reels-creator

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Multiple Python and Node.js scripts in the references/code/ directory (e.g., host_reel.py, host_lipsync.py, render.mjs) use subprocess.run or child_process.spawn to interface with ffmpeg and ffprobe. These calls consistently use list-based argument passing, which prevents shell command injection vulnerabilities.
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download of media assets from fal.ai generation endpoints and provides instructions for installing standard dependencies from the official PyPI (fonttools, brotli, numpy, Pillow) and NPM (playwright) registries. These operations are transparent, documented, and essential for the skill's primary function of video production.
  • [INDIRECT_PROMPT_INJECTION]: While the skill processes external data such as transcriptions (scribe.json) and edit configurations (edit.json), it mitigates prompt injection risks by rendering textual content directly into image frames via the Pillow library rather than passing it to a high-level shell or interpreter. This architecture ensures that data content cannot be misinterpreted as executable commands.
  • [DYNAMIC_EXECUTION]: The skill uses Playwright's evaluate function to control browser-based animation scenes. This is a standard and necessary practice for headless browser automation and is restricted to the local context of the animation scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 07:17 AM