concept-fan
Pass
Audited by Gen Agent Trust Hub on Jul 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is entirely instructional and descriptive. It provides a multi-step workflow for the agent to guide a user through a cognitive exercise.
- [PROMPT_INJECTION]: The instructions do not contain any patterns intended to bypass safety guardrails or override system instructions. It includes clear 'What NOT to do' constraints which serve as legitimate operational boundaries for the technique.
- [DATA_EXFILTRATION]: There are no commands or instructions that access the file system, environment variables, or perform network requests.
- [COMMAND_EXECUTION]: Neither the primary skill file nor the reference example contains shell commands or scripts for execution. References to 'npx' in the worked example are used as illustrative analogies for software distribution and are not instructions for the agent to run code.
Audit Metadata