youtube-knowledge-learner

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests YouTube captions/transcripts from public videos (see SKILL.md "Inputs" and "Transcript Acquisition" and scripts/fetch_youtube_knowledge.py which uses yt-dlp to download caption text) and instructs the agent to read and apply that user-generated/public transcript as authoritative context for downstream tasks, so untrusted third‑party content could inject instructions that influence actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The helper script calls yt-dlp on user-supplied YouTube URLs (e.g. https://www.youtube.com/watch?v=dQw4w9WgXcQ) and then urllib.request.urlopen() to fetch the caption file URL returned by yt-dlp at runtime, injecting the downloaded transcript text into the agent's context which directly controls model input.