The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous commands for the yt-dlp utility to download video, audio, and metadata. These are standard operations for media management tasks.- [DATA_EXFILTRATION]: The skill includes instructions for using --cookies-from-browser, which allows the tool to access browser session data to bypass age restrictions or access private content. While this involves sensitive data access, it is a documented feature of the legitimate tool described.- [PROMPT_INJECTION]: The skill is designed to process untrusted external data in the form of YouTube transcripts and metadata.
Ingestion points: Untrusted data enters the context via yt-dlp transcript downloads (SRT files) and API responses from YouTube.
Boundary markers: No specific boundary markers or 'ignore' instructions are provided when passing transcript text to the agent for summarization.
Capability inventory: The skill has the ability to execute shell commands (yt-dlp), write files, and make network requests (requests).
Sanitization: The provided Python script uses regular expressions to remove timestamps and indices from SRT files, but it does not filter for malicious instructional content.

youtube-content