mage-video-remake

Pass

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided video files to extract pacing, captions, and visual descriptions. This external content is used to automatically plan and prompt generative models for a remake.\n
  • Ingestion points: Video files processed by rea-video to generate metadata in REEL.md and timeline.md.\n
  • Boundary markers: The skill instructs the agent to present a plan and cost estimate to the user and "Wait for a yes" before starting generation.\n
  • Capability inventory: The skill can execute shell commands (ffmpeg, pipx), perform network uploads, and invoke generative AI tools.\n
  • Sanitization: There is no explicit sanitization of extracted OCR or VLM text before it is interpolated into generative prompts.\n- [EXTERNAL_DOWNLOADS]: The skill requires the installation of the rea-video Python package from a specific commit on a GitHub repository (JuneYaooo/rea-video).\n
  • The installation bypasses standard package registries by fetching code directly from a specific git commit hash.\n
  • The tool also downloads external machine learning models for audio separation and optical character recognition during its initial execution.\n- [COMMAND_EXECUTION]: The skill makes extensive use of local shell commands to perform its core functionality.\n
  • It uses ffmpeg for complex media editing, including joining clips, applying caption overlays, and mixing audio.\n
  • It uses rea-video for automated decomposition and analysis of source video files.\n
  • It uses pipx and python environment tools for software installation and setup.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 13, 2026, 03:49 PM