nano-banana-pro

The Nano Banana Pro skill facilitates image synthesis and manipulation via the Gemini 3 Pro Image API. It operates directly from your active directory, ensuring all generated PNG assets remain local to your project workspace. The tool accepts natural language prompts to create original visuals from scratch or apply modifications to existing local image files. By standardizing filename conventions with embedded timestamps and mapping user language to resolution presets (1K, 2K, 4K), it simplifies the image creation pipeline for developers and designers. The tool requires a valid API key, either passed as a command argument or set via environmental variables, and handles execution through the uv package manager. This implementation is optimized for rapid iteration without requiring manual file management or external platform uploads.

When to Use This Skill

• •Creating placeholder assets for web or mobile UI prototypes
• •Modifying existing illustrations to match a specific art style
• •Generating custom social media banners from textual descriptions
• •Iterative prototyping of visual concepts during design brainstorming

How to Invoke This Skill

Example prompts that trigger this skill in Claude Code, Cursor, or Antigravity:

• “Generate a 4K image of a futuristic city skyline
• “Create a 2K image of a minimalist desk setup
• “Edit this photo to make the background look like a watercolor painting
• “Use the Nano Banana tool to generate a new graphic for the landing page
• “Add storm clouds to this existing mountain photo

Pro Tips

• 💡Experiment with detailed and descriptive prompts for image generation to achieve precise results, specifying styles, colors, and content.
• 💡When editing, provide clear and concise instructions, referencing specific elements in the input image for accurate modifications.
• 💡Utilize the `4K` resolution option for high-fidelity outputs, especially for final assets or prints, but be mindful of processing time for larger resolutions.

What this skill does

• •Generates high-resolution PNG images from text prompts
• •Performs stylistic or content-based edits on local image files
• •Supports 1K, 2K, and 4K output resolution toggles
• •Applies timestamped, descriptive naming conventions to output files
• •Provides direct execution via uv without path-switching overhead

When not to use it

• ✕Tasks requiring real-time video generation or animation
• ✕Scenarios needing extreme image upscaling beyond 4K parameters
• ✕Projects that require image analysis or metadata extraction instead of generation

Example workflow

1. User identifies a need for a new asset in the current directory
2. User executes the generation script with a specific descriptive prompt
3. Agent selects the appropriate resolution based on the user's request
4. Script generates the file and saves it with a timestamped filename
5. Agent confirms the saved file path back to the user
6. User proceeds to integrate the saved image into their project

Prerequisites

• –Valid Gemini API key
• –uv package manager installed
• –Read/write permissions for the current working directory

Pitfalls & limitations

• !Failing to provide an input image path during an edit request
• !Over-describing filenames leading to long or truncated labels
• !Environment variable conflicts if both a key argument and system variable exist

FAQ

How it compares

Unlike manual web-based generation, this tool automates file naming, directory placement, and parameter mapping, allowing for direct asset integration into your local code environment.

Source & trust

Recommended Rules

Recommended Workflows

Recommended MCP Servers