behavioral-modes

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Prompt Injection] (SAFE): The instructions are legitimate behavioral guidelines (e.g., BRAINSTORM, IMPLEMENT) intended to focus the AI's output. No bypass or jailbreak patterns were detected.
  • [Data Exposure & Exfiltration] (SAFE): No credentials, secrets, or sensitive file paths (like .ssh or .aws) are present. The allowed tools (Read, Glob, Grep) are restricted to local file access without network capabilities.
  • [Unverifiable Dependencies & Remote Code Execution] (SAFE): The skill does not reference external packages, installation scripts, or remote URLs for code execution.
  • [Indirect Prompt Injection] (SAFE): While the skill defines modes for reviewing or debugging code (which could be untrusted), the lack of high-privilege tools or network access mitigates the risk of automated exploitation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:45 PM