An MCP server that fetches random cat images and uses LLM sampling to identify the breed and describe the scene.

Built with the Go MCP SDK.

1. The describe_kitten tool is called by an MCP client
2. A random cat image is fetched from The Cat API
3. The image is sent to the client via an MCP sampling request, asking the LLM to identify the breed and describe the scene
4. The description and image URL are returned as structured output

• Go 1.25+
• A Cat API key

# Clone the repo
git clone https://github.com/omgitsads/go-sdk-kitten-description.git
cd go-sdk-kitten-description

# Build
go build -o kitten-description-server .

Set your Cat API key and run the server over stdio:

export CAT_API_KEY="your-api-key"
./kitten-description-server

Add the server to your MCP client config (e.g. Claude Desktop, Copilot, etc.):

{
  "mcpServers": {
    "kitten-description": {
      "command": "./kitten-description-server",
      "env": {
        "CAT_API_KEY": "your-api-key"
      }
    }
  }
}

Fetches a random cat image and asks the LLM to identify the breed and describe the scene.

• Input: None
• Output:
  • description — LLM-generated breed identification and scene description
  • image_url — The original image URL from The Cat API

Note: This tool uses MCP sampling, so the connected client must support CreateMessage.