# AI for Recruiting: 31 Claude Skills to Hire the Right People AI for recruiting: 31 free Claude skills, in-house or agency: intake, job ads, sourcing, structured interviews, scorecards, offers and an AI recruiting policy. - Page: https://lantern.meet-polar-bear.com/skills/ai-for-recruiting-pack - Download: https://lantern.meet-polar-bear.com/skills/ai-for-recruiting-pack.zip - Source: https://github.com/polar-bear-org/claude-skills/tree/main/packs/ai-for-recruiting-pack - Price: free ## Install in Claude Code ``` /plugin marketplace add polar-bear-org/claude-skills /plugin install ai-for-recruiting-pack@polar-bear-skills ``` ## The 31 skills you get The set follows the search, in eight parts: briefing the role, keeping candidates in the loop, attracting, sourcing and reaching out, screening and interviewing, deciding and closing, the agency desk, and running the desk. ### 1 · Brief the role - **Intake Meeting** — use when: The manager cannot describe the job and your search drifts. Output: A question plan for the hiring manager, a calibration on three to five anonymised sample profiles, a must-have versus trainable list, and a one-page intake summary to sign off. - **Job Scorecard** — use when: Two jobs hide in one title and “senior” means five things. Output: First-year outcomes, four to six job-related criteria with what evidence of each looks like, the must-have versus trainable split, and the stage where each criterion is assessed. - **Salary Range Brief** — use when: You fear “we didn't budget for this” after four rounds. Output: The proposed range from the evidence you supply, where the range must appear, what to say to candidates above or below it, and pay transparency questions for an adviser. - **Hiring Process Plan** — use when: There is always one more interview and the pipeline goes cold. Output: A stage map with the criterion assessed at each step, a maximum number of rounds, feedback and decision deadlines, the decision meeting date and the candidate touchpoints. ### 2 · Keep candidates in the loop - **Candidate Communication Plan** — use when: Candidates hear nothing for weeks and the good ones go silent. Output: A promise per stage naming who writes and by when, the “no update yet” message, a close-the-loop rule, and keep-warm notes from offer to first day. - **Interview Scheduling Messages** — use when: A screen was cancelled twelve minutes before it started. Output: An invitation that says what to expect and offers adjustments, then confirmation, reminder and reschedule, and apologies for a late cancellation or a missed slot. - **Rejection Email** — use when: You have decided no and need to say it kindly and fast. Output: Messages per stage with feedback lines drawn only from the documented criteria, a short phone script for finalists, and a calm reply to a hostile response. - **Candidate Experience Survey** — use when: Strong candidates decline because of how the process felt. Output: Five to seven questions per stage, when to send it, how to read the results in aggregate, and a fix list for the process owner. ### 3 · Attract - **Job Description** — use when: You are about to repost last year's description. Output: An outcomes-first description built from the scorecard, must-haves only, a pay range line, an adjustments line, and a wording check that flags gendered, age-coded and exclusionary words. - **Job Ad** — use when: Hundreds apply who ignored the rules, and the right people scroll past. Output: A two-line hook, the real job including its hard parts, pay, location and work authorisation rules up front, how to apply, and what happens next by when. - **Employee Referral Program** — use when: Inbound is full of fakes and you want warm introductions. Output: The ask to staff built from the scorecard, a guide to what makes a good referral, bonus rules as placeholders, the same assessment for referred people, and a thank-you and update loop. ### 4 · Source and reach out - **Boolean Search Strings** — use when: Your strings return thousands or nobody. Output: Title and skill synonym sets, strings for LinkedIn, Google X-ray and job boards in a broad and a narrow version, and a test-and-tighten log. - **Talent Mapping** — use when: The manager wants “our exact industry” and you need evidence. Output: The target companies and teams where the profile works, adjacent titles, feeder fields and a pool size estimate from public information, with no files on individuals. - **Sourcing Strategy** — use when: Inbound is unusable and replies have collapsed. Output: A channel mix across outbound, referrals, communities, past applicants, boards and events, free routes when you have no paid seat, weekly targets per channel, and a list of what to stop. - **Recruiting Email Templates** — use when: Everyone can spot a message pasted from a chatbot. Output: A first message and two follow-ups built only from facts you supply, with subject lines, a short InMail version, and a what-not-to-say list. ### 5 · Screen and interview - **Screening Questions** — use when: Hundreds of applications a week and a person still reads each one. Output: Evidence questions for the application form in place of knockouts, a phone screen guide asked the same way to everyone, and an identity and work-authorisation step done by a person. - **STAR Interview Questions** — use when: Candidates answer “tell me about a time” with what they would do. Output: Behavioural questions per scorecard criterion, STAR probes for vague or “I would” answers, one situational question per criterion, and a never-ask list. - **Structured Interview** — use when: Every interviewer asks their favourite questions. Output: The panel split of criteria, the running order with timings and fixed probes, the opening and closing scripts, note-taking rules, and what the candidate is told in advance. - **Work Sample Test** — use when: CVs no longer tell you who can do the work. Output: A short task that mirrors the job, the time limit and payment decision, the candidate instructions, and an evidence guide interviewers use to mark the work. - **Interview Scorecard** — use when: You need evidence written down before anyone says “I liked her”. Output: One blank form per interviewer, a section per criterion with anchored descriptions of weak to strong evidence, a notes box before each level box, and no total score. ### 6 · Decide and close - **Interview Debrief** — use when: The loudest voice decides, or the manager asks to “see a few more”. Output: An agenda that collects written scorecards before anyone speaks, a walk through the evidence criterion by criterion, and a record of the hiring manager's decision and reasons. - **Reference Check Questions** — use when: You want references that tell you something. Output: A consent step, questions tied to the scorecard criteria and the debrief's open questions, probes past polite answers, a never-ask list and a note template. - **Offer Letter** — use when: The offer is approved and must go out today. Output: A letter drafted from terms a named person approved, the conditions and start date, an offer call script, and a clause list for adviser review. - **Counter Offer Plan** — use when: Declines and counter-offers keep taking your accepted candidates. Output: Questions to ask before the offer, resignation conversation prep, honest answers when the current employer counters, and a keep-warm plan from notice period to day one. ### 7 · Agency desk - **Recruiter Business Development** — use when: Mass outreach to decision makers is dead. Output: A target account list from hiring signals you gather, a one-page approach per account, first-call questions, and a four-touch outreach plan. - **Agency Terms of Business** — use when: Clients ignore fees or hire someone you introduced. Output: Each term in plain language — fee basis, retained versus contingency, guarantee or rebate, introduction period, payment terms — a one-page client summary, and a clause checklist for an adviser. - **Candidate Submittal** — use when: You want the client to read evidence, not adjectives. Output: A consent-to-represent record, the facts against each scorecard criterion in the candidate's own words, and motivation, pay expectation, notice and availability, with no fit score. - **Client Search Update** — use when: The client goes quiet, chases daily, or rejects everyone. Output: Activity by stage, what the market is saying about why candidates say no, a proposed brief change when the evidence supports it, and decisions requested by a date. ### 8 · Run the desk - **Recruiting Metrics** — use when: You are judged on roles that were dead on arrival. Output: Time to fill and time to hire, source of hire, offer acceptance, pass-through by stage, and requisition ageing with the reason each role is stuck, at requisition and channel level only. - **AI Recruiting Policy** — use when: A vendor promises AI screening and you need a line to defend. Output: Allowed and banned uses, the candidate notice text, the human review steps, and the questions to put to vendors and a qualified adviser. - **Interviewer Training** — use when: Interviewers improvise and the debrief has no evidence. Output: A 45-minute session for hiring managers and panels on the structured interview, note-taking, independent scorecards and questions never to ask, with a one-page cheat sheet. ## Workflows - **Brief the role** (A new role, and a brief nobody has agreed yet.) Intake Meeting → Job Scorecard → Job Description → Job Ad - **Keep candidates informed** (The search is running long and candidates are waiting.) Candidate Communication Plan → Interview Scheduling Messages → Rejection Email → Candidate Experience Survey - **Source and reach out** (You need to go and find people.) Job Scorecard → Boolean Search Strings → Talent Mapping → Sourcing Strategy → Recruiting Email Templates - **Screen and interview** (Every candidate assessed on the same criteria, stage by stage.) Screening Questions → STAR Interview Questions → Structured Interview → Interview Scorecard - **Decide and close** (The interviews are done and a decision is due.) Interview Debrief → Reference Check Questions → Offer Letter → Counter Offer Plan - **Run the agency desk** (You recruit for clients.) Recruiter Business Development → Agency Terms of Business → Candidate Submittal → Client Search Update ## The quality bar - Every skill starts from the job: outcomes and criteria agreed with the hiring manager, not a description copied from last year - Every candidate meets the same questions, the same stages and the same messages - Every output has a fixed shape, a finish line, and a named person who decides next and by when - Messages use only the facts you supply: nothing is scraped, enriched, invented or guessed about a person - Legal and regulatory points — pay transparency, work authorisation, automated decision rules, agency terms — are flagged for a qualified adviser rather than answered - No skill screens, ranks or scores a candidate: scorecards ship blank and carry no total, metrics stop at requisition, stage and channel level, and talent maps stay at company and team level - The red line: Claude writes the search, the questions and the message, never the verdict — a person reads every application and makes every hiring decision ## Questions people ask ### What is in the AI for Recruiting pack? 31 Claude skills that follow one search in eight parts: brief the role, keep candidates in the loop, attract, source, screen and interview, decide and close, the agency desk, and running the desk. Each skill is one named method or one artifact. ### How do I install these Claude skills? In Claude Code, two lines install all 31 at once. In Claude.ai or Claude Desktop, download the zip, open Settings, Skills, Add, and upload the skills you want. Either way it takes about five minutes. ### Do I need Claude Code to use them? No. The skills work in Claude.ai, Claude Desktop and Claude Code. In the app you upload them yourself; in Claude Code one command installs the whole pack. ### Do the skills work with ChatGPT or other assistants? They are built as Claude Skills, and installing them is a Claude feature. Each one is a plain Markdown file, so you can read it and reuse the method elsewhere, but it will not load and trigger by itself outside Claude. ### Will Claude decide who we hire? No. Claude writes the search, the questions and the message, never the verdict. It does not screen, rank or score a candidate. A person reads every application and makes every hiring decision. ### Is the pack free? Yes. It is free to use inside your company and in client work, and not for resale. There is no signup: the download and the install command are on this page. ## About Written by Polar Bear (https://meet-polar-bear.com/). Published on Lantern (https://lantern.meet-polar-bear.com/), which turns your notes, documents and call transcripts into social post drafts you review and publish.