skills/imraywang/rayskills/ray-broll/Gen Agent Trust Hub

ray-broll

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill makes extensive use of the subprocess module to call ffmpeg and ffprobe. These calls are used for media processing tasks, such as determining clip durations, scaling images to project dimensions, stripping audio from source videos, and concatenating multiple video/audio segments into a final product.
  • [EXTERNAL_DOWNLOADS]: The skill downloads generated video files from remote servers. generate_video.py fetches content from Google's File API (generativelanguage.googleapis.com), while fal_driver.py downloads videos from fal.run after generation via third-party models. Both sources are legitimate for the skill's creative purpose.
  • [DATA_EXFILTRATION]: In its alternative processing path, fal_driver.py transmits user-provided or skill-generated image data to external endpoints at fal.ai. This data is sent as Base64-encoded data URIs within API requests to facilitate video generation, which is a documented feature of the skill.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it interpolates user-supplied text (such as spoken lines or topics) directly into templates used for image and video generation prompts. While no explicit sanitization or safety delimiters are visible in the code, the risk is mitigated by the safety guardrails inherent in the external AI services being queried.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 12:11 PM
Security Audit — agent-trust-hub — ray-broll