Setup Guide

Prerequisites

• A SmarterWeather account with a paid plan
• An API key from your Developer Dashboard
• Node.js 18+ installed

Claude Desktop

Add the following to your Claude Desktop configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "smarterweather": {
      "command": "npx",
      "args": ["-y", "@smarterweather/mcp-weather"],
      "env": {
        "SW_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Cursor

Add the MCP server to your Cursor MCP configuration at .cursor/mcp.json in your project or global config:

{
  "mcpServers": {
    "smarterweather": {
      "command": "npx",
      "args": ["-y", "@smarterweather/mcp-weather"],
      "env": {
        "SW_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

Generic MCP Client

For any MCP-compatible client, use the following transport configuration:

# Stdio transport (default)
npx @smarterweather/mcp-weather

# Environment variables
SW_API_KEY=your_key       # Required: API authentication
SW_MCP_TAGS=forecast      # Optional: tool sets to enable
                           # Values: forecast, data, reference, web
                           # Default: forecast

Tool Tags

Control which tool sets are loaded via the SW_MCP_TAGS environment variable:

Verify the Connection

After configuration, try a simple prompt in your AI client:

What's the weather in Denver, CO right now?

The AI should invoke the geocode_location and get_weather tools to answer with real-time weather data.