model-resources-collection

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

• Third-party content exposure detected (medium risk: 0.30). 该skill在运行时会根据（受sources.md预置的）多个URL进行联网访问/无头浏览（Playwright）并读取 document.body.innerText（以及通过网页抓取/搜索结果抽取文本），因此若其中包含外部可被操纵的页面内容，LLM会摄入外部作者的自由文本。

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

• Potentially malicious external URL detected (high risk: 0.90). The Playwright fetch script (scripts/fetch-js-pages.js) will visit and render external pages such as https://www.volcengine.com/activity/codingplan (and the other URLs listed in that script), causing remote JavaScript to be fetched and executed in a headless browser and the skill explicitly requires this runtime behavior to obtain data.