AI Job Post Writer and Applicant Screener for Small Business Hiring

AI Prompts to Build This

Copy these build prompts into Claude, Cursor, or your AI coding tool. Create a free account to unlock the full research behind them.

1. 1. Project Scaffold

   Create a Next.js 14 App Router + TypeScript + Tailwind project called "HireBrief" for small-business hiring. Provision Supabase with these tables:
   
   - employers(id, user_id, company_name, slug UNIQUE, plan, stripe_customer_id)
   - jobs(id, employer_id, title, description_md, must_haves JSONB, nice_to_haves JSONB, pay_min, pay_max, schedule, status)
   - applicants(id, job_id, name, email, phone, resume_url, source, applied_at)
   - screening_results(id, applicant_id, fit_tier, summary_text, matched_must_haves JSONB, missing_must_haves JSONB)
   
   Enable row-level security scoped by employer_id everywhere. Add env vars for AFFINDA_API_KEY, ANTHROPIC_API_KEY, RESEND_API_KEY, STRIPE_SECRET_KEY. Install the Vercel AI SDK and set up Stripe Billing with three products: Free, Starter ($29/mo), Growth ($79/mo).

2. 2. Job Post Generator + Must-Have Checklist

   Build a Server Action generateJobPost(brief) that calls Claude with structured JSON output, where brief holds title, mustHaves, niceToHaves, payMin, payMax, schedule, and location.
   
   System prompt: "You write clear, bias-checked small-business job descriptions. Return strict JSON: title, description_md, and must_have_checklist (an array of items with id, label, and weight of required or preferred). Rules: never use gendered or age-coded language, keep description under 350 words, put pay range and schedule near the top since SMB applicants bounce without it, and translate every must-have from the brief into one checklist item with a short, screenable label such as '2+ years line-cook experience' rather than 'experienced'."
   
   Save the returned description_md and must_have_checklist to the jobs table. Render a live preview pane next to the input form so the employer sees the generated post update as they edit their brief.

3. 3. Resume Screening Pipeline

   Build the applicant screening pipeline, triggered when a new row is inserted into applicants.
   
   Step 1 — Parse: POST the resume file to the Affinda API and retrieve structured JSON (work history, education, certifications, skills, total years experience).
   
   Step 2 — Screen: Send the parsed resume plus the job's must_have_checklist to Claude with this system prompt: "You screen job applicants for a small business owner with no HR background. Given a parsed resume and a list of must-have requirements, return strict JSON with fit_tier (strong, possible, or not_fit), matched, missing, and summary. The summary must be 3-4 plain-English sentences a non-HR person can read in 10 seconds — cite specific resume facts, never output a numeric score, and flag anything worth confirming in a call rather than silently rejecting borderline candidates."
   
   Step 3 — Store the result in screening_results and update the applicant's row. Batch new applicants from the last 24 hours into a single Resend digest sent once daily, sorted strong-fit first, each with a one-line summary and a link to the full write-up.