gemini-viral-shorts

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

• Third-party content exposure detected (high risk: 0.80). The skill's main script (scripts/gemini_viral_shorts_post.py) explicitly downloads a public YouTube video via yt-dlp and uploads that video file to Google Gemini for analysis, so untrusted, user-generated content from arbitrary YouTube URLs is ingested and can influence the model's outputs (enabling indirect prompt injection).

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

• Potentially malicious external URL detected (high risk: 0.80). The skill downloads a user-provided YouTube URL (e.g., https://www.youtube.com/watch?v=...) at runtime via yt-dlp, uploads that fetched video to Google Gemini, and injects the video as input to the model.generate_content call, so the externally fetched content directly controls the model's output and is a required runtime dependency.