nano-banana

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md includes a "With Google Search Grounding" example and an enable_google_search input option that tell the app to fetch and use real-time Google Search results (public web pages) as grounding, so the agent will ingest and act on open, third‑party web content that could carry injected instructions.

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

• Potentially malicious external URL detected (high risk: 0.90). The Quick Start explicitly runs "curl -fsSL https://cli.inference.sh | sh", which fetches and executes a remote installer (from https://cli.inference.sh and its dist.inference.sh downloads) that is required to install the infsh CLI used by the skill, so https://cli.inference.sh is a runtime external dependency that executes remote code.