skills/juliusbrussee/caveman/caveman/Gen Agent Trust Hub

caveman

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill defines a behavioral persona for token efficiency but does not attempt to override core safety guidelines or extract system instructions. It features specific 'Auto-Clarity' rules that prioritize safety over the persona during high-risk interactions.
  • [REMOTE_CODE_EXECUTION]: No patterns for downloading external scripts, installing packages, or executing dynamic code were found.
  • [DATA_EXFILTRATION]: There is no evidence of sensitive file access, network exfiltration patterns, or hardcoded credentials.
  • [SAFE]: The skill is a benign personality modifier with clear boundaries for preserving technical accuracy and safety.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 02:35 PM
Security Audit — agent-trust-hub — caveman