video-download

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). The skill explicitly fetches content from open/public URLs using yt-dlp (e.g., "Download video and audio from URLs using yt-dlp" and commands like yt-dlp "URL" --dump-json and --write-subs), which exposes untrusted, user-generated metadata/subtitles/descriptions that the agent may read and which could influence subsequent actions.