Google Cloud

Run Agent Gateway on GCP to leverage Vertex AI, GKE, and Google Cloud services.

Deployment options

Google Kubernetes Engine (GKE)

For GKE deployments, use kgateway which provides native Kubernetes Gateway API support, dynamic configuration, and MCP service discovery.

Cloud Run

Run Agent Gateway as a serverless container on Cloud Run.

gcloud run deploy agentgateway \
  --image ghcr.io/agentgateway/agentgateway:latest \
  --port 3000 \
  --set-env-vars "GOOGLE_CLOUD_PROJECT=my-project" \
  --service-account [email protected] \
  --allow-unauthenticated

GCP integrations

Integration Purpose
Vertex AI Access Gemini and other models
Google Gemini Direct Gemini API access
GCP Secret Manager Secure API key storage
Cloud Load Balancing Global load balancing with SSL
Cloud Trace Distributed tracing
Cloud Monitoring Metrics and alerting

IAM permissions

Create a service account with these roles:

# Create service account
gcloud iam service-accounts create agentgateway \
  --display-name "Agent Gateway"

# Grant Vertex AI access
gcloud projects add-iam-policy-binding my-project \
  --member "serviceAccount:[email protected]" \
  --role "roles/aiplatform.user"

# Grant Secret Manager access
gcloud projects add-iam-policy-binding my-project \
  --member "serviceAccount:[email protected]" \
  --role "roles/secretmanager.secretAccessor"

Learn more