bilibili-analyzer

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

• Third-party content exposure detected (low risk: 0.10). The workflow ingests outsider-authored video content (Bilibili videos) and extracts frames that are analyzed by the LLM as free text/images containing text, which constitutes low-risk indirect prompt injection via fetched third-party content.

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

• Potentially malicious external URL detected (medium risk: 0.30). The skill makes programmatic HTTP calls to the Bilibili API endpoints (https://api.bilibili.com/x/web-interface/view and https://api.bilibili.com/x/player/playurl) at runtime via scripts/prepare.cs to fetch video metadata and download media streams without strong publisher provenance or cryptographic verification.