Skip to main content

Wing SDK

This topic includes contribution guidelines related to the Wing SDK, the package containing the language's standard library.

πŸ”¨ How do I build just the SDK?​

The SDK resides in libs/wingsdk. It's written in TypeScript, and is published to npm.

The SDK is built using a couple of extra libraries and tools:

  • CDK for Terraform ("cdktf") is a framework for defining Terraform infrastructure. The SDK uses it to generate the Terraform files that users deploy.
  • JSII is a tool we used to compile the SDK. JSII is a wrapper over TypeScript that makes it possible to use the SDK in other languages like Python, Java, C#, and Go. This is made possible through extra type checks. In practice, the main difference from ordinary TypeScript is that you cannot use advanced TypeScript types like Partial or generics in public APIs.
  • Projen is a tool used to manage project configuration files. It uses the .projenrc.ts file to generate package.json and other files. You can modify it and run pnpm projen to regenerate the resources. If you are not touching configuration files, you can totally ignore this.

Everything in the SDK can be built by running pnpm build from libs/wingsdk. You can also run pnpm test to just run tests.

In order to work on the source code, you will need to the build at least once so that TypeScript bindings for Terraform resources will be automatically generated. These files are not checked in because they are quite large.

(If you have any issues building the package, please open an issue and let us know!)

πŸ—ΊοΈ How is the SDK organized?​

Check out this document for a summary of what's contained in each module of the SDK's source code.

🧱 How do I add a dependency to the SDK?​

If you need to add a new npm dependency to the SDK, you can do so by editing the .projenrc.ts file and running pnpm projen after making the edits. If the dependency is a JSII library[2], you should add it to the list named peerDeps - otherwise, you should add it to bundledDeps.

Here is an example of adding a package named "fast-json-stringify" pinned to major version 5 through the projenrc file:

--- a/libs/wingsdk/.projenrc.ts
+++ b/libs/wingsdk/.projenrc.ts
@@ -17,6 +17,7 @@ const project = new cdk.JsiiProject({
bundledDeps: [
+ "fast-json-stringify@^5",

[2] JSII libraries are npm packages that are compiled with JSII. They are usually published to npm with the cdk keyword, and they will have a .jsii file at their root.

🧱 How do I add a new Terraform provider for use in the SDK?​

The SDK uses CDK for Terraform to generate Terraform files. This means that you can generate bindings for any Terraform provider and reference in in the SDK using TypeScript.

To add a new provider, go to libs/wingsdk/.projenrc.ts and edit the section that says "CDKTF_BINDINGS" to add the new provider. Then, run pnpm projen to update the project. One that has finished, you can run pnpm build and the new bindings should be generated inside libs/wingsdk/src/.gen.

🧩 How do I add a resource to the SDK?​

A resource in the SDK has several parts:

  • A preflight polycon API that is shared across all cloud targets. Resource polycons are defined in src/cloud. For example, src/cloud/bucket.ts.
  • An interface representing the inflight API common across all cloud targets. By convention, if the resource is named like Gizmo, the inflight interface should be named IGizmoClient. This is usually in the same file as the preflight API.
  • A simulator implementation in src/sim. This includes:
    • A schema with information to simulate the resource and display the resource in the Wing console. Currently these are in src/sim/schema-resources.ts.
    • A class that implements the polycon API and can produce the resource's simulation schema. For example, src/sim/bucket.ts.
    • An class that implements the inflight API and can simulate the resource. For example, src/sim/bucket.sim.ts.
    • Unit tests for the simulator implementation. For example, test/sim/bucket.test.ts.
  • An implementation for each target cloud (currently just AWS). This includes:

If you are implementing a new resource, or implementing an existing resource for a new cloud provider, try to take a look at code for existing resources (Bucket, Function, Queue) to see how to structure your code.

Feel free to create an issue if you have questions about how to implement a resource or want to discuss the design of a resource. You can also join us on our Wing Slack to ask questions (or just say hi)!

🏁 How do I add and run tests to the SDK?​

The SDK uses jest for running unit tests - all of the tests are inside of the libs/wingsdk/test directory, organized by the file they test.

All features and bug fixes should have tests! They're easy to forget, but they pay off by helping prevent breakages in the future.

All tests can be run by running the following command from libs/wingsdk:

pnpm test

During development, you might find it useful to watch for changes and automatically re-run the tests:

pnpm test:watch

To re-run individual tests, you can directly use the vitest command -- for example:

pnpm vitest run test/target-tf-aws/bucket.test.ts

What is the architecture of the Wing SDK?​

SDK Architecture​

The Wing SDK is the standard library for the Wing language. The core of the SDK are its APIs for creating cloud resources.

By using the SDK to specify the desired state of cloud application including resources like API gateways, queues, storage buckets, and so on, the SDK can synthesize an artifact for deploying that application to the cloud. Today the SDK supports either synthesizing a collection of Terraform configuration, or synthesizing a simulator file that the SDK's Simulator API can use to simulate the app within Node.js.


The SDK resources are written using the Constructs Programming Model. constructs are building blocks that can be composed together to represent more complex desired state, most commonly for cloud applications. constructs serves as the low-level foundation of several other infrastructure-as-code frameworks, such as the AWS CDK, cdk8s, and cdktf.

Conceptually, constructs are ordinary classes that additionally have a unique scope (parent construct) and id. By adding constructs as children of other constructs, they can form in-memory trees, where each construct is uniquely addressible based on its location within the tree.

A construct's path is obtained by joining the sequence of construct ids from the tree root to the construct, with the "/" character. For example, if a construct with no parent is declared the root with an id "root", and it has a child named "Child1", the child has a path of "root/Child1". Constructs with the same parent are required to have different ids.

Each construct also has an address, which is just a hexadecimal hash of the construct's path. This value is useful for generating application identifiers in situations where the path cannot be used because of character / punctuation limitations.

When a tree of constructs all implement a method like toTerraform(), then it is possible to traverse the construct tree and aggregate the result of calling the method on each construct in order to synthesize a result or artifact.


In order to model resources that are implemented differently for each cloud provider, the SDK also uses polycons, a dependency injection framework designed to work with constructs.

Using polycons, the SDK resources are structured as follows:

  • Each resource has a polycon class defined in the cloud namespace with the API that is shared across all cloud implementations (e.g. cloud.Bucket). In order to work with polycons, any shared properties and methods expected to exist on all classes must be defined on a base clase like cloud.BucketBase, and then we have cloud.Bucket extending cloud.BucketBase. Each polycon also has a unique polycon type name needed for polycons to perform dependency injection on them.
  • Each cloud target can implement a polycon by defining a class that extends the polycon base class (e.g. tfaws.Bucket extends cloud.BucketBase).
  • Each cloud target defines a polycon factory that defines the concrete mapping from polycon type names to polycon implementations.
  • Each cloud target has a unique App construct that specifies logic for synthesizing a one or more types of constructs. It also registers the cloud target's polycon factory to that node on the construct tree.

Through polycons, when a user writes new cloud.Bucket() within the scope of an AWS App, the constructor of cloud.Bucket will automatically look up the polycon factory associated with the construct tree, and call the factory's resolve method to produce the class instance specific to that cloud target (new tfaws.Bucket()), and return that back to the caller.

Each App class has an automatically registered polycon factory, but it's possible to pass a custom factory in new App(...) that builds on top of (or overrides) the original factory to support more polycons, or different resolution behavior.


Inflights are Wing's distributed computing primitive. They are isolated code blocks which can be packaged and executed on compute platforms in the cloud (such as containers, CI/CD pipelines or FaaS).

When a resource wants to use an inflight in an API, it is represented in the SDK through a resource with a single inflight method named handle.

Currently, the SDK provides a utility class named Inflight that can be used to quickly create an in-memory resource that implements the handle method.

For example, given the following Wing code:

let queue = new cloud.Queue();
let counter = new cloud.Counter();
new cloud.Function(inflight (event: str) => {;

... the Wing compiler will transform it into JavaScript like this (this is not the exact code generated, but it's close enough):

const queue = new, "Queue");
const counter = new, "Counter");
const handler = new sdk.core.Inflight(this, "Inflight", {
code: sdk.core.NodeJsCode.fromInline(
`async handle(event) {
await this.myQueue.push(event);
bindings: {
messageCount: {
resource: counter,
ops: ["inc"],
myQueue: {
resource: queue,
ops: ["push"],
new, "Function", handler);

Every resource added to the bindings field is implicitly added as a dependency of the inflight, and is made available to the inflight code through a field with the same name. (Hence the API calls to and this.myQueue.push passed in the code field above.)

The bindings field requires a resource field with a reference to the original resource object, and an ops field that specifies the operations that the inflight code will use on the resource.

Under the hood, two main functions are performed by the SDK with this information:

1. Resource binding​

First, each referenced resource is "bound" to the inflight host through a _bind method that is defined on all resources. In the example above, cloud.Function is the host, and cloud.Queue is a referenced resource. The function would call queue._bind(this, ["push"]) to bind the queue to the function host, providing information about the methods it expects to use on the queue.

The referenced resource can then perform any necessary setup to allow the host to referenced it during runtime, such as setting up least privilege permissions. For example, when compiling to AWS, the queue can create an IAM role that allows the function to execute sqs:SendMessage API calls on the queue.

In more complex resources, the _bind method will automatically call _bind on any sub-resources based on the operations that the host expects to use. For example, suppose the user defines a TaskList resource with a method named addTask, and addTask calls put on a cloud.Bucket (a child resource). Then whenever TaskList is bound to a host and addTask is one of the operations the inflight code expects to call, then the cloud.Bucket will automatically be bound to the host as well.

2. Inflight code bundling​

Second, the SDK will bundle the inflight code needed for each of the resources that it references. For example, to call push on a queue in a piece of inflight code, the SDK needs to bundle the inflight code of cloud.Queue with the user's code. The inflight host accomplishes this by calling the _toInflight method on each referenced resource, which returns a stringified JavaScript class with the code for performing the inflight operations and API calls. These classes are then bundled together with esbuild, and can be used by the inflight host as necessary for deploying to the target cloud.

In the case of AWS for example, a cloud.Function will create an AWS Lambda function referencing a Terraform asset with a zipfile of the inflight code.

For the simulator, the inflight code is bundled into the .wsim file produced by the compiler.