Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Aggiunge alle repository GitHub dei siti workflow di sicurezza automatici - scansione dipendenze vulnerabili, ricerca di segreti/chiavi nel codice, analisi statica CodeQL e Dependabot. Trigger - "sicurezza repo", "GitHub Actions di sicurezza", "scansione repository", "metti la sicurezza nella repo", "dependabot", "secret scanning".
.claude/skills/ccplugins-sicurezza-github/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 0% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 1% | 0% |
Quando un progetto ha (o avrà) una repository GitHub, installare i file di sicurezza nella directory .github/ della repo. I template pronti sono in references/.
.github/workflows/security.yml — da references/security-workflow.yml: a ogni push/PR e ogni notte esegue scansione segreti (Gitleaks), audit dipendenze (npm/pip) e analisi statica CodeQL..github/dependabot.yml — da references/dependabot.yml: PR automatiche di aggiornamento per dipendenze vulnerabili.Adattare al linguaggio del progetto: lasciare il job npm audit solo se c'è package.json, pip-audit solo se c'è requirements.txt/pyproject.toml, impostare i linguaggi CodeQL corretti (javascript, python, ecc.).
.gitignore escluda .env e credenziali PRIMA del primo push. Se un segreto è già finito nella history, va considerato compromesso: ruotarlo subito (cambiare la chiave dal provider), non basta cancellare il file.Questi workflow proteggono il codice e le dipendenze, non bloccano gli attacchi al sito in esecuzione: per quello servono la skill difesa-attacchi (middleware) e un WAF/CDN davanti al sito.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 16,868 | 13,164 | -22% | 1 | 1 | 0% | 3,116 | 2,866 | -8% | 0 | 0 | — |
case-02 | fail→pass | 19,227 | 16,491 | -14% | 1 | 1 | 0% | 3,290 | 3,184 | -3% | 0 | 0 | — |
case-03 | fail→pass | 20,010 | 17,089 | -15% | 1 | 1 | 0% | 3,154 | 3,260 | +3% | 0 | 0 | — |
case-04 | pass→pass | 12,214 | 10,892 | -11% | 1 | 1 | 0% | 1,845 | 2,164 | +17% | 0 | 0 | — |
case-05 | pass→pass | 9,983 | 11,211 | +12% | 1 | 1 | 0% | 1,573 | 2,018 | +28% | 0 | 0 | — |
case-06 | pass→pass | 16,468 | 63,040 | +283% | 1 | 1 | 0% | 2,366 | 2,339 | -1% | 0 | 0 | — |
case-07 | pass→pass | 13,275 | 11,591 | -13% | 1 | 1 | 0% | 1,909 | 2,135 | +12% | 0 | 0 | — |
case-08 | pass→pass | 17,246 | 13,594 | -21% | 1 | 1 | 0% | 2,326 | 2,428 | +4% | 0 | 0 | — |
case-09 | pass→pass | 10,155 | 6,754 | -33% | 1 | 1 | 0% | 1,522 | 1,447 | -5% | 0 | 0 | — |
case-10 | pass→pass | 11,622 | 7,034 | -39% | 1 | 1 | 0% | 1,677 | 1,443 | -14% | 0 | 0 | — |
case-11 | pass→pass | 14,206 | 11,259 | -21% | 1 | 1 | 0% | 2,212 | 2,375 | +7% | 0 | 0 | — |
case-12 | pass→pass | 5,775 | 2,792 | -52% | 1 | 1 | 0% | 847 | 706 | -17% | 0 | 0 | — |
case-13 | fail→pass | 15,936 | 11,853 | -26% | 1 | 1 | 0% | 2,113 | 2,112 | -0% | 0 | 0 | — |
case-14 | fail→pass | 15,681 | 11,908 | -24% | 1 | 1 | 0% | 2,392 | 2,412 | +1% | 0 | 0 | — |
case-15 | pass→pass | 10,058 | 9,450 | -6% | 1 | 1 | 0% | 1,558 | 1,734 | +11% | 0 | 0 | — |
case-16 | pass→pass | 3,672 | 4,518 | +23% | 1 | 1 | 0% | 485 | 935 | +93% | 0 | 0 | — |
case-17 | pass→pass | 16,155 | 12,984 | -20% | 1 | 1 | 0% | 2,791 | 3,004 | +8% | 0 | 0 | — |
case-18 | pass→pass | 10,904 | 10,612 | -3% | 1 | 1 | 0% | 1,492 | 2,086 | +40% | 0 | 0 | — |
case-19 | pass→pass | 15,425 | 11,179 | -28% | 1 | 1 | 0% | 2,181 | 2,067 | -5% | 0 | 0 | — |
case-20 | pass→pass | 20,769 | 14,600 | -30% | 1 | 1 | 0% | 3,254 | 2,577 | -21% | 0 | 0 | — |
case-21 | pass→pass | 15,517 | 13,285 | -14% | 1 | 1 | 0% | 2,543 | 2,857 | +12% | 0 | 0 | — |
case-22 | pass→pass | 12,588 | 13,362 | +6% | 1 | 1 | 0% | 2,384 | 2,790 | +17% | 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.