gemini-image-gen

Gemini Image Gen Plugin for Claude Code

Generate stunning AI images using Google's Gemini image generation models with professional prompt optimization based on analysis of 1,186 viral prompts.

Features

• 🎨 Smart Prompt Optimization - 6 rules extracted from viral prompts
• 🎯 Genre Detection - Auto-detects food, portrait, product, cinematic, 3D, design
• 🚀 Lightweight Agent - Keeps main context clean (~500 tokens vs 50K)
• 🔑 Flexible Setup - Works with direct API key OR proxy
• 📚 Template Library - Pre-built patterns from top-performing prompts

Quick Start

Option 1: Direct API Key (Recommended)

Windows (PowerShell):

$env:GEMINI_API_KEY = "your_api_key_here"

Windows (Git Bash / CMD):

export GEMINI_API_KEY=your_api_key_here
# Or for CMD:
set GEMINI_API_KEY=your_api_key_here

macOS / Linux:

export GEMINI_API_KEY=your_api_key_here

Then install the plugin (see Installation section below).

Option 2: With Proxy

# Start the antigravity-claude-proxy
cd antigravity-claude-proxy && npm start

# Install the plugin (see Installation section below)
# Use it in Claude Code: /image a bowl of ramen

Installation

From GitHub

Windows (PowerShell):

git clone https://github.com/AryanXPatel/gemini-image-gen.git
Copy-Item -Recurse gemini-image-gen "$env:USERPROFILE\.claude\plugins\gemini-image-gen"

Windows (Git Bash / MINGW64):

git clone https://github.com/AryanXPatel/gemini-image-gen.git
cp -r gemini-image-gen ~/.claude/plugins/
# Or if you're inside the repo:
cd .. && cp -r gemini-image-gen ~/.claude/plugins/

macOS / Linux:

git clone https://github.com/AryanXPatel/gemini-image-gen.git
cp -r gemini-image-gen ~/.claude/plugins/

Already inside the repo?

# Go up one directory first
cd ..
cp -r gemini-image-gen ~/.claude/plugins/

# Or copy current directory
cp -r . ~/.claude/plugins/gemini-image-gen

From Claude Code Marketplace

# Add the marketplace first
/plugin marketplace add AryanXPatel/aryanxpatel-plugins

# Then install the plugin
/plugin install gemini-image-gen@aryanxpatel

Usage

Basic Generation

/image a bowl of ramen

The plugin will:

1. Spawn a lightweight agent
2. Detect genre (food)
3. Apply optimization rules
4. Generate and save the image
5. Return just the result

With Refinement

/image a sunset over mountains
# "Make it more dramatic"
/image --refine add volumetric god rays and deeper colors

Raw Mode (No Optimization)

/image --raw "your exact prompt here without any changes"

Template Mode

/image --template food "steaming ramen bowl"
/image --template portrait "woman in rain"
/image --template cinematic "car chase scene"

The 6 Optimization Rules

Based on analysis of 1,186 viral AI-generated images:

Rule	Description	Example
1. Professional Terms	Replace feeling words with specific terminology	"cinematic" → "Wong Kar-wai aesthetics, Kodak Vision3 500T"
2. Quantified Parameters	Use exact technical specs	"professional" → "90mm lens, f/1.8, shallow DOF"
3. Negative Constraints	State what NOT to include	"No text, no watermarks, no distortion"
4. Sensory Stacking	Add tactile, motion, temperature	"steam rising, texture tangible, crisp cold"
5. Group & Cluster	Organize complex prompts	Visual Style / Lighting / Technical / Constraints
6. Genre Adaptation	Apply genre-specific patterns	Food: 45° overhead, warm tones, macro

Configuration

Environment Variables

Variable	Description	Default
GEMINI_API_KEY	Your Google AI Studio API key	(none - uses proxy)
GEMINI_MODEL	Model to use	gemini-3-pro-image
PROXY_HOST	Proxy hostname	localhost
PROXY_PORT	Proxy port	8080

Getting a Google AI API Key

1. Go to Google AI Studio
2. Click "Get API Key"
3. Create a new API key
4. Set it: export GEMINI_API_KEY=your_key_here

Using the Proxy (Alternative)

If you prefer using the proxy for multi-account support:

1. Clone antigravity-claude-proxy
2. Configure your Google accounts
3. Run: npm start
4. The plugin will auto-detect and use the proxy

Plugin Structure