ima-skill

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and download arbitrary public URLs and HTTP responses (see knowledge-base/SKILL.md "添加 URL 到知识库" and related examples using curl -sI -L, downloading to a temp dir, and root/notes SKILL.md which require handling "来自 WebFetch / HTTP 请求的内容"), thereby ingesting untrusted third‑party web content that the agent reads and uses to decide actions (media_type detection, uploads, import_urls, creating/appending notes), so such content could carry indirect prompt injections that influence behavior.