One Skill To Rule Them All

**Judgement:** Strong, purpose-built security auditing skill for OpenClaw SKILL.md files. It is defensive in nature and aligns well with platform safety goals. **Key strengths:** - Explicitly targets a wide range of relevant threats (prompt injection, data exfiltration, obfuscation, privilege escalation, persistence, metadata poisoning, indirect injection, time-delayed attacks). - Provides structured, severity-based verdicts with line-numbered evidence and remediation advice. - Can optionally generate a cleaned version of a suspicious SKILL.md, with warnings about limitations. **Main risks / limitations:** - As a heuristic, text-based analyzer, it can still miss novel or subtle attacks (false negatives) or over-flag benign patterns (false positives). - Users may develop an overreliance on its “SAFE” verdicts despite the built-in disclaimers; it does not replace expert review. - Automatic “cleaned” outputs, if used without human review, might either break intended functionality or fail to remove all malicious behavior. **Recommended use scenarios:** - As a first-line security scanner in review pipelines for new or updated OpenClaw skills. - For security-conscious users who want an additional check before installing third-party skills. - As a complementary tool alongside manual code review and broader security practices, not as a sole source of truth.