verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Jul 20, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists entirely of instructional markdown and metadata. It does not include any executable scripts, binaries, or configuration files that could pose a direct security risk.
- [PROMPT_INJECTION]: The skill utilizes strong, imperative language (e.g., 'Iron Law', 'Hard gate') to modify the agent's behavior. However, these instructions are aimed at improving the reliability of the agent's task reporting and do not attempt to bypass safety filters, extract sensitive information, or enable unauthorized actions.
- [DATA_EXFILTRATION]: There are no patterns suggesting data harvesting or external exfiltration. The only URLs present are references to the author's domain and the upstream source repository.
- [NO_CODE]: The skill does not provide or execute any code. It functions as a set of logical constraints for the agent to follow during its normal operation.
Audit Metadata