Image Gen MCP Server

@Cleveland-RD

Basic information

About Image Gen MCP Server

Local stdio MCP server for generating and editing images via OpenAI gpt-image-2 or Google Gemini (Nano Banana / Pro image-preview). Saves PNGs to ~/Pictures/image-gen/ and returns inline JPEG thumbnails. Bring your own OpenAI and/or Gemini API key.

#image-generation #openai#gpt-image-2#gemini #nano-banana #stdio #claude-desktop #claude-code

Image Gen MCP Server config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "image-gen": {
      "command": "node",
      "args": [
        "<HOME>/.image-gen-mcp/server/index.js"
      ],
      "env": {
        "OPENAI_API_KEY": "<YOUR_OPENAI_API_KEY>",
        "GEMINI_API_KEY": "<YOUR_GEMINI_API_KEY>"
      }
    }
  }
}

Image Gen MCP Server tools

Generate or edit an image via OpenAI gpt-image-2 or Google Gemini. Accepts a prompt plus optional

Poll a previously-started image generation job.

Overview

What is Image Gen?

Image Gen is an open-source Claude plugin and stdio MCP server for generating and editing images via OpenAI gpt-image-2 and Google Gemini models (gemini-3.1-flash-image-preview / `gemini-3-pro-image-preview)). Users bring their own API keys.

How to use Image Gen?

Install via .mcpb (Claude Desktop), plugin zip (Claude Desktop or Claude Code), or manual configuration. Set at least one API key (OPENAI_API_KEY or GEMINI_API_KEY) as an environment variable. Invoke the generate_image tool with a prompt and optional parameters, or use check_image_job to poll an in-progress job.

Key features of Image Gen?

Supports OpenAI gpt-image-2 and Google Gemini image models
Generates and edits images from text prompts
Optionally includes reference images for edits
Returns inline JPEG thumbnail for immediate display
Saves full-resolution PNGs to ~/Pictures/image-gen/
Provides check_image_job tool for polling long-running generations

Use cases of Image Gen

Generate images from descriptive prompts within Claude conversations
Edit existing images by providing reference images alongside prompts
Integrate AI image generation into Claude-powered workflows

Frequently asked questions

What API keys are required?

At least one API key: either an OpenAI key for gpt-image-2 or a Google Gemini key for gemini-3.1-flash-image-preview / gemini-3-pro-image-preview.

Which models are supported?

OpenAI gpt-image-2 and Google Gemini models: gemini-3.1-flash-image-preview (Nano Banana) and gemini-3-pro-image-preview (Pro).

Where are generated images saved?

Full-resolution PNG files are saved to `~/Pictures/image-gen/`. The tool also returns an inline JPEG thumbnail for immediate rendering.

What operating systems are supported?

Prebuilt `.mcpb` packages are available for darwin-arm64, darwin-x64, and win32-x64. Linux users can build from source.

How do I install the server manually?

Download the platform-matching `.mcpb`, extract it, and register with Claude Code using `claude mcp add` with environment variables for API keys. Alternatively, configure via `claude_desktop_config.json`.

Comments

More Media & Design MCP servers

Your Move Exercise Video API

Your Move

Exercise videos, workouts and training programs for AI agents and apps: 1,413+ exercises with professional HD video, instructions and muscle data, plus workout and program generation and nutrition tools. Hosted MCP, no i

TravelAnimator

Lascade-Co

MCP server and CLI for TravelAnimator — generate animated travel videos, itineraries, and trip media from Claude, Codex, or your terminal

Caimera

Build and run AI product photography workflows from your AI assistant: on-model shots, catalog images, prints and edits, with results returned in chat.

CoDesign

IMG.LY

Give your AI agent a real design engine. Create, edit and export real, editable designs from Claude Code, Codex, Cursor and more.

BulkTranscripts: YouTube Transcripts

Transcripts for one YouTube video or a whole channel or playlist, plus YouTube search, channel and playlist listings, and free new-upload tracking.