dLazy Nano Banana 2.0 Image Generation

Problem

When you need to turn natural-language prompts, reference images, or existing assets into higher-quality images, Nano Banana 2.0 supports text-to-image and image-to-image workflows. The banana2 skill wraps this capability into a callable CLI flow, reducing repetitive work such as assembling API requests, handling local file uploads, and polling async tasks.

How It Works

• Core command: run dlazy banana2 to start an image generation or editing task.
• Input handling: prompts, parameters, and local image paths are sent to dLazy's hosted API; local files are uploaded to dLazy media storage so the model can read them.
• Result delivery: synchronous calls return the generated result; --no-wait returns task.generateId and status, which can be polled with dlazy status --wait; result URLs are served from dLazy media storage.
• Authentication: a dLazy API key is required and can be stored through the CLI or supplied per run via DLAZY_API_KEY.

Boundaries and Notes

This skill is a thin client over dLazy's cloud API, so image inference is not performed locally; local files, when used, are uploaded to the cloud service. If it returns unauthorized, check the API key; if it returns insufficient_balance, top up credits; if an async task fails, read the backend error message for the specific cause.