behavioral-modes

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Prompt Injection] (SAFE): The skill defines operational modes (personas) like BRAINSTORM and IMPLEMENT. These are legitimate behavioral instructions and do not contain patterns intended to bypass safety filters or ignore system constraints.
  • [Data Exposure & Exfiltration] (SAFE): No credentials, sensitive file paths, or exfiltration patterns were detected. Allowed tools (Read, Glob, Grep) are restricted to read-only access.
  • [Remote Code Execution] (SAFE): No remote scripts, package installations, or dynamic execution patterns are present.
  • [Indirect Prompt Injection] (SAFE): Although the skill defines tools that ingest file data, it lacks the high-risk capabilities (like network-write or code execution) that would make this surface exploitable. All tool usage is restricted to the local environment with read-only permissions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:29 PM