Documenting HTTP APIs with OpenAPI standard

What is OpenAPI?

OpenAPI standard is the most broadly adopted industry standard for describing new, and old, APIs. Through it, you can describe which are your routes, its HTTP methods, headers, bodies, responses, information to access and more. It is worth to say the OpenAPI standards cannot describe any kind of API, it has its limitations.

Advances of OpenAPI

According to the getting started of OpenAPI Initiative, using the standard to describe an API empowers you to do the following topics:

OpenAPI best practices

There is a sequence of recommendations given by OpenAPI initiative to build better specification documents. They are: use a Design-First Approach, Keep a Single Source of Truth, **Add OpenAPI Documents to Source Control, Make the OpenAPI Documents Available to the Users, There is Seldom Need to Write OpenAPI Documents by Hand and Working with Big Documents. Check each one following the link.

OpenAPI Tools

There are countless great tools to work with this standard, what you can check here, but I’ll show to you just those what have a better fit with my working flow. Those tools are:

Redocly CLI

This tool helps you to lint files, join, bundle and split them. I’ve chosen it because it allowed me to don’t repeat myself, splitting the specification into many files with parameters, responses, models and so on. You can also join two different versions of a specification to enable the release of private and public versions of the API, for example, and preview your specification through the browser.

The Redocly has a great extension what provides a language server to help you to work with the OpenAPI kind of files. If you use vim/neovim with coc.nvim, I’ve built a simple extension to enable the Redocly extension to work with it.

Check the tool here:

Prism CLI

The Prism CLI provides a way for you to create a mock server, proxy to the real API, following the definition and patterns set on the specification file, and linting, as the Redocly CLI. The mock process is important, for example, when you don’t have the real API implemented and need to test it. Proxy process helps you to guarantee the real API is implemented as the specified.

Prism also have a client what could be used to mock and proxy the API too, allowing to create integration and contract tests, everything from the OpenAPI specification file. I, personally, didn’t do that yet, but it is the next step to implement. I believe I’ll document my experience with it, too.

Check the tool here:

OpenAPI Generator

You’ve learned about the standard, read about the best practices, used tools to build your API documentation, and now you can to generate a server or client; that is the why of OpenAPI Generator. This tool gets as entry the bundled OpenAPI file and generate the client or the server for the language/framework you defined.

It’ll provide to you a fully functional, depending on what generator you’ve chosen, a client layer, to send requests to the API sever, or a server layer to receive those requests. The tool isn’t perfect, some generators doesn’t have some features, what makes it important to select the generator carefully.

Postman and Insomnia

When your specification is ready, you can use it with other tools to test your API. In addition to what I said about Prism client, mock and proxy from a OpenAPI file, you can load that specification file to a tool like Postman or insomnia to enable you to test any requests manually, when necessary.

Check the tool here:


It isn’t possible to talk about OpenAPI without mention Swagger and its tools. As its websites says: swagger is a set of open-source tools built around the OpenAPI Specification that can help you to design, build, document and consume REST APIs.

Swagger contains three greats tools to work with the specification: Swagger UI, Swagger Editor and Swagger Codegen. The Swagger UI renders OpenAPI specs as interactive API documentation, Swagger Editor is a browser-based editor where you can write OpenAPI specs and Swagger Codegen generates server stubs and client libraries from an OpenAPI spec like the OpenAPI generator.


I believe it can empower you to improve your API service in many points. It helps you to check the API’s viability before code anything, saving time, efforts and money, document when the API already exist, expose to public your routes and test to guarantee the API’s stability throughout changes and much more.

That has been my experience using the OpenAPI standard; I’m still on the beginning of the journey, and I have a lot to learn and to do yet. However, I think that experience could help someone to speed up its process to build OpenAPI specification and understand how powerful it is.

Thank you for reading. If you have any feedback; I’d appreciate so much, as one of the goals from that post is to document what I’ve learned and help someone else over the process.