skills/varghq/skills/video-moodboard/Gen Agent Trust Hub

video-moodboard

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external searches (FrameThrower, Refs, X) and user-provided reference images to generate style guides. Evidence chain: 1) Ingestion points: External search results and reference images are processed to extract hex codes and lens metadata. 2) Boundary markers: The skill uses structured templates and a "Never" list of vague adjectives to constrain the agent's output. 3) Capability inventory: The agent can perform network requests and file operations. 4) Sanitization: The instructions mandate human review for all generated stills and boards before final approval.
  • [EXTERNAL_DOWNLOADS]: The skill references and fetches content from vendor-owned infrastructure (varg.ai) and well-known technology services. This includes loading font assets from Google Fonts and querying established film databases (FrameThrower, Refs) for visual references. These operations are within the expected scope of the skill's art direction functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 09:50 PM