arc-security-audit
por trypto1019
Git & GitHubResumen
Comprehensive security audit for an agent's full skill stack.
Descripción
Qué hace esta skill
Runs a full security audit of all installed OpenClaw skills in a single pass. Chains arc-skill-scanner, arc-trust-verifier, and SHA-256 binary integrity checks, then outputs a prioritized risk report with per-skill trust scores. Optionally generates trust attestations for skills that pass all checks.
Replaces running arc-skill-scanner, arc-trust-verifier, and binary checks separately, so nothing is skipped and findings are consolidated into one ranked report.
Cuándo usarla
- Auditing all installed skills before deploying an agent to production
- Checking a newly added third-party skill for suspicious code or binaries
- Exporting a JSON audit report for a compliance or security review
- Identifying which installed skills have the highest risk scores at a glance
- Generating trust attestations for skills that pass all verification steps
Flujo de trabajo de ejemplo
User asks: audit all my installed skills and flag anything critical
- 1
Runs arc-skill-scanner patterns across every installed skill directory
- 2
Assesses provenance and code cleanliness for each skill via arc-trust-verifier
- 3
Verifies binary integrity using SHA-256 checksums
- 4
Ranks all findings by severity and builds a per-skill breakdown
- 5
Surfaces critical actions that require immediate attention
A report listing total skills scanned, severity counts, overall risk level, per-skill trust scores with recommendations, and a critical actions section
Instalación
clawhub install trypto1019/arc-security-auditEjecuta el comando de la CLI o pega el prompt en Claude Code, Cursor, Codex o el agente de programación que uses.