caveman
Pass
Audited by Gen Agent Trust Hub on Apr 28, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious behavior detected. The skill is limited to defining a specific communication style (terse/fragmented) to optimize token consumption.
- [PROMPT_INJECTION]: The instructions establish a persistent persona but do not attempt to override system safety filters or extract the underlying system prompt. The 'Persistence' section refers to maintaining the communication style, not system-level persistence.
- [DATA_EXFILTRATION]: There are no commands or instructions involving network requests, file access, or credential harvesting.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute external scripts or packages.
Audit Metadata