Nexus Tools / Data guides

Company job postings as JSON from the public Greenhouse, Lever and Ashby APIs: endpoints, field differences, a Python example and a free sample

How to read the open jobs of a company that hires through Greenhouse, Lever or Ashby straight from the public JSON endpoints those systems offer for career pages, how to tell which one a company uses, how the fields differ (salary, remote, dates), where the limits are, and when an Actor saves you the code.

Updated: November 2026. Everything below was tested live on 6 November 2026.

The short version

• All three systems publish the jobs of a company as plain JSON over HTTPS, with no account, key or login: boards-api.greenhouse.io/v1/boards/{token}/jobs, api.lever.co/v0/postings/{slug} and api.ashbyhq.com/posting-api/job-board/{name}. They exist so that companies can show their jobs on their own site, and each returns the whole board in one response.
• The hard part is not the request, it is the differences: salary only appears if you pass the right flag (Greenhouse pay_transparency=true, Ashby includeCompensation=true), dates come in three formats, and "remote" is a free-text location, a workplace type or a boolean depending on the system.
• The one rate rule we found in writing is Lever's: its robots.txt says Crawl-delay: 1. Boards are big (Anthropic's Greenhouse board was 9.9 MB in one response), so fetch one company at a time and keep what you need.
• This page gives the endpoints with curl, one Python script (about 90 lines, tested today) that handles all three and writes a CSV, a field comparison, a normalized schema, and a free 20-job sample.

1. Which system does a company use?

Start from the careers URL. A hosted board has a recognizable host, and the first path segment is the id you need:

Many companies hide the board behind their own domain. What we saw on 6 November 2026:

• boards.greenhouse.io/airbnb answers HTTP 200 after redirecting to careers.airbnb.com/positions/. The token still works in the API. Job links on that site carry ?gh_jid=8306554, and the HTML of a job page includes a script boards.greenhouse.io/embed/job_board/js?for=airbnb: the for= value is the token. A gh_jid parameter in a job URL is a good Greenhouse hint.
• The static HTML of discord.com/careers contains JavaScript that calls api.greenhouse.io/v1/boards/${board}/jobs?content=true for a list of boards.
• The static HTML of palantir.com/careers/ links to jobs.lever.co/palantir, and those of notion.com/careers and linear.app/careers link to jobs.ashbyhq.com/notion and jobs.ashbyhq.com/Linear.
• Some pages give no hint in the HTML you fetch: stripe.com/jobs and spotifyjobs.com showed nothing. Stripe's token stripe works in the Greenhouse API (766 jobs) and Spotify's slug spotify in Lever's (85 postings), but we only know that by trying. When the HTML is silent, open the browser developer tools on the careers page and look at the network requests for one of the hosts above.
• Guessing is cheap and safe: an unknown id returns HTTP 404 on all three (Greenhouse {"status":404,"error":"Job not found"}, Lever {"ok":false,"error":"Document not found"}, Ashby the plain text Not Found). Ids are not always the brand name, so a 404 does not prove the company is on another system.
• Lever also has an EU region (api.eu.lever.co). We did not find a real EU account to test, so we cannot tell you how those slugs behave.

2. The three endpoints, with curl

Greenhouse Job Board API

Documentation: docs.greenhouse.io/job-board.html. It states that Job Board data is public and that no authentication is needed for any GET endpoint (only the application-submission POST needs a key).

# list: id, title, location, dates, URL (155,048 bytes for Airbnb's 202 jobs)
curl -s "https://boards-api.greenhouse.io/v1/boards/airbnb/jobs"

# the same list plus description HTML, departments and offices (2,909,857 bytes)
curl -s "https://boards-api.greenhouse.io/v1/boards/airbnb/jobs?content=true"

# salary ranges: add pay_transparency=true
curl -s "https://boards-api.greenhouse.io/v1/boards/discord/jobs?content=true&pay_transparency=true"

# one job
curl -s "https://boards-api.greenhouse.io/v1/boards/discord/jobs/{id}?pay_transparency=true"

• The response is {"jobs": [...], "meta": {"total": 202}}: no paging, everything in one response (Anthropic: 673 jobs, 9,857,703 bytes with content=true).
• departments and offices only appear with content=true. The content field is HTML escaped as entities (it starts with <div): unescape it before use.
• Salary. Without pay_transparency=true we found no pay_input_ranges on any board. With it, Discord returned a range on 26 of 26 jobs and Anthropic on 584 of 673; Stripe on 0 of 766, because Stripe publishes none. Each range is {"min_cents": 20000000, "max_cents": 22500000, "currency_type": "USD", "title": "Base Salary", "blurb": "..."}. The documentation lists this parameter on the single-job endpoint; it also works on the list endpoint, as the Discord and Anthropic numbers show.
• Dates: first_published and updated_at, ISO 8601 with the company's UTC offset (2026-10-28T10:52:05-04:00).
• Remote: there is no field. location.name is free text ("San Francisco Bay Area or Remote (U.S.)"), and some boards add a custom metadata entry, e.g. Airbnb's "Workplace Type": "Remote".

Lever Postings API

Documentation: github.com/lever/postings-api.

# every published posting of a company (Palantir: 311 postings, 6,089,826 bytes)
curl -s "https://api.lever.co/v0/postings/palantir?mode=json"

# paging and filters: limit, skip, and team / department / location / commitment
curl -s "https://api.lever.co/v0/postings/outreach?mode=json&limit=2&skip=1"

• The response is a JSON array. limit and skip work (we got 2 items back for limit=2&skip=1); without them you get everything.
• Description as description (HTML) and descriptionPlain, plus lists[] (responsibilities, requirements) and additional/additionalPlain.
• Date: createdAt in epoch milliseconds (UTC). There is no update date.
• categories holds commitment ("Full-time"), department, team, location and allLocations. workplaceType was remote, hybrid or onsite in every posting we looked at (Spotify: 28 remote, 45 hybrid, 12 onsite of 85).
• Salary: salaryRange {"min": 123000, "max": 168000, "currency": "USD", "interval": "per-year-salary"} plus salaryDescriptionPlain, only on postings where the company published one. Zoox: 216 of 238 postings; Outreach: 12 of 22; Palantir: 0 of 311; Spotify: 0 of 85. interval is a Lever string such as per-year-salary; Zoox also has an hourly posting (the Actor maps it to period hour, 43.25 to 52.5 USD).
• api.lever.co/robots.txt says User-agent: *, Allow: /, Crawl-delay: 1. Honor it: at most one request per second, one at a time.

Ashby public job posting API

# every listed job, with compensation (Ramp: 168 jobs, 3,024,560 bytes)
curl -s "https://api.ashbyhq.com/posting-api/job-board/ramp?includeCompensation=true"

• The response is {"apiVersion": ..., "jobs": [...]}, everything in one response. Without includeCompensation=true the compensation key was absent from all 168 Ramp jobs; with it, present on all 168.
• Compensation is a structure: compensationTierSummary ("$211.4K – $290.6K • Offers Equity"), scrapeableCompensationSalarySummary ("$211.4K - $290.6K") and summaryComponents[] with compensationType (Salary, EquityPercentage...), interval ("1 YEAR"), currencyCode, minValue, maxValue. The summary was non-empty on 157 of Ramp's 168 jobs.
• Date: publishedAt, ISO 8601 UTC with milliseconds. No update date.
• Remote: isRemote (boolean) and workplaceType (Remote, Hybrid, OnSite). They do not mean the same thing: of Ramp's 168 jobs, 134 were Hybrid with isRemote: true, 17 Remote with true and 17 OnSite with false. Trust workplaceType when you need a strict remote filter.
• Others: employmentType (FullTime, Intern, Temporary), department, team, location plus secondaryLocations[] with a postal address, jobUrl, applyUrl, isListed. Titles can carry a leading space (one Ramp title did), so strip them.

3. The three side by side

A note on Ashby and robots.txt: jobs.ashbyhq.com/robots.txt disallows /api/ on the careers site host, but the documented posting API lives on api.ashbyhq.com, whose robots.txt answers 401 Unauthorized (no rules). We use the second host only.

4. A normalized schema

If you combine companies from several systems you need one record shape. This is the one the Actors below output (schema version 1.0), with the source field each value comes from:

Two rules keep it honest: a missing value is null, never a guess, and the word "remote" can make isRemote true but is never used to set it to false.

5. DIY: three systems to one CSV in Python

This script takes careers URLs, detects the system, fetches the board, reduces every job to eight columns and writes jobs.csv. Run on 6 November 2026 on Discord (Greenhouse), Zoox (Lever) and Ashby's Ramp it returned 432 rows (26 + 238 + 168), of which 399 had a salary range: 26 of 26 for Discord, 216 of 238 for Zoox, 157 of 168 for Ramp. It needs pip install requests. Put a real contact in the User-Agent.

import csv
import re
import time
from datetime import datetime, timezone
from urllib.parse import urlparse

import requests

HEADERS = {"User-Agent": "my-script/1.0 (contact: you@example.com)"}


def detect(url):
    """Careers URL -> (ats, slug), or None if it is not a hosted board URL."""
    u = urlparse(url if "//" in url else "https://" + url)
    host, parts = (u.hostname or "").lower(), [p for p in u.path.split("/") if p]
    if host in ("boards.greenhouse.io", "job-boards.greenhouse.io") and parts:
        return "greenhouse", parts[0]
    if host == "jobs.lever.co" and parts:
        return "lever", parts[0]
    if host == "jobs.ashbyhq.com" and parts:
        return "ashby", parts[0]
    return None


def get(url, **params):
    r = requests.get(url, params=params, headers=HEADERS, timeout=60)
    r.raise_for_status()
    return r.json()


def iso_ms(ms):
    return datetime.fromtimestamp(ms / 1000, timezone.utc).isoformat(timespec="seconds")


def job(ats, slug, title, loc, remote, posted, url, salary):
    return {"ats": ats, "company": slug, "title": (title or "").strip(), "location": loc,
            "remote": remote, "posted": posted, "url": url, "salary": salary}


def greenhouse(slug):
    data = get(f"https://boards-api.greenhouse.io/v1/boards/{slug}/jobs", pay_transparency="true")
    for j in data["jobs"]:
        pr = (j.get("pay_input_ranges") or [None])[0]
        sal = f"{pr['min_cents'] / 100:.0f}-{pr['max_cents'] / 100:.0f} {pr['currency_type']}" if pr else None
        loc = (j.get("location") or {}).get("name")
        yield job("greenhouse", slug, j["title"], loc, None, j.get("first_published"), j["absolute_url"], sal)


def lever(slug):
    for j in get(f"https://api.lever.co/v0/postings/{slug}", mode="json"):
        sr, cat = j.get("salaryRange"), j.get("categories") or {}
        sal = f"{sr['min']}-{sr['max']} {sr['currency']} ({sr['interval']})" if sr else None
        remote = j.get("workplaceType") == "remote" if j.get("workplaceType") else None
        yield job("lever", slug, j["text"], cat.get("location"), remote, iso_ms(j["createdAt"]), j["hostedUrl"], sal)


def ashby(slug):
    data = get(f"https://api.ashbyhq.com/posting-api/job-board/{slug}", includeCompensation="true")
    for j in data["jobs"]:
        sal = (j.get("compensation") or {}).get("scrapeableCompensationSalarySummary")
        remote = j["workplaceType"] == "Remote" if j.get("workplaceType") else None
        yield job("ashby", slug, j["title"], j.get("location"), remote, j.get("publishedAt"), j["jobUrl"], sal)


FETCH = {"greenhouse": greenhouse, "lever": lever, "ashby": ashby}


def main(urls):
    rows = []
    for url in urls:
        found = detect(url)
        if not found:
            print("not a Greenhouse/Lever/Ashby board URL:", url)
            continue
        ats, slug = found
        rows += list(FETCH[ats](slug))
        print(ats, slug, len(rows), "jobs so far")
        if ats == "lever":
            time.sleep(1)          # Lever's robots.txt asks for Crawl-delay: 1
    with open("jobs.csv", "w", newline="", encoding="utf-8-sig") as f:
        w = csv.DictWriter(f, fieldnames=list(rows[0]))
        w.writeheader()
        w.writerows(rows)


main(["https://job-boards.greenhouse.io/discord",
      "https://jobs.lever.co/zoox",
      "https://jobs.ashbyhq.com/ramp"])

Things the script does on purpose: it sleeps one second after each Lever company (the Crawl-delay), uses workplaceType and not Ashby's isRemote for the remote column, and leaves remote empty for Greenhouse because the API has no such field. Description fields are not kept; add content=true (Greenhouse) and read descriptionPlain (Lever, Ashby) if you need them.

Limits worth stating: we did not find a published rate limit for Greenhouse or Ashby and did not stress them. Boards are one response of up to several MB, so do not poll a hundred companies in parallel. Postings are what each employer chose to publish; unlisted or internal jobs are not in these feeds. The postings are the employers' content: do not use them to contact individuals, and check the employer's terms for your use.

6. When the Actors save time

For one company the script above is enough, and free. The Apify Actors by Nexus Scrapers run the same public endpoints and add what you would otherwise build:

• Greenhouse Jobs Scraper, Lever Jobs Scraper and Ashby Jobs Scraper: one per system, for when you only need that one.
• Career Site Jobs Scraper: the same schema for Greenhouse, Lever, Ashby and also Workable, Recruitee, Personio and Workday (see our Workday guide).

What they do:

• Companies mode: give career URLs, ats:id entries (lever:palantir) or a company website (the Actor fetches that one page and looks for an embedded board, robots.txt respected; it finds nothing if the page builds its job list with JavaScript).
• Discovery mode: leave the company list empty and it scans the bundled company directory (biggest boards first, up to maxCompanies, default 200) and filters live by title keyword, location, remote flag and posting date.
• The normalized schema above, in JSON, CSV or Excel, with scheduling, webhooks and API/MCP access from Apify.
• Politeness built in: honest User-Agent, robots.txt read per host, Lever limited to one request per second, retries with backoff on 429/5xx, no proxies, no browser.
• No personal data is extracted: no recruiter names or e-mail addresses are copied into fields.

Pricing is pay per result; the Actor page shows the current price.

Since build 0.1.2 the Actors request the Greenhouse feed with pay_transparency=true, so salary is filled for Greenhouse jobs whenever the company publishes a range (the 20-job sample on this page was taken before that change). What they do not do: Ashby's isRemote is passed through as published, so hybrid Ramp jobs show isRemote: true with workplaceType: hybrid; filter on workplaceType. There is no "only jobs new since the last run" mode yet: schedule runs and compare jobIds, or use postedWithinDays.

7. Real sample: 20 normalized jobs

These 20 jobs were produced on 6 November 2026 by running the Actor's own code locally on the live endpoints (no Apify platform involved): the newest postings of six public boards, Airbnb 4 and Discord 3 (Greenhouse), Outreach 4 and Zoox 3 (Lever), Ramp 3 and PostHog 3 (Ashby). They are the first postings in newest-first order, not hand-picked. Description text is cut at 1,500 characters and the HTML description is left out to keep the files small. A scrub pass for e-mail addresses ran over every text field and none remain in the files. Salary shows only where the board published one and the Actor read it (see the note above for Greenhouse).

Download: sample.json (full records, UTF-8) and sample.csv (flattened, lists joined with | , opens in Excel).

"Remote" is the normalized isRemote: yes, no, or - when the system does not say.

8. Terms

These are public endpoints that each vendor offers for building career pages (Greenhouse and Lever document them for that purpose, Ashby calls its endpoint the public job posting API), but the postings belong to the employers and we reviewed no terms of service beyond that. Use job data to display or analyse openings, not to harvest personal data, and read the terms that apply to your use. This is not legal advice.

Links

• Career Site Jobs Scraper on Apify (Greenhouse, Lever, Ashby, Workable, Recruitee, Personio, Workday)
• Greenhouse Jobs Scraper, Lever Jobs Scraper, Ashby Jobs Scraper
• Greenhouse Job Board API documentation
• Lever Postings API documentation
• The companion guide: Workday career sites
• The same kind of guide for EU tenders (TED)