Security audit skill — multi-phase security analysis adapted to app type, with confidence gating and false positive suppression
Multi-phase security analysis. Runs automatically on Full tier, available on-demand for all tiers.
/z:secure — full daily audit (8/10 confidence gate)/z:secure --comprehensive — monthly deep scan (2/10 bar)/z:secure --diff — branch changes only (combinable with any flag)/z:secure --deps — dependency audit only/z:secure --owasp — OWASP Top 10 only/z:secure --secrets — secrets scan only.zstack/project.json for type, language, framework, runtime.Using Grep tool (not bash grep), find:
Output: Attack Surface Map with counts for each category.
AKIA, sk-, ghp_, gho_, xoxb-, xoxp-.env files tracked by git (not .example/.sample/.template).env tracked..env.local in .gitignore is expected.npm audit, pip audit, cargo audit, etc.)node-gyp install scripts expected.pull_request_target (dangerous: fork PRs get write access)${{ github.event.* }} in run stepsUSER (runs as root), secrets as ARG, .env copied inpostgres://, mysql://, mongodb://, redis://)"*" in IAM actions, hardcoded secrets in .tf/.tfvarshostNetwork, hostPIDScan for:
Map all findings to OWASP Top 10 2021 categories:
For each component identified in Phase 0:
These are critical for preventing alert fatigue:
"your_", "changeme", "TODO") excluded from secretsnode-gyp/cmake install scripts expected (MEDIUM not HIGH).env.local in .gitignore is expectedSECURITY POSTURE REPORT
══════════════════════════════════════
App type: [type]
Mode: [daily | comprehensive]
Scope: [full | diff | deps | owasp | secrets]
ATTACK SURFACE: [N] entry points
FINDINGS: [N] total
CRITICAL: [N]
HIGH: [N]
MEDIUM: [N]
[For each finding:]
[SEVERITY] [confidence/10] — Title
Location: file:line
What: [description]
Exploit: [concrete exploit scenario]
Fix: [remediation steps]
OWASP: [category]
VERDICT: SECURE | NEEDS_REMEDIATION | CRITICAL_RISK
══════════════════════════════════════