bilibili-page-reader

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s required runtime workflow uses browser-evaluated providers to fetch and parse outsider-authored Bilibili page content (e.g., 投稿字幕 and danmaku JSON from BE providers / in-page API responses), which is then decoded as free text/JSON and returned into the agent’s LLM context via the evaluate result.