Blockers - [file:line] - [vulnerability] - [exact fix] #### 🟡 Warnings - [risk] - [next action] #### ✅ Verified - [verified control]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 10,667 | 11,761 | +10% | 1 | 1 | 0% | 1,302 | 2,236 | +72% | 0 | 0 | — |
case-02 | fail→fail | 15,643 | 12,273 | -22% | 1 | 1 | 0% | 1,974 | 2,112 | +7% | 0 | 0 | — |
case-03 | fail→fail | 15,300 | 16,110 | +5% | 1 | 1 | 0% | 1,864 | 2,716 | +46% | 0 | 0 | — |
case-04 | fail→pass | 9,559 | 4,163 | -56% | 1 | 1 | 0% | 1,631 | 1,445 | -11% | 0 | 0 | — |
case-05 | fail→fail | 11,245 | 5,190 | -54% | 1 | 1 | 0% | 1,796 | 1,535 | -15% | 0 | 0 | — |
case-06 | pass→pass | 8,138 | 2,633 | -68% | 1 | 1 | 0% | 1,336 | 1,020 | -24% | 0 | 0 | — |
case-07 | pass→pass | 5,605 | 2,574 | -54% | 1 | 1 | 0% | 918 | 1,094 | +19% | 0 | 0 | — |
case-08 | fail→pass | 12,922 | 2,221 | -83% | 1 | 1 | 0% | 2,158 | 1,066 | -51% | 0 | 0 | — |
case-09 | fail→pass | 7,083 | 3,461 | -51% | 1 | 1 | 0% | 1,164 | 1,308 | +12% | 0 | 0 | — |
case-10 | pass→pass | 10,605 | 4,993 | -53% | 1 | 1 | 0% | 1,677 | 1,610 | -4% | 0 | 0 | — |
case-11 | pass→pass | 10,579 | 5,628 | -47% | 1 | 1 | 0% | 1,686 | 1,705 | +1% | 0 | 0 | — |
case-12 | pass→pass | 8,962 | 4,459 | -50% | 1 | 1 | 0% | 1,558 | 1,356 | -13% | 0 | 0 | — |
case-13 | pass→pass | 10,237 | 6,180 | -40% | 1 | 1 | 0% | 1,699 | 1,745 | +3% | 0 | 0 | — |
case-14 | pass→pass | 10,673 | 3,049 | -71% | 1 | 1 | 0% | 1,767 | 1,198 | -32% | 0 | 0 | — |
case-15 | fail→pass | 6,027 | 3,148 | -48% | 1 | 1 | 0% | 1,027 | 1,020 | -1% | 0 | 0 | — |
case-16 | fail→pass | 8,586 | 1,871 | -78% | 1 | 1 | 0% | 1,489 | 1,005 | -33% | 0 | 0 | — |
case-17 | pass→pass | 13,626 | 4,551 | -67% | 1 | 1 | 0% | 2,339 | 1,352 | -42% | 0 | 0 | — |
case-18 | pass→pass | 4,096 | 3,046 | -26% | 1 | 1 | 0% | 659 | 1,206 | +83% | 0 | 0 | — |
case-19 | pass→fail | 10,131 | 2,611 | -74% | 1 | 1 | 0% | 1,767 | 1,071 | -39% | 0 | 0 | — |
case-20 | pass→pass | 10,371 | 3,223 | -69% | 1 | 1 | 0% | 1,645 | 1,196 | -27% | 0 | 0 | — |
case-21 | fail→pass | 11,473 | 14,518 | +27% | 1 | 1 | 0% | 1,978 | 3,386 | +71% | 0 | 0 | — |
case-22 | fail→fail | 13,221 | 13,196 | -0% | 1 | 1 | 0% | 1,735 | 2,086 | +20% | 0 | 0 | — |
case-23 | fail→fail | 17,803 | 29,724 | +67% | 1 | 1 | 0% | 3,031 | 3,902 | +29% | 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 +22 percentage points is the difference between those two pass rates over the 23 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.