Stability AI

You can use this MCP server to connect MCP clients to Stability AI’s Stable Diffusion image manipulation capabilities, enabling generate, edit, upscale, and related operations through a simple, file-forward workflow. By supplying your Stability AI API key and a storage directory for output images, you can run a local MCP server that exposes these powerful image tools through your preferred MCP client.

How to use

You interact with the Stability AI MCP Server through an MCP client. After you start the server, you can request image generation, background removal, outpainting, and other image editing tasks by issuing MCP-like commands from your client. All processed images are saved to your configured image storage directory and opened for preview, so you can review results immediately. Copy the image files you want to process into the storage directory and reference them by filename in your conversations with the MCP client to avoid inefficient in-chat image data transfers.

How to install

Prerequisites you need on your machine before you begin:

• Node.js and npm or another runtime you plan to use to execute the MCP server (the setup uses npx in examples)
• Access to a Stability AI API key from stability.ai
• A local directory where generated images will be stored

1. Create a storage directory for generated images. For example:

/Users/yourname/stability-ai-images

2. Obtain a Stability AI API key from stability.ai and keep it handy.

3. Configure the MCP server using the provided configuration snippet. This snippet defines the server as a stdio MCP server started via npx with the appropriate environment variables.

{
  "mcpServers": {
    "stability-ai": {
      "command": "npx",
      "args": [
        "mcp-server-stability-ai"
      ],
      "env": {
        "STABILITY_AI_API_KEY": "sk-1234567890",
        "IMAGE_STORAGE_DIRECTORY": "/Users/admin/Downloads/stability-ai-images"
      }
    },
  },
}

4. Update the config path to your environment:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json Then add the MCP server entry under mcpServers in your Claude Desktop configuration. Restart Claude Desktop to load the new server.

Configuration example and starting the server

The following inline configuration shows how to register the MCP server in your Claude Desktop configuration. This example uses the same values described above. Copy and paste into your claude_desktop_config.json and restart Claude Desktop.

{
  "mcpServers": {
    "stability-ai": {
      "command": "npx",
      "args": [
        "mcp-server-stability-ai"
      ],
      "env": {
        "STABILITY_AI_API_KEY": "sk-1234567890",
        "IMAGE_STORAGE_DIRECTORY": "/Users/admin/Downloads/stability-ai-images"
      }
    },
  },
}

Notes and tips

• All generated or edited images are saved in the directory you specify in IMAGE_STORAGE_DIRECTORY and opened for quick preview.

• Keep your API key secure. Do not expose STABILITY_AI_API_KEY in public repos or share it in unsecured channels.

Examples of common tasks

Generate an image for a cat, remove its background, and upscale the result in a single workflow from your MCP client by chaining commands in a single conversation.

Available tools

generate-image

Generate a high quality image based on a prompt with optional parameters such as style presets and aspect ratios.

remove-background

Remove the background from an image to create a transparent or clean cutout.

outpaint

Extend an image in any direction while maintaining visual continuity.

search-and-replace

Replace elements in an image by describing what to replace and with what.

upscale-fast

Increase image resolution by 4x with a faster upscale.

upscale-creative

Upscale image quality up to 4K with enhanced details.

control-sketch

Translate a hand-drawn sketch into a production-ready image.