agent-visual-verification

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

• Third-party content exposure detected (low risk: 0.10). The skill’s runtime workflow explicitly treats page text/pixels as untrusted but it only performs visual screenshot capture via a typed URL/selector path (using tools like web_extract/web_search/vision_analyze as directed by the task), so outsider-authored free text is ingested only when the user requests capture of an external page and selects the target.