Skip to main content

AWS CDK in IaCM

Last updated on

AWS Cloud Development Kit (CDK) lets you define cloud infrastructure using general-purpose programming languages instead of a domain-specific language like HCL. The code you write compiles to CloudFormation templates, which AWS uses to provision and manage the underlying resources. Harness IaCM currently supports Python and TypeScript, with JavaScript, Java, C#, and Go support coming in subsequent releases.

Harness IaCM treats CDK as a first-class provisioner alongside Terraform, OpenTofu, and Terragrunt. You create a CDK workspace, connect your AWS account and Git repository, and run a default deploy pipeline that manages the full CDK lifecycle. You can use pipeline controls such as approval steps with CDK.

Go to Get started with IaCM to create your first CDK workspace and run a deploy pipeline.

The CDK deployment lifecycle​

CDK workspaces in Harness IaCM run four pipeline steps in sequence. Each step corresponds to a phase of the CDK workflow:

Bootstrap provisions the CDK toolkit resources in your target AWS account and region. The toolkit consists of an S3 bucket to store CloudFormation assets and an ECR repository to store container image assets. Bootstrap is a one-time operation per account and region. Once the toolkit exists, subsequent pipeline runs skip or fast-forward this step.

Synth compiles your CDK app code into one or more CloudFormation templates. This step validates your code, resolves constructs, and produces the asset manifest that the deploy step will use. If your app contains errors, synth fails before any AWS resources are touched.

Diff compares the synthesized CloudFormation template against the currently deployed stack and reports what will change: resources to add, modify, or remove. It shows intent without executing anything.

Deploy executes the CloudFormation change set and provisions your infrastructure. Harness records the result in the workspace activity history, broken down by stack.

Go to CDK pipeline steps to read the full reference on each step's inputs and configuration options.

Stacks​

A CDK app can define one or more stacks. Each stack maps to a CloudFormation stack and is the unit of deployment. Resources inside a stack are provisioned and managed together.

When you run a CDK deploy pipeline in Harness, all stacks in your app are deployed by default. If you want to deploy a single stack (for example, when you have independent stacks per environment or service), set the Stack Path field on your pipeline steps to the identifier of the target stack.

The workspace activity history in Harness shows deployment results broken down by stack, so you can tell which stacks changed and their status within a single pipeline run.

Supported languages​

CDK supports multiple programming languages because it compiles to CloudFormation regardless of which language you use. Harness IaCM installs the required language runtime and package manager dynamically at pipeline execution time, rather than requiring a pre-baked Docker image.

When you create a CDK workspace, you select:

  • AWS CDK version: The CDK version to run your app against.
  • Programming Language: Python or TypeScript.
  • Language Version: The runtime version to install.
  • Package Manager: The package manager to use to install dependencies.
  • Package Manager Version: The package manager version to install.

CDK plugin environment variables​

CDK pipeline steps read plugin environment variables (the same PLUGIN_* pattern as other IaCM steps, for example PLUGIN_AWS_SESSION_DURATION). You must always set PLUGIN_AWS_REGION explicitly (for example as a manual variable on the workspace Connectors and Variables tab, or on the stage, pipeline, or step Environment Variables). Set PLUGIN_AWSCDK_STACKS only when you want the run to process specific stacks instead of every stack in the app; if you omit it, all stacks are included. Harness is working on a more seamless experience for stack selection.

VariableRequiredDescription
PLUGIN_AWS_REGIONYesTarget AWS Region for the deployment (for example us-east-1). Must be set for CDK steps to know where to deploy.
PLUGIN_AWSCDK_STACKSNoComma-separated logical stack IDs when you need only certain stacks processed (for example SQSStack,S3Stack). Omit to process every stack in the app.

You can still target a single stack from the pipeline UI using the Stack Path field on CDK steps; use PLUGIN_AWSCDK_STACKS when you want to drive a comma-separated list from variables or YAML. Go to AWS connector configuration (optional) to review optional region and role overrides at the step level in CD.

FAQs​

Do I need to run bootstrap every time I deploy?

No. Bootstrap is a one-time operation per AWS account and region. Once the CDK toolkit resources exist, subsequent pipeline runs detect them and proceed without re-bootstrapping.

Can I use CDK with an existing CloudFormation stack?

Yes, but you need to import the existing stack into your CDK app first. CDK does not automatically manage resources that were not created through it.

How does Harness track CDK state?

CDK state is managed by AWS CloudFormation, not by Harness. Harness tracks activity history and stack-level deployment results, but there is no Harness-managed state file for CDK workspaces.