verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of instructional text without any executable scripts, external dependencies, or network-enabled commands. All procedures described are meta-instructions for the agent's internal reasoning process.
- [PROMPT_INJECTION]: The content uses strong imperative language and psychological framing (e.g., 'If you lie, you'll be replaced') to enforce compliance. While these techniques are common in prompt engineering to improve model reliability, they do not attempt to bypass safety guidelines, exfiltrate data, or redirect the agent to unauthorized actions.
Audit Metadata