nano-banana-2

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly supports real-time Google search grounding via the "enable_google_search" input and accepts arbitrary image URLs in the "images" field, so the agent will fetch and interpret public web/search results and user-hosted images (untrusted third-party content) as part of its workflow, which can materially influence outputs.