MCP Cloud Run Sample - Google Store Mock

This project demonstrates how to deploy an MCP server to Google Cloud Run. This specific implementation mocks a Google Store API, providing data about phones, earbuds, watches, and more.

Setup GCP Project

Configure your working project:

gcloud config set project [PROJECT_ID]

Enable required APIs:

gcloud services enable \
  run.googleapis.com \
  artifactregistry.googleapis.com \
  cloudbuild.googleapis.com

Init project with uv

uv init --description "Example of deploying a Google Store Mock MCP server on Cloud Run" --bare --python 3.13

Create service account

gcloud iam service-accounts create mcp-server-sa --display-name="MCP Server Service Account"

server.py

The server uses FastMCP to define tools and resources for mocking Google Store data.

import asyncio
import logging
import os
from typing import List, Dict, Any, Optional

from fastmcp import FastMCP

logger = logging.getLogger(__name__)
logging.basicConfig(format="[%(levelname)s]: %(message)s", level=logging.INFO)

mcp = FastMCP("Google Store Mock Server 📱⌚🏠")

# Mock data for Google Store
STORE_PRODUCTS = [
    # Phones
    {
        "id": "p9pxl",
        "category": "phones",
        "name": "Pixel 9 Pro XL",
        "description": "The most powerful Pixel yet, with a pro-level camera and Gemini AI.",
        "price": 1099.00,
        "currency": "USD",
        "specs": ["6.8-inch Super Actua display", "Triple rear camera system", "Tensor G4 chip"]
    },
    # ... (see server.py for full data)
]

@mcp.tool()
def list_categories() -> List[str]:
    """Lists all available product categories in the Google Store."""
    # ...

@mcp.tool()
def get_products_by_category(category: str) -> List[Dict[str, Any]]:
    """Retrieves all products for a specific category."""
    # ...

@mcp.tool()
def get_product_details(product_name: str) -> Optional[Dict[str, Any]]:
    """Retrieves detailed information about a specific product."""
    # ...

@mcp.tool()
def get_active_offers() -> List[Dict[str, Any]]:
    """Retrieves currently active offers and promotions."""
    # ...

if __name__ == "__main__":
    port = int(os.environ.get("PORT", 8080))
    asyncio.run(
        mcp.run_async(
            transport="streamable-http", 
            host="0.0.0.0", 
            port=port
        )
    )

Package Installation

uv add fastmcp==2.12.4 --no-sync

Deploy to Cloud Run

export GOOGLE_CLOUD_PROJECT=your-project-id

gcloud run deploy google-store-mcp-server \
    --service-account=mcp-server-sa@$GOOGLE_CLOUD_PROJECT.iam.gserviceaccount.com \
    --no-allow-unauthenticated \
    --region=europe-west1 \
    --source=. \
    --labels=dev-tutorial=codelab-mcp

Troubleshooting Local uv Issues

If you encounter issues with the uv.lock file referring to private registries:

# Delete the old lockfile
rm uv.lock

# Regenerate it forcing the public index
uv lock --default-index https://pypi.org/simple

# Authenticate for local development if needed
gcloud auth login
gcloud auth application-default login
uv pip install keyrings.google-artifactregistry-auth
export UV_KEYRING_PROVIDER=subprocess
```# mcp-cloudrun-gstore