Choose a Task

The task decides what the assessment can tell you. Pick it for the signal you want, not for the technology on the CV.

Categories

Backend — CRUD APIs, reservation services, payment flows, authentication, permissions, inventory, queue processing, rate limiting, data modelling.

Frontend — dashboards, form workflows, data fetching, state management, accessibility, performance, component architecture.

Full stack — a small feature end to end, API and frontend integration, authentication flows, search, admin interfaces.

Engineering judgement — tasks built around a specific trap: a race condition, a security flaw, weak tests, poor error handling, an over-engineered abstraction, or an incorrect AI suggestion waiting to be accepted.

Greenfield or existing codebase

Most real engineering is modifying systems that already exist, and a large share of the library reflects that. An existing-codebase task shows you whether a candidate reads before they write.

Greenfield tasks are cleaner to compare across candidates. Existing-codebase tasks are closer to the job.

Hidden engineering concerns

Library tasks give candidates business requirements, not a checklist of best practices. A task might say only "implement the reservation endpoint" while the rubric expects the candidate to think about race conditions, transactions, overselling, idempotency, validation, authorisation, and failure handling.

That gap is the assessment. Do not add the checklist to the brief — it removes the thing you are measuring.

Match difficulty to seniority

A junior task with a senior rubric produces a flat, unhelpful distribution. Each library task lists the seniority it targets and what it is designed to surface. Trust that before you trust the title.