Assessments

An assessment is a graded test built in the assessment builder. Participants take it through a link; every attempt is graded at submission and lands in the workspace as a submission you can review, grade, and analyze.

Layout is up to you: everything on a single page, multi-step with several questions per page, or strictly one question at a time.

Most types grade automatically; essay and code writing always go to manual review. Types marked with partial credit can award fractional points when partial scoring is enabled.

Every answer is graded the moment the assessment is submitted. Auto-graded questions score instantly; essay and code-writing answers — and any auto-graded question left unconfigured — are flagged for manual review, where a grader assigns points from a dedicated grading screen.

Partial credit

With partial scoring enabled, supported types award fractional points — three of four correct matches earns three quarters of the question's points. Each question can override the assessment-wide setting either way.

Negative scoring

Optionally deduct a percentage of a question's points for a fully wrong answer. Only submitted answers are penalized — blanks never are — and the assessment total never drops below zero.

Passing score

Set a passing percentage and optionally show participants a pass or fail result on their completion screen. You also control how much detail participants see afterwards: nothing, the score only, or a detailed per-question review.

The grading screen

Manual grading happens on a dedicated screen that walks through the submission question by question. Every answer carries a status — correct, incorrect, partial credit, no answer, or needs review — and a summary panel tracks grading progress, the running score, and whether the participant currently passes.

A final review step warns about questions still awaiting review or left blank and shows exactly how many points the participant will receive. Grades stay editable after submitting — a grade is never locked in.

Alongside standard assessments you can create a coding assessment — a dedicated type chosen when the assessment is created. Instead of mixed question types, it is an ordered set of coding problems, and it counts against the same assessments-per-workspace limit as any other assessment.

Problems & test cases

Each problem has a problem statement, one language chosen by the author, optional starter code participants begin from, and an optional reference solution visible only to authors. Grading is driven by up to 20 test cases per problem — each pairs an input (stdin) with the expected output.

Test cases can be visible or hidden. Visible ones act as samples participants can run their code against; hidden ones run only at grading, so the full test suite is never exposed.

Supported languages

Thirteen languages are supported: JavaScript, TypeScript, Python, Java, C, C++, C#, Go, Rust, Ruby, PHP, Swift, and Kotlin. Each problem targets exactly one of them, chosen by the author.

The participant experience

Participants work in a full-screen IDE-style environment: one problem per screen, the statement beside a code editor with syntax highlighting, autocompletion, in-code search, and adjustable editor preferences. Before submitting they can run their code against the visible sample tests as many times as they like, or execute it with custom input in the built-in console — hidden tests stay out of reach until grading.

How coding problems are graded

At submission, each solution is compiled and executed server-side in a sandboxed runner against every test case — visible and hidden. With partial credit enabled, the score is the problem's points multiplied by the share of passing tests; otherwise it is all-or-nothing.

Output comparison trims surrounding whitespace and ignores letter case by default; both can be configured per problem. If code execution is unavailable or a problem has no test cases, the submission routes to manual grading instead of scoring zero.

Everything else works like a standard assessment: time limits, attempts and retake policies, participant identity, password protection, live monitoring, anomaly detection, custom themes, analytics, and exports all apply.

Assessment-level settings control the shape of every attempt:

Choose how participants identify themselves: fully anonymous, name only, or username-and-password credentials you manage per participant. For anonymous assessments you can optionally record IP addresses.

Identified participants make attempt limits, retake delays, and score policies meaningful — the platform knows which attempts belong to the same person.

Every attempt appears in the submissions list with its score, pass or fail state, grading status, and any proctoring flags. Open a submission to review it answer by answer, or jump into the grading screen for manual questions.

The analytics dashboard aggregates performance across submissions — score distributions, pass rates, and per-question statistics — and everything exports to CSV, Excel, JSON, or PDF on every plan. The Analytics & Exports guide covers both in detail.