The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches video content from remote URLs provided by the user via the 'yt-dlp' utility.
[COMMAND_EXECUTION]: Executes shell commands including yt-dlp, ls, and mkdir to manage downloads and verify files.
[PROMPT_INJECTION]: Vulnerable to indirect prompt injection because the skill ingests untrusted metadata (such as video titles and descriptions) from external sites which could contain malicious instructions.
Ingestion points: Video metadata and filenames returned by 'yt-dlp' after processing a URL (SKILL.md).
Boundary markers: Uses double quotes to wrap the URL argument in shell commands.
Capability inventory: Accesses Bash to run yt-dlp, ls, and mkdir.
Sanitization: None detected; the skill relies on standard shell quoting and does not filter or sanitize the fetched metadata before it enters the agent's context.

download-video