The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/main.py executes the yt-dlp command-line utility using the subprocess module.
Evidence: Multiple calls to subprocess.run appear throughout scripts/main.py (e.g., in get_video_info, download, audio, transcript, and playlist functions) to invoke the external yt-dlp binary.
Context: The implementation uses lists for command arguments and avoids shell=True, which is a security best practice to prevent command injection. Filenames are also passed through a sanitize_filename function before use.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external source (YouTube) that is subsequently presented to the agent.
Ingestion points: scripts/main.py fetches video titles, descriptions, tags, and transcripts using the yt-dlp tool.
Boundary markers: No specific boundary markers or instructions are provided to the agent to treat this external content as untrusted data.
Capability inventory: The skill possesses the ability to write files to the local system and execute the yt-dlp binary.
Sanitization: While filenames are sanitized to prevent path traversal or shell issues, the textual content of metadata and transcripts is not filtered for potential prompt injection patterns before being returned to the agent context.

youtube-downloader