NanoBanana-2 (Google AI Pro Native Image Generation)
This skill guides the AI to use its native image generation capability to satisfy image requests.
Workflow
- When the user requests an image generation using "nanobanana 2", "nanobanana-2", "蕉蕉图", or "香蕉生图", the agent MUST immediately invoke the native
generate_imagetool. - Use the following parameters for
generate_image:Prompt: The prompt provided by the user (optimized for Gemini's image generation model).ImageName: A clean, snake_case filename describing the content.AspectRatio: Default to '1:1' unless the user specifies otherwise.
- CRITICAL WORKSPACE SYNC:
- Immediately after generation, the agent MUST copy the generated image from the sandbox brain directory to the user's workspace root directory (
d:/Zero Tools/). - Use a clear filename such as
nanobanana_out.jpgor a descriptive name related to the prompt. - Run a powershell command to copy it, e.g.:
Copy-Item "C:/path/to/generated/image.jpg" "d:/Zero Tools/filename.jpg"
- Immediately after generation, the agent MUST copy the generated image from the sandbox brain directory to the user's workspace root directory (
- Present the resulting image to the user and always provide a clickable file:// link to the copied image in the workspace root so the user can download and view it easily.