imagen-mcp

A Model Context Protocol (MCP) server for intelligent multi-provider image generation.

Features

• Auto Provider Selection - Analyzes prompts to choose the best provider
• Multi-Provider Support - OpenAI GPT-Image-1 and Google Gemini
• Reference Images - Up to 14 images for character/style consistency (Gemini)
• Real-time Data - Google Search grounding for current info (Gemini)
• Conversational History - Iteratively refine images with context (Gemini)
• High Resolution - Up to 4K output (Gemini)
• Flexible Storage - Save to ~/Downloads/images/ or custom locations

Architecture

flowchart TB
    subgraph Clients["MCP Clients"]
        CD[Claude Desktop]
        CC[Claude Code CLI]
        GC[Gemini CLI]
        CX[Codex CLI]
    end

    subgraph Server["imagen-mcp Server"]
        MCP[MCP Protocol Layer]

        subgraph Tools["MCP Tools"]
            GI[generate_image]
            CI[conversational_image]
            LP[list_providers]
            LM[list_gemini_models]
        end

        subgraph Core["Core Components"]
            PS[Provider Selector]
            PR[Provider Registry]
        end

        subgraph Providers["Image Providers"]
            OAI[OpenAI Provider<br/>GPT-Image-1]
            GEM[Gemini Provider<br/>Nano Banana Pro]
        end
    end

    subgraph APIs["External APIs"]
        OAPI[OpenAI API]
        GAPI[Google Gemini API]
    end

    subgraph Storage["Local Storage"]
        DL[~/Downloads/images/]
    end

    CD & CC & GC & CX --> MCP
    MCP --> Tools
    GI & CI --> PS
    PS --> PR
    PR --> OAI & GEM
    OAI --> OAPI
    GEM --> GAPI
    OAI & GEM --> DL

Provider Comparison

Use OpenAI for: Text-heavy images, menus, infographics, comics, diagrams

Use Gemini for: Portraits, product photography, 4K output, reference images

Available Models

OpenAI Models

Gemini Models

Installation

git clone https://github.com/michaeljabbour/imagen-mcp.git
cd imagen-mcp
pip install -r requirements.txt
chmod +x run.sh

Configuration

At least one API key is required. Both are recommended for auto-selection.

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "imagen": {
      "command": "/path/to/imagen-mcp/run.sh",
      "args": [],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "GEMINI_API_KEY": "AI..."
      }
    }
  }
}

Note: Claude Desktop doesn't support cwd, so use the run.sh wrapper script which handles the directory change.

Restart Claude Desktop (Cmd+Q, then reopen) after editing.

Claude Code CLI

Use the CLI to add the server:

claude mcp add -s user imagen /path/to/imagen-mcp/run.sh

Then add environment variables by editing ~/.claude.json:

{
  "mcpServers": {
    "imagen": {
      "type": "stdio",
      "command": "/path/to/imagen-mcp/run.sh",
      "args": [],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "GEMINI_API_KEY": "AI..."
      }
    }
  }
}

Verify with:

claude mcp list

Reference: Claude Code MCP Documentation

Gemini CLI

Edit ~/.gemini/settings.json:

{
  "mcpServers": {
    "imagen": {
      "command": "/path/to/imagen-mcp/run.sh",
      "args": [],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "GEMINI_API_KEY": "AI..."
      }
    }
  }
}

Reference: Gemini CLI MCP Documentation

OpenAI Codex CLI

Edit ~/.codex/config.toml:

[mcp_servers.imagen]
command = "/path/to/imagen-mcp/run.sh"
args = []

[mcp_servers.imagen.env]
OPENAI_API_KEY = "sk-..."
GEMINI_API_KEY = "AI..."

Or use the CLI:

codex mcp add imagen -- /path/to/imagen-mcp/run.sh

Reference: Codex MCP Documentation

Generic MCP Client

For any MCP-compatible client:

The Wrapper Script

The run.sh script handles the working directory requirement:

#!/bin/bash
cd /path/to/imagen-mcp
exec python3 -m src.server "$@"

This is necessary because the server runs as a Python module (-m src.server) which requires being in the project directory.

Usage

Auto Provider Selection

The server analyzes your prompt and selects the best provider:

"Create a menu card for an Italian restaurant"  → OpenAI (text rendering)
"Professional headshot with studio lighting"    → Gemini (photorealism)
"Infographic about climate change"              → OpenAI (diagram + text)
"Product shot of perfume on marble"             → Gemini (product photography)

Manual Provider Selection

Override auto-selection with the provider parameter:

generate_image(prompt="...", provider="openai")
generate_image(prompt="...", provider="gemini")

Save Location

Specify a custom save path (directory or filename) with output_path:

# Save to specific directory (auto-generated filename)
generate_image(prompt="...", output_path="~/Desktop/logos/")

# Save to specific file
generate_image(prompt="...", output_path="~/Desktop/logos/my-logo.png")

If output_path is omitted, images are saved to ~/Downloads/images/{provider} by default (openai or gemini). Override the base directory with the OUTPUT_DIR environment variable (supports ~ and env vars).

Logs are written to ~/Downloads/images/logs/ by default (or OUTPUT_DIR/logs/ when OUTPUT_DIR is set).

Gemini-Specific Features

# High resolution
generate_image(prompt="...", size="4K")

# Specific model
generate_image(prompt="...", gemini_model="gemini-2.0-flash-exp-image-generation")

# Reference images (base64 encoded)
generate_image(prompt="...", reference_images=["base64..."])

# Real-time data
generate_image(prompt="Current weather in NYC", enable_google_search=True)

MCP Tools

Development

# Install dev dependencies
pip install -r requirements.txt
pip install pytest pytest-asyncio

# Run tests
pytest tests/ -v

# Test server loads
python3 -c "from src.server import mcp; print('Server loads')"

# Test providers
python3 -c "from src.providers import get_provider_registry; print(get_provider_registry().list_providers())"

# Check logs (macOS)
tail -f ~/Library/Logs/Claude/mcp-server-imagen.log

Project Structure

imagen-mcp/
├── src/
│   ├── server.py              # MCP entry point
│   ├── config/
│   │   ├── constants.py       # Provider constants
│   │   └── settings.py        # Environment configuration
│   ├── providers/
│   │   ├── base.py            # Abstract provider interface
│   │   ├── openai_provider.py # OpenAI implementation
│   │   ├── gemini_provider.py # Gemini implementation
│   │   ├── selector.py        # Auto-selection logic
│   │   └── registry.py        # Provider factory
│   └── models/
│       └── input_models.py    # Pydantic input models
├── tests/
│   ├── test_selector.py       # Provider selection tests
│   ├── test_providers.py      # Provider unit tests
│   └── test_server.py         # Server integration tests
├── .github/
│   └── workflows/
│       └── ci.yml             # GitHub Actions CI
├── run.sh                     # Wrapper script for MCP clients
├── requirements.txt
├── CLAUDE.md
└── README.md

Environment Variables

Requirements

mcp>=1.16.0
fastmcp>=2.12.5
pydantic>=2.12.3
httpx>=0.24.0
google-genai>=1.52.0
pillow>=10.4.0

License

MIT

Sources

• Claude Code MCP Documentation
• Gemini CLI MCP Documentation
• Codex MCP Documentation
• Model Context Protocol