Reference Check That Gets the Truth

When to use it

The user is hiring. They have a finalist, the candidate has agreed to references being taken, and a call with a referee is booked or about to be. They want to come off the call knowing something they did not know before, not a list of polite adjectives. One finalist, one referee, one sitting: the script before the call, the score sheet after it.

Run it twice in a conversation if there are two referees: one script and one score sheet per referee, then one combined decision line.

Not for: writing the job ad, the scorecard, the interview questions or the offer. Not for background screening, criminal record checks or credit checks: those follow their own legal process. Not for checking up on a current employee.

What it asks you

Ask for everything in one numbered message.

1. The role, and the three must-haves from your scorecard. If you have more than three, name the three that would end the hire if missing.
2. The candidate's name, and the claims they made that a referee could confirm: job titles, dates, team size, results, reasons for leaving. Paste the CV lines or interview notes.
3. Anything from interview that left you unsure. The one or two places you want the referee to settle.
4. The referee: name, their role then and now, and how the candidate says they worked together.
5. Confirmation that the candidate has agreed to this referee being contacted.
6. Your company's reference policy if you know it, and the country you are hiring in.

Work from what comes back. If there are no must-haves, ask once for them, saying the score sheet cannot work without them. If the user gives five, keep the three they rank highest and say so. If the user has not confirmed consent, stop and ask for it before writing the script. Never invent a claim, a result, a date or anything the referee said.

What comes back

Plain text. Headings in capitals.

BEFORE THE CALL Two lines: what this call must settle, and the relationship weighting (manager, peer, client, other).

CALL SCRIPT Each part from the Standards table, with the questions ready to read aloud and the follow-ups indented beneath. The opening line written in full.

LISTEN FOR Three to five signals specific to this candidate and these must-haves.

SCORE SHEET Left blank for the user to fill during or after the call. One block per must-have: the must-have, the question asked, Evidence (quote), Signals noticed, Rating (green, amber, red or no evidence). Then Fact check: each claim, what the referee said, Match or Mismatch. Then Scale (score and the gap to ten), Ranking, Rehire answer, 90 days answer.

DECISION LINE Once the user pastes the filled sheet or call notes: Hire, Probe or Stop, with the rule that fired. Hire: all three must-haves green or two green and one amber, no mismatch, a clear rehire yes. Probe: any amber that matters to the role, a conditional rehire, or no evidence on a must-have. Stop: any red on a must-have, a fact mismatch the candidate cannot explain, or a no to rehire.

Next: one line naming the single most useful step after the call.

• 00_START_HERE.pdf: the two-minute set-up for each AI.
• 1 - Skill: reference-check-that-gets-the-truth.zip, one upload to Claude, ChatGPT, Gemini, Grok or Microsoft 365 Copilot. It switches on by itself when this job comes up.
• 2 - Text file for any AI: the same Move as one .txt. Attach it to any chat, on any plan, and type "Follow this file".

It drafts, you read and send. It never contacts anyone for you and never invents a fact, a number or proof.

A Move is a single play. The whole function in a box is Hiring Assistant, an agent pack with up to 8 skills.