Coding sandbox and assessments
Real code, real tests, in the same tab.
Candidates write and run code against your test suite. You see the diff, the runtime, and what they did between keystrokes — not just whether it went green.
Coding sandbox and assessments
Real code, real tests, in the same tab.
Candidates write and run code against your test suite. You see the diff, the runtime, and what they did between keystrokes.
Real runtimes
Python, JavaScript, Java, Go, C and C++, with the standard library available.
Question bank or your own
Vetted tasks per role and seniority, or paste in the problem your team actually cares about.
Anti-cheat that shows its work
Paste bursts, tab blur, keystroke rhythm and full replay, flagged with timestamps rather than a verdict.
Non-engineering assessments
Spreadsheet, writing, support-reply and sales-call tasks, scored on the same scale.
How it works
Set the task, read the evidence.
Step 01
Choose or paste a task
Vetted tasks per role and seniority, or your own problem with your own tests. Set the runtime, the time limit and whether assistants are allowed.
Step 02
They code in the browser
A real sandbox with the standard library available. They run the tests as often as they like and see exactly what you will see.
Step 03
You read the replay
Tests passed, runtime, the diff, and a replay of how the solution was arrived at. Flags for paste bursts and tab blur sit alongside it, timestamped.
What you get
The process, not just the result.
Real runtimes
Python, JavaScript, Java, Go, C and C++, with the standard library available.
Your tests, your bar
Hidden tests, visible tests, or both. Candidates see the ones you want them to see, and the report shows every case with its result.
Anti-cheat that shows its work
Paste bursts, tab blur, keystroke rhythm and a full replay — flagged with timestamps rather than reduced to a verdict you cannot inspect.
Non-engineering assessments
Spreadsheet, writing, support-reply and sales-call tasks, scored on the same scale so the shortlist is comparable across functions.
Feeds the same report
The coding result is a section of the candidate report, next to the AI interview and the panel scorecards, weighted by your rubric.
Accessible by default
Keyboard-only navigation, screen-reader labels on the editor chrome, adjustable time for candidates who request it.
Questions about coding assessments
- Can candidates use an AI assistant?
- You decide per task. Where it is allowed, that is recorded on the report. Where it is not, paste bursts and keystroke rhythm are flagged with timestamps and the full replay is there for you to judge — we flag, we do not accuse.
- Do you use the same puzzle for everyone?
- Only if you want to. The bank has vetted tasks per role and seniority, and you can paste in the problem your team actually cares about — a bug from your own tracker usually beats a puzzle.
- What about roles that do not write code?
- Spreadsheet, writing, support-reply and sales-call tasks run in the same flow and score on the same scale, so a sales hire and an engineer are comparable on the same report.
- How long should a task be?
- Thirty to forty-five minutes is where completion rates hold up. The sandbox shows candidates the remaining time and saves continuously, so a closed tab does not lose the work.
Keep reading
Get started
Stop guessing from a CV whether they can build it.
Set one task, read the replay, and know before the first call.
No card. 1 job free forever. Cancel by closing the tab.