Greenhouse, Lever & Ashby Job Board Scraper with Change Feed

Enter company job boards on Greenhouse, Lever or Ashby and get every open job as one flat JSON row: title, location, department, published salary, plain-text description and URL, plus isNew / isChanged flags since your last run.

Built for niche job sites, recruiting tools and market researchers who sync the same list of company boards every day. It reads the three platforms' official public job-board APIs with one request per board. No browser, no proxy, no login.

What you get

• One schema for three ATS platforms. The same 26 fields for Greenhouse, Lever and Ashby; a value a platform does not publish is null.
• Change feed. Give the run a history name and each row says whether the job is new or changed. Turn on onlyChanges to store only those.
• Salaries as numbers, only when published. salaryMin, salaryMax, salaryCurrency. Nothing is estimated from the title, and equity is never counted as salary.
• Plain-text descriptions. Greenhouse's escaped HTML is converted to text; e-mail addresses are masked.
• Honest status per board. Every board ends as completed, partial, failed, blocked or budgetLimited in the free RUN_SUMMARY record. A failed board is never reported as "no open jobs".
• Pay per job row. Failed boards are free.

What this is not

• Not a company database or job search engine. You supply the boards. The Actor does not discover which companies use Greenhouse, Lever or Ashby, and keyword only filters the boards you listed.
• Only Greenhouse, Lever and Ashby. No Workday, SmartRecruiters, LinkedIn or Indeed.
• No "job closed" signal. A job missing from a later run is not reported as closed.

Quick start

1. Open a company's careers page and copy the board URL, for example https://boards.greenhouse.io/stripe, https://jobs.lever.co/spotify or https://jobs.ashbyhq.com/Ashby.
2. Paste the URLs into Company job boards and run.
3. For a daily feed, set a History name (for example my-job-feed), turn on Save only new or changed jobs and schedule the Actor.

Input

Only boards is required.

Field	Type	Default	Description
boards	array	required	Up to 100 boards per run. Each item is a board URL, "source:slug", or an object (see below).
maxJobsPerBoard	integer	100	Store at most this many jobs from each board (1 to 1000). A board with more is partial.
keyword	string	none	Keep only jobs whose title, location or department contains this text.
historyStoreName	string	none	Name of a key-value store in your account that remembers saved jobs. Turns on change tracking.
onlyChanges	boolean	false	Store only new or changed jobs. Needs historyStoreName.
maxItems	integer	1000	Maximum rows stored per run (up to 10,000).
maxRequests	integer	300	Maximum HTTP requests. Each board needs one.
maxRunSeconds	integer	600	Time budget. Boards not reached are budgetLimited.
requestIntervalMs	integer	1000	Pause between two requests (minimum 500).

A board object:

Key	Description
source	greenhouse, lever or ashby. Required.
slug	The company part of the board URL (stripe in boards.greenhouse.io/stripe). Required. Ashby slugs are case-sensitive.
company	Optional display name written to the company field.
region	eu for Lever boards on jobs.eu.lever.co.

Example 1: snapshot of three boards

{
  "boards": [
    "https://boards.greenhouse.io/stripe",
    "https://jobs.lever.co/spotify",
    "https://jobs.ashbyhq.com/Ashby"
  ],
  "maxJobsPerBoard": 20
}

Example 2: daily feed of new and changed jobs

{
  "boards": [
    { "source": "greenhouse", "slug": "airbnb" },
    { "source": "greenhouse", "slug": "figma" },
    { "source": "ashby", "slug": "notion", "company": "Notion" },
    { "source": "lever", "slug": "palantir", "company": "Palantir" }
  ],
  "maxJobsPerBoard": 1000,
  "historyStoreName": "my-job-feed",
  "onlyChanges": true
}

Example 3: only data roles

{
  "boards": ["greenhouse:databricks", "ashby:openai"],
  "keyword": "data",
  "maxJobsPerBoard": 200
}

Output

One row per job. All fields are always present; a value the board does not publish is null.

Real rows from a run on 2026-10-02 (descriptions shortened here):

{
  "recordKey": "greenhouse|global|airbnb|8184174",
  "source": "greenhouse",
  "board": "airbnb",
  "company": "Airbnb",
  "jobId": "8184174",
  "title": "Account Manager",
  "location": "London, United Kingdom",
  "country": null,
  "department": "Business Development",
  "employmentType": null,
  "workplaceType": null,
  "salaryMin": 46000,
  "salaryMax": 54000,
  "salaryCurrency": "GBP",
  "salaryInterval": null,
  "salaryText": "United Kingdom Annual Pay Range",
  "description": "Airbnb was born in 2007 when two hosts welcomed three guests to their San Francisco home, and has since grown to over 5 million hosts who have welcomed over 2 b...",
  "publishedAt": "2026-09-09T04:35:19-04:00",
  "updatedAt": "2026-09-29T20:00:54-04:00",
  "url": "https://careers.airbnb.com/positions/8184174?gh_jid=8184174",
  "availability": "listed",
  "schemaVersion": 1,
  "observedAt": "2026-10-02T07:00:16.968Z",
  "isNew": true,
  "isChanged": null,
  "previousObservedAt": null
}

{
  "recordKey": "ashby|global|Ashby|7458d4e9-da2e-47bd-98cb-adfda43d42b2",
  "source": "ashby",
  "board": "Ashby",
  "company": "Ashby",
  "jobId": "7458d4e9-da2e-47bd-98cb-adfda43d42b2",
  "title": "Engineering Manager - EU",
  "location": "Remote - European Union",
  "country": "European Union",
  "department": "Engineering",
  "employmentType": "FullTime",
  "workplaceType": "Remote",
  "salaryMin": 110000,
  "salaryMax": 185000,
  "salaryCurrency": "EUR",
  "salaryInterval": "1 YEAR",
  "salaryText": "€110K - €185K",
  "description": "Hi 👋 I’m Colin https://uk.linkedin.com/in/colinhoweuk, Director of Engineering, Europe. How do you feel about software engineers writing product specs, making p...",
  "publishedAt": "2024-03-04T14:29:08.532+00:00",
  "updatedAt": null,
  "url": "https://jobs.ashbyhq.com/Ashby/7458d4e9-da2e-47bd-98cb-adfda43d42b2",
  "availability": "listed",
  "schemaVersion": 1,
  "observedAt": "2026-10-02T07:00:16.968Z",
  "isNew": true,
  "isChanged": null,
  "previousObservedAt": null
}

Field	Type	Greenhouse	Lever	Ashby	Description
recordKey	string	yes	yes	yes	Stable key: source, region, board, job id. Join runs on it.
source, board	string	yes	yes	yes	Platform and board slug.
company	string	published name	your label or slug	your label or slug	The company you gave wins.
jobId, title, url	string	yes	yes	yes
location	string | null	yes	yes	yes	As published.
country	string | null	null	ISO code	address country
department	string | null	yes (joined with |)	department or team	yes
employmentType	string | null	null	yes	yes	As published, not normalized.
workplaceType	string | null	null	yes	yes	As published (Remote, hybrid, ...).
salaryMin, salaryMax	number | null	single pay range only	when published	single salary component only
salaryCurrency	string | null	yes	yes	yes
salaryInterval	string | null	null	yes	yes	For example 1 YEAR.
salaryText	string | null	range title	salary description	summary text
description	string | null	plain text	plain text	plain text	E-mail addresses become [email].
publishedAt	string | null	first published	created	published	ISO 8601.
updatedAt	string | null	yes	null	null
availability	string	listed	listed	listed	The job was on the board at observedAt.
schemaVersion	integer				1.
observedAt	string				Run observation time, ISO 8601 UTC.
isNew	boolean				Not in the history before (always true without a history name).
isChanged	boolean | null				Any field differs from the last saved observation. null when new.
previousObservedAt	string | null				Time of the last saved observation.

Dataset views: Jobs (core columns), Salaries and Change feed.

Salary coverage we saw

In the 3,970-job test run: Ashby 890 of 1,219 rows had a salary range (73%), Greenhouse 550 of 2,351 (23%), Lever 0 of 400 (the two Lever companies tested publish none). When a Greenhouse job lists several pay ranges (for example one per US pay zone), the salary fields stay null instead of picking one.

Run summary (free)

The key-value record RUN_SUMMARY has one entry per board. Shortened real example from the first test run (14 boards, 50 jobs per board):

{
  "status": "partial",
  "stopReason": "someTargetsIncomplete",
  "targetsRequested": 14,
  "targetsCompleted": 1,
  "targetsPartial": 10,
  "targetsFailed": 3,
  "rowsSaved": 530,
  "requests": 14,
  "targets": [
    { "target": "greenhouse:stripe", "status": "partial", "reason": "Board observation cap; no expiry inference", "observed": 50, "saved": 50, "unchanged": 0, "requests": 1 },
    { "target": "ashby:linear", "status": "completed", "reason": null, "observed": 30, "saved": 30, "unchanged": 0, "requests": 1 },
    { "target": "lever:plaid", "status": "failed", "reason": "HTTP 404", "observed": 0, "saved": 0, "unchanged": 0, "requests": 1 }
  ]
}

Board status	Meaning	Jobs charged
completed	The whole board was read and stored. A board with no open jobs is completed with 0 rows.	The rows stored
partial	The board has more jobs than maxJobsPerBoard (or maxItems was reached).	The rows stored
failed	Unknown board (HTTP 404), response larger than 16 MB, or malformed data.	None
blocked	The API answered HTTP 403 or 429.	None
budgetLimited	Not reached within maxRequests, maxRunSeconds or your maximum cost.	The rows stored before the limit

The run itself fails only when every board failed.

Change feed

• Without historyStoreName every run is an independent snapshot: all rows have isNew: true.
• With it, the Actor keeps a hash of each saved job in a key-value store of that name in your account. Next run, isNew marks jobs not seen before and isChanged marks jobs where any field differs (title, location, salary, description, Greenhouse updatedAt, ...).
• onlyChanges: true stores only new or changed jobs. A quiet day can give zero rows; RUN_SUMMARY.rowsUnchanged shows how many jobs were identical.
• Jobs cut off by maxJobsPerBoard are not in the history. For a full feed set maxJobsPerBoard to 1000.
• Only rows that were stored move the baseline. If a dataset write fails, the history is not advanced and the run fails.
• Use one history name per board list, and do not run two runs with the same name at the same time. Up to 10,000 jobs are remembered per history.

Pricing

Pay per event: $0.001 for each job row stored (event name job), plus the standard Actor start event ($0.00005 per run). That is $1 per 1,000 jobs.

What you run	Job rows	Cost
3 boards, 20 jobs each, once	60	$0.06
1 board with 400 jobs, once	400	$0.40
14 boards, all jobs, first sync	3,970	$3.97
The same 14 boards daily with onlyChanges, if 2% of jobs are new or changed per day	about 80 per day	about $0.08 per day

Not charged: boards that end as failed or blocked; unchanged jobs skipped by onlyChanges; the run summary.

Good to know:

• If you set a maximum cost per run, rows are trimmed to the budget before they are stored, so stored rows, reported rows and charged rows are always the same number. Boards that no longer fit are not fetched.
• maxJobsPerBoard and maxItems cap what a run can cost.

Reliability: what we measured

Cloud runs on 2026-10-02, 256 MB, no proxy:

Run	Boards	Result	Time
14 boards, 50 jobs per board	6 Greenhouse, 3 Lever, 5 Ashby	530 rows. 11 boards read; 1 unknown slug (HTTP 404); 2 boards failed because their response was above the old 8 MB limit	42 s
14 boards, all jobs (limit raised to 16 MB)	6 Greenhouse, 3 Lever, 5 Ashby	3,970 rows, 14 of 14 boards read (one valid board had 0 open jobs), no duplicates	95 s
The same 14 boards again with onlyChanges	6 Greenhouse, 3 Lever, 5 Ashby	0 rows stored, 3,970 reported unchanged, 14 of 14 boards read	56 s
The two largest boards only	OpenAI (Ashby), Databricks (Greenhouse)	1,705 rows, peak memory 120 MB of 256 MB	6 s

The largest boards tested were OpenAI on Ashby (833 jobs, 14.6 MB) and Databricks on Greenhouse (872 jobs, 10.3 MB). No HTTP 403 or 429 was seen. This is a same-day sample of 14 boards.

Limits and honest notes

• Boards above 16 MB fail explicitly. Greenhouse and Ashby return the whole board in one response; the Actor runs in 256 MB. Such a board ends as failed with the reason and is free.
• Lever boards are read in one page of up to maxJobsPerBoard + 1 postings.
• No proxy, no retries. One request per board, at least 500 ms apart. If an API answers 403 or 429 the board is blocked.
• Salary only when the company publishes a single comparable range. Several ranges or equity-only compensation give null.
• employmentType, workplaceType and salaryInterval are passed through as published, so values differ by platform.
• Descriptions are public text. E-mail addresses and Korean mobile numbers are masked; this is not complete anonymization of free text.
• The public APIs are meant for showing a company's jobs. Check each company's terms before republishing descriptions.

Sources: Greenhouse Job Board API, Lever Postings API, Ashby Job Postings API.

Use from the API

curl -X POST "https://api.apify.com/v2/acts/magenta_courser~niche-job-feed-scraper/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"boards":["https://boards.greenhouse.io/stripe"],"maxJobsPerBoard":20}'

Schedule it in Apify Console (Schedules) or call it from n8n, Make or Zapier with the same JSON.

Use with AI agents (MCP)

The Actor works as a tool through the Apify MCP server. One required input (boards, a list of board URLs), flat output fields, numbers as numbers and null for unknown values. A good instruction: "Get the open jobs of these boards with niche-job-feed-scraper and list the ones with salaryMin above 150000 USD."

FAQ

How do I find a company's slug? Open its careers page and look at the job links: boards.greenhouse.io/<slug>, job-boards.greenhouse.io/<slug>, jobs.lever.co/<slug> or jobs.ashbyhq.com/<slug>. You can paste the URL as it is.

Why HTTP 404? The company has no public board with that slug on that platform, or a Lever EU board needs "region": "eu".

Why is salary null? The company does not publish it through the board API, or publishes several ranges.

Can it tell me when a job closes? Not in this version. Compare recordKey lists between full snapshots yourself, and only for boards that ended as completed.

Can I search all companies for a keyword? No. keyword filters the boards you listed.