The current release of the BuiltForProd repositories is version 1.0.0 in every edition, as the README of each infrastructure and code repository states, and its layout is Terragrunt Stacks: every unit is defined once under units/, every account, subscription, project, region or stage is described by a terragrunt.stack.hcl, and Terragrunt generates the deployable tree. This page lists what ships in that release for the AWS, Azure and GCP Enterprise Baselines and their blueprints, and the exact versions each pins.
What ships
AWS edition
| Repository | Version | Contents |
|---|
acme-aws-platform-baseline | 1.0.0 | The landing zone: Organization, 15 accounts in two OUs, SCPs and tag policy, hub-and-spoke network with IPAM, security services, identity, GitHub OIDC, state backend, three workflows |
acme-aws-blueprint-webapp-infra | 1.0.0 | Web platform per stage: EKS with Karpenter and Cluster Autoscaler, DocumentDB, ElastiCache, load balancer controller, CloudFront, ArgoCD, External Secrets, Fluent Bit (16 units per stage) |
acme-aws-blueprint-webapp-code | 1.0.0 | Flask API and React frontend, Dockerfile, CI and the three delivery workflows |
acme-aws-blueprint-webapp-gitops | 1.0.0 | Helm chart and per-stage values ArgoCD syncs |
acme-aws-blueprint-etl-infra | 1.0.0 | Data platform per stage: S3 data lake, Lake Formation, Glue, Lambda trigger (6 units per stage) |
acme-aws-blueprint-etl-code | 1.0.0 | Glue job script and Lambda container with CI and CD |
acme-aws-blueprint-secrets | 1.0.0 | SOPS-encrypted secrets per stage, synced to SSM Parameter Store |
Azure edition
| Repository | Version | Contents |
|---|
acme-azure-platform-baseline | 1.0.0 | The landing zone: three management groups, 15 subscriptions, Azure Policy guardrails, hub VNet with Azure Firewall, DNS, Defender for Cloud, immutable audit storage, Entra RBAC, federated identities, Container Apps runners, state backend, four workflows |
acme-azure-blueprint-webapp-infra | 1.0.0 | Web platform per stage: AKS with Node Auto Provisioning, managed ingress, Front Door, Cosmos DB for MongoDB vCore, Azure Managed Redis, application Key Vault, Application Insights, External Secrets, cert-manager, ArgoCD (13 units per stage) |
acme-azure-blueprint-webapp-code | 1.0.0 | Flask API and React front end, Dockerfile, CI and the delivery workflows |
acme-azure-blueprint-webapp-gitops | — | Helm chart and per-stage values ArgoCD syncs |
acme-azure-blueprint-etl-infra | 1.0.0 | Data platform per stage: ADLS Gen2 lake, Azure Databricks workspace, Unity Catalog, Databricks job, Event Grid and Container Apps trigger (5 units per stage) |
acme-azure-blueprint-etl-code | 1.0.0 | PySpark transform and trigger container with CI and CD |
acme-azure-blueprint-secrets | — | SOPS-encrypted secrets per stage, synced to the application Key Vault of each stage |
GCP edition
| Repository | Version | Contents |
|---|
acme-gcp-platform-baseline | 1.0.0 | The landing zone: organization, three folders, 15 projects, organization policies and an IAM deny policy, Shared VPC with firewall policies and Cloud NAT, DNS, Security Command Center, aggregated audit logs, groups and Privileged Access Manager, Workload Identity Federation, Artifact Registry, state backend, five workflows |
acme-gcp-blueprint-webapp-infra | 1.0.0 | Web platform per stage: GKE, External Secrets, external-dns (public and internal), cert-manager, ArgoCD, Firestore, Memorystore for Valkey, the Cloud Run front end and the API edge (11 units per stage) |
acme-gcp-blueprint-webapp-code | 1.0.0 | Flask API and React front end, Dockerfile, CI and the delivery workflows |
acme-gcp-blueprint-webapp-gitops | — | Helm chart with the GKE Gateway and per-stage values ArgoCD syncs |
acme-gcp-blueprint-etl-infra | 1.0.0 | Data platform per stage: Spark identity, Cloud Storage lake, BigQuery datasets, Dataplex, Eventarc-driven Cloud Run trigger (5 units per stage) |
acme-gcp-blueprint-etl-code | 1.0.0 | PySpark transform and trigger service with CI and CD |
acme-gcp-blueprint-secrets | — | SOPS-encrypted secrets per stage, synced to Secret Manager in each stage project |
The GitOps and secrets repositories carry no version line in their README; they ship with the release of their blueprint.
The Terragrunt Stacks layout
Every repository defines each unit type once and describes each folder with a stack file. Templates group the units a workload stage always has:
| Template | AWS edition | Azure edition | GCP edition |
|---|
| Workload stage of the Baseline | stacks/plat-account (seven units) | stacks/plat-subscription (eight units) | stacks/plat-project (four units) |
| Shared includes | stacks/includes/account-common | stacks/includes/subscription-common, subscription-link | stacks/includes/project-common |
| Web App Blueprint stage | stacks/webapp-stage (sixteen units) | stacks/webapp-stage (thirteen units) | stacks/webapp-stage (eleven units) |
| Data and ETL Blueprint stage | stacks/etl-stage (six units) | stacks/etl-stage (five units) | stacks/etl-stage (five units) |
Terragrunt generates the units into git-ignored .terragrunt-stack/ folders, and root.hcl keeps every state key equal to the folder path. scripts/check-stack-layout.py guards the layout in pre-commit and CI, and the plan and apply workflows upload a JSON run report per run. The concept is explained under Terragrunt units and stacks.
Pinned versions
| Component | Version | Where it is pinned |
|---|
| OpenTofu | 1.12.6 | root.hcl (>= 1.12.6, < 2.0.0); TF_VERSION in every workflow |
| Terragrunt | 1.1.6 | root.hcl (>= 1.1.6); TG_VERSION in every workflow |
| AWS provider | ~> 6.65 | root.hcl generate "versions", repeated in module versions.tf files that need aliases or extra providers |
| Helm provider | ~> 3.3 | Module versions.tf (web platform) |
| Kubernetes provider | ~> 3.2 | Module versions.tf (web platform) |
| tflint | v0.64.0 | plan.yml (landing zone) |
| tflint AWS ruleset | 0.48.0 | .tflint.hcl |
| SOPS | 3.13.3 | SOPS_VERSION in the secrets workflows |
| GitHub Actions | Commit SHA per action, version in the trailing comment | .github/workflows/*.yml in every repository |
| Component | Version | Where it is pinned |
|---|
| OpenTofu | 1.12.6 | root.hcl (>= 1.12.6, < 2.0.0); TF_VERSION in every workflow; the runner image |
| Terragrunt | 1.1.6 | root.hcl (>= 1.1.6); TG_VERSION in every workflow; the runner image |
hashicorp/azurerm | ~> 5.7 | root.hcl generate "versions" in every infrastructure repository |
hashicorp/azuread, Azure/azapi | ~> 3.10, ~> 2.13 | root.hcl (Baseline) |
hashicorp/time | ~> 0.13 | Four Baseline module versions.tf files |
hashicorp/helm, hashicorp/kubernetes | ~> 3.3, ~> 3.2 | Module versions.tf (web platform) |
hashicorp/random | ~> 3.7 | root.hcl (web platform) |
databricks/databricks | ~> 1.135 | Module versions.tf (data platform) |
| tflint | v0.64.0 | plan.yml |
tflint-ruleset-azurerm | 0.32.0 | .tflint.hcl |
| SOPS, yq | 3.13.3, v4.47.1 | SOPS_VERSION, YQ_VERSION in the secrets workflows; the runner image |
| kubectl, Helm, kubelogin | v1.36.5, v3.22.0, v0.2.20 | runner-image/Dockerfile, checksum-verified |
| GitHub Actions | Commit SHA per action, version in the trailing comment | .github/workflows/*.yml in every repository |
| Component | Version | Where it is pinned |
|---|
| OpenTofu | 1.12.6 | root.hcl (>= 1.12.6, < 2.0.0); TF_VERSION in every workflow |
| Terragrunt | 1.1.6 | root.hcl (>= 1.1.6); TG_VERSION in every workflow |
hashicorp/google, hashicorp/google-beta | ~> 8.5 | root.hcl generate "versions", repeated in module versions.tf files with aliases |
hashicorp/helm | ~> 3.3 | Module versions.tf (Baseline runners, web platform) |
hashicorp/kubernetes | ~> 3.2 | Module versions.tf (web platform) |
| tflint | v0.64.0 | plan.yml |
tflint-ruleset-google | 0.40.0 | .tflint.hcl |
| SOPS, yq | 3.13.3, v4.47.1 | SOPS_VERSION, YQ_VERSION in the secrets workflows |
| GitHub Actions | Commit SHA per action, version in the trailing comment | .github/workflows/*.yml in every repository |
Registry modules (AWS edition)
| Module | Version | Where it is used |
|---|
terraform-aws-modules/vpc/aws | 6.7.3 | units/vpc-hub, units/vpc-spoke, units/vpc-runner; modules/vpc-endpoints submodule (landing zone) |
terraform-aws-modules/transit-gateway/aws | 3.3.1 | units/transit-gateway (landing zone) |
terraform-aws-modules/ecr/aws | 3.2.1 | modules/ecr (landing zone) |
terraform-aws-modules/s3-bucket/aws | 5.16.1 | modules/s3 (landing zone, ETL); modules/frontend-cdn (web platform) |
github-aws-runners | 7.11.0 | modules/github-runner (landing zone) |
terraform-aws-modules/eks/aws | 21.25.1 | modules/eks and its karpenter submodule in modules/karpenter (web platform) |
terraform-aws-modules/eks-pod-identity/aws | 2.9.0 | modules/alb-controller, app-namespace, cluster-autoscaler, external-dns, external-secrets (web platform) |
terraform-aws-modules/elasticache/aws | 1.11.1 | modules/elasticache (web platform) |
terraform-aws-modules/cloudfront/aws | 6.7.1 | modules/frontend-cdn (web platform) |
terraform-aws-modules/lambda/aws | 8.8.2 | modules/lambda (ETL) |
The design decisions record every pin as the latest release as of 2026-09-19. The landing zone pins vpc and transit-gateway in the unit definitions; every other registry module is pinned in the module's main.tf.
The Azure and GCP repositories use no registry modules: every module is custom and lives in its repository, so there are no module versions to track.
Runtime versions set in the units
| Component (AWS edition) | Version | Where |
|---|
| Kubernetes (EKS) | 1.36 | units/eks (web platform) |
| DocumentDB engine | 5.0.0 | units/documentdb |
| ElastiCache Redis engine | 7.1 | units/elasticache |
| Glue | 4.0, PySpark | modules/glue-job (ETL) |
| Component (Azure edition) | Version | Where |
|---|
| Kubernetes (AKS) | 1.36 at creation, then the patch channel | units/aks (web platform) |
| Cosmos DB for MongoDB vCore server | 8.0 | modules/mongo-cluster (web platform) |
| ArgoCD chart | 9.5.22 | modules/argocd (web platform) |
| cert-manager chart | v1.21.2 | modules/cert-manager (web platform) |
| External Secrets chart | 2.6.0 | modules/external-secrets (web platform) |
| Databricks runtime | The latest long-term-support release | modules/databricks-job (data platform) |
| Component (GCP edition) | Version | Where |
|---|
| Kubernetes (GKE) | 1.35 minimum at creation, then the REGULAR release channel | units/gke and modules/gke (web platform) |
| Memorystore for Valkey | VALKEY_8_0 | modules/valkey (web platform) |
| ArgoCD chart | 10.9.6 | modules/argocd (web platform) |
| cert-manager chart | v1.21.2 | modules/cert-manager (web platform) |
| external-dns chart | 1.23.0 | modules/external-dns (web platform) |
| External Secrets chart | 2.11.0 | modules/external-secrets (web platform) |
| Dataproc Serverless runtime | 2.3 (Spark 3.5, Python 3.11) | stacks/etl-stage and modules/cloud-run-trigger (data platform) |
The Azure and GCP Baseline version pages list every GitHub Action and pre-commit hook revision of the Baseline repositories.
How versions change
Pins are updated one component at a time, planned in dev before staging and prod, with a risk level and an approver per kind of change; see the versioning policy. The procedures that keep a running deployment current are indexed under upgrade guides.