verification-before-completion

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [General] (SAFE): The skill consists solely of a markdown file (SKILL.md) and does not include any executable scripts, binaries, or automation configuration.
  • [Prompt Injection] (SAFE): Although the content uses imperative language (e.g., 'The Iron Law', 'Non-negotiable'), these instructions are focused on operational accuracy and do not attempt to bypass AI safety filters, ignore system instructions, or extract sensitive data.
  • [Indirect Prompt Injection] (SAFE): The skill identifies a surface where the agent processes external inputs such as version control diffs and third-party reports, but it does not provide code to perform these actions and explicitly mandates a defensive verification strategy to ensure the integrity of those inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:36 PM