mcp-cloud-server by datavtar - MCP Server

MCP Cloud Server

A Model Context Protocol (MCP) server built with FastMCP that exposes comprehensive weather tools, resources, and prompts via Server-Sent Events (SSE). Designed for deployment on Google Cloud Run.

Features

US Weather (NWS): Alerts, forecasts, hourly forecasts, current conditions, radar stations, hurricane tracking
Global Weather (Open-Meteo): Worldwide forecasts, air quality, UV index, marine conditions
Geocoding (Nominatim): Convert addresses to coordinates and vice versa
Utilities: Sunrise/sunset times, weather comparison, comprehensive summaries
MCP Resources: Weather glossary, station lists, national alert summaries
MCP Prompts: Travel weather, severe weather analysis, clothing recommendations, outdoor activity suitability
SSE Transport: Uses Server-Sent Events for real-time MCP communication
Cloud Run Ready: Configured for Google Cloud Run deployment

Quick Start

Local Development

# Create virtual environment
python -m venv .venv
source .venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Run server
python server.py

The server starts on http://localhost:8080 by default. Set the PORT environment variable to change it.

Docker Deployment

# Build image
docker build -t mcp-cloud-server .

# Run container
docker run -p 8080:8080 mcp-cloud-server

Google Cloud Run Deployment

Using the deploy script (recommended):

# Basic deployment
python gcp_deploy.py --project-id <PROJECT_ID>

# Custom region and service name
python gcp_deploy.py -p <PROJECT_ID> -r europe-west1 -s weather-mcp

# Redeploy without rebuilding
python gcp_deploy.py -p <PROJECT_ID> --skip-build

Or manually with gcloud:

# Build and push to Google Container Registry
gcloud builds submit --tag gcr.io/<PROJECT_ID>/mcp-server

# Deploy to Cloud Run
gcloud run deploy mcp-server \
    --image gcr.io/<PROJECT_ID>/mcp-server \
    --platform managed \
    --region <REGION> \
    --allow-unauthenticated

Replace <PROJECT_ID> with your GCP project ID and <REGION> with your preferred region (e.g., us-central1).

Testing with MCP Inspector

# Test local server
npx @modelcontextprotocol/inspector \
    npx -y @modelcontextprotocol/server-sse-client \
    --url http://localhost:8080/sse

# Test deployed server
npx @modelcontextprotocol/inspector \
    npx -y @modelcontextprotocol/server-sse-client \
    --url https://<SERVICE_URL>/sse

Replace <SERVICE_URL> with your Cloud Run service URL (e.g., mcp-server-xxxxx.us-central1.run.app).

MCP Tools

US Weather (NWS API)

Tool	Description
`get_alerts`	Get weather alerts for a US state
`get_forecast`	Get multi-day forecast for US location
`get_hourly_forecast`	Get hourly forecast (next 24h)
`get_current_conditions`	Get current conditions from nearest station
`get_radar_stations`	List nearby radar stations
`get_active_hurricanes`	Get active tropical storms/hurricanes

Global Weather (Open-Meteo)

Tool	Description
`get_global_forecast`	7-day forecast (worldwide)
`get_global_hourly`	Hourly forecast (worldwide)
`get_air_quality`	Air quality index and pollutants
`get_uv_index`	UV index current and forecast
`get_marine_forecast`	Wave height, direction, period

Geocoding (Nominatim)

Tool	Description
`geocode_location`	Convert place name/address to coordinates
`reverse_geocode`	Convert coordinates to address

Utilities

Tool	Description
`get_sunrise_sunset`	Sunrise/sunset times for any date
`compare_weather`	Compare weather between two locations
`get_weather_summary`	Comprehensive weather summary

MCP Resources

Resource URI	Description
`weather://glossary`	Weather terminology definitions
`weather://stations/{state}`	List weather stations in a US state
`weather://alerts/national`	Summary of all active US alerts

MCP Prompts

Prompt	Description
`travel_weather`	Travel weather briefing between two locations
`severe_weather_summary`	Severe weather analysis for a US state
`clothing_recommendation`	What to wear based on weather
`outdoor_activity`	Is weather suitable for an activity?

Client Integration

Claude Desktop App

Add to your Claude Desktop config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "weather": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sse-client",
        "--url",
        "https://<SERVICE_URL>/sse"
      ]
    }
  }
}

Replace <SERVICE_URL> with your Cloud Run service URL, then restart Claude Desktop.

Python SDK

pip install mcp

import asyncio
from mcp.client.sse import sse_client
from mcp.client.session import ClientSession

SERVER_URL = "https://<SERVICE_URL>/sse"

async def main():
    async with sse_client(SERVER_URL) as (read, write):
        async with ClientSession(read, write) as session:
            await session.initialize()

            # List available tools
            tools = await session.list_tools()
            for tool in tools.tools:
                print(f"- {tool.name}: {tool.description}")

            # Call a tool
            result = await session.call_tool("get_global_forecast", {
                "latitude": 28.6139,
                "longitude": 77.2090
            })
            print(result.content[0].text)

asyncio.run(main())

TypeScript/Node.js SDK

npm install @modelcontextprotocol/sdk

import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { SSEClientTransport } from "@modelcontextprotocol/sdk/client/sse.js";

const client = new Client({ name: "weather-client", version: "1.0.0" });
const transport = new SSEClientTransport(
  new URL("https://<SERVICE_URL>/sse")
);

await client.connect(transport);

// List tools
const tools = await client.listTools();

// Call a tool
const result = await client.callTool("geocode_location", {
  query: "Tokyo, Japan"
});

APIs Used

All APIs are free and require no API keys:

National Weather Service API - US weather data
Open-Meteo - Global weather, air quality, marine data
Nominatim - Geocoding (OpenStreetMap)
Sunrise-Sunset.org - Sunrise/sunset times

datavtar/mcp-cloud-server