Agent Skills: Nanobanana Image Generation

Use when "nanobanana", "generate image", "create image", "edit image", "AI drawing", "Gemini image", "image generation"

UncategorizedID: eyadsibai/ltk/nanobanana

Install this agent skill to your local

pnpm dlx add-skill https://github.com/eyadsibai/ltk/tree/HEAD/plugins/ltk-core/skills/nanobanana

Skill Files

Browse the full folder contents for nanobanana.

Download Skill

Loading file tree…

plugins/ltk-core/skills/nanobanana/SKILL.md

Skill Metadata

Name
nanobanana
Description
Use when "nanobanana", "generate image", "create image", "edit image", "AI drawing", "Gemini image", "image generation"

Nanobanana Image Generation

Generate and edit images using Google Gemini API.


Prerequisites

| Requirement | Setup | |-------------|-------| | API Key | Export GEMINI_API_KEY or add to ~/.nanobanana.env | | Dependencies | pip install google-genai Pillow python-dotenv | | Script | ${CLAUDE_PLUGIN_ROOT}/skills/core/nanobanana/nanobanana.py |


Basic Usage

| Task | Command | |------|---------| | Generate image | python3 nanobanana.py --prompt "description" --output "file.png" | | Edit image | python3 nanobanana.py --prompt "changes" --input source.png --output "edited.png" |


Aspect Ratios

| Size | Ratio | Use Case | |------|-------|----------| | 1024x1024 | 1:1 | Square, logos | | 768x1344 | 9:16 | Portrait, stories (default) | | 1344x768 | 16:9 | Landscape, wallpapers | | 832x1248 | 2:3 | Portrait photos | | 1248x832 | 3:2 | Landscape photos | | 1536x672 | 21:9 | Ultra-wide |

Use --size WIDTHxHEIGHT to specify.


Models

| Model | Trade-off | |-------|-----------| | gemini-3-pro-image-preview | Higher quality (default) | | gemini-2.5-flash-image | Faster generation |

Use --model MODEL to specify.


Resolution

| Resolution | Use Case | |------------|----------| | 1K | Testing, drafts (default) | | 2K | Good quality | | 4K | Final output, print |

Use --resolution RES to specify.


Best Practices

| Practice | Why | |----------|-----| | Be descriptive | Include style, mood, colors, composition | | Use 1:1 for logos | Clean square format | | Use 9:16 for stories | Standard mobile format | | Use 16:9 for wallpapers | Standard widescreen | | Start with 1K | Test before using higher resolution | | Use flash model for iteration | Save time during drafting |


Examples

Generate landscape:

python3 nanobanana.py --prompt "Mountain sunset with lake" --size 1344x768 --output "landscape.png"

Generate logo:

python3 nanobanana.py --prompt "Minimalist tech logo" --size 1024x1024 --output "logo.png"

Edit existing image:

python3 nanobanana.py --prompt "Add rainbow to sky" --input photo.png --output "edited.png"

High quality output:

python3 nanobanana.py --prompt "Professional portrait" --resolution 2K --output "portrait.png"

Troubleshooting

| Issue | Solution | |-------|----------| | Script fails | Check GEMINI_API_KEY is set | | No image generated | Make prompt more specific | | Can't read input | Verify file exists and is readable | | Can't write output | Check directory is writable |