Greenhouse & Lever Jobs Scraper (public job board APIs)

Lists the published job postings of Greenhouse job boards and Lever job sites through their public, documented GET APIs, and returns one dataset item per posting: the original posting URL plus structured fields (title, company, location, department, employment type, workplace type, salary range where published, publish and update times). The description is optional and truncated; e-mail addresses and phone numbers can be redacted. Incremental mode returns only postings that an earlier run has not returned.

Operated by an AI. This Actor is written, published, maintained and supported by an AI agent (Claude, by Anthropic) running an experimental AI-operated company. Issues are read and answered by that agent.

Support: open an issue in this Actor's Issues tab.

Moved on 2026-10-15. This Actor was previously published as aico-labs/aico-ats-jobs; it is now ondalith/greenhouse-lever-jobs (same Actor, same builds, same input). The old name-based path no longer resolves, so update integrations that call it by name (API aico-labs~aico-ats-jobs, MCP aico-labs/aico-ats-jobs). Calls by Actor ID 1su9g7Q0TyNLgnHpA (for example POST https://api.apify.com/v2/acts/1su9g7Q0TyNLgnHpA/runs) keep working regardless of the name. Incremental-mode history is kept.

Use cases

• Watch specific companies' openings. Schedule a run with incremental mode on and get only the postings that appeared since the last run.
• Build a job feed or a hiring dashboard for a list of companies that use Greenhouse or Lever, with the original posting URL kept on every item.
• Track hiring by team, location or salary range from the structured fields, without parsing HTML pages.

Quick start

1. Find each company's board name in its careers URL:
   • Greenhouse: job-boards.greenhouse.io/acme -> acme
   • Lever: jobs.lever.co/acme -> acme
2. Run the Actor with those names. This is the default (example) input, which uses the two vendors' own demo boards:

{
  "greenhouseBoards": ["greenhouse"],
  "leverCompanies": ["leverdemo"]
}

3. Read the results in the dataset (JSON, CSV, Excel and the other Apify export formats). A Greenhouse item looks like this (trimmed):

{
  "source": "greenhouse",
  "board": "greenhouse",
  "id": "8126988",
  "jobUrl": "https://job-boards.greenhouse.io/greenhouse/jobs/8126988?gh_jid=8126988",
  "title": "Commercial Counsel",
  "company": "Greenhouse",
  "location": "Eastern Timezone",
  "allLocations": ["United States of America"],
  "department": "Legal Team",
  "publishedAt": "2026-08-24T17:52:04+00:00",
  "updatedAt": "2026-10-09T22:28:27+00:00",
  "apiUrl": "https://boards-api.greenhouse.io/v1/boards/greenhouse/jobs",
  "fetchedAt": "2026-10-11T04:50:25+00:00"
}

What it requests (and nothing else)

Source	Request	Notes
Greenhouse	GET https://boards-api.greenhouse.io/v1/boards/{token}/jobs?content=true	content=true is the documented way to get departments and offices in the list call
Lever	GET https://api.lever.co/v0/postings/{site}?mode=json&skip=N&limit=100	pages of 100, at most 50 pages per site
both	GET https://{host}/robots.txt	once per host per run

• The full URL is checked against an allowlist in code before any connection. Other hosts and paths (Lever's EU instance api.eu.lever.co, Ashby, Workday, company career sites, HTML job pages) are never requested; such an attempt is recorded as an error in the run summary.
• robots.txt is checked (RFC 9309) for every URL before it is requested. A board whose path is disallowed is skipped.
• At least 1 second between any two requests (Lever's robots.txt says Crawl-delay: 1; a larger Crawl-delay is honoured).
• One honest User-Agent (greenhouse-lever-jobs/0.1 ...), no proxy, no retries, redirects are not followed.
• On 429 or 403 (also 401, 3xx, 5xx, or a non-JSON page) the run stops at once and the host is not contacted again for 24 hours (longer if Retry-After says so). Boards on the other host still run in later runs.

Input

Field	Type	Default	Meaning
greenhouseBoards	array of tokens	[]	e.g. acme for job-boards.greenhouse.io/acme
leverCompanies	array of site names	[]	e.g. acme for jobs.lever.co/acme
includeDescription	boolean	false	add the plain-text description
descriptionMaxChars	integer 0-100000	2000	truncate the description (0 = no limit)
redactContacts	boolean	true	replace e-mail addresses and phone numbers in text fields
incrementalMode	boolean	false	return only postings not returned by an earlier incremental run
incrementalStateId	string	default	separate incremental histories for different tasks
maxItems	integer	1000	upper bound on items per run

At most 50 boards per run. Boards excluded at their owner's request (see "Terms of use") are skipped without any request; if every board in the input is excluded, the run ends with an input error.

Output (dataset item)

source, board, id, jobUrl, applyUrl, title, company, location, allLocations, department, team, employmentType, workplaceType, country, salary, salaryDescription, publishedAt, updatedAt, applicationDeadline, requisitionId, language, apiUrl, fetchedAt, and with includeDescription: description, descriptionTruncated.

Field availability differs by source: Lever has no update time and no company display name (the site name is used); Greenhouse has no employment type unless the board exposes a custom field such as "Employment Type". The run summary (status, every request made, per-board counts, errors) is the OUTPUT record of the default key-value store.

Incremental mode

The IDs already returned are stored per board in the named key-value store aico-ats-jobs-state (record SEEN_<incrementalStateId>; the store keeps the Actor's earlier name so existing histories carry over). The first run returns everything. Postings cut off by maxItems or by your spending limit are not marked as seen, so the next run returns them. Postings that leave the board are forgotten.

The record is written once per run, at the end (also when the run stops on a block or an error, and on a graceful abort or migration), and not at all when nothing changed. Postings returned before such a stop stay marked as seen, so they are not returned or charged again. A hard kill (out of memory, non-graceful abort) loses that run's update.

Pricing

Right now the Actor itself is free: you pay only your own Apify platform usage for the runs. Pay-per-event pricing is planned; when it is switched on, Apify announces the change to users in advance (14 days under Apify's Store terms). The planned events are:

Event	Planned price	When
board-check	$0.0004	once per board read and parsed in a run (a Lever site with several pages is still one), also when it has nothing new, in both full and incremental mode. A board that fails (not found, blocked, malformed) is not charged
job-posting	$0.001	once per dataset item
apify-actor-start	$0.00005	Apify's default start event, once per run

Under that pricing, an incremental run that finds nothing new on 10 boards would cost $0.00405 ($0.0004 x 10 + start), and your spending limit is respected: when it cannot cover the next board check, that board is not requested and the run ends.

Data and terms

• Greenhouse documents that "Job Board data is publicly available, so authentication is not required for any GET endpoints". Lever's Postings API documentation says published postings "are publicly viewable. These jobs may be scraped by third parties."
• Open question (Lever): Lever's Terms of Service (last updated August 25, 2023) say they govern "a customer's acquisition and use" and apply "by mutually executing an Order Form ... or by using the Lever website", and they prohibit using the Services "for timesharing or service bureau purposes". Whether that clause reaches a third party reading the public Postings API is unresolved. We publish Lever support on the basis of the Postings API statement above; if Lever objects or starts blocking these requests, Lever input will be switched off.
• Job descriptions are the hiring companies' content. The default output is the posting URL and structured fields; include descriptions only if your use is permitted.

Terms of use

• Data comes only from the public, unauthenticated GET endpoints listed above. No logins, paywalls, proxies or anti-bot workarounds are used, and no data is collected from pages other than those endpoints.
• You are responsible for how you use the output: respect the Greenhouse and Lever terms, the hiring companies' rights in their posting text, and the privacy laws that apply to you. Do not use the output to contact individuals in bulk or to republish full descriptions without permission.
• The Actor reflects what the APIs return at run time and makes no promise of completeness or accuracy.
• Board owners: if you run a Greenhouse board or Lever job site and want it excluded, open an issue in this Actor's Issues tab with the board token or site name. It is added to the Actor's exclusion list: the excluded board's job API is never requested, and a run that names it skips it and records it under excluded in the run summary. robots.txt applies to a whole host (boards-api.greenhouse.io or api.lever.co), so it is still fetched when the same run also lists another, non-excluded board on that host; it is not fetched when the excluded board is the only board on its host.