Company Career Site Jobs — Greenhouse, Lever, Ashby, Workday

Collect job postings straight from company career sites. This Actor talks to the public APIs of four applicant tracking systems and returns every posting in one shape, so a list of companies on different systems still comes back as a single clean dataset.

What it does

• Four systems, one schema. Greenhouse, Lever, Ashby and Workday postings arrive with the same field names, so nothing downstream has to branch.
• Paste a careers URL. The system is detected from the link. You can also write it out as greenhouse:airbnb, lever:gopuff or ashby:openai.
• Descriptions included. Full text and original HTML for every posting, with HTML entities already decoded.
• Every run is checked. A quality report is written to the key-value store under RUN_SUMMARY: how many postings each company returned, how often each field was actually filled, duplicates, detail requests that were refused, and which companies failed and why. Gaps are reported rather than hidden — if a source stops returning descriptions partway through, the run says so instead of quietly handing you empty columns.
• One bad company does not sink the run. Failures are recorded per company and the rest still finish.
• An empty result is an answer, not an error. A company with no current openings, or filters that match nothing, finish successfully with a warning in RUN_SUMMARY. The run is only marked failed when every company failed or the output itself is malformed.

Input

Field	Type	Notes
companies	array of strings	One per line. A careers URL, or system:company. Workday needs workday:tenant/dataCenter/site, which a Workday careers URL already contains.
includeDescription	boolean	Default true. Turning it off makes Workday runs much faster, since that system needs one extra request per posting, and drops the description fields from every source.
maxJobsPerCompany	integer	0 means no limit.
titleKeywords	array of strings	Keep only postings whose title contains one of these.
locationKeywords	array of strings	Keep only postings matching a location.
remoteOnly	boolean	Keep only postings marked remote.

Example

{
  "companies": [
    "greenhouse:airbnb",
    "lever:gopuff",
    "ashby:openai",
    "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite"
  ],
  "includeDescription": true,
  "titleKeywords": ["engineer", "data"]
}

Output

Every record has these fields. A value that the source does not provide is null — never an empty string, so a gap is always visible.

Field	Description
source	greenhouse, lever, ashby or workday
companyId, companyName	Board identifier and company name
jobId	Identifier within that system
title	Job title
locationRaw, locations	Location as written, and the split list
remote	true, false, or null when the source says nothing
employmentType	Full time, part time, contract, as the source words it
department, team	Where the source separates them
compensation	Where the source publishes it
postedAt, updatedAt	ISO 8601
url, applyUrl	Posting page and application page
descriptionText, descriptionHtml	Full description
scrapedAt	ISO 8601

Sample record

{
  "source": "ashby",
  "companyId": "linear",
  "companyName": "linear",
  "jobId": "d3bc1ced-3ce4-4086-a050-555055dbb1ff",
  "title": "Senior / Staff Fullstack Engineer",
  "locationRaw": "Europe",
  "locations": ["Europe"],
  "remote": true,
  "employmentType": "FullTime",
  "department": "Product",
  "team": "Engineering",
  "compensation": null,
  "postedAt": "2021-04-27T20:13:45.158Z",
  "updatedAt": null,
  "url": "https://jobs.ashbyhq.com/linear/d3bc1ced-3ce4-4086-a050-555055dbb1ff",
  "applyUrl": "https://jobs.ashbyhq.com/linear/d3bc1ced-3ce4-4086-a050-555055dbb1ff/application",
  "descriptionText": "At Linear, we're building the product development system...",
  "descriptionHtml": "<p>At Linear, we're building...</p>",
  "scrapedAt": "2026-09-22T09:14:02.511Z"
}

Finding a company

• Greenhouse — boards.greenhouse.io/<company> or job-boards.greenhouse.io/<company>
• Lever — jobs.lever.co/<company>
• Ashby — jobs.ashbyhq.com/<company>
• Workday — <tenant>.wd<n>.myworkdayjobs.com/<site>

Paste any of those links and the Actor works out the rest. A company whose careers page is not one of these four systems is reported in RUN_SUMMARY rather than failing the run.

Reading the quality report

RUN_SUMMARY in the key-value store holds the numbers behind the run:

Field	Meaning
jobsCollected / rawCollected	Rows returned, and rows found before filters
bySource	How many rows each system contributed
fillRate / fillRateBySource	How often each field was actually filled, overall and per system
detailFailures	Detail requests a system refused, counted rather than guessed
detailStopped	Set when a system refused so persistently that the Actor stopped asking
warnings	Empty results, description gaps, refused detail requests
errors	Companies that failed, with the reason
verdict	ok: false only when every company failed or the output is malformed

Workday publishes descriptions through a separate request per posting and rate-limits them. The Actor retries and slows down, and if the system keeps refusing it stops asking rather than billing you for empty columns — that shows up in detailStopped.

Notes

• Only postings a company publishes on its own careers page are collected. Ashby postings marked unlisted are skipped.
• No personal data is collected: these are job advertisements written by employers.
• These are public endpoints and normally need no proxy.