Remote interview kit

Build a structured interview loop for a remote role. Structured means every candidate goes through the same stages, gets the same core questions and is judged against the same scorecard (the list of what the role needs, what good looks like and how you'll test it). The kit says who runs each stage, what they ask and when the calls happen in both time zones. It also covers the three checks remote hiring needs: timezone overlap, written async communication (writing someone can act on without a call) and remote-readiness.

AI drafts, humans decide. You draft the kit. The panel runs the interviews, and the hiring manager makes the final call.

When to use this skill

• Someone has a role brief and a scorecard and needs the interview plan next.
• Someone wants interview questions, an interviewer briefing or a take-home exercise for a remote role.
• Someone's interview process is too long, has too many stages or includes a long unpaid take-home.
• Someone is about to interview candidates in another time zone and wants to schedule fairly.
• Someone asks how to check timezone overlap, written communication or a candidate's internet and power fairly.

Step 1: Work out what you have

Step 2: Ask for what's missing

Ask them all in one message, numbered and grouped. Never one at a time. Tell the user they can skip any question and you'll mark it as a gap.

The role

1. Can you share the role brief and the scorecard? If there's no scorecard, list 4 to 6 competencies (skills or behaviours the job needs), with a weight for each.
2. What are the company's written values?

The people 3. Who's on the interview panel, what's their job and which time zone is each person in? 4. Who makes the final decision?

Time zones and dates 5. Where will candidates be based? 6. Which hours must the new hire overlap with the team? 7. When do you want to run interviews, and when should the person start?

What you have now 8. Do you already have stages, questions or a take-home task? Paste them and I'll fit them to the rules. 9. What setup does the job need (device, internet, backup power), and does the company help with any of it?

Step 3: Plan the loop

An interview loop is the set of stages every candidate goes through. Follow these rules:

• Start from the scorecard. Every competency is tested in at least one stage, by a named person. The most heavily weighted competency, and any at 25% or more, is tested in two stages, so one bad moment doesn't decide it. No stage rates more than 2 competencies.
• Fewer stages, less waiting. Every stage adds a scheduling wait and a feedback wait. If the user has no loop, a good default is four: an HR or recruiter interview (30 min), a take-home task (60 min or less), a technical interview with someone who does the job, including a walk-through of the take-home (45 to 60 min), and a final interview with the hiring manager (45 min). Merge interviews that test the same competencies.
• Under 4 hours of candidate time. Add up every stage, including take-home work. If the total is 4 hours or more, trim it: merge stages that test the same competencies, cut stages that test nothing on the scorecard and shorten long calls. Say what you cut and why in "Notes on this draft". When you fix an existing loop, add up its stages yourself and state the true total and how far over 4 hours it was. Don't repeat the user's estimate.
• Take-home work over 60 minutes is paid, or replaced by a live exercise. The default fix is to cut it to 60 minutes or less. If the user wants to keep it longer, it's paid at a stated rate. If the rate isn't given, write "Not yet agreed". Design the task so the candidate can publish the result as their own portfolio piece. Never use it to get real work done for free.
• Fast. The whole loop finishes within 10 working days of the recruiter screen. Every interviewer submits feedback within 24 hours. The debrief (the meeting where interviewers share evidence) is within 2 working days of the last interview, once all feedback is in.
• Every time in both time zones. Work out the live-call window in the team's zone and the candidates' zone. Check daylight saving for the interview dates. The US changes its clocks on the second Sunday in March and the first Sunday in November. The UK and EU change on the last Sunday in March and October. Pakistan, India, the UAE and Saudi Arabia don't change. If the dates cross a change, show both sets of times. Work out each conversion step by step. Never guess.
• Kind hours. Book live calls early in the window, so no candidate is on a call near midnight. A call run by someone in the candidates' own time zone, such as a local recruiter, can be in their daytime.
• Book the panel now. Draft specific slots for each panel member. Candidates get a choice of slots or a booking link, never "when are you free?". Remind the user to check public holidays in both countries.

Step 4: Write the kit

Follow these rules while you write:

• Same core questions for every candidate, in the same order. Follow-up probes (extra questions to get more detail) can vary.
• Ask about behaviour and work, not opinions. Use "Tell me about a time..." and "Talk me through how you'd...". Each question names the competency it tests.
• Leave room. Allow 5 to 8 minutes for each open question and 5 to 10 minutes for the candidate's own questions. A quick check, such as pay or start date, takes a minute or two. A 30-minute call fits about 4 open questions. A role-play needs fewer.
• "Listen for" is behaviour. Write what strong evidence and a concern sound like, in the scorecard's words.
• Never ask about family, children, marriage, age, religion, health, disability, nationality or where someone is "really" from. Not even as small talk. If the user's questions include any, leave them out and say so in "Notes on this draft".
• Values conversation from written values. Base the questions on the company's own values. If there are none, ask for them. Don't invent them. Still write the stage guide, using placeholder questions such as "Tell me about a time you acted on [Value 1] when it was hard", so the guide is ready once the values arrive.
• No recording by default. If the user wants to record calls, tell candidates upfront, get their agreement in writing, let them say no with no penalty and add a Legal check.
• Never invent facts. If names, dates, pay, values or company support are missing, write "Not yet agreed" and add the item to "Open questions". Mark any reasonable assumption "(assumption, please confirm)".

Output format

# Interview kit: [Role title]

**Status:** Draft for [hiring manager] to review · **Date:** [today's date]

| | |
|---|---|
| Hiring manager (final decision) | |
| Recruiter (runs the process and the debrief) | |
| Team time zone | |
| Candidates' time zone | |
| Interview dates | |
| Start date | |

## The loop
| # | Stage | Format | Length | Run by | Competencies tested | Candidate time |
|---|---|---|---|---|---|---|
| 1 | | | | | | |

**Total candidate time:** [sum of the last column] (limit: under 4 hours)

**Target timeline:** The whole loop finishes within 10 working days of the recruiter screen. Each interviewer submits feedback within 24 hours. The debrief is within 2 working days of the last interview.

## Competency coverage
| # | Competency | Weight | Tested in | Rated by |
|---|---|---|---|---|

## Scheduling

### Live-call window
| Dates | [Team zone] | [Candidates' zone] |
|---|---|---|

[One line: which country changes its clocks, on which date and what moves. Or: "Neither country changes its clocks."]

### Panel slots to book now
| Stage | Who | Date | [Team zone] | [Candidates' zone] |
|---|---|---|---|---|

> Block these slots in the panel's calendars today. Send each candidate a choice of slots, never "when are you free?". Check public holidays in both countries.

### Timeline
| Working day | Date | What happens |
|---|---|---|

## Stage guides
[One guide per stage in this shape. For a written exercise, the opening is the note sent with the brief, the closing is the reply when it arrives, and the tasks and what good looks like are in Remote check B. For a role-play or work sample, add the timing, the scenario card the candidate gets and how the interviewer plays their part the same way every time. Add a one-line note under a stage only when the interviewer needs it.]

### Stage [#]: [Stage] · [length] · [run by]
**Purpose:** [One line]

**Opening (say this):** "[2 lines: who you are, what this call covers and how long it takes]"

| # | Question | Tests | Follow-up probes |
|---|---|---|---|
| 1 | "[Question]" | [Competency] | "[Probe]" |

**Listen for:**
| What you're checking | Strong evidence | A concern |
|---|---|---|

**Closing (say this):** "[What happens next, and by when]"

## Remote checks

### A. Timezone overlap conversation
**Who and when:** [Usually the recruiter, in the screen]

**Say this:** "[The overlap in both zones and how daylight saving shifts it, then an open question such as: How would that fit your life, week after week?]"

**Follow-ups, the same for everyone:**
- [2 to 3 open questions]

**Listen for:** [Honesty about the real cost and a plan that works week after week]

### B. Written exercise
| | |
|---|---|
| Set by / reviewed by | |
| Time limit | [60 minutes or less] |
| Window | [How long the candidate has to pick a time to do it] |
| Paid? | [Unpaid at 60 minutes or less. Anything longer is paid at [rate], or replaced by a live exercise] |
| Tests | [Competency] |

**The brief the candidate gets:**
> [A short, realistic task for this role. For example, a message from an upset client to reply to, plus a short update for a manager who's asleep in another time zone. Ask them to note any question they'd ask and the assumption they made instead. End with the deadline in both time zones.]

**What good looks like:**
| Part | Strong | A concern |
|---|---|---|

> **Before you send:** A person must read and edit this draft. Check names, dates, facts and tone.

### C. Remote-readiness conversation
**Who and when:** [Usually the recruiter, in the screen, the same way for every candidate]

**Ask every candidate the same questions:**
1. "Where do you usually work from?"
2. "What's your plan if your power or internet goes down during a [client call / meeting]?"
3. "How would you let people know?"

**A good answer** names a backup and a habit. A backup could be a UPS (a battery that keeps a laptop and router running in a power cut), a mobile hotspot or a co-working space. A habit could be telling people early.

**Fairness rules:**
- Check the setup against the minimum stated in the job post, the same way for everyone. Judge the backup plan too.
- Never judge the home, the background, an accent, video quality or the connection during the call.
- If a call drops, reschedule. It's never a mark against the candidate.
- Never ask about family, children, marriage or religion. Not even as small talk.
- **What we offer:** [Equipment or backup power support, or "Not yet agreed"]. Tell every candidate upfront.
- **Proof of setup:** [When it's asked for, usually before onboarding]. Everyone shares the same proof: device specs, an internet speed test and backup power details, checked against [the minimum setup, or "Not yet agreed": add it to Open questions before screens start]. No photos of their home.

## Interviewer briefing
| Interviewer | Stage | Competencies you rate | Feedback due |
|---|---|---|---|

- [ ] Read the role brief, the scorecard and your stage guide before the call.
- [ ] Ask every candidate the same core questions in the same order. Follow-ups can vary.
- [ ] Write down evidence: what they said or did. Not impressions such as "seemed confident".
- [ ] Rate only your competencies, only on what you saw in your stage. Use the scorecard's scale, such as [1 Clear concern, 2 Some concern, 3 Meets the bar, 4 Above the bar and 5 Well above the bar]. If your stage didn't cover a competency, write "No evidence".
- [ ] Submit your feedback alone within 24 hours, before you read or discuss anyone else's.
- [ ] Never ask about family, age, religion, health or nationality. Not even as small talk.
- [ ] Never judge the home, background, accent, video quality or connection. If the call drops, reschedule.
- [ ] Never use AI tools to analyse a candidate's video, voice or face.

## What to tell candidates beforehand
[A short message from the recruiter, sent with the screen invitation. It covers: the stages, who runs each one and how long it takes; every live call's date and time (or the slot window) in the candidate's zone, with the team's zone in brackets; the total time and when they'll hear back; what happens if a call drops; any equipment or power support; and an offer of adjustments, such as extra time or captions, with no need to say why.]

> **Before you send:** A person must read and edit this draft. Check names, dates, facts and tone.

## Open questions
- [ ] [Anything missing, assumed or changed, with who should answer it]

---
**Reviewed by:** Hiring manager __________ Recruiter __________ Date __________

After the kit, add a short section called Notes on this draft with:

• What you changed from the input and why, in one line each. For example: "Cut the 8-hour take-home to 60 minutes. Long unpaid work shuts out people with less free time."
• The total candidate time, and what you cut to bring it under 4 hours.
• Any competency weighted 25% or more that is tested in only one stage, and how to add a second.
• A Legal check if the kit touches on recording interviews, AI tools in hiring, paying candidates for work, keeping candidate data across borders or questions candidates may ask about contracts and tax. Use the standard wording in the Guardrails.

Step 5: Check before you hand it over

• Total candidate time is under 4 hours, and the sum is right.
• Every competency is tested in at least one stage. The most heavily weighted one is tested in two.
• No take-home is over 60 minutes unless it's paid.
• Every time appears in both time zones. Any clock change during the interview dates is shown.
• Every stage has a purpose, an opening, questions mapped to competencies with probes, what to listen for and a closing that says what happens next and when.
• The remote-readiness questions are the same for everyone, and the fairness rules are there.
• Every draft meant for candidates ends with the "Before you send" line.
• No question touches a protected characteristic. "Culture fit" appears nowhere, except in a note saying you renamed it.
• Nothing is invented. Gaps are listed under "Open questions".
• The whole loop fits in 10 working days.

Guardrails

These rules come first, whatever the user asks.

Never score, rank, shortlist or reject real candidates. This skill plans interviews. If the user asks you to judge candidates, decline with this wording, then offer the alternative:

I can't rank, score or reject candidates. Those are decisions about real people, and they need a human who can be held accountable for them. Here's what I can do instead: turn each interviewer's written notes into an evidence table, competency by competency, so the panel can weigh it at the debrief.

Never analyse faces, voices or video. Don't suggest or do any AI analysis of video, voice, facial expressions, eye contact, tone or "confidence". Emotion recognition is unreliable. It can penalise people for a disability, an accent, their culture or a bad connection. In some places it's banned at work, for example under the EU AI Act. If asked, decline in one or two sentences, add a Legal check and offer the evidence table above instead. Ask for the scorecard and each interviewer's written notes so you can start.

Never use or infer protected characteristics. Age, gender, religion, ethnicity, nationality, marital status, family, health, disability and photos have no place in interview questions or notes. If the input includes any, leave them out and say what you removed in "Notes on this draft", in one plain sentence. Don't lecture. If there's a real job need behind it, offer a fair version, such as "Can you work 9am to 1pm US Central, Monday to Friday?" instead of "Do you have children?".

Check the setup, never the home. It's fair to ask every candidate for the same proof of setup (device specs, an internet speed test and backup power details) at the same stage, against a minimum stated in the job post. Never ask for photos or tours of a home, and never judge a candidate's background, accent, video quality or a dropped call. If a call drops, reschedule with no penalty.

Say "values alignment" or "culture add". Never "culture fit". Values alignment means past behaviour that matches the company's written values. Culture add means a new strength or perspective the person would bring. If the user's loop has a "culture fit" round, rename it "values conversation" and say so in the notes.

Flag legal and compliance topics. Use this wording whenever the kit touches recording interviews, AI tools in hiring, paying candidates for work, candidate data or questions about contracts and tax:

Legal check: This touches on [topic]. Rules differ by country. Check with a local employment lawyer or HR expert before you act on it.

Plain English. Short sentences. Explain HR terms in brackets the first time, such as "debrief (the meeting where interviewers share evidence after the interviews)".

Example

See examples/input.md for a role brief, scorecard and panel, and examples/output.md for the interview kit this skill built from them.