mcp-weather

denhamparry/mcp-weather

3.2

If you are the rightful owner of mcp-weather and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

An MCP server designed to provide weather conditions and forecasts.

mcp-weather

An MCP server to answer weather conditions for UK locations using OpenWeatherMap API.

Setup

  1. Get a free API key from OpenWeatherMap

  2. Set the environment variable:

    export OPENWEATHER_API_KEY="your_api_key_here"

You can also set the environment variable in the .envrc file and use direnv to load it automatically.

Usage

This server provides two tools:

get_forecast

Get weather forecast for a location using latitude and longitude coordinates.

Parameters:

  • latitude: Latitude of the location (-90 to 90)
  • longitude: Longitude of the location (-180 to 180)

Example UK coordinates:

  • London: 51.5074, -0.1278
  • Manchester: 53.4808, -2.2426
  • Edinburgh: 55.9533, -3.1883
  • Cardiff: 51.4816, -3.1791

get_alerts

Get weather alerts for a location using latitude and longitude coordinates.

Parameters:

  • latitude: Latitude of the location (-90 to 90)
  • longitude: Longitude of the location (-180 to 180)

This project uses Task to build and run the project. You can install it with brew install task.

Build

npm install && npm run build

Run

Stdio

To run the MCP server as a standard MCP server:

node build/index.js

HTTP Server

To run the MCP server as an HTTP service for local testing:

# Build and start the HTTP server
npm run build && node build/http.js

The server will start on port 3000. You can then test it using the MCP Inspector:

  1. Run MCP Inspector locally:

    npx @modelcontextprotocol/inspector

  2. In the inspector interface:

    • Select "HTTP" as the transport type
    • Enter http://localhost:3000/mcp as the URL
    • Click "Connect"

  3. You can now test the get_forecast and get_alerts tools with UK coordinates directly in the inspector interface.

Docker

The project provides two Docker configurations:

  • Docker/Dockerfile.stdio: For standard MCP server with stdio communication
  • Docker/Dockerfile.http: For HTTP-based MCP server

Build Docker images

# Build stdio version
docker build -f Docker/Dockerfile.stdio -t mcp-weather:stdio .

# Build HTTP version
docker build -f Docker/Dockerfile.http -t mcp-weather:http .

Run with Docker directly

# Run stdio version (standard MCP server)
docker run -it --rm -e OPENWEATHER_API_KEY="your_api_key_here" mcp-weather:stdio

# Run HTTP version (web server on port 3000)
docker run -p 3000:3000 --rm -e OPENWEATHER_API_KEY="your_api_key_here" mcp-weather:http

The stdio container runs the MCP server which communicates via stdio. Make sure to run it in interactive mode (-it) to enable proper communication.

The HTTP container runs a web server on port 3000 for testing with MCP Inspector.

Using Task (task runner)

If you have Task installed, you can use these commands:

# Build and push both Docker images to registry (requires REGISTRY_USER env var)
# Note: This pushes to docker.io/REGISTRY_USER/mcp-weather:latest-stdio and :latest-http
REGISTRY_USER=your-dockerhub-username task build

# Run stdio version locally (requires OPENWEATHER_API_KEY env var)
OPENWEATHER_API_KEY="your_api_key_here" task run-stdio

# Run HTTP version locally (requires OPENWEATHER_API_KEY env var)
OPENWEATHER_API_KEY="your_api_key_here" task run-http

# Clean up Docker images
task clean

Kubernetes

The project includes Kubernetes manifests for deploying the MCP server in a Kubernetes cluster. The deployment uses the HTTP variant for easier connectivity.

Kubernetes Resources

The kubernetes/ directory contains:

  • pod.yaml: Pod definition for the MCP weather server
  • service.yaml: ClusterIP service to expose the pod internally
  • secret.yaml: Example secret for the OpenWeatherMap API key

Deploy to Kubernetes

Using Task (recommended)
# Deploy the MCP server with your API key
OPENWEATHER_API_KEY="your_api_key_here" task kubernetes-apply

# Port forward to access the service locally
task kubernetes-port-forward

# Clean up the deployment
task kubernetes-delete
Using kubectl directly
# Create the secret with your API key
kubectl create secret generic openweather-api-key \
  --from-literal=OPENWEATHER_API_KEY="your_api_key_here"

# Apply the Kubernetes manifests
kubectl apply -f kubernetes/

# Port forward to access the service (runs on localhost:3000)
kubectl port-forward svc/mcp-weather 3000:3000

# Delete the deployment when done
kubectl delete -f kubernetes/

Accessing the Service

Once deployed, you can:

  1. Port forward to test locally:

    kubectl port-forward svc/mcp-weather 3000:3000

    Then access the MCP server at http://localhost:3000/mcp

  2. Test with MCP Inspector:

    npx @modelcontextprotocol/inspector

    Connect to http://localhost:3000/mcp while port forwarding is active

Kubernetes Configuration Notes

  • The pod uses the denhamparry/mcp-weather:latest-http image
  • API key is stored as a Kubernetes secret for security
  • Service exposes port 3000 internally within the cluster
  • Pod has imagePullPolicy: Always to ensure latest image is used

Test

To test weather functionality:

# Set your API key first
export OPENWEATHER_API_KEY="your_api_key_here"

# Run the Cardiff weather test
npm test

Examples

The examples/ directory contains configuration examples for:

  • Claude Code integration (examples/claude code/)
  • Claude Desktop integration (examples/claude desktop/)

Note

This server uses the OpenWeatherMap API which provides weather data for worldwide locations including the UK. The forecast shows temperature in Celsius and wind speeds in meters per second.