verification-before-completion
Pass
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses authoritative framing and strict directives (e.g., 'Iron Law', 'Gate Function') to override the agent's default conversational behavior. It employs high-pressure language and instructions to ignore internal success reports in favor of external evidence.
- [COMMAND_EXECUTION]: The skill mandates the execution of verification commands (such as tests or builds) to substantiate any claims of completion. It requires identifying, running, and reading the full output of these commands as a prerequisite for communication.
- [SAFE]: While the skill uses strong behavioral constraints, its directives are defensive and aimed at ensuring honesty and technical accuracy. No malicious patterns or security bypasses were detected.
Audit Metadata