The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external media files (video/audio) using FFmpeg, which creates a potential surface for indirect prompt injection if the agent processes files with malicious metadata or specially crafted structures.
Ingestion points: External files are ingested via the -i parameter in ffmpeg and ffprobe commands.
Boundary markers: No specific delimiters or instructions to ignore embedded metadata are included in the command templates.
Capability inventory: The skill provides instructions for executing ffmpeg, ffprobe, and shell commands (echo, for loops) that can read and write files to the local system.
Sanitization: There is no explicit sanitization or validation of input filenames or file content before processing.
[EXTERNAL_DOWNLOADS]: The skill references an external project repository for feedback and contributions.
Evidence: Points to the GitHub repository github.com/digitalsamba/claude-code-video-toolkit for collaborative updates and issues.

ffmpeg