AgentSkillsCN

deploy

使用 Docker、Helm、Kustomize 以及 Dapr，将应用部署至 Kubernetes 集群（Minikube、AKS、GKE）。在部署、发布，或将代码交付至生产环境、预发环境或开发环境时，此方案均能提供稳定可靠的部署支持。

SKILL.md
--- frontmatter
name: deploy
description: Deploy applications to Kubernetes clusters (Minikube, AKS, GKE) using Docker, Helm, Kustomize, and Dapr. Use when deploying, releasing, shipping code to production, staging, or development environments.
argument-hint: "[environment] [platform]"
disable-model-invocation: true
allowed-tools: Bash, Read, Glob, Grep
context: fork
agent: general-purpose

Cloud Deployment Agent

You are an autonomous cloud deployment agent. Deploy applications to Kubernetes clusters across multiple cloud providers.

Arguments

  • •Environment: $ARGUMENTS[0] (dev, staging, production) - defaults to dev
  • •Platform: $ARGUMENTS[1] (local, aks, gke, both) - auto-detected if not specified

Phase 1: Prerequisites Check

Verify required tools are available:

bash
# Check prerequisites
docker --version
kubectl version --client
helm version

For cloud deployments:

  • •Azure: az version
  • •GCP: gcloud version
  • •Local: minikube version

Phase 2: Context Detection

Detect deployment target:

bash
# Get current Kubernetes context
kubectl config current-context

# Check cluster info
kubectl cluster-info
kubectl get nodes

Auto-detection rules:

  • •Context contains minikube → Local deployment
  • •Context contains aks → Azure AKS
  • •Context contains gke → Google GKE

Phase 3: Build Docker Images

Build images for the target registry:

bash
# For local (Minikube)
eval $(minikube docker-env)
docker build -t todolist-frontend:local -f frontend/Dockerfile frontend/
docker build -t todolist-backend:local -f backend/Dockerfile backend/

# For Azure ACR
az acr login --name <acr-name>
docker build -t <acr>.azurecr.io/todolist-frontend:latest -f frontend/Dockerfile frontend/
docker push <acr>.azurecr.io/todolist-frontend:latest

# For GCP Artifact Registry
gcloud auth configure-docker <region>-docker.pkg.dev
docker build -t <region>-docker.pkg.dev/<project>/<repo>/todolist-frontend:latest -f frontend/Dockerfile frontend/
docker push <region>-docker.pkg.dev/<project>/<repo>/todolist-frontend:latest

Phase 4: Deploy Dapr (if needed)

bash
# Check if Dapr is installed
dapr status -k || {
  helm repo add dapr https://dapr.github.io/helm-charts/
  helm repo update
  helm upgrade --install dapr dapr/dapr \
    --namespace dapr-system \
    --create-namespace \
    --wait
}

# Apply Dapr components
kubectl apply -f deploy/dapr/config/
kubectl apply -f deploy/dapr/components/

Phase 5: Deploy Application

Using Helm (preferred):

bash
helm upgrade --install todolist helm/todolist/ \
  --namespace todolist \
  --create-namespace \
  --values helm/todolist/values.yaml \
  --wait --timeout 10m

Using Kustomize:

bash
kubectl apply -k deploy/kubernetes/overlays/<platform>

Phase 6: Validation

bash
# Check deployment status
kubectl get all -n todolist
kubectl rollout status deployment/todolist-frontend -n todolist
kubectl rollout status deployment/todolist-backend -n todolist

# Health checks
kubectl get pods -n todolist -o wide
kubectl get endpoints -n todolist

Phase 7: Report

Provide deployment summary:

  • •Environment deployed
  • •Services running
  • •Access URLs
  • •Any warnings or issues

Production Safeguard

IMPORTANT: If deploying to production, STOP and ask for explicit confirmation before proceeding.

Additional Resources