The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user input and images, which is a necessary function for its primary purpose of image generation.
Ingestion points: User-provided text is ingested via the --prompt argument, and existing images are ingested via the --reference argument in scripts/image.py.
Boundary markers: Absent; user-provided text is concatenated with internal aspect ratio instructions before being transmitted to the model.
Capability inventory: The skill performs network requests to Google's API and has the capability to write generated files to the local file system.
Sanitization: No sanitization is performed on the prompt text or the content of reference images before they are processed by the generative model.
[EXTERNAL_DOWNLOADS]: The skill utilizes well-known, official libraries from established organizations to facilitate its functionality.
The scripts/image.py script identifies dependencies on google-genai (an official Google library) and pillow for image handling, which are managed by the uv tool.

nano-banana-pro