skill-security-auditor
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is a security scanner. Findings identified by automated tools (such as destructive commands or prompt injection patterns) are contained within the detection engine's regex patterns or as educational examples in the threat model documentation.
- [COMMAND_EXECUTION]: The main auditor script uses
subprocess.runwith list arguments to clone git repositories for auditing. This is a secure implementation that avoids shell injection risks. - [PROMPT_INJECTION]: The skill documents and scans for prompt injection vectors like 'Ignore previous instructions'. These patterns are used to identify risks in other skills and do not represent malicious behavior by this skill itself.
- [SAFE]: The auditor includes a specific finding category to detect its own suppression directives (
noqa: SEC-AUDITOR). This prevents malicious skills from using these comments to hide payloads from the auditor, which is a robust security design choice. - [REMOTE_CODE_EXECUTION]: While the skill can download content from remote git URLs provided by the user, it does so solely to perform a static analysis. It does not execute the code it downloads.
Audit Metadata