Multi-ATS Jobs Search Scraper

The current public input and rich dataset are validated for public Greenhouse boards. The historical handler modules for other ATS products remain in the source tree, but they are not reachable from the public input until each platform has a matching source-specific dataset and access review.

Search a board

{
  "atsPlatform": "auto",
  "mode": "search",
  "companyUrl": "https://boards.greenhouse.io/fivetran",
  "searchQuery": "engineer",
  "maxItems": 3,
  "requestTimeoutMs": 15000,
  "deadlineMs": 60000
}

The Actor reads one official public Greenhouse Board API response with content=true, applies the all-word query to published job title, company, location, department, and office names, then emits complete detail records. single and multiple modes select exact job IDs or exact Greenhouse absolute_url values from that same public board response.

Dataset

Records retain Greenhouse job and requisition identifiers; title; employer; location, offices, and departments; employee type and metadata; dates; language; public compliance and AI-disclosure fields; salary only when it appears in source content; the sanitized full description; sectioned jobFacts for company overview, role summary, technologies, responsibilities, qualifications, bonus skills, benefits, work arrangement, and experience; canonical job URL; source payload; and request provenance.

Only source-published values are emitted. The record must contain at least 21 distinct populated source facts. The quality counter excludes IDs and URLs, aliases, duplicated description formats, request receipts, and runtime metadata; individual items in source-published skills, responsibilities, technologies, and benefits lists count as distinct facts.

Run evidence is stored separately from job rows:

• RUN_SUMMARY
• RUN_DIAGNOSTICS
• RUN_REQUESTS
• RUN_HEALTH
• RUN_METADATA

Access behavior

The source request is direct HTTPS with ordinary headers, one attempt, and redirects disabled. An HTTP 401, 403, 429, 451, or visible source-authored access challenge stops the run, discards job rows, and records SKIPPED with the diagnostic and receipt. HTTP 407, unverified transport errors, malformed responses, and un-followed redirects are DEFERRED. A successful search with no matching complete jobs is NO_RESULTS.

The public input rejects proxy, retry, browser, fixture, and other unsupported controls. Descriptions are always retained for complete records. No Cloud push or run is part of local validation.

Local QA

npm ci --ignore-scripts --no-audit --no-fund
npm run check
npm test
npm run lint
apify validate-schema .actor/input_schema.json
apify validate-schema .actor/dataset_schema.json
apify validate-schema .actor/key_value_store_schema.json
apify validate-schema .actor/output_schema.json
$env:APIFY_LOCAL_STORAGE_DIR = 'storage-batch08-run'
apify run --resurrect --input-file ./INPUT.json
npm run validate -- storage-batch08-run

Use a fresh actor-relative APIFY_LOCAL_STORAGE_DIR per run. Do not use --purge or reuse a prior store.