wechat-article-writer

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

• Third-party content exposure detected (high risk: 0.90). The skill's Step 2 ("搜索资料") explicitly instructs the agent to parallel-search and ingest content from open/public sources (e.g., official docs, X/Twitter, Reddit, 技术论坛) and "深度总结后再进入撰写", so untrusted, user-generated third‑party content is read and used to drive writing and publishing decisions—meeting the criteria for potential indirect prompt injection.