nano-banana-2-imagegen
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a specialized Python script (
scripts/generate_image.py) intended for command-line execution. It allows users to generate images from prompts or edit existing ones by passing file paths and text via arguments. - [EXTERNAL_DOWNLOADS]: The skill utilizes the official
google-genailibrary from Google and the well-knownPillowimage processing library. It facilitates communication with Google's Gemini API to perform image generation tasks. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process user-provided prompts and images, which constitutes an attack surface for indirect prompt injection. A malicious prompt or an image containing hidden instructions could potentially influence the model's output.
- Ingestion points: Data enters the context via the
promptargument, the--prompt-filepath, and the--input-imagefiles inscripts/generate_image.py. - Boundary markers: The script does not implement specific boundary markers or 'ignore' instructions to isolate user input from the rest of the model's instructions.
- Capability inventory: The skill possesses the capability to write files to the local disk (
--output,--state-file) and perform network requests to the official Google Gemini API. - Sanitization: No specific content sanitization or filtering is applied to the input prompt or images before they are passed to the Gemini API.
Audit Metadata