Build an Assessment

An assessment is the thing you send. It holds a task, a time limit, a rubric, and its settings. Each candidate who takes it produces an attempt — one timed session, recorded end to end.

One assessment, many attempts. That is what makes candidates comparable: everyone faced the same task under the same constraints.

What you configure

Task. One task per assessment. From the library, or your own.

Time limit. See Time Limits and the Timer.

Rubric and weighting. Every task ships with a rubric. You can reweight it toward what matters for the role — see Rubrics and Score Weighting.

Runtime profile. Set by the task. It determines the language, the tooling, and whether a database is available. See Runtime Profiles.

Name it for the role

Assessments are reusable. Name them so that a year from now the shortlist still makes sense: role, seniority, and focus. "Backend / Mid / Data modelling" is a name you can act on.

Before you send it widely

Run it yourself, or have an engineer on your team run it. You are checking three things:

• Is the task actually doable in the time limit?
• Does it surface the thing you care about for this role?
• Is anything in the brief ambiguous in a way you did not intend?

Ambiguity is sometimes the point — real requirements are ambiguous. Just make sure it is deliberate.