carousel-benchmark

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

• Third-party content exposure detected (medium risk: 0.30). The skill workflow involves scanning public TikTok search results, profiles, and posts as well as fetching public slideshow banks (autovirality.com/tiktok/slideshows), all of which ingest outsider-authored free text and content.

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

• Potentially malicious external URL detected (medium risk: 0.30). The skill instructs fetching content from autovirality.com/tiktok/slideshows as a public slideshow bank for candidate discovery, representing an unverified third-party web dependency tiering as medium.