How does this skill handle large image files in Claude Code?

By setting returnInlineImage to false, the skill saves images directly to your disk, preventing base64 data from filling your context window and slowing down the session.

Which API key is required for this skill?

This skill requires a GEMINI_API_KEY from Google AI Studio, which must be added to your MCP server environment configuration.

Why should I use narrative prompts instead of keywords?

Narrative prompts provide better context for the AI, achieving approximately 94% coherence compared to only 61% when using comma-separated keyword lists.

Can I modify an image after it's been generated?

Yes, you can use the edit_image and continue_editing tools to make specific adjustments to the previous output, allowing for an iterative design process.

Nano Banana Image Generation

Name: Nano Banana Image Generation
Author: daveremy

bydaveremy

•

デザインツール

Generates and edits high-quality images and technical diagrams using narrative-driven prompts and iterative refinement tools.

This skill empowers Claude to create, modify, and refine visual assets using the Nano Banana 2 MCP server. It moves beyond simple keyword-based generation by utilizing narrative prompting techniques that significantly improve image coherence and spatial accuracy. Whether you need photorealistic imagery, flat-style UI icons, or complex educational diagrams, this skill provides granular control over resolution (up to 4K), aspect ratios, and spatial reasoning levels. It is specifically optimized for developer workflows, offering local file saving to manage context window limits and an iterative editing process to perfect visuals directly within Claude Code.

主な機能

01Context-aware image handling that saves files to disk to prevent token overflow

02Narrative-driven prompting for 94% coherence in complex visual scenes

037 GitHub stars

04Iterative editing workflow with continue_editing for precise refinements

05Advanced 'thinking' modes for improved spatial logic and text rendering

06Support for multiple resolutions (1K, 2K, 4K) and common aspect ratios

ユースケース

01Prototyping marketing banners and photorealistic conceptual art within a chat interface

02Creating UI placeholders, icons, and design assets during frontend development

03Generating technical diagrams, flowcharts, and infographics for documentation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add daveremy/nano-banana-2-mcp generate-image

For use in Claude.ai and ChatGPT

主な機能

01Context-aware image handling that saves files to disk to prevent token overflow

02Narrative-driven prompting for 94% coherence in complex visual scenes

037 GitHub stars

04Iterative editing workflow with continue_editing for precise refinements

05Advanced 'thinking' modes for improved spatial logic and text rendering

06Support for multiple resolutions (1K, 2K, 4K) and common aspect ratios

ユースケース

01Prototyping marketing banners and photorealistic conceptual art within a chat interface

02Creating UI placeholders, icons, and design assets during frontend development

03Generating technical diagrams, flowcharts, and infographics for documentation