nano-banana-2

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references the installation of the belt CLI from the vendor's GitHub repository and the inferencesh Python package from a standard registry. These are official resources provided by the skill author for interacting with their platform.
  • [COMMAND_EXECUTION]: The skill instructions heavily utilize the belt CLI tool for authenticating, running image generation apps, and sampling configurations. The allowed-tools metadata restricts command execution to this specific toolset, reducing the attack surface.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an interface for the agent to process user-supplied strings and image URLs to generate or edit content. While this creates a surface for instructions embedded in data, the risk is inherent to the image generation use case and is categorized as low due to the use of structured JSON inputs and the specific scope of the model's output.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 11:59 PM
Security Audit — agent-trust-hub — nano-banana-2