Does editing an image overwrite the original file?

No, each edit creates a new file, ensuring the original image is always preserved for versioning and safety.

Where are the generated images stored by default?

All generated and edited images are saved to the ~/nanobanana-images/ directory to keep your workspace organized.

How do I ensure my prompts get the best results?

Follow the structured pattern: [Subject] + [Style] + [Composition] + [Lighting/Mood] + [Details] for maximum prompt clarity.

Which Gemini model is best for rendering text in images?

The gemini-3-pro-image-preview model is recommended for complex prompts involving text rendering and highest overall quality.

What are the file size and format limitations?

The skill supports JPEG, PNG, WebP, and GIF inputs under 20MB, and always outputs high-quality PNG files.

Gemini Image Generation Prompting

Name: Gemini Image Generation Prompting
Author: DojoCodingLabs

byDojoCodingLabs

0•

Design Tools

Optimizes AI image generation and editing using Gemini models through structured prompting and iterative refinement.

This skill enhances Claude's ability to interface with the Nano-Banana-MCP server for secure, Gemini-powered image generation. It provides a specialized framework for constructing high-performing prompts, offers expert guidance on selecting the right Gemini model for specific tasks like text rendering or rapid prototyping, and outlines a robust iterative workflow for image editing. By integrating security-hardened file pathing and technical constraints, it ensures that visual asset creation remains safe, organized, and high-quality within your development environment.

Key Features

01Structured prompting formula covering subject, style, composition, and mood.

02Iterative editing workflows for granular, non-destructive image refinements.

030 GitHub stars

04Intelligent model selection between Gemini Flash and Pro for speed vs. quality.

05Support for advanced style transfer using reference image parameters.

06Secure file management constrained to authorized home and temporary directories.

Use Cases

01Iteratively refining marketing graphics with specific text and color adjustments.

02Rapidly prototyping visual concepts using high-speed Gemini Flash models.

03Generating high-fidelity UI placeholders and visual assets during frontend development.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dojocodinglabs/nanobanana-mcp image-prompting

For use in Claude.ai and ChatGPT

Key Features

01Structured prompting formula covering subject, style, composition, and mood.

02Iterative editing workflows for granular, non-destructive image refinements.

030 GitHub stars

04Intelligent model selection between Gemini Flash and Pro for speed vs. quality.

05Support for advanced style transfer using reference image parameters.

06Secure file management constrained to authorized home and temporary directories.

Use Cases

01Iteratively refining marketing graphics with specific text and color adjustments.

02Rapidly prototyping visual concepts using high-speed Gemini Flash models.

03Generating high-fidelity UI placeholders and visual assets during frontend development.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dojocodinglabs/nanobanana-mcp image-prompting

For use in Claude.ai and ChatGPT