DEV Community

Cover image for Rapidly prototype your APIs on Kubernetes with Kusk Gateway
Abdallah Abedraba for Kubeshop

Posted on • Originally published at

Rapidly prototype your APIs on Kubernetes with Kusk Gateway

OpenAPI (f.k.a Swagger) has introduced a set of standardized specifications for REST APIs that, among many things, allows producers and consumers of APIs to work together in designing an API before even writing a single line of code!

This design-first approach has improved the experience of API developers by giving them the opportunity to use tools like OpenAPI generator which takes an OpenAPI definition and generates scaffolding code for backenders, making the development of APIs much faster. 

But consumers of APIs still didn’t have a way of really taking advantage of this design-first approach as they still depended on the backend teams to implement and deploy APIs.

How could consumers of APIs then take advantage of OpenAPI and the design-first approach? By making the API gateway aware of the OpenAPI definition. Let me explain how!

An API gateway sits in front of your microservices routing incoming traffic that your Kubernetes cluster receives to the corresponding microservice. By having the API gateway be aware of your OpenAPI definition it can then see what a response would look like and generate a mock (or fake) response for the consumers instead.

This approach would allow API consumers to start using the API with mocked results and meanwhile (in parallel!) the backend team can start implementing the actual API; it essentially loses the dependency between consumer and producer teams and allows for quicker prototyping of the API. With no code whatsoever, just your run-of-the-mill OpenAPI definition. 

In this tutorial, you will learn how to use Kusk Gateway, an open-source OpenAPI-driven API gateway, to enable this rapid prototyping development! Let’s make a simple example of how to achieve this:

Schema of using OpenAPI and Kusk Gateway


1. Install Kusk Gateway CLI

brew install kubeshop/kusk/kusk
Enter fullscreen mode Exit fullscreen mode

For other installation methods, check the installation docs.

2. Install Kusk Gateway in your Kubernetes cluster

kusk install
Enter fullscreen mode Exit fullscreen mode

3. Create an OpenAPI definitionCreate the file openapi.yaml with the content:

This is a simple common OpenAPI definition for a single /hello route that accepts a GET method. There are a couple of important details here: 

1. The x-kusk extension 

This section is an extension to the OpenAPI specification. We use it to configure Kusk Gateway to enable in this case CORS access and mocking of the API. This approach makes OpenAPI the definition of your API and your gateway configuration(!) essentially having a single source of truth. 

2. The example field

To allow Kusk Gateway to generate mock responses, you'll need to define examples of your responses under OpenAPI's example field. In this case we have an example message "Hello from a mocked response!". 

4. Apply the API configuration to the cluster

kusk api generate -i openapi.yaml kusk-gateway-envoy-fleet | kubectl apply -f -
Enter fullscreen mode Exit fullscreen mode

This applies the configuration to the gateway and deploys the API. 

5. Test your mocked API

Now you are ready to test the API and build against mocked responses!

First, you need to get the External IP of the gateway, i.e. the entry point of your APIs. For that run:


kubectl get -n kusk-system svc/kusk-gateway-envoy-fleet
Enter fullscreen mode Exit fullscreen mode

Note: if you're running a local cluster with Minikube, the ExternalIP might be in pending state; in case it is, you can port-forward to the gateway by running kubectl port-forward -n kusk-system svc/kusk-gateway-envoy-fleet 8080:80, so your external IP becomes localhost:8080

And now test the /hello path:


curl $EXTERNAL_IP/hello{"message": "Hello from a mocked response!"}
Enter fullscreen mode Exit fullscreen mode

And that's it! You can see that this is the same response that was defined in our OpenAPI definition. Now your API consumer teams can already start working with the mocked API!

Once the API team has implemented the service, you will stop mocking your endpoint and connect it to a microservice in your cluster. That's what we'll cover in the next sections.

6. Deploy your application to the cluster

Deploy this example application that we will connect with the API gateway:


kubectl apply -f
Enter fullscreen mode Exit fullscreen mode

7. Connect your application to Kusk Gateway

Now, you will need to disable the mocking for your API and connect the upstream. Update your openapi.yaml file with the following changes:

8. Test your implemented application

With the same External IP from step 5, run

curl $EXTERNAL_IP/hello{"message": "Hello from an implemented service!"}
Enter fullscreen mode Exit fullscreen mode

Voilá! You've just implemented and connected a service to your gateway!

You can find the code for the article on Github.


Kusk Gateway enabled you to take advantage of OpenAPI and a design-first approach to allow your teams to work in parallel once they've agreed on a specific API design. Nothing should have be stopping them anyway!

Having an OpenAPI-aware gateway allows you to reap more advantages of the OpenAPI lifecycle, like for example automatic validation of your APIs and more!

Head over to the Kusk Gateway documentation to try it out and explore all the benefits of an OpenAPI-driven gateway! We would love to hear your thoughts and ideas in our Discord server or just drop by if you to say hi :) 

Thank you!

Top comments (0)