Greenhouse Jobs Scraper

Extract public jobs from hosted Greenhouse company boards for recruiting research, job feeds, hiring analytics, and vacancy monitoring. Supply company board slugs or hosted board URLs; filter by title keyword, location, and department. Export the default dataset as JSON, CSV, Excel, or via the Apify API.

{"boards":["anthropic"],"keyword":"Engineer","location":"","department":"","maxItems":20,"maxPagesPerBoard":20,"includeDescription":true}

Output

Each unique board/job ID produces one record: jobId, boardToken, title, companyName, location, department, url, updatedAt, publishedAt, requisitionId, descriptionHtml, descriptionText, salaryRanges, source, and scrapedAt. Nullable fields remain null when unavailable. Company names and descriptions require detail fetching. Salary ranges are source data; salaries are never inferred. SUMMARY contains outcome, counts, limits, warnings, request attempts, stop reason, and charged result events. Diagnostics never become job records.

Limits and source coverage

Supports standard hosted job-boards.greenhouse.io, legacy boards.greenhouse.io board URLs (normalized to the current host), and job-boards.eu.greenhouse.io URLs. Individual job URLs and arbitrary websites are rejected. Custom careers-site redirects are unsupported and reported explicitly. No global company discovery or historical/closed jobs. Listings reflect the public board at run time and can change while pagination runs.

The scraper reads embedded page-owned structured data and follows bounded pagination. Defaults: 20 jobs, 20 pages per board, descriptions enabled. Maximum: 20 boards, 5,000 jobs, 100 pages per board; overall request attempts are capped at 250 and runtime at four minutes. These budgets can stop extraction before the requested count. Requests are sequential, with at most three attempts for transient errors. Deduplication state is checkpointed every 20 records and on SDK persistence/lifecycle events. Details that fail leave a valid listing record and a warning. Missing boards/access failures are distinct from truthful no-match results; complete source failure exits unsuccessfully. Verified free-plan production runs deliver at most five jobs; paid and unknown local environments use the requested bound.

Pay per event

All prices include platform usage; no platform-usage pass-through. One synthetic start fee per run and one primary dataset-item event per delivered job. No manual synthetic charging or duplicate value event. The SDK enforces the total charge cap before writes. Live tiered pricing is active and includes platform usage. The Console and authenticated API both verify the following prices:

Event	FREE	BRONZE (2.5%)	SILVER (5%)	GOLD (20%)
Start	$0.005	$0.005	$0.005	$0.005
Job	$0.002	$0.00195	$0.0019	$0.0016

FREE-tier totals: one job $0.007, five $0.015, twenty $0.045. GOLD twenty-job total $0.037. PLATINUM and DIAMOND use the GOLD rates. Start fees apply even to empty or rejected runs if the container starts; no job events for empty results. Set a run charge cap appropriate to your expected volume. At least $0.007 permits a start and one job at the highest tier price. Page/request/runtime limits also bound usage when no jobs match.

Development

Node.js 22, Apify SDK 3, Cheerio; no browser, external provider, or proxy dependency. npm ci, npm test, npm run check, and apify run --purge. The live tier contract is recorded in .actor/pay_per_event.json and .actor/live_pay_per_event.json. Deployment and repeatable private Tasks are managed by scripts/cloud.mjs; cloud evidence is saved in qa-report.json.

Uses publicly available job advertisements only. Respect source terms and applicable law. No authentication, applications, applicant data, or access-control bypass. This independent tool is not affiliated with Greenhouse or the employers. Actor and example Tasks remain private until the owner reviews and publishes them.