AWS CDK vs Kusion

AWS CDK is a framework for defining AWS cloud application infrastructure with familiar programming languages and deploying it through AWS CloudFormation. Stable bindings are available for TypeScript, JavaScript, Python, Java, C#/.NET, and Go. The AWS Construct Library supplies reusable constructs for AWS services and features. CDK Pipelines supports CI/CD deployment to testing, integration, and production environments. In Visual Studio Code, the AWS CDK Explorer preview can browse projects, stacks, resources, and policies. The tool supports AWS only and provides managed state, policy as code, drift detection, importing existing resources, and self-hosted execution. CDK itself has no charge, although deployed AWS resources may incur charges. Developers need Node.js 22.x or later regardless of application language. Deployments use configured AWS credentials and IAM roles. AWS recommends IAM security practices, and custom constructs cannot guarantee compliance because general-purpose code can bypass their checks.

Kusion is a free, open-source platform orchestrator for internal developer platforms. Its AppConfiguration lets developers describe application workloads and resource dependencies without setting environment-specific values. The declarative model combines workload, accessory, pipeline, and dependency concepts. The documented process moves through Write, Generate, and Apply, with a three-way diff preview before changes are applied. Kusion currently applies operations to Kubernetes, Terraform, and on-premises infrastructure. Platform engineers can create reusable modules to standardize deployments and support developer self-service. A built-in delivery pipeline can be extended with customized plugin steps managed by platform owners. The project describes security and compliance support through base models, policy-as-code guardrails, and automatic injection of accessory secrets into workloads. External secret-store guides cover AWS, Alibaba Cloud, and Viettel Cloud. Kusion is a lightweight client-side solution with APIs for integration and automation. Its CLI installation methods cover macOS, Linux, and Windows. The project calls itself early and rapidly iterated; help and feedback are available through Slack or GitHub issues.