caveman
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFE
Full Analysis
- Instructional Output Formatting: The skill defines rules for a terse 'Caveman' response style. These are standard behavioral guidelines and do not include attempts to override safety protocols or bypass system instructions.
- Security-First Clarity Rules: The skill explicitly includes boundaries where compression is disabled, such as when communicating vulnerability disclosures or confirming destructive actions. This preserves the integrity of safety-critical communications.
- No External Capabilities: The skill does not perform network requests, access the file system, or execute code. It relies solely on the conversation transcript for state, minimizing its attack surface.
Audit Metadata