AI Prompt Pack: Candidate Screening
Five copy-paste prompts that take a CV from inbox to client-ready submittal. Built to pair with the screening scorecard in this kit.
How to use: paste a prompt into ChatGPT, Claude, or whatever model your team uses. Replace anything in [brackets]. For best results, paste the full job spec once at the start of the conversation and reuse it across candidates.
Prompt 1 — The 60-second CV summary
For triaging a full inbox fast. One CV in, six lines out.
You are a recruitment researcher. Summarize the CV below in exactly this format:
CURRENT: [current title, company, tenure]
TRACK RECORD: [2 lines — most relevant achievements, with numbers if present]
CORE SKILLS: [comma-separated, max 8]
GAPS/UNKNOWNS: [1 line — what the CV doesn't tell us]
LOGISTICS: [location, work authorization if stated, notice period if stated]
ONE-LINER: [how a recruiter would pitch them in one sentence]
Do not infer anything not supported by the CV. Mark guesses as "unclear".
CV:
[paste CV text]
Prompt 2 — Score a CV against the spec
Outputs the exact six criteria from the scorecard, so you can type the numbers straight into the sheet.
You are screening candidates for the role below. Score the CV against the
spec on each criterion from 1 (poor fit) to 5 (excellent fit), using only
evidence from the CV. For each score, give the single strongest piece of
evidence in one short line.
Criteria:
1. Must-have skills — the non-negotiables in the spec
2. Experience depth — years and seniority doing this actual job
3. Industry fit — has done it in our client's sector or one adjacent
4. Tenure & stability — reasonable stints, explainable moves
5. Logistics fit — location/remote, work authorization, salary band if stated
6. Communication — clarity and professionalism of the CV itself
Output format:
| Criterion | Score | Evidence |
Then: OVERALL READ: [2 sentences — would you put this person in front of
the client, and what would you check first on a call?]
JOB SPEC:
[paste spec]
CV:
[paste CV text]
Prompt 3 — Red-flag and gap scan
Review this CV as a skeptical recruiter. List:
1. RED FLAGS — unexplained gaps over 6 months, overlapping dates,
title inflation patterns, vague ownership language ("involved in",
"exposure to"), qualifications that don't match the timeline.
2. QUESTIONS TO ASK — the 5 questions a phone screen must answer
before we could submit this candidate.
3. NOT RED FLAGS — anything that looks unusual but has a common
innocent explanation, so we don't ding them unfairly.
Be specific and cite the CV. If it's clean, say so.
CV:
[paste CV text]
Prompt 4 — Client submittal draft
Turns a screened candidate into a submittal your client can read in 60 seconds. Edit the voice sample once, reuse forever.
Write a client submittal for the candidate below. Match this voice: direct,
warm, no recruiter clichés ("rockstar", "hit the ground running"), no
adjectives without evidence. Here is a sample of our agency's writing so
you can match tone:
[paste 2-3 sentences from a real submittal you've sent]
Structure (keep under 180 words):
SNAPSHOT: name, current role, one-line positioning for THIS role
WHY THEY FIT: 3 bullets, each tied to a specific requirement in the spec,
each with evidence (metric, project, or tenure)
LOGISTICS: availability, notice period, location/remote, rate expectation
NEXT STEP: one sentence proposing the interview slot
JOB SPEC: [paste spec]
CANDIDATE NOTES: [paste your screening notes / scorecard row]
CV: [paste CV text]
Prompt 5 — Interview questions from the gaps
Based on this candidate's CV and the job spec, write 6 interview questions:
- 3 that probe the specific gaps or unknowns in their CV (name the gap)
- 2 behavioral questions tied to the hardest part of this role
- 1 that tests the must-have skill with a concrete scenario
For each, add one line: "Good answer sounds like…"
JOB SPEC: [paste spec]
CV: [paste CV text]
Responsible use
The AI does the first read — a human makes every decision. Never let a model auto-reject candidates. Don't feed it protected characteristics (age, ethnicity, disability, family status) or ask it to infer them. Keep the scoring evidence-based and job-related — the same standard you'd defend to a client or a regulator. If you automate this workflow at volume, check your local rules on automated employment decisions (e.g. NYC Local Law 144).
Doing this 300 times a week? This exact workflow — score inbound CVs, update the ATS, draft the submittal — is what we build as custom software you own outright. One week, flat fee. [modusstudio.co]