naval-perspective
Pass
Audited by Gen Agent Trust Hub on Apr 12, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill implements a robust persona-based roleplay system. It uses specific trigger phrases and clear instructions for entering and exiting the character mode. These instructions are focused on personality simulation and do not attempt to override the underlying agent's safety protocols or system instructions.\n- [EXTERNAL_DOWNLOADS]: The skill uses an installation mechanism (
npx skills add) that fetches data from the author's public repository. As these resources originate from the identified vendor 'alchaincyf' and are part of the documented setup process, they are considered standard functional components.\n- [INDIRECT_PROMPT_INJECTION]: The skill includes an 'Agentic Protocol' that directs the agent to perform web searches to verify facts before responding. This category flags the following attack surface:\n - Ingestion points: Web search tool output (SKILL.md, Step 2).\n
- Boundary markers: None explicitly defined in the prompt instructions to delimit search results.\n
- Capability inventory: Persona-based response generation; no file-system or subprocess execution capabilities are exposed to the gathered content.\n
- Sanitization: No specific validation or filtering of external web content is mentioned.
Audit Metadata