@visionengine/image-generate

VisionEngine Image Generate MCP Server - Generate and edit images using AI models.

Features

• Text-to-Image Generation - Generate images from text descriptions using AI
• Image Editing - Modify existing images based on text prompts
• Auto-save - Generated images are automatically saved to WORKDIR
• Multiple Image Formats - Support for JPG, PNG, GIF, WebP input

Installation

As MCP Server

Add to your MCP client configuration:

{
  "mcpServers": {
    "ve-image-generate": {
      "type": "local",
      "command": "npx",
      "args": ["-y", "@visionengine/image-generate@latest"],
      "transport": "stdio",
      "env": {
        "API_URL": "https://openrouter.ai",
        "API_KEY": "your_api_key_here",
        "WORKDIR": "./generated-images"
      }
    }
  }
}

As NPM Package

npm install -g @visionengine/image-generate

Configuration

Environment variables:

| Variable | Description | Default | |----------|-------------|---------| | API_URL | API endpoint | https://openrouter.ai | | API_KEY | Your API key (required) | - | | MODEL | Image generation model | bytedance-seed/seedream-4.5 | | WORKDIR | Output directory for generated images | ./ | | HTTP_REFERER | HTTP Referer header | https://visionengine-tech.com | | X_TITLE | X-Title header | VEC AI |

Tools

generate

Generate an image from a text description.

Parameters:

• prompt (string, required) - Text description of the image to generate

Returns:

• Relative path of the generated image file (based on WORKDIR)

Example:

// Generate a landscape image
await generate({
  prompt: "A beautiful sunset over mountains with orange and purple sky"
});
// Returns: "generated_1704067200000_a1b2c3d4.png"

// Generate a character
await generate({
  prompt: "A cute cartoon cat wearing a wizard hat, digital art style"
});

edit

Edit/modify an existing image based on a text prompt.

Parameters:

• imagePath (string, required) - Image file path to edit (relative to WORKDIR or absolute path)
• prompt (string, required) - Text description of how to modify the image

Returns:

• Relative path of the edited image file (based on WORKDIR)

Example:

// Add elements to an image
await edit({
  imagePath: "./photo.jpg",
  prompt: "Add a rainbow in the sky"
});
// Returns: "edited_1704067200000_e5f6g7h8.png"

// Change style
await edit({
  imagePath: "./portrait.png",
  prompt: "Convert this photo to oil painting style"
});

Usage Examples

MCP Client

Once configured as an MCP server, the tools are available through your MCP client:

> Use generate tool to create an image of a futuristic city
> Use edit tool to add flying cars to the city image

Direct Usage

# Install globally
npm install -g @visionengine/image-generate

# Set environment variables
export API_KEY="your_api_key"
export WORKDIR="./output"

# Run the server
ve-image-generate

Output

Generated images are saved to WORKDIR with unique filenames:

• generated_<timestamp>_<randomId>.png - For text-to-image generation
• edited_<timestamp>_<randomId>.png - For image editing

Development

Build

pnpm build

Test

pnpm test

Local Testing

# Build first
pnpm build

# Run locally
node dist/index.js

Support

For issues and questions:

• Email: [email protected]
• Website: https://visionengine-tech.com