nano-banana
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
beltCLI command via the Bash tool to perform model inference, searching, and application management. This is the intended behavior for interacting with the specified inference platform.- [EXTERNAL_DOWNLOADS]: The skill references external resources including installation scripts and documentation from the officialinference.shwebsite and its GitHub organization. These are standard resources for the service.- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text prompts and external image URLs to be processed by Gemini models. While this represents a surface for untrusted data, it is a core and expected part of the image generation functionality and is mitigated by the provider's native safety filters.
Audit Metadata