skill-audit
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script
scripts/audit.pyto perform deterministic security checks (Unicode anomalies, Base64 detection, structural validation) on skill directories as part of its auditing workflow. - [COMMAND_EXECUTION]: Includes a 'Companion check' feature that runs
lson common AI agent skill directories (e.g.,~/.claude/skills/) to identify installed tools and suggest missing recommended companions from the same vendor. - [PROMPT_INJECTION]: The skill contains various prompt injection strings such as 'ignore previous instructions' and 'DAN'. These are explicitly used as training examples, test cases in
evals.json, or as target patterns for the auditing logic, and do not represent attempts to manipulate the agent's own behavior. - [SAFE]: The skill's external references and installation instructions (
npx skills add) are associated with the vendor's official resources (AbsolutelySkilled) and are consistent with the skill's stated purpose as a management and auditing tool.
Audit Metadata