grilling
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary purpose is to help users evaluate decisions by walking through decision trees and resolving dependencies. No malicious intent was identified in the instructions or configuration.
- [PROMPT_INJECTION]: There are no instructions aimed at bypassing safety filters, extracting system prompts, or overriding agent constraints. The directions are purely behavioral for the defined task.
- [DATA_EXFILTRATION]: No network operations, hardcoded credentials, or sensitive file access patterns were found. While the skill mentions looking up facts via filesystem tools, this is part of the agent's expected functionality and does not include exfiltration logic.
- [COMMAND_EXECUTION]: The skill does not invoke any dangerous shell commands, privilege escalation techniques, or persistence mechanisms.
- [REMOTE_CODE_EXECUTION]: No external scripts are downloaded or executed. The skill does not define any third-party dependencies.
Audit Metadata