The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/denoise.py script executes FFmpeg commands using subprocess.call().
The implementation passes arguments as a list of strings, which inherently prevents shell injection by avoiding the use of a shell interpreter (equivalent to shell=False).
Command arguments such as input/output paths and filter parameters are handled as distinct list elements.
[EXTERNAL_DOWNLOADS]: The skill provides guidance on downloading external model files (e.g., .pb, .rnnn) required for advanced FFmpeg filters.
The documentation in SKILL.md and references/filters.md points users toward well-known and reputable sources, specifically the official Xiph.Org RNNoise repository and the FFmpeg organization's project pages.
These downloads are documented for legitimate feature support and do not involve automated execution of untrusted remote scripts.
[PROMPT_INJECTION]: A review of the instructions and metadata confirms the absence of patterns designed to override agent behavior, bypass safety filters, or extract system prompts.
[DATA_EXFILTRATION]: No code or instructions were found that attempt to access sensitive files (like credentials or SSH keys) or transmit data to external unauthorized domains.

ffmpeg-denoise-restore