caveman
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill exclusively contains natural language instructions for persona modification and token optimization. It is intended to help the agent save context space by removing linguistic filler.
- [NO_CODE]: There are no scripts, executables, or code files associated with this skill. The examples provided (SQL and React) are for illustrative purposes within the instructions and are not executed.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user input to trigger the persona change, but it lacks any dangerous capabilities (such as network access, file system writes, or command execution) that could be leveraged for an attack. Additionally, the skill explicitly instructs the agent to drop the 'caveman' persona for security warnings and action confirmations, maintaining clarity for critical safety information.
Audit Metadata