Multi-ATS Jobs Scraper (Greenhouse, Lever, Ashby)

Multi-ATS Jobs Scraper: every open role on a Greenhouse, Lever or Ashby board, in one schema

Give it a list of company board slugs and it returns every open posting on each one, with the full description text, the department, the apply link and the posting date. No keyword search, no login, no API key.

The three systems agree on almost no field names, so the rows are normalised into one shape. That also means some fields are simply absent on one system and present on another, and the table below says which.

🗂️ What Multi-ATS Jobs Scraper does

Greenhouse, Lever and Ashby each publish a company's whole careers board as open JSON. This reads those feeds directly, so what you get is the company's own posting data rather than a scrape of a rendered careers page.

You name the boards as ats:token. The token is the slug in the company's careers URL: job-boards.greenhouse.io/<slug> (the older boards.greenhouse.io/<slug> works too), jobs.lever.co/<slug> or jobs.ashbyhq.com/<slug>. Several boards run a few at a time, and the same posting reachable from two boards is delivered once.

Every row carries the full description as plain text, not a snippet. That is the part most job feeds make you fetch one page at a time.

📥 What you give it

{
  "companies": ["greenhouse:stripe", "lever:spotify", "ashby:ramp"],
  "maxItems": 0
}

One malformed entry stops the whole run. A spec that is not ats:token, or names a system other than the three, ends the run with a single BAD_INPUT row before any board is read. Check the list before a large run.

locations and remoteOnly work on the whole board each system sends, so they cost no extra reading, and jobs they leave out are never charged. A strict filter can return fewer jobs than maxItems, or none, in which case the board gets a free NO_RESULTS row saying so.

📤 What you get back

A real row from a recent run, with the description cut short:

{
  "ok": true,
  "ats": "greenhouse",
  "company": "stripe",
  "title": "Abuse Investigator",
  "location": "Dublin",
  "department": "8611 Security Analytics",
  "employmentType": null,
  "url": "https://stripe.com/jobs/search?gh_jid=8172487",
  "postedAt": "2026-09-12T13:25:25-04:00",
  "description": "Who we are\n\nAbout Stripe\n\nStripe is a financial infrastructure platform for businesses...",
  "salary": null
}

🧾 Reading the output

Two kinds of row can land in your dataset.

Each board reports for itself, so a run over ten boards can deliver jobs for nine and one diagnostic row for the tenth. A diagnostic row is never charged.

A wrong slug is by far the commonest problem, and it shows up as NOT_FOUND. The default table view does not show errorCode, so switch the dataset to All fields when a board comes back empty.

▶️ How to run it

1. Open Multi-ATS Jobs Scraper and click Try for free. 2. Open a company's careers page and copy the slug out of its URL. 3. Add one line per board to Companies (ats:token), like greenhouse:airbnb. 4. Leave Max jobs per company at 0 for the whole board, then click Start. 5. Download the dataset as JSON, CSV or Excel, or read it from the Apify API.

💰 How much does it cost?

$3.00 per 1,000 jobs, which is $0.003 each. Flat on every Apify plan, no volume tiers.

You pay per posting delivered. Duplicate postings reachable from two boards are charged once, jobs your locations or remoteOnly filter leaves out are not charged, diagnostic rows are not charged, and if every board comes back empty, no rows are charged.

💡 What people use it for

• Watching a list of competitors' boards on a schedule and spotting the day a new team opens.
• Building a startup jobs feed without writing three different clients for three different systems.
• Pulling full descriptions for keyword analysis, which needs the whole text rather than a snippet.
• Checking headcount direction by department, using department and the row count per company.

🚧 What it does not do

• No keyword search. You name boards, not roles. Filter the rows afterwards.
• Remote on Greenhouse is read from the location text. Greenhouse has no remote field, so a

remote job whose location never says "remote" is missed by remoteOnly.

• Only three systems. Greenhouse, Lever and Ashby. A company on anything else will not resolve.
• You have to know the slug. There is no company-name lookup; copy it from the careers URL.
• Greenhouse has no salary and no employment type, and its date is the last edit rather than the

original posting date. That is what the board publishes.

• Open postings only. Nothing closed, filled or draft, and no history of what used to be there.
• Many large boards at once is the risky run. Rows are held together before they are written,

so a long list of very large companies is the case most likely to run out of room. Split it.

• No applicant-side anything. No salary benchmarks, no company profile, no reviews.

🧭 Which jobs scraper do you need?

❓ Questions people ask

Where do I find the board token? In the careers URL. job-boards.greenhouse.io/stripe gives greenhouse:stripe, jobs.lever.co/spotify gives lever:spotify, jobs.ashbyhq.com/ramp gives ashby:ramp.

Does maxItems cap the run or each company? Each company. Five boards with maxItems at 20 is up to 100 rows.

Why is salary empty on so many rows? Greenhouse does not publish one at all. Lever and Ashby do when the company filled it in, which many do not.

How many boards can I put in one run? There is no fixed cap, but keep an eye on the size of the companies. A dozen mid-sized boards is comfortable; a dozen of the largest is not.

Can I schedule it? Yes. Daily on a fixed list works well, and url is a stable key for spotting which postings are new.

Is this legal? These are public job feeds the companies publish themselves. Postings can still carry personal data, which GDPR and similar laws cover, so have a reason for collecting it. Apify's write-up on scraping and the law is a good starting point, and we are not lawyers.

🆘 If something breaks

Open the Issues tab on the actor page. Send the run ID and the exact ats:token line that went wrong. The errorCode and hint on the diagnostic row usually name the problem on their own.