Together AI Image Server
@zym9863
The server acts as a bridge between MCP-compatible assistants and Together AI's image generation capabilities.
Overview
What is Together AI Image Server?
Together AI Image Server is a TypeScript-based MCP (Model Context Protocol) server that generates images using the Together AI API. It provides a simple interface for Claude and other MCP-compatible assistants to create images from text prompts.
How to use Together AI Image Server?
Install the server by cloning the repository and running npm install. Set the TOGETHER_API_KEY environment variable or create a .env file with your Together AI API key. Build the server with npm run build, then configure Claude Desktop by adding the path to the built index.js in the mcpServers section of the Claude Desktop configuration file.
Key features of Together AI Image Server
- Exposes a
generate_imagetool for text-to-image generation - Requires only a text prompt; optional steps (1-4) and number of images (1-4)
- Returns image URLs and local file paths
- Built with TypeScript for reliable MCP integration
- Debug using the MCP Inspector via
npm run inspector
Use cases of Together AI Image Server
- Generate images directly from Claude Desktop conversations
- Create multiple image variations from a single prompt
- Cache generated images locally for later use
- Integrate image generation into any MCP-compatible workflow
FAQ from Together AI Image Server
What runtime and API key are required?
Node.js v14 or later and a Together AI API key are required. The API key is set as the environment variable TOGETHER_API_KEY or in a .env file.
Where are generated images stored?
The server returns URLs to the images and saves local copies to paths reported in the response.
What are the limits for image generation?
The steps parameter accepts 1–4 diffusion steps (default 4), and the n parameter accepts 1–4 images (default 1).
How does the server communicate with clients?
The server uses the MCP protocol over standard I/O (stdio), as is typical for MCP servers.
How can I debug the server?
Use the MCP Inspector by running npm run inspector, which provides a URL for browser-based debugging tools.
