GC-201b · Module 1

Google Cloud Integration

3 min read

Gemini CLI's deepest integrations are with Google's own ecosystem. This is not surprising — Gemini models power the tool, and Google Cloud provides the infrastructure. The integration surfaces include Firebase (database, hosting, functions), Cloud Run (container deployment), BigQuery (analytics queries), Vertex AI (model management), and Google Workspace (Docs, Sheets, Drive). These are not generic API connections — they are purpose-built integrations that understand Google Cloud's authentication, project structure, and resource hierarchy.

Authentication for Google Cloud integrations uses your existing gcloud credentials. If you are authenticated via gcloud auth login, Gemini CLI's Google Cloud extensions inherit those credentials automatically. For service accounts in CI/CD, set the GOOGLE_APPLICATION_CREDENTIALS environment variable pointing to a service account key file. The credential chain follows Google's standard Application Default Credentials (ADC) flow.

# Authenticate with Google Cloud (one-time setup)
gcloud auth login
gcloud auth application-default login

# Install Google Cloud extensions
gemini extensions install firebase
gemini extensions install cloud-run
gemini extensions install bigquery

# Now Gemini can interact with your Google Cloud resources:
# "Deploy this function to Firebase"
# "Query the users table in BigQuery"
# "Show me the Cloud Run service logs for the last hour"

Do This

  • Use gcloud auth for local development — credentials flow automatically to extensions
  • Install Google Cloud extensions for services your project actually uses
  • Use service account credentials in CI/CD via GOOGLE_APPLICATION_CREDENTIALS

Avoid This

  • Hardcode Google Cloud credentials in settings.json or GEMINI.md
  • Install every Google Cloud extension when you only use Firebase and Cloud Run
  • Use generic API calls via run_shell_command when purpose-built extensions exist