Dev-time security audit of KarvyLoop's OWN source (twin of /code-review, but for security). Use when asked to "audit the security of these changes", "审这轮改动的安全", "security review this endpoint/diff", or before landing anything that touches the untrusted-input frontier (new API endpoints, URL/file fetch, tool authorization, sandbox, LLM output that reaches a persistent store or the shell). Runs a multi-agent adversarial audit — parallel discovery agents, then ≥3 refutation passes per finding (keep only if ≥2/3 say REAL) — and REPORTS ONLY; it never edits code. Not a KarvyLoop product feature.
$npx -y skills add caprista/karvyloop --skill security-reviewInstalls into the current project.
Run `npx skills use "https://github.com/caprista/karvyloop" --skill "caprista/karvyloop/security-review"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/caprista/karvyloop" that are relevant to the current task. Run `npx skills add "https://github.com/caprista/karvyloop"` and select the relevant skills, then follow their instructions.