Home / MCP / Draw Things MCP Server

Draw Things MCP Server

Provides an MCP integration that generates images via Draw Things API for Cursor.

javascript
Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
    "mcpServers": {
        "draw_things": {
            "command": "draw-things-mcp-cursor",
            "args": []
        }
    }
}

Draw Things MCP provides a Cursor integration that lets you generate images through the Draw Things API using the Model Context Protocol. This enables you to generate visuals directly from Cursor with configurable prompts, image sizes, and model options.

How to use

You use the Draw Things MCP server by loading it into your MCP client setup and then invoking the generateImage capability from the client. Your prompts drive the image generation, and you can adjust dimensions, model, and steps to refine results. The server handles communication with the Draw Things API and returns image data that you can display or save.

How to install

Prerequisites: you need Node.js version 14.0.0 or higher and Draw Things API running locally at http://127.0.0.1:7888.

Install the MCP server globally or run it directly using the package you need:

# Install globally
npm install -g draw-things-mcp-cursor

# Or run directly
npx draw-things-mcp-cursor

Configuration and usage with Cursor

Add the MCP server configuration for Cursor to connect to the Draw Things MCP tool. Create or edit your Cursor MCP config file to register the local command that launches the server.

{
  "mcpServers": {
    "draw-things": {
      "command": "draw-things-mcp-cursor",
      "args": []
    }
  }
}

Parameters and defaults

When you use the MCP tool from Cursor, you can pass the following parameters to customize image generation. If a parameter is omitted, a default value is used.

{
  "prompt": "a happy smiling dog, professional photography",
  "negative_prompt": "ugly, deformed, blurry",
  "width": 360,
  "height": 360,
  "steps": 4
}

MCP tool integration details

The MCP integration exposes a tool named generateImage that you can call from Cursor. The tool accepts the following fields.

{
  prompt: string;       // Required - The prompt to generate the image from
  negative_prompt?: string;  // Optional - The negative prompt
  width?: number;       // Optional - Image width (default: 360)
  height?: number;      // Optional - Image height (default: 360)
  model?: string;       // Optional - Model name
  steps?: number;       // Optional - Number of steps (default: 8)
}

Output and storage

Generated images are saved to the images directory with a filename format of <sanitized_prompt>_<timestamp>.png so you can easily locate and reuse results.

Troubleshooting

If you encounter issues, check the following areas to diagnose problems quickly.

    Available tools

    generateImage

    Invokes image generation using a prompt, with optional negative prompt, dimensions, model, and steps. Returns an image payload that Cursor can display or store.