nano-banana

The nano-banana skill provides a streamlined interface for developers and creators to bridge Python scripting with the Gemini image generation API. By leveraging the uv package manager and heredoc-based execution patterns, users can perform rapid image generation, image editing, and custom data processing tasks directly within their CLI environment without the overhead of creating persistent project files. This skill is optimized for one-off tasks where speed and reproducibility are essential, effectively turning your terminal into a generative workspace.

• Perform AI-driven image generation using the gemini-3-pro-image-preview model with configurable aspect ratios and resolutions.

• Execute ad-hoc Python scripts using uv run with inline dependency management via TOML-style script metadata.

• Support for image-to-image workflows, allowing users to load local files and apply transformations using the Pillow library.

• Integrated error handling and status reporting, ensuring that generated images are saved consistently to the tmp/ directory for immediate visual review.

• Stateless execution model that ensures each script runs independently, preventing side-effect accumulation during complex iterative design sessions.

• Ideal for software developers, creative technologists, and data practitioners who need quick visual output or automated file processing.

• Trigger this skill when requesting AI art, image manipulation, or executing short Python snippets using phrases like 'generate an image', 'draw', or 'run python script'.

• Utilize the default 'Nano Banana Pro' configuration for high-quality production assets unless cost or speed constraints necessitate a simpler model.

• Follow the recommended loop of writing, running, evaluating, and refining to handle multi-step image generation tasks efficiently.

• Ensure the GOOGLE_API_KEY environment variable is configured correctly to enable seamless API interactions with Google Generative AI services.