Bay Wheels MCP Server

This is an MCP server that provides access to Bay Wheels realtime bikeshare data.

Features

• Find nearest bike (standard or ebike)

• Find nearest dock with available spaces

• Supports checking for free bikes (dockless) when looking for a single bike

Related MCP server: Bixi MCP Server

Setup

For Claude Desktop (Local Development)

Add this to your Claude Desktop configuration file:

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

{
  "mcpServers": {
    "bay-wheels": {
      "command": "/opt/homebrew/bin/uv",
      "args": [
        "--directory",
        "/path/to/bay-wheels-mcp",
        "run",
        "server.py"
      ]
    }
  }
}

Make sure to update the path to match your local installation directory.

Manual Testing (stdio)

You can run the server directly for testing with Claude Desktop:

uv run server.py

Deployment

Docker Deployment

Quick Start

# Build the image
docker build -t bay-wheels-mcp .

# Run the container
docker run -p 8000:8000 bay-wheels-mcp

# Test the health check
curl http://localhost:8000/health

Using Docker Compose

# Start the server
docker-compose up -d

# Check logs
docker-compose logs -f

# Stop the server
docker-compose down

Environment Variables

• PORT - Server port (default: 8000)

• HOST - Bind host (default: 0.0.0.0)

Health Check

The server exposes a health check endpoint at /health for container orchestration:

curl http://localhost:8000/health
# Response: {"status":"healthy","service":"bay-wheels-mcp","version":"0.1.0"}

Platform-Specific Deployment

AWS ECS/Fargate

1. Push image to ECR:

docker build -t bay-wheels-mcp .
docker tag bay-wheels-mcp:latest <aws-account>.dkr.ecr.<region>.amazonaws.com/bay-wheels-mcp:latest
docker push <aws-account>.dkr.ecr.<region>.amazonaws.com/bay-wheels-mcp:latest

2. Create ECS task definition with health check enabled

3. Deploy as ECS service with load balancer

Google Cloud Run

gcloud builds submit --tag gcr.io/<project-id>/bay-wheels-mcp
gcloud run deploy bay-wheels-mcp --image gcr.io/<project-id>/bay-wheels-mcp --port 8000

Fly.io

fly launch --dockerfile Dockerfile
fly deploy

Azure Container Instances

Use Azure Portal or Azure CLI to deploy the Docker image with port 8000 exposed.

Connecting Mobile Apps

The deployed server uses StreamableHTTP transport. Configure your MCP client to connect to:

URL: https://your-deployed-server.com/mcp

Note: The MCP endpoint is at /mcp, not the root path.

See MCP documentation for client integration details.

Testing

Testing the Deployed Server

The simplest way to test is using the health check endpoint:

# Test that the server is running
curl https://your-server.com/health
# Should return: {"status":"healthy","service":"bay-wheels-mcp","version":"0.1.0"}

For full MCP protocol testing, use an MCP client (Claude Desktop, mobile app, or custom client). The StreamableHTTP transport requires session management and proper header negotiation which is best handled by official MCP clients.

Testing with MCP Clients

The best way to test the deployed server is to configure it in your MCP client:

Claude Desktop (Remote Server)

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "bay-wheels-remote": {
      "url": "https://your-server.com/mcp",
      "transport": "streamable-http"
    }
  }
}

Mobile App

Configure your mobile app's MCP client to connect to:

https://your-server.com/mcp

Then test the tools by asking Claude:

• "Find me the nearest Bay Wheels bike near the Ferry Building in SF"

• "Where can I return a bike near Dolores Park?"

Tools

find_nearest_bike

Finds the nearest bike availability.

• latitude: float

• longitude: float

• count: int (default 1)

• bike_type: str (optional, "classic_bike" or "electric_bike")

find_nearest_dock_spaces

Finds the nearest dock with return spaces.

• latitude: float

• longitude: float

• count: int (default 1)