film-breakdown

Pass

Audited by Gen Agent Trust Hub on Oct 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • External Command Execution: The skill heavily utilizes the ffmpeg and ffprobe command-line tools via node:child_process (spawn/spawnSync) to perform video decoding, frame extraction, and analysis. This is a core requirement for its functionality but warrants oversight of the input filenames passed to these tools.
  • External Audio Processing: The skill optionally uses faster-whisper via a Python bridge (whisper_words.py) for audio transcription. It prompts the user to install these dependencies if they are missing (pip install faster-whisper).
  • Indirect Prompt Injection Surface: The skill processes external video files and extracts metadata (codec, duration, audio streams) which are then interpolated into generated markdown reports (SHOTLIST.draft.md). While this could technically be a vector for indirect injection, the impact is low as the output is intended for human review and further manual editing by the developer.
  • Dynamic Context and Execution: Scripts like breakdown.mjs and compare.mjs dynamically generate configuration files (timeline.draft.js) based on video analysis. These files are intended to be manually integrated into the user's project timeline by the developer.
  • File System Operations: The scripts perform extensive file system operations (read/write/delete) within a designated /workspace/ directory to manage caches, review images, and analysis results.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 28, 2026, 05:18 AM