The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run to interface with ffmpeg and ffprobe for video processing tasks in make_video.py and probe_moments.py. All calls use list-based arguments rather than shell strings, which effectively mitigates the risk of command injection from external filenames or configuration values.
[EXTERNAL_DOWNLOADS]: The scripts/bootstrap.sh utility creates a local Python virtual environment and installs the Pillow and NumPy libraries via pip. These are well-known, industry-standard packages for image manipulation and numerical computation.
[INDIRECT_PROMPT_INJECTION]: The skill processes external media files and a structured film.json configuration file. The analysis is limited to technical pixel differences and metadata extraction, presenting a minimal attack surface. Ingestion points include the film.json configuration and the source video files; capability inventory is restricted to local video processing and file writes for the generated output; sanitization is handled via structured JSON parsing.

shared-perf-comparison-video