ATS job board APIs: endpoints, fields and limits, measured

Eleven applicant tracking systems publish a company's open roles where anyone can read them: Ashby, BambooHR, Breezy HR, Greenhouse, JazzHR, Lever, Personio, Recruitee, Rippling, Workable and Workday. For each one this file gives the public endpoint with a curl that works, what a posting holds, how paging works, the limits we measured against the live endpoint, and where it stops. Every reading was taken between 22 September and 27 October 2026 on public, logged-out endpoints. Ten of the sections link to the full page on adderleydata.com, which has the tables and the runs behind the numbers, five systems have a runnable example in examples/, and one more script searches every board at once.

Quick reference

ATS	Public endpoint	Key to read	Paging	Posting text	Pay
Ashby	GET api.ashbyhq.com/posting-api/job-board/{name}	None	None: one response	In the list	Structured, with includeCompensation=true
BambooHR	GET {company}.bamboohr.com/careers/list (undocumented)	None for the list; a key for the documented API	None: one response	Job record only	compensation, the company's own words, in the job record
Breezy HR	GET {company}.breezy.hr/json (undocumented)	None for the feed; a token for the v3 API	None: one response	Posting page only	A sentence, such as "$60,000 – $70,000 / year"
Greenhouse	GET boards-api.greenhouse.io/v1/boards/{token}/jobs	None	None: one response	With content=true	In the text; structured per job with pay_transparency=true
JazzHR	HTML page {company}.applytojob.com/apply	None for the page; a key for the API	None: one page	Posting page only	On some posting pages' structured data
Lever	GET api.lever.co/v0/postings/{site} (EU: api.eu.lever.co)	None	Optional skip and limit	In the list	salaryRange, set on 26% of 2,376 postings read
Personio	GET {company}.jobs.personio.de/xml?language=en (XML)	None	None: one response	In the feed, in the language asked for	salaryInformation, rare
Recruitee	GET {company}.recruitee.com/api/offers/	None until 10 February 2027	None: one response	In the list	Structured salary
Rippling	GET api.rippling.com/platform/api/ats/v2/board/{slug}/jobs	None	pageSize up to 1,000	Job record only	payRangeDetails, in the job record
Workable	GET www.workable.com/api/accounts/{account}?details=true	None for this endpoint	None: one response	With details=true	No field; in the text
Workday	POST {tenant}.{wd}.myworkdayjobs.com/wday/cxs/{tenant}/{site}/jobs	None	20 a page, total capped at 2,000	Posting record only	No field; in the text

Ashby

curl "https://api.ashbyhq.com/posting-api/job-board/ashby?includeCompensation=true"

• The name is the first part of the path in a company's Ashby links, jobs.ashbyhq.com/{board_name} or its /embed. It is not case-sensitive, but the API writes it into every jobUrl exactly as you sent it, so send it in lower case. An unknown name answers HTTP 404 with the plain-text body Not Found.
• A posting holds title, department, team, location, secondaryLocations, address, workplaceType (OnSite, Remote or Hybrid), isRemote, employmentType (FullTime, PartTime, Intern, Contract or Temporary), descriptionHtml, descriptionPlain, publishedAt, jobUrl, applyUrl and isListed, plus compensation with includeCompensation=true. The work arrangement and the employment type are fields, and pay is data rather than text.
• No paging: one request returns every posting with its full text.
• Measured on Ashby's own board: 62 postings on 6 October 2026, in one request. On 30 September it listed 67, every one with a yearly pay range, in seven currencies.
• Where it stops: it answers what is on the board now, not what changed since your last read. Leave out postings with isListed false.

Full page: Ashby job board API, measured · Actor: Ashby Jobs Scraper

BambooHR

curl -H "Accept: application/json" "https://demo.bamboohr.com/careers/list"
curl -H "Accept: application/json" "https://demo.bamboohr.com/careers/1/detail"

• Not a documented API. It is the JSON BambooHR's own careers pages load. BambooHR's documented API authenticates every request, with an API key a user creates inside that company's account or through OAuth, and lists no public job endpoint (checked 27 October 2026). demo is BambooHR's own demo site, with one opening.
• The company name is the subdomain in {company}.bamboohr.com/careers. A name with no account behind it answers HTTP 302 to https://www.bamboohr.com on every path, and an expired account answers 302 to its own /settings/account/expired.php, so read the redirect rather than follow it. An unknown job id answers HTTP 404 with {"type":"not_found","title":"Resource not found.",…}.
• The list answers {"meta": {"totalCount"}, "result": […]}, every opening with the same 10 keys: id, jobOpeningName (the title), departmentId, departmentLabel, employmentStatusLabel (the company's own label, such as "Full Time"), employmentType, location (city, state), atsLocation (country, state, province, city), isRemote and locationType ("0" on-site, "1" remote, "2" hybrid). No date, no pay, no text.
• The job record, /careers/{id}/detail, answers {"meta", "result": {"jobOpening", "formFields"}}. jobOpening has 17 keys, among them datePosted (a date, YYYY-MM-DD), compensation (the company's own words, such as "$70,000 - $100,000 per year" or "$40 per hour"), description as HTML, minimumExperience and jobOpeningShareUrl. No field records an edit or a closing date.
• Measured directly, not through Apify, on two public boards on 27 October 2026: 138 and 13 openings, 153 requests (each list plus every record), all at the first attempt, in 2 minutes 32 seconds one at a time. isRemote was null on all 151 and locationType set on all; atsLocation named a country on 7 of the smaller board's 13 and on none of the larger board's 138. Pay on 117 of the 151, all as written: 114 with a dollar sign and no currency code, 2 naming the code ("150,000 NZD", "$110,000 – $140,000 CAD per year") and one bare "20".
• Where it stops: the list is cheap and thin; the date, the pay and the text cost one request per opening, the pay is free text whose currency is often only a sign, and nothing says which openings changed.

Actor: BambooHR Jobs Scraper · Example: examples/bamboohr.mjs

Breezy HR

curl "https://{company}.breezy.hr/json"

• Two ways in. Breezy's documented API, under api.breezy.hr/v3, needs a token belonging to a user of that company's account (a personal access token starts breezy_pat_), allows 100 requests in each 60 seconds per token, and answers HTTP 400 missingAccessToken without one. The career portal's /json feed needs no token and is not in Breezy's developer documentation (checked 12 October 2026).
• The subdomain is the identifier. An unknown one answers HTTP 404 with an HTML page titled "Career portal not found | Breezy HR"; a posting id the board does not have answers HTTP 302, to the portal's home page.
• A position holds the same 11 keys: id, friendly_id, name, url, published_date, type, location, department, salary, company and locations. salary is a sentence with the currency's sign, not its code. The primary place is the one flagged primary: true in locations, which is not always first. remote_details says which kind of remote: remote, remote-location or hybrid.
• No posting text in the feed. It is on https://{company}.breezy.hr/p/{friendly_id}, whose schema.org JobPosting block also gives the currency's ISO code where pay is stated.
• Measured on a public board of 69 positions on 12 October 2026: the feed in one request; the feed plus every posting page, 70 requests in 24 seconds. Posting pages refused 8 of 78 attempts with HTTP 403 from datacentre addresses; the feed answered at the first attempt.
• Where it stops: the feed is undocumented, so it can change without notice; the text and the currency code cost one request per position; nothing records an edit.

Full page: Breezy HR job board API, measured · Actor: Breezy HR Jobs Scraper · Example: examples/breezy-hr.mjs

Greenhouse

curl "https://boards-api.greenhouse.io/v1/boards/greenhouse/jobs?content=true"

• No key. Greenhouse's documentation: "Job Board data is publicly available, so authentication is not required for any GET endpoints."
• The board token is the first part of the path in boards.greenhouse.io/{token}, job-boards.greenhouse.io/{token}/jobs/{id} and the embed boards.greenhouse.io/embed/job_board?for={token}. An unknown token answers HTTP 404 with {"status":404,"error":"Job not found"}, so a typo never reads as an empty board.
• Endpoints, all under https://boards-api.greenhouse.io: /v1/boards/{token}/jobs (content=true adds the description, departments and offices), /v1/boards/{token}/jobs/{id} (questions=true adds the application questions, pay_transparency=true adds pay_input_ranges), /v1/boards/{token}, and /departments and /offices.
• A job holds id, internal_job_id, title, absolute_url, location.name, requisition_id, first_published and updated_at. The list has no work-arrangement field, and a structured pay range costs one more request per job. metadata holds each company's custom fields, which can be anything.
• Measured on Greenhouse's own board: 19 roles on 6 October 2026 in one request, in 2 seconds. On 30 September it listed 20: a country on all 20, a city on 1, a pay range in the text on 13.

Full page: Greenhouse job board API, measured · Actor: Greenhouse Jobs Scraper

JazzHR

curl "https://{company}.applytojob.com/apply"

• The documented API, under api.resumatorapi.com/v1, takes the key from the company's own JazzHR account as an apikey parameter and returns at most 100 results a call (/page/2 and on for more). Without a key it answers HTTP 401 {"error":"apikey not set"}. It reads the account itself, drafts and applicants included, so it is not a way to read another company's openings.
• The public career page is HTML. There is no JSON or XML feed. It lists every open posting in one response, without paging (412 on the largest of 49 boards read on 25 September): the title, a link of the form https://{company}.applytojob.com/apply/{id}/{Slug}, the location, and the department when set.
• Status codes: an unknown subdomain answers HTTP 302 to JazzHR's page for job seekers; a closed account answers HTTP 200 with a page titled "JazzHR - Inactive Career Page"; a posting no longer on the board answers HTTP 410. Two page layouts are in use (48 of 49 boards had the current one).
• The posting's page adds the employment type in words, the text, and on most pages a schema.org JobPosting block with datePosted, employmentType, the address's locality, region and postcode, and sometimes baseSalary. Its validThrough was exactly 90 days after datePosted on every block read, so it is not a closing date.
• Measured on a public board of 125 postings on 12 October 2026: 126 requests, the list plus one page per posting, all at the first attempt, in 1 minute 14 seconds. A JobPosting block on 118 of the 125 pages; pay on 33, 15 of them in the block.

Full page: JazzHR job board API, measured · Actor: JazzHR Jobs Scraper

Lever

curl "https://api.lever.co/v0/postings/employ?mode=json"

• Two hosts. A board on Lever's EU instance is the same path on api.eu.lever.co. One site name can be two different boards, one on each host, so read the host the company's own links use (jobs.lever.co/{site} or jobs.eu.lever.co/{site}). Lever documents the API in its postings-api README on GitHub, which asks for a key only to submit applications.
• Case counts: employ answers and Employ answers HTTP 404. An unknown site answers HTTP 404 with {"ok":false,"error":"Document not found"}.
• Paging is optional. skip and limit page, but without them the largest board we tried returned all 922 postings in one response (13 MB, 3 seconds). location, commitment, team, department and level filter; group groups.
• A posting holds id, text (the title), categories (location, commitment, team, department, allLocations), country, workplaceType, opening, description and descriptionBody as HTML with plain-text twins ending in Plain, lists, additional, salaryRange (currency, interval, min, max), salaryDescription, hostedUrl and applyUrl.
• What the README leaves out, from 2,376 postings on eight boards on 23 September 2026: createdAt, in epoch milliseconds, is the only date, and nothing records an edit; workplaceType read onsite, without the README's hyphen; salaryRange was set on 26%; descriptionPlain can lag the HTML; no field names the company.
• Measured on Employ's boards (Employ is the company Lever is part of) on 6 October 2026: 4 postings on the global host and 2 different ones on the EU host, two requests, 3 seconds.

Full page: Lever job board API, measured · Actor: Lever Jobs Scraper

Personio

curl "https://personio.jobs.personio.de/xml?language=en"

• personio is Personio's own job page. Personio documents the feed for companies that list their jobs on their own website ("Integration via code" on developer.personio.de), and its help centre says the feed needs no credentials and changes as soon as a job is edited in Personio (both read 26 October 2026). The same board answers on {company}.jobs.personio.com; Personio's own board returned the same bytes on both hosts.
• The company name is the part before .jobs.personio.de in any of its job links; {company}.personio.de, without jobs, is the company's own sign-in. A name with no board answered HTTP 404 with an HTML "Page Not Found" page on both hosts on 26 October 2026; on 25 September the same request was redirected to personio.com, so test for both.
• Send the language. language picks the text. Without it, both boards read answered as they did with language=de; a code Personio does not have, xx, answered 200 with the same positions and no text. A position with no text in the language asked for keeps its other fields and has an empty <jobDescriptions>.
• A position holds, in the feed's order: id, subcompany (the legal entity), office, additionalOffices, department, recruitingCategory, name (the title), jobDescriptions (one <jobDescription> per section: <name> the heading, <value> its HTML in CDATA), employmentType, seniority, schedule, yearsOfExperience, keywords, occupation, occupationCategory and createdAt (ISO 8601 with an offset), with salaryInformation where the company has set it. A position's page is /job/{id} on the board's host.
• No paging: one request returns every position with its text.
• Measured on Personio's own job page (1 position) and a public board of 27 on 26 October 2026: 2 requests, both at the first attempt, in 2 seconds; the 27-position feed was 113,249 bytes in English. Text in English on 19 of the 27, pay on none of them.
• Where it stops: nothing records an edit or a closing date, remote work shows only in the free-text office label, salaryInformation is rare, and some companies add a section to the text naming a recruiter, with an email address and a phone number.

Full page: Personio job board API, measured · Actor: Personio Jobs Scraper · Example: examples/personio.mjs

Recruitee

curl "https://jobs.recruitee.com/api/offers/"

• jobs is Tellent's own careers site (Tellent makes Recruitee). The reference documents two filters, department and tag, and no paging; /api/offers/{id} returns one offer as {"offer": {…}}. An unknown name or id answers HTTP 404 with {"error":"Not Found"}.
• The token deadline. Recruitee's authentication page (read 12 October 2026) gives 10 February 2027 as the date calls without a token stop working, with HTTP 401. The token goes in an X-Careers-Sites-Token header and is created inside the company's own Recruitee account, so reading many companies' boards would need one token from each.
• An offer holds the same 56 keys. The ones a table needs: salary as {"min", "max", "currency", "period"} with the amounts as strings (all null, or "0" and "0", means none stated); remote, hybrid and on_site as separate booleans, more than one of which can be true; city, state_name, postal_code, country_code and country for the first place and locations for every place; published_at, updated_at, created_at and close_at, written as 2026-10-05 13:25:03 UTC (close_at was null on every offer read); description and requirements as HTML; employment_type_code.
• Watch for trial accounts under big brands' names that hold only Recruitee's template offer, "Senior Marketer (Sample)", tagged Sample; and custom careers domains, which do not reliably lead back to a company's Recruitee name.
• Measured on Tellent's board: 13 offers on 12 October 2026, one request, 3 seconds; a yearly pay range in euros on all 13.

Full page: Recruitee job board API, measured · Actor: Recruitee Jobs Scraper · Example: examples/recruitee.mjs

Rippling

curl "https://api.rippling.com/platform/api/ats/v2/board/rippling/jobs?groupJobsByLocation=true&page=0&pageSize=1000"
curl "https://api.rippling.com/platform/api/ats/v1/board/rippling/jobs/{id}"

• Two endpoints, the ones the public board calls. Rippling's developer site has a Job Board API page, but it asks for a Rippling sign-in (checked 6 October 2026). A board lives at ats.rippling.com/{slug}/jobs.
• The list answers {"items", "page", "pageSize", "totalItems", "totalPages"}, each item an id, name (the title), url, department.name, language and locations, every location with a label, a country and its ISO code, a state and its code, a city and a workplaceType of ON_SITE, HYBRID or REMOTE. pageSize goes to 1,000; 5,000 answered HTTP 400. The older v1 list ignores paging and returns one entry per posting and location: 601 entries for 328 postings. The slug is case-sensitive, and an unknown one answers HTTP 404 with {"error_code":"RESOURCE_NOT_FOUND","message":"Job Board not found"}.
• The record has what the list lacks: createdOn (no field records an edit), payRangeDetails (up to three ranges, each with a location label, currency, frequency, rangeStart, rangeEnd and isRemote), employmentType (its label holds the code and its id the words), companyName, department, workLocations and description in two HTML parts.
• Measured on Rippling's own board: 336 postings on 6 October 2026. A full read took 337 requests, the list plus one record per posting, all at the first attempt, in 3 minutes 17 seconds at 90 requests a minute. Pay on 164; 59 of them give two or three ranges, by location tier or office.
• Where it stops: the date and the pay live only in the record, so every full read costs one request per posting, and nothing says which postings changed.

Full page: Rippling job board API, measured · Actor: Rippling Jobs Scraper · Example: examples/rippling.mjs

Workable

curl -L "https://www.workable.com/api/accounts/careers?details=true"

• careers is Workable's own account. Workable documents this endpoint for companies building their own careers page, in the help-centre article Using the Workable API to create a careers page, with /locations and /departments under the same address. It answers HTTP 302 to https://apply.workable.com/api/v1/widget/accounts/{account}?details=true, so follow the redirect. Workable's full API, under {subdomain}.workable.com/spi/v3, needs the account's own token.
• The account is the first part of the path on the careers page, apply.workable.com/{account}/. An unknown one answers HTTP 404 with the plain-text body Not Found. Each job's url, apply.workable.com/j/{shortcode}, does not name the account.
• The rate limit. After about eight requests in forty seconds from one address, apply.workable.com answered HTTP 429 with Cloudflare's one-line page error code: 1015 in place of JSON. The help article gives no limit.
• A job entry holds the same 20 keys, in this order: title, shortcode, code, employment_type, telecommuting, department, url, shortlink, application_url, published_on, created_at, country, city, state, education, experience, function, industry, locations and description. A job open in several places appears once per place with the same shortcode (one board had 70 entries for 27 jobs); locations flags places the company hides from its careers page with hidden; telecommuting false means only "not remote"; there is no pay field.
• Measured on Workable's own board: one job on 6 October 2026, one request, 2 seconds.

Full page: Workable job board API, measured · Actor: Workable Jobs Scraper

Workday

curl -X POST "https://acme.wd5.myworkdayjobs.com/wday/cxs/acme/External/jobs" \
  -H "Content-Type: application/json" \
  -d '{"appliedFacets": {}, "limit": 20, "offset": 0, "searchText": ""}'

• Not a published API. It is the JSON every careers site on myworkdayjobs.com loads in the visitor's browser. For https://acme.wd5.myworkdayjobs.com/External the tenant is acme, the host wd5 and the site External; nothing public maps a company's name to those three, so start from its careers link.
• Limits, measured on 24 September 2026: 20 a page (a limit of 21 or more answers HTTP 400); total only on the first page on most boards (two of thirteen read on 29 October 2026 report it on every page); total stops at 2,000 (a board whose filter counts summed to 2,616 reported 2,000), so a larger board is read filter by filter through appliedFacets; an offset past the end serves the first page again on some boards and an empty page on others, so paging stops at the first page's total; the tenant is case-sensitive and the site is not; a locale segment answers 405, the wrong wd host or an unknown tenant 422, an unknown site 404.
• The list is thin: title, externalPath, locationsText, postedOn (relative, up to "Posted 30+ Days Ago"), remoteType and bulletFields. The record, GET /wday/cxs/{tenant}/{site}{externalPath}, holds jobPostingInfo with startDate (the only absolute date), timeType, location, additionalLocations, jobRequisitionLocation.country.alpha2Code, jobDescription as HTML, jobReqId, remoteType and externalUrl, and hiringOrganization.name. Neither response has a pay field.
• Measured on Workday's own careers site: 362 roles in 24 countries on 1 October 2026 took 381 requests, 19 list pages plus one record per role, all at the first attempt, in 4 minutes 21 seconds at 90 requests a minute. A pay range in the text on 235 of the 362.

Full page: Workday job board API, measured · Actor: Workday Jobs Scraper

Runnable examples

Five of the systems have a script in examples/ that reads one board and prints one JSON object per posting, with a summary line on stderr: Node 22 or later, nothing to install, no key. Each answers a missing board with a one-line message and exit code 1, and makes its per-posting requests one at a time, at most about 85 a minute. Each was run as printed below on 27 October 2026.

Script	Run	Requests	Read on 27 October 2026
bamboohr.mjs	node examples/bamboohr.mjs demo --details 1	1, plus 1 a job with --details	BambooHR's demo site: 1 opening with its date and pay, 2 requests
breezy-hr.mjs	node examples/breezy-hr.mjs breezy	1	Breezy HR's own portal: 3 placeholder positions with pay sentences, 1 request
personio.mjs	node examples/personio.mjs personio --text	1	Personio's own job page: 1 position with 4 text sections, 1 request
recruitee.mjs	node examples/recruitee.mjs jobs	1	Tellent's careers site: 12 offers, each with a pay range, 1 request
rippling.mjs	node examples/rippling.mjs rippling --records 3	1 a page of 1,000, plus 1 a posting with --records	Rippling's own board: 379 postings, 3 with their records, 4 requests

Search every board at once

The sections above read one company's board at a time. To find postings by title and place without a list of companies, search-every-board.mjs runs Adderley Data's Career Site Jobs Scraper in search mode with one request to Apify's API. The search reads Adderley Data's index of the public boards of eight of these systems: Greenhouse, Lever, Ashby, Workday, Recruitee, Personio, Pinpoint and JazzHR. The index is refreshed by its own crawl and holds each posting's list facts, not its text, and no names, emails or phone numbers.

APIFY_TOKEN=... node examples/search-every-board.mjs "data engineer" Germany --max 20

• What it needs: Node 22 or later and nothing to install, plus an Apify account. The script reads your API token from APIFY_TOKEN (Apify Console, Settings, API & Integrations) and sends it in a header, never in the address.
• What it costs: the run is charged to your Apify account at the Actor's price, which on 9 November 2026 was US$2.00 per 1,000 postings saved on Apify's Free and Starter plans and US$0.00005 a run, so the command above costs about US$0.04. --max is the cap.
• What it sends: one POST to https://api.apify.com/v2/acts/adderleydata~career-site-jobs-scraper/run-sync-get-dataset-items with the search as JSON, and the matching rows come back in the same response. The input is the Actor's own: keywords and locations (several of each, separated by commas), postedWithinDays (--days 7), sources (--sources greenhouse,lever) and workArrangement (--remote).
• What it prints: one JSON object per posting, with its title, company, system, board, city, region, country, location as written, work arrangement, posted date, pay as written and link, and a summary line on stderr.
• Measured on 9 November 2026 with the command's input, run through Apify: 142 postings matched, 20 saved, from 13 boards on 4 systems, in 23 seconds. Six of the 20 were Workday postings, whose own records were read for their dates, and every request was answered at the first attempt.
• Where it stops: this call waits for the run, and Apify ends the wait after 300 seconds with HTTP 408 while the run carries on, so the script caps --max at 1,000. For bigger searches, or to save only new and changed postings each day in incremental mode, start the run without waiting and read its dataset when it finishes.

Watching boards over time

Each endpoint answers what one company is advertising now. None of the eleven says which postings changed since the last read, and Lever, Rippling, Workable, Breezy HR, Personio and BambooHR record no edit time at all, so seeing a change means keeping each read and comparing it with the next. Country, city, pay and work arrangement often sit in free text, and every company is its own identifier on its own system. The Actors linked above do that work on Apify, one per system, each returning one row per posting in the same schema and, in incremental mode, returning and charging for only postings that are new, changed, back or gone. The Career Site Jobs Scraper takes company names or careers links across fifteen systems and returns one table.

How these were measured

Logged out, on public endpoints only. The measured runs went through Apify's datacentre addresses at the Actors' default pace, 4 requests at a time and 90 a minute, unless a section says otherwise. Parameters and fields are as each vendor documented them on the date given and as the live endpoint answered; where the two differ, the section says so. Every reading is dated, and the pages on adderleydata.com are the ones kept current.

No personal data. Job postings describe roles, and nothing here is read for people. Where a response can carry someone's details (Greenhouse's custom metadata, Recruitee's mailbox_email, a Personio text section that names a recruiter, the application forms on Recruitee offers and JazzHR posting pages), the Actors linked here never read it, and they redact email addresses and phone numbers inside posting text by default.

Not affiliated with, endorsed by or sponsored by any company named here. Ashby, BambooHR, Breezy HR, Greenhouse, JazzHR, Lever, Personio, Recruitee, Rippling, Workable and Workday are trade marks of their owners, used only to say which API each section describes.

Written by Adderley Data, Melbourne. The text and the scripts in examples/ are licensed CC BY 4.0: reuse them with a link back.