nano-banana-pro
Pass
Audited by Gen Agent Trust Hub on Mar 21, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill transmits the user's image prompts and input images to
ai-gateway.happycapy.aifor processing. This network communication is a core component of the skill's image generation functionality and targets the author's own infrastructure. The request includes an Origin header forhttps://trickle.soas required by the gateway.- [PROMPT_INJECTION]: The skill processes user-supplied prompts and images to generate model outputs. While the skill possesses the surface for indirect prompt injection, the risks are mitigated by the specific nature of the output (image files) and the lack of high-privilege system capabilities. - Ingestion points: User input provided via the
--promptand--input-imageCLI arguments inscripts/generate_image.py. - Boundary markers: User input is combined with image data and passed directly to the API without specific delimitation or behavioral constraints.
- Capability inventory: The skill's operations are limited to image generation, format conversion using the
PILlibrary, and writing the resulting file to a user-specified path. It does not perform shell execution or modify sensitive system files. - Sanitization: The skill does not implement custom filtering or sanitization of the input prompt strings.
Audit Metadata