Memgen MCP Server

A Model Context Protocol server that generates memes from user input using OpenAI's image editing capabilities.

Overview

This server implements the Model Context Protocol (MCP) to provide meme generation functionality. It takes a text description as input, refines it using AI, and generates a meme image by editing a base template. The generated images are uploaded to MinIO S3-compatible storage and returned as URLs.

Features

- MCP-compliant server for seamless integration with MCP clients
- Prompt refinement using AI to optimize meme descriptions
- Image generation using OpenAI's GPT-image-1 model
- Automatic upload of generated images to MinIO S3 storage

Prerequisites

- Node.js >= 22.0.0
- MinIO S3-compatible storage
- OpenAI API key

Installation

1. Clone the repository:

   git clone https://github.com/sint-ai/image_mcp_server.git

   cd image_mcp_server

   

2. Install dependencies:

   npm install

   

3. Copy the example environment file and fill in your credentials:

   cp .env.example .env

   

4. Edit .env with your configuration:
- PORT: Server port
- OPENAI_API_KEY: Your OpenAI API key
- MINIO_ENDPOINT: MinIO server endpoint
- MINIO_PORT: MinIO server port
- MINIO_USE_SSL: Whether to use SSL for MinIO connection
- MINIO_ACCESS_KEY: MinIO access key
- MINIO_SECRET_KEY: MinIO secret key
- MINIO_BUCKET_NAME: MinIO bucket name
- MINIO_BUCKET_URL: Public URL for accessing the MinIO bucket

Usage

Start the server:

npm start

The server will start running on the configured port (default: 3000).

API Endpoints

- POST /mcp: Main MCP endpoint for processing requests

MCP Tooling

This server provides the following MCP tool:

- create_meme: Generates a meme from a text description
- Input: String describing the desired meme
- Output: URLs to the generated images

Development

Check TypeScript types:

npm run check-types