gcp
π―SkillInstallation
npx skills add g1joshi/agent-skills --skill gcpRun it in a terminal; it asks which AI coding tools to install it for (Claude Code, Codex, Cursor and more).
Uses skills.sh Β· Node.js 22.20+
After installing
Ask for related work and the agent uses it on its own, or call it yourself.
/gcpInstall by hand
Copy the whole skill folder from GitHub to the place below. Copying only SKILL.md can leave out files the skill uses.
Open the skill folder β- This project
.claude/skills/gcp/ - All projects
~/.claude/skills/gcp/
These paths are for Claude Code. For other tools, the command above puts it in the right place.
SKILL.md
View on GitHub βExpert Google Cloud Platform (GCP) assistance covering Cloud Run, GKE, Cloud Functions, BigQuery, Pub/Sub, and IAM. Use when architecting and operating scalable, enterprise cloud applications on Google Cloud.
Google Cloud Platform (GCP)
Google Cloud Platform (GCP) provides scalable cloud infrastructure, featuring managed container runtimes (Cloud Run, GKE), BigQuery analytics, and Vertex AI foundation model tooling.
When to Use
- Enterprise Google Cloud Infrastructure: Compute Engine (GCE), Google Kubernetes Engine (GKE), and Cloud Run.
- Big Data Analytics & AI/ML: BigQuery data warehousing, Vertex AI foundation models, and Cloud Pub/Sub streaming.
- Global Anycast Networking & CDN: Google Cloud Armor, Cloud CDN, and Cloud Load Balancing with single global anycast IPs.
- Serverless Event-Driven Microservices: Cloud Run, Cloud Functions, and Eventarc.
Quick Start
# Deploy container directly to serverless Cloud Run
gcloud run deploy web-service \
--image us-docker.pkg.dev/cloudrun/container/hello \
--platform managed \
--region us-central1 \
--allow-unauthenticated
Core Concepts
Cloud Run Container Service Architecture
Deploying containerized microservices that scale to zero and scale up dynamically:
# cloud-run-service.yaml
apiVersion: serving.knative.dev/v1
kind: Service
metadata:
name: order-service
labels:
cloud.googleapis.com/location: us-central1
spec:
template:
metadata:
annotations:
autoscaling.knative.dev/minScale: "1"
autoscaling.knative.dev/maxScale: "50"
run.googleapis.com/cpu-throttling: "true"
spec:
containerConcurrency: 80
timeoutSeconds: 300
containers:
- image: us-central1-docker.pkg.dev/my-project/apps/order-service:v2.1
resources:
limits:
cpu: "1000m"
memory: "1Gi"
env:
- name: DB_USER
valueFrom:
secretKeyRef:
name: db-credentials
key: username
Pub/Sub Event Streaming with Python
Publishing and subscribing to real-time events with high throughput:
from google.cloud import pubsub_v1
import json
project_id = "my-gcp-project"
topic_id = "telemetry-events"
publisher = pubsub_v1.PublisherClient()
topic_path = publisher.topic_path(project_id, topic_id)
data = json.dumps({"sensor_id": 401, "temp": 24.5, "status": "OK"}).encode("utf-8")
future = publisher.publish(topic_path, data, origin="edge-gateway")
message_id = future.result()
print(f"Published message ID: {message_id}")
Zero-Trust VPC Service Controls & IAM
Restricting access to GCP APIs within secure service perimeters:
[Internet Client] βββΊ [Global External HTTPS Load Balancer] (Cloud Armor WAF)
β
βΌ
[GKE / Cloud Run] (Private VPC)
β (Private Google Access)
βΌ
[VPC Service Controls Perimeter]
βββ BigQuery Datasets
βββ Cloud Storage Buckets
Common Patterns
Event-Driven Architecture with Cloud Pub/Sub and Cloud Run
Problem: Decoupling synchronous service dependencies to absorb traffic surges without dropping requests.
Solution: Publish events to Pub/Sub and consume via Cloud Run push subscriptions:
# 1. Create Pub/Sub Topic
gcloud pubsub topics create order-events
# 2. Create Cloud Run push subscription targeting invoice worker service
gcloud pubsub subscriptions create order-events-sub \
--topic order-events \
--push-endpoint https://invoice-service-xyz-uc.a.run.app/events \
--push-auth-service-account invoice-invoker@my-project.iam.gserviceaccount.com
Best Practices
Do:
- Deploy serverless workloads to Cloud Run for automated scale-to-zero and high concurrency per container.
- Protect applications with Google Cloud Armor security policies to neutralize DDoS and OWASP Top 10 exploits.
- Configure VPC Service Controls to prevent data exfiltration from BigQuery and Cloud Storage.
- Enable Secret Manager integration directly into Cloud Run and GKE pods rather than passing raw env strings.
Don't:
- Use standard service account keys; use Workload Identity for GKE pods and Cloud Run services.
- Assign
roles/editororroles/ownerto service accounts; adhere strictly to least-privilege IAM roles. - Expose BigQuery datasets publicly without explicit authorized views.
Troubleshooting
| Error | Cause | Solution |
|---|---|---|
Cloud Run: The user-provided container failed to start and listen on the port | Container listening on localhost or ignoring $PORT environment variable. | Ensure web server binds to 0.0.0.0 and listens on process.env.PORT (default 8080). |
403 Forbidden: Caller does not have required permission | Service account lacks IAM role on target resource. | Grant appropriate IAM role: roles/run.invoker or roles/storage.objectViewer. |
Cloud Build: Step failed with code 1 | Build failure inside container build step. | Inspect build logs in GCP Console under Cloud Build > History. |
References
More from this repository10
A template skill providing a starter framework for creating new agent skills with structured sections for triggers, quick start examples, and capability descriptions.
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Skill