Recruiting Prompt Library

Source

Find candidates who actually match the role, including the ones searching the obvious title misses.

1. Boolean / X-ray sourcing prompt

You are a technical sourcer. Given a role title, 3-5 must-have
skills, and a location/remote constraint, generate:
1. A LinkedIn X-ray search string (site:linkedin.com/in
   formatted, with skill synonyms and title variants OR'd together)
2. A boolean string formatted for Indeed/Google Jobs
3. 3 adjacent job titles candidates might actually hold instead
   of the literal title searched (people rarely have your exact
   title)
Do not pad with generic "passionate self-starter" language —
output only the search strings and the adjacent-title list.

Most sourcing prompts stop at the obvious title. The adjacent-title list is where the actual candidates are.

Qualify

Turn the job description into a shared, evidence-based bar before anyone looks at a resume.

2. Job description → weighted scorecard

Convert this job description into a structured evaluation
scorecard. Separate must-have from nice-to-have. For each
criterion, define what "strong," "adequate," and "weak" evidence
looks like in a resume or interview answer — concretely, not as
vague adjectives. Weight must-haves higher than nice-to-haves.
Flag any criterion in the JD that's aspirational rather than
actually required for success in month one.
[paste job description]

This is the single biggest bias-reduction move in hiring — a shared rubric before anyone looks at a resume, not after.

3. Job description bias & inclusivity audit

Review this job description for language that narrows the
applicant pool without improving job-fit signal: gendered
phrasing, unnecessarily inflated requirements (e.g. "10 years"
for a tool that's existed 5), degree requirements where
experience should suffice, and jargon that filters out qualified
non-traditional candidates. Rewrite each flagged section with a
tighter, requirement-justified alternative. Keep everything that's
a genuine job requirement — this isn't about softening the bar.
[paste job description]

Most inclusivity audits either do nothing or gut the requirements. This keeps the bar and removes the noise.

Screen

Score against the rubric, not a gut read.

4. Resume screening against the scorecard

Using this scorecard [paste scorecard] and this resume [paste
resume], score the candidate against each criterion with the
specific line from the resume that supports or contradicts it.
Output: Strong / Possible / Pass, plus what's missing that you'd
need to confirm in a screen call. Never infer soft skills from
resume formatting or school prestige — score only what's
evidenced.

Ties every judgment back to the scorecard, not a gut read — and blocks the two most common bias vectors.

Interview

Keep every interviewer grading the same test.

5. Structured interview question generator

For each competency in this scorecard [paste scorecard], generate
one behavioral question ("tell me about a time...") and one
situational question ("what would you do if..."). For each, give
a one-line example of what a strong answer contains versus a
generic-but-plausible-sounding weak answer, so different
interviewers score consistently.

Consistency across interviewers is where most panels quietly fall apart — this keeps everyone grading the same test.

6. Passive-candidate outreach prompt

Draft a first-touch LinkedIn message to a passive candidate for
[role]. Lead with a specific, genuine observation about their
actual work — not a compliment about their title. Ask for
nothing in this first message. Keep it under 300 characters. The
job-specific ask comes only after they respond, not in the opener.

The standard "impressive background, exciting opportunity" opener reads as a template because it is one. A real observation is what actually gets replies.

Decide

Surface disagreement instead of averaging it away.

7. Reference check question generator

Using this scorecard [paste scorecard] and the specific gaps
flagged during the interview [paste interview notes/gaps], generate
reference check questions that target exactly those gaps — not a
generic reference script. For each question, note what a genuinely
strong reference sounds like versus a diplomatically vague one
(references rarely say no outright; they go quiet on specifics).

Most reference checks confirm what you already believe. This is built to test the parts you're actually unsure about.

8. Interview panel debrief synthesizer

Here are scorecard ratings and notes from [N] interviewers for one
candidate [paste all]. Synthesize into: where the panel agrees,
where it genuinely disagrees (not just phrasing differences), and
the specific question to ask in the debrief meeting to resolve the
disagreement. Do not average disagreeing scores into a false
consensus — surface the conflict so the panel actually discusses it.

Averaging away disagreement is how panels accidentally hire the candidate nobody was sure about.

Close

The last stretch is where good candidates are most often lost — and how every candidate is treated on the way out.

9. Not-selected candidate message

Draft a rejection message for a candidate who made it to [stage].
Specific enough that they know this wasn't a form letter — one
real, respectful reason tied to the role, not their worth as a
professional. No false "we'll keep your resume on file" unless
that's genuinely true. Warm, brief, no corporate hedging.

Candidates who got this far talk about how they were treated, even when rejected. This is the message most companies get wrong.

10. Offer & close conversation prep

Given this candidate's likely motivations [paste what you know:
current comp, stated priorities, competing offers if known],
anticipate their top 3 objections to this offer [paste offer
details] and draft a response to each that's honest, not just
persuasive — including what you'd say if the honest answer is
"we can't move on that."

The reps who lose good candidates at offer stage are usually the ones who only prepared the pitch, not the pushback.

Advanced agents

Persistent, ongoing prompts rather than one-shot asks — for running an agent continuously against a live pipeline instead of re-running a prompt by hand each time.

11. Multi-channel candidate sourcing agent Agent

You are a persistent candidate-sourcing agent for [role]. Run
continuous sourcing across LinkedIn, GitHub (if technical), and
relevant portfolio platforms. Maintain a running pipeline, not a
one-time list: flag new candidates as they appear, note when a
previously-passive candidate shows job-search signals (open-to-work,
recent profile updates), and surface adjacent-title candidates the
literal search misses. Output a pipeline table, not a one-off list.

A one-time search goes stale in days. This keeps the pipeline current without a human re-running it.

12. Continuous qualification & re-scoring agent Agent

You maintain a live-scored candidate pipeline against [scorecard].
As new information arrives — an updated resume, a screen-call note,
an interview scorecard — re-score the affected candidate and flag
what changed and why, rather than requiring a manual re-review.
Surface any candidate whose score moved enough to change their
priority tier.

Scores decay the moment new information arrives and nobody updates the sheet. This keeps the ranking honest in real time.

13. Candidate authenticity screening agent Agent

Screen this candidate's application materials for authenticity red
flags: work history that doesn't align with company timelines on
LinkedIn, portfolio pieces that appear elsewhere online under a
different name, inconsistent narrative between resume and interview
answers, and signs of AI-generated or plagiarized writing samples
where original work was requested. Flag only, never auto-reject —
these are prompts for a human to investigate, not verdicts.

Flags for a human to check, never a verdict — the point is catching what's worth a second look, not automating rejection.

14. Talent-market benchmarking agent Agent

For [role] at [company], pull who peer/competitor companies of
similar size and stage actually hired for the equivalent role in
the last 18 months. Compare their backgrounds against what this
JD asks for: is the requirement realistic for who's actually
available and taking these roles, or aspirational relative to the
real market. Separately flag any strong candidates at those peer
companies who look like plausible passive targets.

Tests whether the job description matches who the market actually produces, not just what sounds good in a requisition.

Recruiting prompt library