media-processing

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill interacts with system binaries (ffmpeg, ffprobe, and magick) using the Python subprocess module. All script implementations, including scripts/video_optimize.py, scripts/batch_resize.py, and scripts/media_convert.py, correctly avoid the use of shell=True and pass arguments as lists. This approach effectively mitigates potential shell injection vulnerabilities arising from malicious media filenames or parameters.
  • [EXTERNAL_DOWNLOADS]: Documentation within SKILL.md provides instructional references to official download pages for media tools. These links target established and well-known domains such as ffmpeg.org and imagemagick.org, or recommend standard package managers like brew and apt for binary installation.
  • [PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection because it reads and processes data from external media files.
  • Ingestion points: Untrusted data enters the context through media file parsing in scripts/video_optimize.py (via ffprobe) and scripts/batch_resize.py (via magick).
  • Boundary markers: The skill lacks specific delimiters or 'ignore' warnings for the data extracted from these files.
  • Capability inventory: The skill maintains the capability to execute system binaries and write to the filesystem via subprocess.run calls located in all primary script files.
  • Sanitization: While technical sanitization against shell injection is present through the use of argument lists, semantic sanitization of the processed metadata is not implemented.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 07:09 AM