Overview
What is MCP Image Generator?
MCP Image Generator is a Model Context Protocol (MCP) server that generates images using Together AI or Replicate image generation models. It is designed for developers who want to integrate AI image generation into MCP‑compatible applications, and can be run locally via stdio or as an SSE endpoint in Docker.
How to use MCP Image Generator?
Clone the repository, install dependencies with pnpm install, and configure your MCP client with the appropriate provider (replicate or together) and API token. Then invoke the generate_image tool with a text prompt and optional parameters for width, height, and number of images.
Key features of MCP Image Generator
- Supports Together AI and Replicate as image generation providers.
- Provides a
generate_imagetool with configurable prompt, width, height, and number of images. - Can run locally via stdio or as an SSE endpoint inside Docker.
- Defaults to
black-forest-labs/flux-schnellmodel if not specified. - Works with any MCP client that supports stdio or SSE transports.
Use cases of MCP Image Generator
- Generate images dynamically from text prompts in chat applications.
- Integrate image creation into AI‑powered content generation pipelines.
- Automate image production for prototyping or design workflows.
FAQ from MCP Image Generator
What API keys are required?
You need the TOGETHER_API_KEY environment variable for Together AI, or the REPLICATE_API_TOKEN for Replicate. Set the PROVIDER variable to replicate or together accordingly.
How do I configure the model used for generation?
Set the MODEL_NAME environment variable (default is black-forest-labs/flux-schnell).
What parameters does the generate_image tool accept?
It accepts prompt (required), plus optional width, height (both default 512), and numberOfImages (default 1).
Can I run the server as an SSE endpoint?
Yes. Build the Docker image using the provided Dockerfile.server and run it on port 3000. Then configure your MCP client with the SSE URL and environment variables.
What are the transport options?
The server supports stdio (local execution via pnpx tsx) and SSE (Dockerized HTTP server). The README only shows these two transports; no authentication is documented beyond API tokens.
