bezos-perspective
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses detailed role-play instructions to establish a strategic advisor persona. These instructions are specific to the skill's intended purpose of providing business advice and do not contain patterns for bypassing safety filters, system prompt extraction, or ignoring previous instructions.
- [DATA_EXFILTRATION]: There is no evidence of sensitive file access (such as credentials, SSH keys, or environment files) or network exfiltration attempts. The skill operates purely within the context of conversational interaction.
- [COMMAND_EXECUTION]: No shell commands, binary scripts, or system-level execution patterns were identified within the skill's body or reference materials.
- [EXTERNAL_DOWNLOADS]: The documentation includes standard references to public GitHub repositories and platform-specific installation commands. No unauthorized remote code execution or runtime script downloads from untrusted sources are present.
- [SAFE]: The skill consists of high-level prompting logic and educational material related to Amazon's leadership principles, posing no technical risk to the environment.
Audit Metadata