The AWS CDK Toolkit provides the cdk command-line interface that can be used to work with AWS CDK applications.

• Bundling
• MFA Support
• SSO Support
• Configuration
  • Running in CI

This module is part of the AWS Cloud Development Kit project.

Outputs the URL to the documentation for the current toolkit version, and attempts to open a browser to that URL.

$ # Open the documentation in the default browser (using 'open')
$ cdk docs
https://docs.aws.amazon.com/cdk/api/latest/

$ # Open the documentation in Chrome.
$ cdk docs --browser='chrome %u'
https://docs.aws.amazon.com/cdk/api/latest/

Creates a new CDK project.

$ # List the available template types & languages
$ cdk init --list
Available templates:
* app: Template for a CDK Application
   └─ cdk init app --language=[csharp|fsharp|java|javascript|python|typescript]
* lib: Template for a CDK Construct Library
   └─ cdk init lib --language=typescript
* sample-app: Example CDK Application with some constructs
   └─ cdk init sample-app --language=[csharp|fsharp|java|javascript|python|typescript]

$ # Create a new library application in typescript
$ cdk init lib --language=typescript

Lists the stacks modeled in the CDK app.

$ # List all stacks in the CDK app 'node bin/main.js'
$ cdk list --app='node bin/main.js'
Foo
Bar
Baz

$ # List all stack including all details (add --json to output JSON instead of YAML)
$ cdk list --app='node bin/main.js' --long
-
    name: Foo
    environment:
        name: 000000000000/bermuda-triangle-1
        account: '000000000000'
        region: bermuda-triangle-1
-
    name: Bar
    environment:
        name: 111111111111/bermuda-triangle-2
        account: '111111111111'
        region: bermuda-triangle-2
-
    name: Baz
    environment:
        name: 333333333333/bermuda-triangle-3
        account: '333333333333'
        region: bermuda-triangle-3

Synthesizes the CDK app and produces a cloud assembly to a designated output (defaults to cdk.out)

Typically you don't interact directly with cloud assemblies. They are files that include everything needed to deploy your app to a cloud environment. For example, it includes an AWS CloudFormation template for each stack in your app, and a copy of any file assets or Docker images that you reference in your app.

If your app contains a single stack or a stack is supplied as an argument to cdk synth, the CloudFormation template will also be displayed in the standard output (STDOUT) as YAML.

If there are multiple stacks in your application, cdk synth will synthesize the cloud assembly to cdk.out.

$ # Synthesize cloud assembly for StackName and output the CloudFormation template to STDOUT
$ cdk synth MyStackName

$ # Synthesize cloud assembly for all the stacks and save them into cdk.out/
$ cdk synth

$ # Synthesize cloud assembly for StackName, but don't include dependencies
$ cdk synth MyStackName --exclusively

$ # Synthesize cloud assembly for StackName, but don't cloudFormation template output to STDOUT
$ cdk synth MyStackName --quiet

See the AWS Documentation to learn more about cloud assemblies. See the CDK reference documentation for details on the cloud assembly specification

Computes differences between the infrastructure specified in the current state of the CDK app and the currently deployed application (or a user-specified CloudFormation template). This command returns non-zero if any differences are found.

$ # Diff against the currently deployed stack
$ cdk diff --app='node bin/main.js' MyStackName

$ # Diff against a specific template document
$ cdk diff --app='node bin/main.js' MyStackName --template=path/to/template.yml

Deploys a stack of your CDK app to its environment. During the deployment, the toolkit will output progress indications, similar to what can be observed in the AWS CloudFormation Console. If the environment was never bootstrapped (using cdk bootstrap), only stacks that are not using assets and synthesize to a template that is under 51,200 bytes will successfully deploy.

$ cdk deploy --app='node bin/main.js' MyStackName

Before creating a change set, cdk deploy will compare the template and tags of the currently deployed stack to the template and tags that are about to be deployed and will skip deployment if they are identical. Use --force to override this behavior and always deploy the stack.

If a resource fails to be created or updated, the deployment will roll back before the CLI returns. All changes made up to that point will be undone (resources that were created will be deleted, updates that were made will be changed back) in order to leave the stack in a consistent state at the end of the operation. If you are using the CDK CLI to iterate on a development stack in your personal account, you might not require CloudFormation to leave your stack in a consistent state, but instead would prefer to update your CDK application and try again.

To disable the rollback feature, specify --no-rollback (-R for short):

$ cdk deploy --no-rollback
$ cdk deploy -R

NOTE: you cannot use --no-rollback for any updates that would cause a resource replacement, only for updates and creations of new resources.

You can have multiple stacks in a cdk app. An example can be found in how to create multiple stacks.

In order to deploy them, you can list the stacks you want to deploy. If your application contains pipeline stacks, the cdk list command will show stack names as paths, showing where they are in the pipeline hierarchy (e.g., PipelineStack, PipelineStack/Prod, PipelineStack/Prod/MyService etc).

If you want to deploy all of them, you can use the flag --all or the wildcard * to deploy all stacks in an app. Please note that, if you have a hierarchy of stacks as described above, --all and * will only match the stacks on the top level. If you want to match all the stacks in the hierarchy, use **. You can also combine these patterns. For example, if you want to deploy all stacks in the Prod stage, you can use cdk deploy PipelineStack/Prod/**.

--concurrency N allows deploying multiple stacks in parallel while respecting inter-stack dependencies to speed up deployments. It does not protect against CloudFormation and other AWS account rate limiting.

Pass parameters to your template during deployment by using --parameters (STACK:KEY=VALUE). This will apply the value VALUE to the key KEY for stack STACK.

Example of providing an attribute value for an SNS Topic through a parameter in TypeScript:

Usage of parameter in CDK Stack:

new sns.Topic(this, 'TopicParameter', {
    topicName: new cdk.CfnParameter(this, 'TopicNameParam').value.toString()
});

Parameter values as a part of cdk deploy

$ cdk deploy --parameters "MyStackName:TopicNameParam=parameterized"

Parameter values can be overwritten by supplying the --force flag. Example of overwriting the topic name from a previous deployment.

$ cdk deploy --parameters "ParametersStack:TopicNameParam=blahagain" --force

⚠️ Parameters will be applied to all stacks if a stack name is not specified or * is provided. Parameters provided to Stacks that do not make use of the parameter will not successfully deploy.

⚠️ Parameters do not propagate to NestedStacks. These must be sent with the constructor. See Nested Stack documentation

Write stack outputs from deployments into a file. When your stack finishes deploying, all stack outputs will be written to the output file as JSON.

Usage of output in a CDK stack

const fn = new lambda.Function(this, "fn", {
  handler: "index.handler",
  code: lambda.Code.fromInline(`exports.handler = \${handler.toString()}`),
  runtime: lambda.Runtime.NODEJS_14_X
});

new cdk.CfnOutput(this, 'FunctionArn', {
  value: fn.functionArn,
});

Specify an outputs file to write to by supplying the --outputs-file parameter

$ cdk deploy --outputs-file outputs.json

Alternatively, the outputsFile key can be specified in the project config (cdk.json).

The following shows a sample cdk.json where the outputsFile key is set to outputs.json.

{
  "app": "npx ts-node bin/myproject.ts",
  "context": {
    "@aws-cdk/core:enableStackNameDuplicates": "true",
    "aws-cdk:enableDiffNoFail": "true",
    "@aws-cdk/core:stackRelativeExports": "true"
  },
  "outputsFile": "outputs.json"
}

The outputsFile key can also be specified as a user setting (~/.cdk.json)

When the stack finishes deployment, outputs.json would look like this:

{
  "MyStack": {
    "FunctionArn": "arn:aws:lambda:us-east-1:123456789012:function:MyStack-fn5FF616E3-G632ITHSP5HK"
  }
}

⚠️ The key of the outputs corresponds to the logical ID of the CfnOutput. Read more about identifiers in the CDK here

If multiple stacks are being deployed or the wild card * is used to deploy all stacks, all outputs are written to the same output file where each stack artifact ID is a key in the JSON file

$ cdk deploy '*' --outputs-file "/Users/code/myproject/outputs.json"

Example outputs.json after deployment of multiple stacks

{
  "MyStack": {
    "FunctionArn": "arn:aws:lambda:us-east-1:123456789012:function:MyStack-fn5FF616E3-G632ITHSP5HK"
  },
  "AnotherStack": {
    "VPCId": "vpc-z0mg270fee16693f"
  }
}

By default, stack deployment events are displayed as a progress bar with the events for the resource currently being deployed.

Set the --progress flag to request the complete history which includes all CloudFormation events

$ cdk deploy --progress events

Alternatively, the progress key can be specified in the project config (cdk.json).

The following shows a sample cdk.json where the progress key is set to events. When cdk deploy is executed, deployment events will include the complete history.

{
  "app": "npx ts-node bin/myproject.ts",
  "context": {
    "@aws-cdk/core:enableStackNameDuplicates": "true",
    "aws-cdk:enableDiffNoFail": "true",
    "@aws-cdk/core:stackRelativeExports": "true"
  },
  "progress": "events"
}

The progress key can also be specified as a user setting (~/.cdk.json)

For more control over when stack changes are deployed, the CDK can generate a CloudFormation change set but not execute it. The default name of the generated change set is cdk-deploy-change-set, and a previous change set with that name will be overwritten. The change set will always be created, even if it is empty. A name can also be given to the change set to make it easier to later execute.

$ cdk deploy --no-execute --change-set-name MyChangeSetName

You can pass the --hotswap flag to the deploy command:

$ cdk deploy --hotswap [StackNames]

This will attempt to perform a faster, short-circuit deployment if possible (for example, if you only changed the code of a Lambda function in your CDK app, but nothing else in your CDK code), skipping CloudFormation, and updating the affected resources directly; this includes changes to resources in nested stacks. If the tool detects that the change does not support hotswapping, it will fall back and perform a full CloudFormation deployment, exactly like cdk deploy does without the --hotswap flag.

Passing this option to cdk deploy will make it use your current AWS credentials to perform the API calls - it will not assume the Roles from your bootstrap stack, even if the @aws-cdk/core:newStyleStackSynthesis feature flag is set to true (as those Roles do not have the necessary permissions to update AWS resources directly, without using CloudFormation). For that reason, make sure that your credentials are for the same AWS account that the Stack(s) you are performing the hotswap deployment for belong to, and that you have the necessary IAM permissions to update the resources that are being deployed.

Hotswapping is currently supported for the following changes (additional changes will be supported in the future):

• Code asset (including Docker image and inline code), tag changes, and configuration changes (only description and environment variables are supported) of AWS Lambda functions.
• AWS Lambda Versions and Aliases changes.
• Definition changes of AWS Step Functions State Machines.
• Container asset changes of AWS ECS Services.
• Website asset changes of AWS S3 Bucket Deployments.
• Source and Environment changes of AWS CodeBuild Projects.
• VTL mapping template changes for AppSync Resolvers and Functions

⚠ Note #1: This command deliberately introduces drift in CloudFormation stacks in order to speed up deployments. For this reason, only use it for development purposes. Never use this flag for your production deployments!

⚠ Note #2: This command is considered experimental, and might have breaking changes in the future.

The watch command is similar to deploy, but instead of being a one-shot operation, the command continuously monitors the files of the project, and triggers a deployment whenever it detects any changes:

$ cdk watch DevelopmentStack
Detected change to 'lambda-code/index.js' (type: change). Triggering 'cdk deploy'
DevelopmentStack: deploying...

 ✅  DevelopmentStack

^C

To end a cdk watch session, interrupt the process by pressing Ctrl+C.

What files are observed is determined by the "watch" setting in your cdk.json file. It has two sub-keys, "include" and "exclude", each of which can be either a single string, or an array of strings. Each entry is interpreted as a path relative to the location of the cdk.json file. Globs, both * and **, are allowed to be used. Example:

{
  "app": "mvn -e -q compile exec:java",
  "watch": {
    "include": "src/main/**",
    "exclude": "target/*"
  }
}

The default for "include" is "**/*" (which means all files and directories in the root of the project), and "exclude" is optional (note that we always ignore files and directories starting with ., the CDK output directory, and the node_modules directory), so the minimal settings to enable watch are "watch": {}.

If either your CDK code, or application code, needs a build step before being deployed, watch works with the "build" key in the cdk.json file, for example:

{
  "app": "mvn -e -q exec:java",
  "build": "mvn package",
  "watch": {
    "include": "src/main/**",
    "exclude": "target/*"
  }
}

Note that watch by default uses hotswap deployments (see above for details) -- to turn them off, pass the --no-hotswap option when invoking it.

By default watch will also monitor all CloudWatch Log Groups in your application and stream the log events locally to your terminal. To disable this feature you can pass the --no-logs option when invoking it:

$ cdk watch --no-logs

You can increase the concurrency by which watch will deploy and hotswap your stacks by specifying --concurrency N. --concurrency for watch acts the same as --concurrency for deploy, in that it will deploy or hotswap your stacks while respecting inter-stack dependencies.

$ cdk watch --concurrency 5

Note: This command is considered experimental, and might have breaking changes in the future. The same limitations apply to to watch deployments as do to --hotswap deployments. See the Hotswap deployments for faster development section for more information.

Sometimes you want to import AWS resources that were created using other means into a CDK stack. For some resources (like Roles, Lambda Functions, Event Rules, ...), it's feasible to create new versions in CDK and then delete the old versions. For other resources, this is not possible: stateful resources like S3 Buckets, DynamoDB tables, etc., cannot be easily deleted without impact on the service.

cdk import, which uses CloudFormation resource imports, makes it possible to bring an existing resource under CDK/CloudFormation's management. See the list of resources that can be imported here.

To import an existing resource to a CDK stack, follow the following steps:

1. Run a cdk diff to make sure there are no pending changes to the CDK stack you want to import resources into. The only changes allowed in an "import" operation are the addition of new resources which you want to import.
2. Add constructs for the resources you want to import to your Stack (for example, for an S3 bucket, add something like new s3.Bucket(this, 'ImportedS3Bucket', {});). Do not add any other changes! You must also make sure to exactly model the state that the resource currently has. For the example of the Bucket, be sure to include KMS keys, life cycle policies, and anything else that's relevant about the bucket. If you do not, subsequent update operations may not do what you expect.
3. Run the cdk import - if there are multiple stacks in the CDK app, pass a specific stack name as an argument.
4. The CLI will prompt you to pass in the actual names of the resources you are importing. After you supply it, the import starts.
5. When cdk import reports success, the resource is managed by CDK. Any subsequent changes in the construct configuration will be reflected on the resource.

This feature is currently in preview. Be aware of the following limitations:

• Importing resources in nested stacks is not possible.
• Uses the deploy role credentials (necessary to read the encrypted staging bucket). Requires a new version (version 12) of the bootstrap stack, for the added IAM permissions to the deploy-role.
• There is no check on whether the properties you specify are correct and complete for the imported resource. Try starting a drift detection operation after importing.
• Resources that depend on other resources must all be imported together, or one-by-one in the right order. The CLI will not help you import dependent resources in the right order, the CloudFormation deployment will fail with unresolved references.

Deletes a stack from it's environment. This will cause the resources in the stack to be destroyed (unless they were configured with a DeletionPolicy of Retain). During the stack destruction, the command will output progress information similar to what cdk deploy provides.

$ cdk destroy --app='node bin/main.js' MyStackName

Deploys a CDKToolkit CloudFormation stack into the specified environment(s), that provides an S3 bucket and ECR reposity that cdk deploy will use to store synthesized templates and the related assets, before triggering a CloudFormation stack update. The name of the deployed stack can be configured using the --toolkit-stack-name argument. The S3 Bucket Public Access Block Configuration can be configured using the --public-access-block-configuration argument. ECR uses immutable tags for images.

$ # Deploys to all environments
$ cdk bootstrap --app='node bin/main.js'

$ # Deploys only to environments foo and bar
$ cdk bootstrap --app='node bin/main.js' foo bar

By default, bootstrap stack will be protected from stack termination. This can be disabled using --termination-protection argument.

If you have specific needs, policies, or requirements not met by the default template, you can customize it to fit your own situation, by exporting the default one to a file and either deploying it yourself using CloudFormation directly, or by telling the CLI to use a custom template. That looks as follows:

# Dump the built-in template to a file
$ cdk bootstrap --show-template > bootstrap-template.yaml

# Edit 'bootstrap-template.yaml' to your liking

# Tell CDK to use the customized template
$ cdk bootstrap --template bootstrap-template.yaml

Inspect the current command-line environment and configurations, and collect information that can be useful for troubleshooting problems. It is usually a good idea to include the information provided by this command when submitting a bug report.

$ cdk doctor
ℹ️ CDK Version: 1.0.0 (build e64993a)
ℹ️ AWS environment variables:
  - AWS_EC2_METADATA_DISABLED = 1
  - AWS_SDK_LOAD_CONFIG = 1

CDK Notices are important messages regarding security vulnerabilities, regressions, and usage of unsupported versions. Relevant notices appear on every command by default. For example,

$ cdk deploy

... # Normal output of the command

NOTICES

16603   Toggling off auto_delete_objects for Bucket empties the bucket

        Overview: If a stack is deployed with an S3 bucket with
                  auto_delete_objects=True, and then re-deployed with
                  auto_delete_objects=False, all the objects in the bucket
                  will be deleted.

        Affected versions: <1.126.0.

        More information at: https://github.com/aws/aws-cdk/issues/16603


17061   Error when building EKS cluster with monocdk import

        Overview: When using monocdk/aws-eks to build a stack containing
                  an EKS cluster, error is thrown about missing
                  lambda-layer-node-proxy-agent/layer/package.json.

        Affected versions: >=1.126.0 <=1.130.0.

        More information at: https://github.com/aws/aws-cdk/issues/17061


If you don’t want to see an notice anymore, use "cdk acknowledge ID". For example, "cdk acknowledge 16603".

You can suppress warnings in a variety of ways:

• per individual execution:

  cdk deploy --no-notices

• disable all notices indefinitely through context in cdk.json:

  {
    "notices": false,
    "context": {
      ...
    }
  }

• acknowleding individual notices via cdk acknowledge (see below).

To hide a particular notice that has been addressed or does not apply, call cdk acknowledge with the ID of the notice:

$cdk acknowledge 16603

Please note that the acknowledgements are made project by project. If you acknowledge an notice in one CDK project, it will still appear on other projects when you run any CDK commands, unless you have suppressed or disabled notices.

List the notices that are relevant to the current CDK repository, regardless of context flags or notices that have been acknowledged:

$ cdk notices

NOTICES

16603   Toggling off auto_delete_objects for Bucket empties the bucket

        Overview: if a stack is deployed with an S3 bucket with
                  auto_delete_objects=True, and then re-deployed with
                  auto_delete_objects=False, all the objects in the bucket
                  will be deleted.

        Affected versions: framework: <=2.15.0 >=2.10.0

        More information at: https://github.com/aws/aws-cdk/issues/16603


If you don’t want to see a notice anymore, use "cdk acknowledge <id>". For example, "cdk acknowledge 16603".

By default asset bundling is skipped for cdk list and cdk destroy. For cdk deploy, cdk diff and cdk synthesize the default is to bundle assets for all stacks unless exclusively is specified. In this case, only the listed stacks will have their assets bundled.

If mfa_serial is found in the active profile of the shared ini file AWS CDK will ask for token defined in the mfa_serial. This token will be provided to STS assume role call.

Example profile in ~/.aws/config where mfa_serial is used to assume role:

[profile my_assume_role_profile]
source_profile=my_source_role
role_arn=arn:aws:iam::123456789123:role/role_to_be_assumed
mfa_serial=arn:aws:iam::123456789123:mfa/my_user

If you create an SSO profile with aws configure sso and run aws sso login, the CDK can use those credentials if you set the profile name as the value of AWS_PROFILE or pass it to --profile.

On top of passing configuration through command-line arguments, it is possible to use JSON configuration files. The configuration's order of precedence is:

1. Command-line arguments
2. Project configuration (./cdk.json)
3. User configuration (~/.cdk.json)

Some of the interesting keys that can be used in the JSON configuration files:

{
    "app": "node bin/main.js",        // Command to start the CDK app      (--app='node bin/main.js')
    "build": "mvn package",           // Specify pre-synth build           (--build='mvn package')
    "context": {                      // Context entries                   (--context=key=value)
        "key": "value"
    },
    "toolkitStackName": "foo",        // Customize 'bootstrap' stack name  (--toolkit-stack-name=foo)
    "toolkitBucketName": "fooBucket", // Customize 'bootstrap' bucket name (--toolkit-bucket-name=fooBucket)
    "versionReporting": false,         // Opt-out of version reporting      (--no-version-reporting)
}

If specified, the command in the build key will be executed immediately before synthesis. This can be used to build Lambda Functions, CDK Application code, or other assets. build cannot be specified on the command line or in the User configuration, and must be specified in the Project configuration. The command specified in build will be executed by the "watch" process before deployment.

The following environment variables affect aws-cdk:

• CDK_DISABLE_VERSION_CHECK: If set, disable automatic check for newer versions.
• CDK_NEW_BOOTSTRAP: use the modern bootstrapping stack.

The CLI will attempt to detect whether it is being run in CI by looking for the presence of an environment variable CI=true. This can be forced by passing the --ci flag. By default the CLI sends most of its logs to stderr, but when ci=true it will send the logs to stdout instead.