Build an interview kit

A kit keeps the exercise and rubric consistent across interviews. At Lanefold, a fictional freight company, you might build one around a backend engineer's design for Dispatch Service.

Kits are available on every plan once Interviews is enabled. Reading kits requires interviews:read; creating, editing and publishing them requires interviews:manage.

Start a kit

1. Open Interviews and select Browse kits.
2. Select New kit, or use the duplicate icon on a starter kit to make a copy. Select Open builder to edit an existing kit.
3. Set Kit name and Minutes. Enable Live, Async, or both for the modes you intend to use.

Write the stages

1. On Stages, select the Add stage icon.
2. Set Title and write the Candidate prompt. Put the problem and constraints the candidate should read here.
3. Set Reveal mode and, optionally, Stage budget in minutes.
4. Write Interviewer notes and Probes for the panel. Add Strong signals, Weak signals or Red flags where they give the interviewer something concrete to observe.
5. Check Candidate preview. Stage edits save after you leave a field; wait for Saved before moving on.

For a Dispatch Service exercise, the first prompt could ask for a job assignment design. A later stage could introduce duplicate delivery events. Keep advice about detecting retries in Interviewer notes so it doesn't appear in the candidate prompt.

Manual stages wait for an interviewer to reveal them. Timed stages use Reveal at session minute, measured from the session start. Auto sequential stages advance when the preceding stage is complete; the first such stage can appear at session start. Async kits can't contain manual stages.

Define the rubric

1. Open Rubric and select the Add competency icon.
2. Set Name, Weight and, optionally, Description.
3. Under Anchored rating scale, write a label and observable descriptor for each level from 1 to 5.
4. Select Save competency.
5. Return to each stage and select the relevant items under Competencies exercised.

A trade-off competency, for instance, might distinguish naming a queue from explaining how retries affect ordering. The panel chooses ratings using these anchors; the kit doesn't rate the candidate.

Prepare the canvas

Use Seed canvas for the starting drawing shared across the interview. Choose Draw it to draw directly, DSL YAML to enter YAML, or None to start without a seed.

Drawing a seed canvas requires a viewport at least 768px wide.

Each stage also has a Stage seed canvas with the same choices. Its drawing is added when that stage is revealed. It doesn't replace what the candidate has already drawn. Switching away from Draw it discards that seed drawing after confirmation.

Creating an interview copies its kit and seed content. Later kit edits don't change that existing session.

Draft an empty kit with AI

On a draft with no stages or competencies, select Draft with AI. This also requires Team or above and ai:use, and consumes AI credits.

Enter Role, Level, Duration in minutes and Focus areas, then select Generate draft. Review the proposed prompts and rubric before selecting Use this draft. Generating alone doesn't add the proposal to the kit.

Publish the kit

Resolve the publish issues shown at the top, then select Publish. A kit needs at least one stage and one competency, five rating anchors per competency, and a competency mapping for every stage. Stage budgets can't exceed the session duration. Timed stages need a reveal minute, and at least one run mode must be enabled.

Once published, the kit appears in the picker when you schedule an interview.