summarise-youtube

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

• Third-party content exposure detected (medium risk: 0.30). The required workflow ingests outsider-authored free text because the user-provided YouTube URL determines which English .vtt subtitles are fetched via yt-dlp and then read by the LLM at runtime for summarisation.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill runs yt-dlp at runtime to download subtitles from the supplied YouTube '' and then reads/injects that fetched transcript into the model to produce summaries, so remote content directly controls the agent's output.