nanobanana
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation suggests installing the
uvtool using a script fromastral.sh, which is the official distribution domain for this well-known Python developer tool. - [COMMAND_EXECUTION]: Provides documented commands for running image generation scripts using the local Python interpreter or the
uvtask runner. - [INDIRECT_PROMPT_INJECTION]: The skill provides an interface for the agent to ingest user-supplied text prompts and reference images, which are then passed to the Google Gemini API.
- Ingestion points: Prompts and image paths are passed via CLI arguments to
scripts/generate.pyandscripts/batch_generate.py. - Boundary markers: The prompt is passed directly as a string to the API client without explicit boundary markers.
- Capability inventory: The skill possesses capabilities to read local image files, write generated images to the filesystem, and perform network requests to Google's API.
- Sanitization: The implementation relies on Google Gemini's internal safety filters to manage and block potentially harmful prompt content.
Audit Metadata