analyze-stock

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md 的 Agent 1/2/3/4/5 在运行时通过 WebSearch/WebFetch 抓取公开网页与新闻原文（外部作者内容），并将其摘要/原文内容与来源 URL 一起喂入各子 Agent 的 LLM 上下文，属于“公共 web 内容/新闻原文”类间接提示注入风险。

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The skill explicitly instructs running Chrome headless with the --no-sandbox flag (bypassing Chrome's sandbox security) and also directs executing local scripts, which encourages bypassing security mechanisms and running arbitrary code on the host.