create-vignette-ad

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Several Python scripts, including composite_variants.py, fire_t2v_variants.py, and render_overlays.py, utilize the subprocess module to execute system commands for ffmpeg, rsvg-convert, and the higgsfield CLI. These tools are integral to the skill's purpose of rendering and compositing video variants.
  • [EXTERNAL_DOWNLOADS]: The skill performs network operations to download media assets and model outputs from well-known platforms including fal.ai, higgsfield.ai, pexels.com, and Shopify CDNs. It also retrieves typography resources from fontshare.com and jsdelivr.net. These activities are documented and consistent with the intended media production workflow.
  • [PROMPT_INJECTION]: The skill contains logic for interpolating user-provided inputs, such as brand URLs and product handles, into shell commands used for scraping metadata and images. This creates a potential surface for indirect prompt injection if user inputs are not properly sanitized before being passed to the execution environment.
  • [SAFE]: The skill adheres to security best practices for secret management by instructing the agent to utilize local environment files for sensitive API keys and maintaining CLI credentials in standard configuration directories.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 03:07 PM
Security Audit — agent-trust-hub — create-vignette-ad