verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION] (SAFE): The skill uses strong, authoritative language (e.g., 'The Iron Law', 'non-negotiable') to enforce operational discipline. While the phrasing is assertive, it does not attempt to override system safety guidelines or bypass AI constraints; rather, it encourages accuracy and honesty in task reporting.
- [COMMAND_EXECUTION] (SAFE): Although the documentation references executing verification commands like tests or builds, it does not provide any specific, hardcoded, or malicious shell commands. It serves as a behavioral guideline rather than a script-carrying module.
- [DATA_EXFILTRATION] (SAFE): No access to sensitive files, environment variables, or network endpoints is requested or present in the instruction set.
- [EXTERNAL_DOWNLOADS] (SAFE): No remote dependencies or external scripts are referenced or executed.
Audit Metadata