island-rescue
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Behavioral Override]: The skill includes a specific directive requiring the agent to modify its natural language output by inserting the word STOP every four words. This is a pattern used to override the default communication style of the agent.
- [Instructional Secrecy]: The skill contains a directive for the agent to hide the formatting rule from the user, which is a common characteristic of behavioral prompts that aim to alter interaction dynamics without the user's explicit knowledge.
Audit Metadata