---

How it compares

	Job Seek	LinkedIn / Indeed	Roll-your-own (JobSpy, JobFunnel)
Source	Direct from employer career pages	Aggregated from job boards	Whatever you wire up
Postings per role	One canonical entry	1–N copies	Depends
Coverage	4,400+ vetted companies	Millions, mostly duplicates	Bring your own
Latency from posting	Typically hours	Days	Your schedule
Search	Typesense, faceted (seniority, stack, locale, salary)	Keyword + location	None bundled
Self-host	Yes — one repo, MIT	Not possible	DIY
License	MIT code, CC BY-NC 4.0 data	Closed	Per-project

What you get on jseek.co

	Explore. Filter every tracked company by seniority, tech stack, salary, location, and posting language. Free, no account.
	Company pages. Live and historical posting counts, similar-company suggestions, every role linked back to the source.

A built-in application tracker moves saved roles through saved → applied → interviewing → offered/rejected, with stats and an interview log. Free for everyone. Pro ($10/month) unlocks unlimited watchlists with email alerts.

Built by Colophon Group, a small team in Switzerland — so German, French, and Italian aren't an afterthought.

---

Add a company

Open issues labelled company-request are companies waiting to be added — most resolve in five minutes with a coding agent.

ws is a tool for your coding agent, not for you directly. Set up Claude Code, Cursor, or another agent first; the agent installs and runs the CLI itself.

pip install jobseek-crawler-setup

Pick an issue, then hand your agent this prompt:

Run ws task --issue <NUMBER> and follow the printed instructions.

ws walks the agent through fetching the issue, finding the career page, choosing the right monitor and scraper from the 40 available, validating the result, and opening a PR. Choosing the right combination by hand is tedious — getting it wrong silently misses postings, which is why the workflow is structured this way.

No open issues for the company you want? Request it. Anyone can.

Agent environment: git, gh (authenticated), Python 3.12+, web access.

---

Run your own

Clone, point at your Postgres + Redis + Typesense, run the crawler and the Next.js app:

git clone https://github.com/colophon-group/jobseek
cd jobseek

# Crawler (Python 3.12+, uv)
cd apps/crawler
uv sync
cp .env.example .env.local      # DATABASE_URL, REDIS_URL, TYPESENSE_*
uv run crawler sync             # CSV → Postgres + Redis + Typesense
uv run crawler run              # start a worker

# Web app (Node 20+, pnpm)
cd ../web
pnpm install
pnpm db:migrate
pnpm dev                        # http://localhost:3000

Architecture overview: AGENTS.md. Search stack — collections, scoped API keys, deployment: docs/11-typesense.md. Production routines: error review, daily labelling.

---

What's in the repo

apps/crawler/          Python pipeline (asyncio, Playwright fallback)
  src/core/monitors/   Monitor types — Greenhouse, Lever, Workday, …
  src/core/scrapers/   Scrapers — JSON-LD, DOM, sitemap, vendor-specific
  src/redis_queue.py   Claim queue with atomic reservation, requeue, reschedule
  src/exporter.py      CDC: local Postgres → Typesense
  src/labeller/        Daily labelling pipeline (HuggingFace upload)
  src/workspace/       `ws` — agent orchestrator for company onboarding
  data/companies.csv   Source of truth — every tracked company is one row
  data/boards.csv      One row per board (monitor + scraper config)

apps/web/              Next.js 16 + Drizzle + Lingui + Better Auth
  app/[lang]/...       Path-prefix i18n (en / de / fr / it)
  src/db/schema.ts     Drizzle schema for the web-owned Postgres boundary

docs/                  Architecture and operational routines
scripts/               Typesense setup, backfill, IndexNow notifications

---

License

• Code — MIT. Use, modify, redistribute, no warranty.
• Job-posting data — CC BY-NC 4.0. Free for research and non-commercial reuse with attribution. Not "open data" by the strict OKD definition. For commercial licensing, get in touch via [email protected].

---