ATS Jobs Search API: Greenhouse, Lever, Ashby avatar

ATS Jobs Search API: Greenhouse, Lever, Ashby

Pricing

from $2.00 / 1,000 job returneds

Go to Apify Store
ATS Jobs Search API: Greenhouse, Lever, Ashby

ATS Jobs Search API: Greenhouse, Lever, Ashby

Search open jobs by keyword across 19,630 company career sites on Greenhouse, Lever, Ashby, Recruitee and Personio. Filter by location, remote and posting date. Each job is read live from the employer's board, with full description and apply link. No login. $2 per 1,000 jobs.

Pricing

from $2.00 / 1,000 job returneds

Rating

0.0

(0)

Developer

Sanjaya Maharjan

Sanjaya Maharjan

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

10 hours ago

Last modified

Categories

Features

Limited permissions

Share

This ATS jobs search API finds open jobs by keyword across the career sites of 19,630 companies that hire through Greenhouse, Lever, Ashby, Recruitee or Personio, such as Discord, Spotify, Linear or Palantir. Type a job title such as data engineer, add a location, remote or posting-date filter, and get every matching job in one clean format: title, company, location, remote flag, salary where published, full description and apply link. Each job is read live from the employer's own job board during your run, so you never get a job that has closed. No login, API key, browser or proxy is needed.

Not affiliated with or endorsed by Greenhouse, Lever, Ashby, Recruitee or Personio; the Actor uses the public job-board endpoints they publish for their customers' careers pages.

How to search jobs across company career sites

  1. In Job title keywords, enter one or more titles, for example data engineer or account executive.
  2. Optionally narrow the search with Locations (Berlin, United Kingdom), Remote jobs only, Posted within the last N days, Applicant tracking systems or Companies.
  3. Set Maximum jobs and click Start. The matching jobs appear in the Output tab. Export them as JSON, CSV or Excel, or read them with the API.
  4. For a daily feed, turn on Only jobs not returned before and schedule the Actor.

What it does

  • One search, 19,630 company job boards. A shared index of every open job on 19,630 company boards (16,828 of them with open jobs, 395,365 jobs in all, on 18 November 2026) is rebuilt from the vendors' public job-board APIs, so a search takes seconds instead of reading thousands of boards.
  • Live, never stale. Matching jobs are re-read from the company's board during your run: closed jobs are dropped, jobs posted since the index was built are added, and descriptions are current.
  • One schema for five ATSs. Title, company, location, remote flag, department, employment type, salary (where published), posting date, job and apply links, plain-text description, the company's board link and a stable job id.
  • Precise keyword matching. Whole words, not substrings: java does not match "JavaScript". Accents and letter case are ignored, engineer* matches "Engineering", and exclude keywords remove what you do not want.
  • Saved searches. With Only jobs not returned before, each run returns only new matches, so a scheduled run is a new-jobs feed. You never pay twice for the same job.
  • No duplicates. A job is returned once per run, even when a company lists it on two boards.
  • Public job-board data only. No login, no browser, no proxies, and no recruiter or hiring-manager contact data in the output.

Use cases

  • Job boards and aggregators. Fill a niche job board (for example remote data jobs in Europe) with fresh roles straight from employers' own boards, every day.
  • Recruiting and sales intelligence. Find which companies are hiring for a role right now. A new posting is a buying signal for recruiters, staffing agencies and B2B sales teams.
  • AI agents. One call answers "who is hiring a Rust engineer in Berlin?" with small, structured results (see "Use with AI agents" below).
  • Labour-market research. Track openings, salary ranges and remote share by title, location or company.

Coverage

The index is rebuilt from each vendor's public job-board API and checked board by board. Numbers from the refresh of 18 November 2026:

ATSCompany boards indexedBoards with open jobsOpen jobs
Greenhouse6,5005,555191,948
Personio4,0943,54031,860
Ashby3,6923,32060,132
Recruitee3,1792,74846,974
Lever2,1651,66564,451
Total19,63016,828395,365

Companies are found in Common Crawl's public web index and kept only while their board answers on the vendor's API; boards whose ATS-hosted pages are switched off are left out. Workable boards are not in the index.

Input example

{
"keywords": ["data engineer", "analytics engineer"],
"excludeKeywords": ["intern"],
"locations": ["Berlin", "Remote"],
"remoteOnly": false,
"postedWithinDays": 30,
"maxResults": 200,
"includeDescription": true
}
FieldDefaultWhat it does
keywordsemptyJob title must match at least one keyword or phrase (rules below).
excludeKeywordsemptyLeave out jobs whose title matches any of these.
locationsemptyLocation must match at least one of these words or phrases.
remoteOnlyfalseOnly jobs the ATS marks remote.
postedWithinDaysemptyOnly jobs posted in the last N days. Jobs without a posting date are left out when this is set.
atsallOnly these ATSs: greenhouse, lever, ashby, recruitee, personio.
companiesemptyOnly these companies, by name (Discord) or board token (discord).
maxResults100Stop after this many jobs (up to 50,000).
maxPerCompanyemptyAt most this many jobs from any one company, so a few large employers cannot fill the results.
includeDescriptiontrueAdd the full plain-text description.
onlyNewfalseReturn only jobs this saved search has not returned before.
stateStoreNameats-jobs-search-stateKey-value store that remembers returned jobs; use one name per saved search.

Set at least one of keywords, locations, companies, remoteOnly or postedWithinDays. A job is returned only if it passes every filter you set.

Matching rules. Keywords match the job title, ignoring letter case and accents, on whole words: data engineer matches "Senior Data Engineer, Platform" and "Data-Engineer (m/w/d)" but not "Data Engineering Lead". End a word with * to match its stems (engineer*). Several keywords are alternatives; the one that matched is returned in matchedKeyword. Locations follow the same whole-word rule on the job's location text.

Output

One dataset item per job. Companies with the most recently posted matches are read first, so with a low Maximum jobs you get mostly recent jobs; within a company, jobs are newest first. A real item from a test run on 18 November 2026 (description shortened):

{
"id": "ashby:benchling:c97e5668-29ee-4810-b7cc-35331ac123f5",
"ats": "ashby",
"boardToken": "benchling",
"company": "Benchling",
"title": "Data Engineer",
"location": "San Francisco, CA",
"remote": false,
"department": "Engineering",
"employmentType": "Full-time",
"salary": "$154K - $208K",
"postedAt": "2026-11-18T01:41:03.029Z",
"updatedAt": null,
"jobUrl": "https://jobs.ashbyhq.com/benchling/c97e5668-29ee-4810-b7cc-35331ac123f5",
"applyUrl": "https://jobs.ashbyhq.com/benchling/c97e5668-29ee-4810-b7cc-35331ac123f5/application",
"description": "We are rebuilding biotech for the AI era.\n\nWhen a breakthrough is delayed, the world waits. ...",
"boardUrl": "https://jobs.ashbyhq.com/benchling",
"matchedKeyword": "data engineer"
}
FieldNotes
idStable id, ats:boardToken:atsJobId. The same job keeps the same id from run to run.
ats, boardToken, boardUrlWhere the job came from, and the company's public job board.
companyCompany name as the ATS publishes it. Lever, Ashby and Personio publish none in their job feeds, so the name is taken once from the company's public board page.
title, locationAs published. Several locations are joined with ; .
remotetrue for remote jobs, from the ATS's own remote or workplace field. Greenhouse has none, so there it is true when the location mentions "remote". Hybrid is not remote.
department, employmentTypeWhere the ATS publishes them.
salaryThe published pay range, for example CAD 35-38 per-hour-wage (Lever), $240K - $320K (Ashby), EUR up to 6000 per month (Recruitee), EUR 14.50-22.50 per hour (Personio). null when not published.
postedAt, updatedAtISO 8601 dates, where the ATS publishes them.
jobUrl, applyUrlJob page and application page on the employer's board.
descriptionPlain text with line breaks and - bullets, as written by the employer. null when descriptions are off.
matchedKeywordThe keyword that matched (null when you gave no keywords).

Run report

Each run writes an OUTPUT record (the Run report output): when the index was built, how many jobs matched in it, how many company boards were read live and how many failed (with the error), jobs returned, duplicates removed, jobs that closed since the index was built and jobs added since, and whether the run's spending limit was reached. A board that fails is skipped, reported and never charged; the run carries on with the next boards.

Saved searches and scheduling

  1. Create a task with your keywords and filters, turn on Only jobs not returned before, and give it its own state store name, for example rust-jobs-berlin.
  2. Add a schedule, for example once a day.
  3. Each run returns only jobs not returned before. Send them on with a webhook or an integration (Slack, Google Sheets, Make, Zapier, n8n).

Jobs not delivered because a spending limit was reached are not marked as returned, so the next run delivers them. To start over, use a new state store name.

Pricing

Pay per event, with Apify platform usage included:

  • USD 2.00 per 1,000 jobs returned (USD 0.002 per job).
  • Apify's Actor start fee of USD 0.00005 per GB of memory per run (USD 0.05 per 1,000 runs at 1 GB).

You pay only for jobs written to your dataset. Jobs left out by filters, duplicates, jobs already returned by the same saved search, and boards that fail cost nothing. Example: 500 matching jobs cost USD 1.00; a daily saved search that finds 20 new jobs costs USD 0.04 a day. Apify's USD 5 of free monthly usage covers about 2,500 jobs.

Use via API

Run a search and get its jobs in one call (replace YOUR_API_TOKEN):

curl -X POST "https://api.apify.com/v2/acts/scaleturn~ats-jobs-search/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \
-H "Content-Type: application/json" \
-d '{"keywords": ["product designer"], "remoteOnly": true, "maxResults": 50, "includeDescription": false}'

The Apify JavaScript and Python clients, webhooks, schedules and integrations work as with any Actor.

Use with AI agents (MCP)

AI agents can call this Actor as a tool through the Apify MCP server. Add the server to your MCP client (Claude Desktop, Cursor, VS Code and others) and give it this Actor:

{
"mcpServers": {
"apify": {
"url": "https://mcp.apify.com?tools=scaleturn/ats-jobs-search",
"headers": { "Authorization": "Bearer YOUR_API_TOKEN" }
}
}
}

Then ask in plain language, for example: "Find remote staff data engineer jobs posted this week at companies on Greenhouse or Ashby, and list title, company and salary." The agent fills in keywords, remoteOnly, postedWithinDays and ats. Tips for agents:

  • Keep maxResults small (10 to 50) and set includeDescription: false for overviews; fetch descriptions only for the jobs you will read.
  • Use companies to answer "is this company hiring for X?" in one call.
  • The output is flat JSON with a stable id, so results can be cited and compared across calls.

Speed

Measured on 18 November 2026 over the full index (19,630 boards, 395,365 jobs) from a single machine; your speed on the platform will differ. A search for data engineer, product designer and account executive with Maximum jobs 500 and Maximum jobs per company 5 matched 12,007 jobs on 3,419 boards, read 136 boards live and returned 500 jobs from 135 companies in 50 seconds, at a peak of 511 MB of memory. On the Apify platform at 1 GB the same search took 108 seconds. Smaller searches finish in seconds.

Limits

  • Five ATSs. Workday, iCIMS, Taleo, SmartRecruiters, Teamtailor, BambooHR and Workable boards are not in the index; neither are LinkedIn, Indeed or other job sites.
  • Companies come from a public web index. A company whose board link has never appeared in Common Crawl is not in the index. The index is rebuilt regularly; new boards are added as new crawls are published.
  • Index rows are refreshed, not live. Matching runs on the latest index, then each matching board is read live. A job posted after the index was built is found only when its board also holds an indexed match.
  • Keyword matching uses the job title. Descriptions are not searched.
  • Recruitee has announced that its Careers Site API will require a token from 10 February 2027; Recruitee boards may drop out of the index then.
  • Personio companies are included only if they have switched on Personio's public job feed. Descriptions come in the language the company published them in, and the remote flag is read from the location text, as Personio's feed has no remote field.
  • Lever asks automated readers to wait one second between requests, which the Actor does; searches that match many Lever boards take longer.
  • Boards with hosted job pages switched off are left out. Some companies keep their ATS job feed but turn off the ATS-hosted job pages and show jobs only on their own site; their job links on the ATS would not open, so those boards are not in the index.
  • Jobs without a description are left out when descriptions are on. A job whose employer published no description text is not output (and not charged); the run report counts them in skippedNoDescription. Turn descriptions off to get them.
  • Data is as published. Fields the employer leaves empty are null. Descriptions may include contact details the employer chose to publish.

FAQ

How is this different from a LinkedIn or Indeed jobs scraper? It reads each employer's own job board, the original posting, not a job site's copy. Jobs appear as soon as the employer posts them, closed jobs never appear, there are no promoted duplicates, and nothing needs a login, cookies or proxies.

How is it different from scaleturn's Greenhouse, Lever, Ashby & Workable Jobs Scraper? That Actor reads the companies you list. This one searches all 19,630 indexed companies by keyword.

Is it legal to use this data? The Actor reads public job-board endpoints that the ATS vendors publish so that their customers' open jobs can be shown on careers sites: Greenhouse documents its Job Board API as public, Lever's Postings API documentation says published jobs "may be scraped by third parties", Ashby documents its public job-posting API, Recruitee's Careers Site API answers without a token today (Recruitee has announced a token requirement from 10 February 2027), and Personio documents its XML job feed as needing no credentials. It respects each site's robots.txt and rate limits, reads job postings only and outputs no personal data about recruiters or hiring managers. How you use the data is your responsibility; this is not legal advice.

Why is a company missing? It uses an ATS not covered here, or its board has not appeared in the public web index yet. The companion Actor above reads any Greenhouse, Lever, Ashby or Workable board you name.

How it is tested

  • 55 unit and end-to-end tests on recorded responses from each ATS: keyword and location matching, remote and date filters, de-duplication, empty results, a failing board, output order with a slow board, the per-company cap, jobs without descriptions, saved searches, restarts and the spending limit (end-to-end with the Apify SDK's pay-per-event simulation).
  • The refreshes of 18 November 2026 checked 30,928 candidate boards on the five ATSs; 19,630 answered with a valid job list and are in the index, 166 Lever, Ashby and Personio boards whose hosted pages answered 404 were left out, and 9 (0.03%) could not be checked for a reason other than "not a board".
  • A three-title search on the full index returned 500 jobs from 135 companies, each with a title, company, job link and description, all ids unique, with no board failing. Searches limited to Lever and to Personio returned 20 jobs each, every one with a description and posting date. A sample of 35 job links across the five ATSs all opened (HTTP 200).

Politeness

At most 4 requests at a time per ATS, one request a second to Lever (its robots.txt crawl delay) and to Personio (which rate-limits faster readers), timeouts, and retries with growing, randomised pauses on rate limits (HTTP 429) and server errors. If an ATS starts refusing requests, the Actor stops calling it for the rest of the run instead of pushing through. A search reads only the boards that hold matches. User agent: ats-jobs-search/1.0.