wechat-article-writer

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md Step 2 explicitly requires the agent to search and ingest multiple public sources including X/Twitter, Reddit and technical forums ("并行搜索多来源（官方文档、X/Twitter、Reddit、技术论坛等）") and instructs the agent to deeply summarize that material before writing, so untrusted user-generated content from the open web will be read and can directly influence actions and outputs.