Testing APIs without a cloud account: a local-first workflow

If you’ve used Postman lately, you’ve probably noticed the pattern: sign in to sync, workspaces in the cloud, your collections living on someone else’s servers. For a lot of day-to-day API work — especially in QA and backend development — that’s more machinery than the job needs, and for some teams it’s a compliance problem.

Here’s the workflow I’ve landed on: collections as plain JSON files on disk, versioned with git.

Why local-first?

• Your requests are code. If collections are plain, readable JSON files, they belong in the same repo as the API they test. Branch, diff, review, merge — the tools you already use.
• No account, no sync, no lock-in. Nothing leaves your machine. If a tool disappears tomorrow, your files are still there and still readable.
• QA and dev share one source of truth. The collection travels with the codebase, so “works on my machine” applies to test suites too.

What the workflow looks like

1. A collection is just a folder. Each request is a small *.rr.json file; subfolders are folders. git init and you’re done.
2. Environments (dev/staging/prod) are JSON files too, with {{variable}} interpolation everywhere.
3. Pre/post-request JavaScript handles auth chaining and assertions:

// Post Response
rr.test('status is 200', () => rr.expect(rr.response.status).toBe(200));
rr.variables.set('token', rr.response.body.token); // flows to the next request

Enter fullscreen mode Exit fullscreen mode

1. For data-driven testing, a runner takes a CSV/JSON file and repeats the sequence once per row — each column becomes a {{variable}}.

Tools

I build RestRuno (Windows/macOS/Linux), which is designed around exactly this workflow — plus Postman/Bruno/cURL import so you don’t start from zero. Bruno shares the same local-first philosophy with a different file format. Whichever you pick, the point is the same: your API tests deserve version control, not a cloud account.

Disclosure: I’m the author of RestRuno. The core app is free; a couple of power features (AI assistant, data-driven runner) are a one-time purchase.