skills/a-canary/arc-skills/caveman/Gen Agent Trust Hub

caveman

Pass

Audited by Gen Agent Trust Hub on Jun 28, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists entirely of natural language instructions to adopt a 'caveman' persona (terse, fragmented speech) to reduce token consumption. No executable scripts, binaries, or configuration files are present.
  • [PROMPT_INJECTION]: The skill includes instructions for persona persistence ('ACTIVE every response once triggered. No revert, no drift.') and specific formatting rules ('AFK frame'). These are standard prompt engineering techniques for persona maintenance and do not attempt to bypass safety filters or override core system instructions.
  • [EXTERNAL_DOWNLOADS]: The SOURCE.md file references a GitHub repository (github.com/mattpocock/skills) for attribution purposes. This is a neutral documentation reference and does not involve automated remote code execution or data fetching.
  • [DATA_EXFILTRATION]: There are no patterns indicating network access or sensitive file reading. The instructions actually include a safety-positive 'Auto-Clarity Exception' that requires the agent to drop the persona and use clear language for security warnings and irreversible actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 28, 2026, 01:39 AM
Security Audit — agent-trust-hub — caveman