career-resume-tailor

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

• Third-party content exposure detected (high risk: 0.78). 通过 Step 1/4 的运行时联网抓取：用户提供的“目标岗位/JD 链接”可能来自公开网页（如 Boss/牛客等）的自由文本正文，被读入并用于生成对齐，从而产生间接的外部提示注入风险。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly fetches user-provided job descriptions at runtime (e.g., https://www.zhipin.com/job_detail/...) and uses the fetched JD text to drive the model's tailoring decisions, so external URLs are read at runtime and their content directly controls the agent's output.