Google Cloud Service Account (Recommended for Production) Steps: bash# Create service account gcloud iam service-accounts create bigquery-finops
--display-name="BigQuery FinOps MCP Server"

Grant permissions

gcloud projects add-iam-policy-binding YOUR_PROJECT_ID
--member="serviceAccount:bigquery-finops@YOUR_PROJECT_ID.iam.gserviceaccount.com"
--role="roles/bigquery.admin"

Create and download key

gcloud iam service-accounts keys create service-account-key.json
--iam-account=bigquery-finops@YOUR_PROJECT_ID.iam.gserviceaccount.com

Move key to your project folder

move service-account-key.json C:\Users\User\bigquery_MCP\

#########################################################

Navigate to directory

cd C:\Users\User\bigquery_MCP example

Install dependencies

python -m pip install --upgrade pip python -m pip install mcp google-cloud-bigquery google-auth pandas numpy python-dotenv

Authenticate with Google Cloud

gcloud auth application-default login gcloud config set project YOUR_PROJECT_ID

Run test

python test_mcp.py

If all tests pass, restart Claude Desktop

Summary: Your Next Steps

Create the folder structure in C:\Users\User\bigquery_MCP
Save the MCP server code as bigquery_finops_mcp.py Create requirements.txt and run pip install Set up Google Cloud auth (choose gcloud or service account) Create config.json with your project settings Update Claude Desktop config with the correct path Run the test script to verify everything works Restart Claude Desktop Test it out by asking me to show your BigQuery costs!