For managers hiring without a recruiter

Read the code they wrote.

Replace the take-home with a short Challenge you can read. Send one link. The candidate solves a small JavaScript Challenge in the browser, with no account, in 20 to 35 minutes. You get a Result page with the submitted code, which tests passed, and how the code arrived.

Saved ResultSample by Portfolio Pro6 of 6 required tests passed6 graded runs in 5 min. First run failed.

Send a hiring link Try a challenge, no signup

Looking for work? Solve one in the browser, then share the Result page as proof of what you wrote.

Open a real sample ResultWritten by Portfolio Pro to show the format, not by a candidate.

Free during early access. No checkout. JavaScript Challenges only.

[ Saved Result ]

Sample by Portfolio Pro

Run tasks with a concurrency limit

Required tests: 6 of 6
Graded runs: 6
First run: failed
First run to save: 5 min

6 graded runs, left to right

19 try {

20 // A task may return a plain value or throw before it returns a promise.

21 running = Promise.resolve(tasks[index]());

22 } catch (err) {

23 fail(err);

24 return;

25 }

Line 21: why wrap this call in Promise.resolve? The Result gives you questions like this one to ask in the interview.

Open the full Result

How it works

One link. One Challenge. A Result you can read.

No vendor score to decode. You get the code and the test results, so you know what to ask in the interview.

01
Pick a Challenge
Choose a task from the library that matches the role.
02
Send one link
The candidate solves it in the browser. Their work runs against a test suite in a sandbox.
03
Review the result
When you send a hiring link, we email you when they save. The Result shows the submitted code, what passed, and how the code arrived.

36 seconds: a candidate solves the debounce Challenge and saves a Result.

The signal

How the code arrived

The server counts every graded run and records how much code each run added. A solution built over many runs looks different from one that appears whole and passes on the first run. Neither shape proves who typed it, so use it as a question in the interview.

Run failed a required testRun passedCode removed

[ Saved Result ]

Sample by Portfolio Pro

Run tasks with a concurrency limit

Bytes added per graded run, left to right

19 try {

20 // A task may return a plain value or throw before it returns a promise.

21 running = Promise.resolve(tasks[index]());

22 } catch (err) {

23 fail(err);

24 return;

25 }

Line 21: why wrap this call in Promise.resolve? The Result gives you questions like this one to ask in the interview.

Required tests: 6 of 6
Graded runs: 6
First run: failed
First run to save: 5 min

Open the full Result

Three real Results

Three saved Results, three shapes.

6 graded runs, left to right

Built up over six runs

Run tasks with a concurrency limit

At save: 6 of 6 required tests passed
First graded run: Failed
Largest single jump: 475 bytes, on run 2
First run to save: 5 min

Open this Result

one graded run

Arrived in one run

Implement debounce

At save: 5 of 5 required tests passed
First graded run: Passed everything
Largest single jump: 593 bytes, the whole solution
First run to save: 13 s

Open this Result

3 graded runs, left to right

Three runs, one test still failing

Implement an LRU cache

At save: 5 of 6 required tests passed
First graded run: Failed
Largest single jump: 197 bytes, on run 2
First run to save: 3 min

Open this Result

Code added in a run that failed a required testCode added in a run that passed

Portfolio Pro wrote these three Results on this site to show the format. They are not candidates. The bars are the runs the server recorded, and each Result page shows the same list beside the code. No shape proves who typed the code. It tells you what to ask about in the interview.

Why this beats a resume

Judge the work yourself.

On a resume

In the margin: what a Result shows

A story they (or an LLM) wrote.
The code they submitted, the tests it passed, and how long the tests took.
You guess from a job description.
You read their actual code and see which tests passed.
Claims you cannot check from a PDF.
Iterations, distinct graded versions, and the largest jump in code between runs, counted by the server.

The library

Start with a Challenge that matches the role.

Ten Challenges, each with its own test suite. Nothing here is a scored submission. Open one to read the prompt and the tests.

Send a link

Get a hiring link

Enter your work email and pick a Challenge. We email you a confirmation link that works for 2 hours. Open it to get the candidate URL, then send that URL to the candidate. The candidate URL works for 14 days.

Hiring every month?

Hiring links are free during early access. If you plan to send more than a few, tell us your monthly budget for candidate Challenges. It decides what we charge. You do not need this to send a link today.

Straight answers

What a Result shows, and what it does not

A Result shows the submitted code, each test, and counts the server measured. A verified inbox is not a government ID and not proof of who typed.

Can a candidate use ChatGPT?: Yes. Nothing stops it. The Result shows how the code arrived: Iterations, distinct graded versions, the time from first run to save, whether the first run already passed, and the largest single addition of code. Code that appears whole and passes on the first run reads differently from code built over twenty runs. Neither shows who typed it.
Is hiring paid today?: No. Hiring links are free during early access. An Early Access Request asks for a monthly budget so we know what a seat is worth to you. There is no checkout.
How long does it take a candidate?: Each Challenge shows a suggested time, 20 to 35 minutes. The time is not enforced. The Result shows the real span from first run to save.
Who can see a Result?: Anyone with its URL. It shows the code, the tests, and the counts. A Result can name the inboxes the server checked. It does not verify a legal identity.

Portfolio Pro

Loading page…

Read the code they wrote.

Saved ResultSample by Portfolio Pro6 of 6 required tests passed6 graded runs in 5 min. First run failed.

Send a hiring link Try a challenge, no signup

Looking for work? Solve one in the browser, then share the Result page as proof of what you wrote.

Open a real sample ResultWritten by Portfolio Pro to show the format, not by a candidate.

Free during early access. No checkout. JavaScript Challenges only.

One link. One Challenge. A Result you can read.

No vendor score to decode. You get the code and the test results, so you know what to ask in the interview.

Pick a Challenge

Choose a task from the library that matches the role.

Send one link

The candidate solves it in the browser. Their work runs against a test suite in a sandbox.

Review the result

When you send a hiring link, we email you when they save. The Result shows the submitted code, what passed, and how the code arrived.

How the code arrived

Run failed a required testRun passedCode removed

Three saved Results, three shapes.

6 graded runs, left to right

Built up over six runs

Run tasks with a concurrency limit

At save: 6 of 6 required tests passed
First graded run: Failed
Largest single jump: 475 bytes, on run 2
First run to save: 5 min

Open this Result

one graded run

Arrived in one run

Implement debounce

At save: 5 of 5 required tests passed
First graded run: Passed everything
Largest single jump: 593 bytes, the whole solution
First run to save: 13 s

Open this Result

3 graded runs, left to right

Three runs, one test still failing

Implement an LRU cache

At save: 5 of 6 required tests passed
First graded run: Failed
Largest single jump: 197 bytes, on run 2
First run to save: 3 min

Open this Result

Code added in a run that failed a required testCode added in a run that passed

Judge the work yourself.

On a resume

In the margin: what a Result shows

A story they (or an LLM) wrote.
The code they submitted, the tests it passed, and how long the tests took.
You guess from a job description.
You read their actual code and see which tests passed.
Claims you cannot check from a PDF.
Iterations, distinct graded versions, and the largest jump in code between runs, counted by the server.