nano-banana-pro
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted user input (text prompts and image data) which is passed directly to an LLM without sanitization, creating an indirect prompt injection surface.\n
- Ingestion points: The
args.promptCLI argument and the content of images loaded viaPILImage.openinscripts/generate_image.py.\n - Boundary markers: Absent; the script interpolates the prompt and images directly into the request payload to the Gemini API.\n
- Capability inventory: The skill performs network calls to Google APIs, reads local files, and writes generated images to the local filesystem.\n
- Sanitization: None; the script does not validate or sanitize the prompt or image content before processing.
Audit Metadata