PerfLoad / curl to k6

curl to k6: Run an Existing curl Request as a k6 Load Test

Converting a curl command to k6 by hand means opening a new file, importing modules, translating headers and a body into JavaScript, and defining options. PerfLoad does that translation for you: it takes the curl request, generates the k6 script, runs it on a k6-backed runner, and hands back the results.

The runner is a Docker image with k6 already bundled — there is nothing to install besides Docker.

How it works

1. Existing curl request
2. PerfLoad
3. k6 runner
4. Performance results

When you start a run, the runner writes a k6 script for your request into that run's directory, starts k6 run against it, and collects k6's summary when the test ends. PerfLoad links out to k6's own live dashboard while the test is in progress, so you see request rate, duration, and VUs from k6 itself. Because each run keeps its generated script alongside its summary, you can see exactly what was executed.

What the translation looks like

Take this request:

curl -X POST 'https://api.example.com/orders' \
  -H 'Content-Type: application/json' \
  -d '{"productId": "101", "qty": 1}'

Written by hand as a k6 script, a minimal equivalent looks like this. It is illustrative — not a verbatim copy of what PerfLoad generates — but it shows the scaffolding you would otherwise write and maintain yourself:

import http from 'k6/http';
import { check, sleep } from 'k6';

export const options = { vus: 10, duration: '30s' };

export default function () {
  const res = http.post(
    'https://api.example.com/orders',
    JSON.stringify({ productId: '101', qty: 1 }),
    { headers: { 'Content-Type': 'application/json' } }
  );
  check(res, { 'status is 2xx': (r) => r.status >= 200 && r.status < 300 });
  sleep(1);
}

That is not much code for one request. The cost shows up around it: getting the quoting right when you copy headers from a terminal, adding variable data, exporting results somewhere you can read them, and repeating all of it for the next endpoint.

What PerfLoad handles for you

What this does and doesn't replace

PerfLoad covers the common case — a request, a load profile, and a readable result. It is not a replacement for everything k6 offers. Custom scenarios, multi-step flows with your own logic, extensions, and threshold-based pass/fail rules written in a script are all things you'd write in k6 directly. Treat PerfLoad as the fast path for getting a load test running from a request you already have, and see the k6 comparison for where each approach fits.

Run Single first, then Run Test

A useful habit when you're converting a request: send it once with Run Single, which shows the full response body, and only then start Run Test. Variables behave differently between the two — Run Single resolves them once using the first value, while Run Test resolves them per virtual user and per iteration — so a request that looks right in the single run can still surprise you once the values start rotating. A short first load run catches that cheaply.

FAQ

Related guides