wechat-article-pipeline

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

• Third-party content exposure detected (medium risk: 0.30). 运行时会把“引导式设计”里由用户填写/选择产生的封面/配图 prompt 等自由文本直接传入 scripts/gen-image.mjs，再由其调用 POST https://doubaoya.com/api/skills/gpt-image-gen/invoke 生成图片（属于外部用户可控文本进入后端推理链路）。

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly performs a runtime fetch of https://doubaoya.com/api/wechat/writing-spec and tells the agent to "按它组织结构再动笔 / 要照着写的就是它", so remote content from that URL directly controls the agent's writing instructions at runtime.