Job Hub

A personal job-search dashboard: pulls new listings from several sources daily, scores each one against your CV with Claude, flags whether the location/remote terms actually work for you, and lets you track status (new → interested → applied → interviewing → rejected/offer) as you work through them. Single-owner, real Supabase auth, deployed on Vercel.

For the full picture of how this app is built — architecture, data model, every feature, known gaps, deployment gotchas — see HANDOFF.md. This file is just setup/day-to-day usage.

Setup

1. Create a Supabase project at supabase.com (free tier).

   • In the SQL Editor, run supabase/schema.sql, then every file in supabase/migrations/ in order (each is a one-time alter table/ insert — see HANDOFF.md §5 for what each one did).
   • In Project Settings > API, copy the Project URL, the service_role key, and the anon/publishable key.
   • On the Authentication > Sign In / Up page (the page-level toggle, not inside a specific provider's settings), turn off "Allow new users to sign up" — this is a single-owner app, not a multi-user one.
   • In Authentication > Users, click "Add user" and create your one account (your email + a password of your choosing). That's your Job Hub login — and also the value for OWNER_EMAIL below, which is a second, code-level check so a session for any other email is rejected too, even if signups ever end up re-enabled by mistake.

2. Install dependencies

   npm install
   

3. Configure environment variables

   cp .env.local.example .env.local
   

   Fill in SUPABASE_URL, SUPABASE_SERVICE_ROLE_KEY, NEXT_PUBLIC_SUPABASE_URL, NEXT_PUBLIC_SUPABASE_ANON_KEY, and OWNER_EMAIL (the email of the user you created above).

   Then sign up for whichever job sources you want (all free, no approval wait): ADZUNA_APP_ID/ADZUNA_APP_KEY at https://developer.adzuna.com/, JOOBLE_API_KEY at https://jooble.org/api/about. Arbeitnow, Remotive, and the company-board sources (Greenhouse/Ashby/SmartRecruiters/Personio) need no key.

   ANTHROPIC_API_KEY powers CV extraction and the two AI analyses (compatibility, location fit) — get one at console.anthropic.com.

4. Run locally

   npm run dev
   

   Open http://localhost:3000 — you'll be redirected to /login. Sign in with the account you created above.

   /demo is a public route with no login required — a frozen sample dataset with edits kept in localStorage only, no Supabase or Anthropic calls. Useful for showing the app off without exposing your real data.

Deploying to Vercel (free tier)

1. Push this repo to GitHub.
2. Import it in Vercel.
3. In Vercel's Project Settings > Environment Variables, add the same variables from .env.local — for each one, enable both Production and Preview (unless you specifically want it scoped narrower), and leave "Git Branch" empty (never commit .env.local itself, it's gitignored).
4. Deploy. /login will gate the live URL the same way it does locally; /demo stays public. vercel.json wires up the daily Cron schedule for each ingestion source automatically.

Day-to-day usage

Upload your CV at /cv (PDF only) — this powers the compatibility scoring. Re-upload any time; every upload is kept, the most recent one is what's used.

Ingestion runs automatically once a day per source (see vercel.json for the schedule). To trigger one manually, sign in in the browser first, then run the fetch from the browser console (it'll carry your session cookie — plain curl can't, since ingestion routes are gated by middleware like everything else):

fetch("/api/ingest/adzuna", { method: "POST" }).then((r) => r.json()).then(console.log)

Response looks like {"jobsFound": 34, "jobsNew": 12}. Vercel Cron itself uses a separate Bearer token (CRON_SECRET), unaffected by any of this — see middleware.ts.

Analyzing a job: new jobs are auto-analyzed at ingestion time if a CV is on file. Use the "Analyze fit" / "Check location" buttons (or "↻ Re-analyze" / "↻ Re-check" once a result exists) to run either one for an individual job on demand — e.g. after uploading a new CV.