safety-guard

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill acts as a security enhancement by providing explicit instructions to the agent to avoid destructive commands such as force-pushes, recursive deletions, or database drops.
  • [SAFE]: The 'freeze' and 'guard' modes implement a least-privilege approach by restricting file-writing capabilities to specific, user-defined directory paths.
  • [SAFE]: No external downloads, remote scripts, or network operations are present. The skill uses local logging to a standard directory for audit purposes.
  • [SAFE]: No credentials, secrets, or obfuscated patterns were detected in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 03:49 AM