nano-banana-2

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly accepts arbitrary image URLs and includes an "enable_google_search" input option and example ("With Google Search Grounding"), meaning it fetches and grounds outputs on open web content that the agent will read and use to influence generation.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The Quick Start instructs users to run "curl -fsSL https://cli.inference.sh | sh", which fetches and executes a remote installer (that then downloads binaries from dist.inference.sh) as a required dependency for running the skill, so remote code is executed at runtime (https://cli.inference.sh).