| name | kro-rgd-pulumi |
| description | Create production-ready KRO ResourceGraphDefinitions using Pulumi TypeScript. Use when users need to define custom Kubernetes APIs, compose resources with KRO, integrate AWS ACK resources, or build platform abstractions using Pulumi infrastructure as code. |
KRO ResourceGraphDefinition with Pulumi TypeScript
Generate production-ready Kubernetes Resource Orchestrator (KRO) ResourceGraphDefinitions using Pulumi TypeScript for creating custom Kubernetes APIs and composing resources with AWS ACK integration.
When to Use This Skill
Use this skill when the user wants to:
- Create custom Kubernetes APIs using KRO ResourceGraphDefinitions
- Compose multiple K8s resources as a single declarative unit
- Integrate AWS resources via ACK (S3, RDS, DynamoDB, etc.) with KRO
- Build platform abstractions for developer self-service
- Generate Pulumi TypeScript code for KRO resources
- Define resource dependencies with automatic orchestration
- Create reusable application templates with CEL expressions (use CEL skill when necessary)
Overview
What is KRO?
KRO (Kube Resource Orchestrator) is an open-source Kubernetes-native project that allows you to:
- Define custom Kubernetes APIs without writing Go controllers
- Compose multiple resources as a directed acyclic graph (DAG)
- Automatically manage resource dependencies and ordering
- Use CEL expressions for dynamic configuration
- Provide lifecycle management with drift detection
What is ResourceGraphDefinition (RGD)?
An RGD is KRO's core custom resource that:
- Defines a schema for your custom API (apiVersion, kind, spec, status)
- Lists resources to create with CEL expression templating
- Automatically generates a CRD when applied
- Deploys a microcontroller to manage instances
Pulumi Integration
Using Pulumi TypeScript to deploy KRO resources provides:
- Type safety for resource definitions
- IDE support with autocomplete and error checking
- Programmatic logic for complex configurations
- GitOps-friendly workflow
- Multi-stack architecture for separation of concerns
Prerequisites
Required Components
kubectl version
helm install kro oci://ghcr.io/kubernetes-sigs/kro/charts/kro \
--namespace kro-system \
--create-namespace
helm install ack-s3-controller \
oci://public.ecr.aws/aws-controllers-k8s/s3-chart \
--namespace ack-system \
--create-namespace
pulumi version
node --version
Pulumi Project Setup
mkdir my-kro-project && cd my-kro-project
pulumi new kubernetes-typescript
npm install @pulumi/kubernetes @pulumi/pulumi
Instructions
Step 1: Understand the Requirements
Before generating code, gather:
- Custom API design: What kind/apiVersion? What fields in spec?
- Resources to compose: Deployments, Services, ConfigMaps, ACK resources?
- Dependencies: Which resources depend on others?
- Conditions: Any conditional resource creation (includeWhen)?
- Status fields: What status to expose to users?
Step 2: Design the Schema
Define the user-facing API:
Step 3: Generate Pulumi TypeScript Code
Use this structure for KRO RGD:
import * as k8s from "@pulumi/kubernetes";
import * as pulumi from "@pulumi/pulumi";
const rgd = new k8s.apiextensions.CustomResource("my-app-rgd", {
apiVersion: "kro.run/v1alpha1",
kind: "ResourceGraphDefinition",
metadata: {
name: "my-application",
namespace: "default",
},
spec: {
schema: {
apiVersion: "v1alpha1",
kind: "Application",
spec: {
name: "string",
image: "string | default=\"nginx:latest\"",
replicas: "integer | default=3",
},
status: {
ready: "${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}",
availableReplicas: "${deployment.status.availableReplicas}",
},
validation: [
{
expression: "self.replicas >= 1 && self.replicas <= 100",
message: "Replicas must be between 1 and 100",
},
],
},
resources: [
{
id: "deployment",
template: {
apiVersion: "apps/v1",
kind: "Deployment",
metadata: {
name: "${schema.spec.name}",
},
spec: {
replicas: "${schema.spec.replicas}",
selector: {
matchLabels: {
app: "${schema.spec.name}",
},
},
template: {
metadata: {
labels: {
app: "${schema.spec.name}",
},
},
spec: {
containers: [
{
name: "app",
image: "${schema.spec.image}",
},
],
},
},
},
},
readyWhen: [
"${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}",
],
},
{
id: "service",
template: {
apiVersion: "v1",
kind: "Service",
metadata: {
name: "${schema.spec.name}-service",
},
spec: {
selector: {
app: "${schema.spec.name}",
},
ports: [
{
port: 80,
targetPort: 8080,
},
],
},
},
},
],
},
});
export const rgdName = rgd.metadata.name;
Step 4: Add ACK Resources (AWS Integration)
For AWS resources via ACK:
const rgdWithAws = new k8s.apiextensions.CustomResource("app-with-aws-rgd", {
apiVersion: "kro.run/v1alpha1",
kind: "ResourceGraphDefinition",
metadata: {
name: "application-with-storage",
},
spec: {
schema: {
apiVersion: "v1alpha1",
kind: "AppWithStorage",
spec: {
name: "string",
image: "string",
bucketName: "string",
},
status: {
bucketArn: "${bucket.status.ackResourceMetadata.arn}",
ready: "${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}",
},
},
resources: [
{
id: "bucket",
template: {
apiVersion: "s3.services.k8s.aws/v1alpha1",
kind: "Bucket",
metadata: {
name: "${schema.spec.bucketName}",
},
spec: {
name: "${schema.spec.bucketName}",
tagging: {
tagSet: [
{
key: "ManagedBy",
value: "KRO",
},
],
},
},
},
readyWhen: [
"${bucket.status.ackResourceMetadata.?arn != null}",
],
},
{
id: "deployment",
template: {
apiVersion: "apps/v1",
kind: "Deployment",
metadata: {
name: "${schema.spec.name}",
},
spec: {
replicas: 3,
selector: {
matchLabels: {
app: "${schema.spec.name}",
},
},
template: {
metadata: {
labels: {
app: "${schema.spec.name}",
},
},
spec: {
containers: [
{
name: "app",
image: "${schema.spec.image}",
env: [
{
name: "S3_BUCKET",
value: "${schema.spec.bucketName}",
},
{
name: "S3_BUCKET_ARN",
value: "${bucket.status.ackResourceMetadata.arn}",
},
],
},
],
},
},
},
},
},
],
},
});
Step 5: Deploy and Test
pulumi up
kubectl get resourcegraphdefinition
kubectl describe resourcegraphdefinition my-application
kubectl get crd | grep kro
kubectl apply -f - <<EOF
apiVersion: v1alpha1
kind: Application
metadata:
name: my-test-app
spec:
name: my-test-app
image: nginx:1.21
replicas: 5
EOF
kubectl get application
kubectl describe application my-test-app
CEL Expression Reference
Referencing Schema Fields
${schema.spec.name} // User-provided spec field
${schema.metadata.name} // Instance name
${schema.metadata.namespace} // Instance namespace
${schema.metadata.uid} // Unique ID
Referencing Other Resources
${deployment.metadata.name} // Resource name
${deployment.spec.replicas} // Spec field
${deployment.status.?endpoint} // Optional status field (use ? for null safety)
Operators and Functions
// Ternary conditional
${schema.spec.env == 'prod' ? 10 : 3}
// String concatenation
${"prefix-" + schema.spec.name}
// Null coalescing
${deployment.status.?ready ?? false}
// List operations
${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}
${schema.spec.containers.map(c, c.name)}
// Type conversion
${string(schema.spec.replicas)}
${int(schema.spec.replicasString)}
Conditional Resource Inclusion
resources:
- id: ingress
includeWhen:
- ${schema.spec.ingress.enabled}
template:
apiVersion: networking.k8s.io/v1
kind: Ingress
Common Patterns
Pattern 1: Web Application Stack
const webAppRgd = createWebAppRgd({
name: "webapp",
supportsIngress: true,
supportsAutoscaling: true,
});
Pattern 2: Database Application
const dbAppRgd = createDatabaseAppRgd({
name: "dbapp",
dbEngine: "postgres",
withBackup: true,
});
Pattern 3: Microservices Bundle
const microservicesRgd = createMicroservicesRgd({
name: "platform",
services: ["api", "worker", "scheduler"],
});
Pattern 4: Multi-Environment App
const envAppRgd = createEnvironmentAppRgd({
name: "envapp",
environments: ["dev", "staging", "prod"],
});
ACK Resource Integration
Supported ACK Resources
| Service | Resource Types | ACK Controller |
|---|
| S3 | Bucket | s3-controller |
| RDS | DBInstance, DBCluster, DBSubnetGroup | rds-controller |
| DynamoDB | Table, GlobalTable | dynamodb-controller |
| SQS | Queue | sqs-controller |
| SNS | Topic, Subscription | sns-controller |
| Lambda | Function | lambda-controller |
| ElastiCache | CacheCluster, ReplicationGroup | elasticache-controller |
ACK Status Fields
// Common ACK status patterns
${bucket.status.ackResourceMetadata.arn} // AWS ARN
${bucket.status.ackResourceMetadata.ownerAccountID} // AWS Account
${dbinstance.status.endpoint.address} // RDS endpoint
${dbinstance.status.endpoint.port} // RDS port
${queue.status.queueURL} // SQS URL
Best Practices
1. Schema Design
spec:
name: string
image: string | default="nginx:latest"
replicas: integer | default=3
environment: string | default="dev"
spec:
n: string
i: string
r: integer
2. Validation Rules
validation:
- expression: "self.replicas >= 1 && self.replicas <= 100"
message: "Replicas must be between 1 and 100"
- expression: "self.environment in ['dev', 'staging', 'prod']"
message: "Environment must be dev, staging, or prod"
- expression: "self.environment == 'prod' ? self.replicas >= 3 : true"
message: "Production requires at least 3 replicas"
3. Readiness Conditions
readyWhen:
- ${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}
- ${deployment.status.readyReplicas == deployment.spec.replicas}
4. Status Exposure
status:
ready: ${deployment.status.conditions.exists(c, c.type == 'Available' && c.status == 'True')}
availableReplicas: ${deployment.status.availableReplicas}
endpoint: ${service.spec.clusterIP}
bucketArn: ${bucket.status.ackResourceMetadata.?arn ?? 'pending'}
5. Naming Conventions
metadata:
name: ${schema.spec.name}
labels:
app.kubernetes.io/name: ${schema.spec.name}
app.kubernetes.io/managed-by: kro
Reference Files
Current Status
KRO Status: v1alpha1 (Alpha)
- Not yet production-ready
- Breaking changes may occur
- Kubernetes 1.25+ required
- Latest release: v0.7.1 (December 2025)
Recommendation: Use in development/testing environments while awaiting v1beta1 for production readiness.
Output Format
When generating KRO RGD with Pulumi, always provide:
- Complete Pulumi TypeScript code with proper types
- RGD specification with schema, resources, and validations
- Example instance YAML showing how to use the custom API
- Deployment instructions for both Pulumi and kubectl
- Testing commands to verify the deployment