Remote Jobs Scraper — RemoteOK, WWR, Himalayas & Arbeitnow

Four remote job boards, one clean spreadsheet. Title, company, salary, tags, location and apply link — with cross-posted duplicates removed automatically.

No API key. No login. Set a filter, press start, export CSV or JSON.

---

Why one Actor instead of four

Every other remote-jobs scraper covers a single board. If you want decent coverage you run four of them, learn four different schemas, and then dedupe by hand — because the same role gets posted to several boards at once.

This does all of that for you:

• One schema across all four sources. salaryMin means the same thing everywhere.
• Cross-board deduplication. Matched on company plus title, keeping the richest copy — the one that actually lists a salary. On a typical run that's ~14 duplicates removed from ~415 jobs.
• Every source carries a source label, so you can still filter back down to one board.

---

What you get

Column	Example
title	Senior Backend Engineer
company	Acme Corp
source	remoteok
url	direct apply link
salaryMin / salaryMax / salaryCurrency	120000 / 160000 / USD
location / locationRestrictions	Anywhere · ["US","EU"]
employmentType / seniority	Full-Time · Senior
tags / categories	["golang","postgres"] · ["Programming"]
excerpt	plain-text summary, HTML stripped
description	full listing
publishedAt / expiresAt	ISO 8601

---

Typical uses

• Job aggregators and newsletters. Schedule it daily, filter by tag, publish the results.
• Recruiting and market research. Who is hiring for what, and at what advertised pay.
• Salary benchmarking. Filter with Only jobs with salary data and analyse the band.
• Lead generation. Companies actively hiring are companies actively spending.
• Feeding a model or an LLM. Consistent field names, one row per job.

---

Example input

{
  "sources": ["remoteok", "weworkremotely", "himalayas", "arbeitnow"],
  "searchQuery": "backend",
  "postedWithinDays": 7,
  "sortBy": "publishedAt",
  "maxItems": 200
}

Salary research only:

{
  "sources": ["remoteok", "himalayas"],
  "withSalaryOnly": true,
  "sortBy": "salary",
  "maxItems": 500
}

A single board, no dedupe:

{ "sources": ["weworkremotely"], "deduplicate": false, "maxItems": 100 }

---

Settings that matter

• sources — keep all four for coverage. Arbeitnow skews EU/DACH; the other three are global.
• deduplicate — leave on unless you specifically want to see which boards carry the same role.
• withSalaryOnly — powerful but aggressive. Most boards don't publish pay, so expect this to keep well under a fifth of listings.
• postedWithinDays — the most useful filter for a scheduled run. Set it to 1 and you get only what's new since yesterday.
• maxItems — your cost ceiling. Set it deliberately.

---

Pricing

Pay per event — you pay for jobs returned, not for runtime.

Event	Price
Actor start	$0.002 per run
Job scraped	$0.002 each

1,000 jobs ≈ $2.00. You're charged per unique job after deduplication, so you never pay twice for the same role appearing on two boards.

---

Reliability

Every source is an official API or public RSS feed — RemoteOK's JSON API, We Work Remotely's RSS, the Himalayas jobs API and the Arbeitnow job board API. No HTML parsing, no headless browser, no bot-detection workarounds. Layout changes on those websites don't affect this Actor.

Requests retry with exponential backoff, and one board being down degrades the run instead of failing it — you still get the other three, and the log says which one was missing.

---

Notes

• Data is aggregated from the four boards' own public interfaces and provided as-is. Job listings change and expire constantly; a row is a snapshot at scrapedAt.
• Salary fields are only populated when the board publishes them, which is a minority of listings. null means "not advertised", not "unpaid".
• Deduplication matches on company and normalised title. Two genuinely different roles with identical titles at the same company will collapse into one — turn deduplicate off if that matters to you.

---

Questions or a bug?

Open an issue on the Issues tab with your input JSON and the run ID. That's usually enough to reproduce it immediately.