---
name: "interview-kit-design"
description: "Use when a role needs its interview loop planned, the panel needs question banks and anchored scorecards, or tomorrow's interviewers need prep packets."
triggers: ["design the interview loop", "build an interview question bank", "write interview scorecards", "prep packet for tomorrow's panel", "what should each interviewer ask", "structured interview kit", "which competency does each slot cover"]
version: "1"
---
# Interview kit design
A structured kit makes a loop fair and a debrief possible: every candidate
for the role meets the same competencies, core questions, and anchored
scale. Build it from the role scorecard, the job description, the loop shape
(stages, slots, lengths), and the panel. Packets also use what the candidate
sent, their public professional work, and earlier scorecards.
## Map competencies to slots
Every must-have on the role scorecard gets exactly one owner, so nothing is
asked twice or skipped. Deal-breakers go early: screen, then the competency
most likely to rule people out, then the deep-dive, with the hiring
manager's close last. Anything a candidate may need room to explain, like a
career change, sits in a later slot.
Before interview coordination books it, confirm every slot has a
competency and the timings fit the day.
## Write the questions
Go one competency at a time, four to six questions per slot. Under each
question, note what a strong answer shows and one probe that tests whether
the candidate did the work themselves: "Which part was yours?", "What number
moved?", "What would you change?"
Drop any question that:
- checks recognition of a tool name rather than use of it;
- signals the wanted answer ("You're fine with ambiguity, right?");
- bundles two questions into one;
- a candidate could answer by reading the posting back.
Tighten and open up the user's own questions before adding new ones.
Give each slot a clock: for 45 minutes, roughly 5 on context, 30 on the
main problem, 10 for the candidate's questions. Warn that the main problem
overruns and eats the candidate's time. Core questions never change between
candidates; follow-ups vary only to clarify.
## Anchor the scorecard
Rate each competency 1 to 5: 1 unsatisfactory, 2 below bar, 3 meets bar, 4
above bar, 5 exceptional. Describe each level in one line of this role's
real work, so a 3 means the same to every interviewer. Add "not assessed"
so nobody guesses at ground they did not cover.
Each rating cites something the candidate said or did. "Seemed junior" is a
label; "could not say how the migration was rolled back" is evidence. The
card closes with hire or no hire and a one-sentence reason. Interviewers
file alone, before talking to the rest of the panel, inside the scorecard
window in memory. One scorecard per role. For the meeting, point the panel
to the running order and challenge line in hiring debrief and decision.
## Build the prep packets
One page per interview, in slot order. Top line: the time in the
candidate's timezone and the user's, the interviewer, and the competency
they own. Then:
- a candidate summary of five lines at most, using only what the user
supplied and the candidate's public professional work, a source on each
line and "not stated" for any gap;
- that competency's questions;
- what earlier rounds covered, and the open question this slot should
close;
- one logistics line: room or video link, who greets them, and the
scorecard deadline.
The evening prep routine runs this pass for tomorrow's loops.
## Keep every question job-related
Nothing may touch race, colour, sex, age or graduation year, pregnancy or
family plans, health or disability, religion, national origin, marital
status, or the language someone speaks at home. Test a language only when
the work happens in it. Work authorisation comes up only if their process
needs it, as a yes or no. Name who handles accommodation requests, and
never ask for medical detail.
When a question drifts, say which one and why, then offer a job-related
version: "Do you have young kids?" becomes "The role travels one week a
month; does that work for you?"
## What needs a yes
Kits and packets are drafts. Nothing reaches the panel or a candidate until
the user approves that particular send.
## What you hand back
The kit (question banks, anchored scorecards, slot timings) in chat and
saved to the hiring folder as one file per role; prep packets saved with
the date and role in the filename; and one line naming any slot missing an
interviewer or a competency.
## Fallbacks
Nothing from or about the candidate: base the packet on the role scorecard
and posting, and flag the summary as thin. No loop shape: propose screen,
knockout competency, deep-dive, hiring-manager close, and take one round of
edits. No role scorecard: run role intake and scorecard first, or draft from
the posting with every anchor marked draft until the hiring manager sees it.