verification-before-completion

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: No security vulnerabilities were identified in the skill instructions. The content is limited to markdown-based procedural guidelines for software development and verification workflows.
  • [NO_CODE]: The skill contains only documentation and instructional text without any executable scripts, binary files, or configuration files that could pose a risk.
  • [PROMPT_INJECTION]: The instructions use strong imperative language to enforce accuracy and honesty; these are behavioral constraints for task integrity and do not attempt to bypass safety guardrails. The skill defines a process for reading command outputs (Ingestion point: command output in Step 3), which presents a standard tool-output injection surface with no specified boundary markers or sanitization. However, this capability is part of the intended verification logic and is assessed as safe in this context.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 06:47 PM