Civil Service Jobs Search Scraper

This Apify Actor extracts current public vacancies from the official Civil Service Jobs portal locally or on Apify Cloud. It uses a normal headless Chrome session, reads the official search results, and enriches each row from its stable jobs.cgi?jcode=... detail page. If the site displays its human-verification quick-check or returns HTTP 401, 403, 429, or 451, the Actor stops without interacting with or retrying the barrier.

Modes

• search: one bounded keyword/location search.
• searchMultiple: several keyword/location searches with stable job-code deduplication.
• single: one jobUrl, url, or numeric jobId.
• multiple: jobUrls, urls, or jobIds.
• startUrls: official search or stable detail URLs.

Every detail row is identity-checked against its public reference/job code and is buffered until the required employer, location, and rich description fields are complete. Published values are omitted when absent; no placeholder, fabricated application URL, diagnostic dataset row, proxy, fingerprint spoofing, or CAPTCHA/WAF bypass is used. Run summaries separate platform status (SUCCEEDED or FAILED) from data resultStatus (COMPLETE, LIMITED, SKIPPED, DEFERRED, NO_DATA, or FAILED). Confirmed source barriers discard staged jobs and report SUCCEEDED/SKIPPED; generic or unverified empty/parser/transport results are SUCCEEDED/DEFERRED; a later non-block failure preserves the successfully stored prefix as SUCCEEDED/LIMITED; fatal zero-row runtime failures use Actor.fail() and report FAILED/FAILED.

Local commands

npm install --no-audit --no-fund
npm test
npm run check
npx --yes apify validate-schema .actor/input_schema.json
npx --yes apify validate-schema .actor/dataset_schema.json
npx --yes apify validate-schema .actor/key_value_store_schema.json
apify run --input-file INPUT.json
npm run validate

The Actor writes jobs only to the default dataset. RUN_SUMMARY, RUN_DIAGNOSTICS, RUN_SKIPS, RUN_HEALTH, RUN_METADATA, and RUN_REQUESTS are KVS-only run state. A handled source barrier is a successful Actor process with a skipped data outcome and zero job rows. Fatal zero-row failures are marked failed; an empty-results outcome is NO_DATA only when every requested search seed visibly reports no matches on a successful page. RUN_SUMMARY records noDataConfirmed, the expected search count, and matching evidence receipts. Local location-filter misses, missing cards, and unparseable pages are DEFERRED, not NO_DATA. Before any local Actor run, follow the repository storage-safety guidance and use isolated storage; this lifecycle-only update did not execute the Actor.

Input examples

{
  "mode": "search",
  "query": "policy advisor",
  "location": "London",
  "maxItems": 3,
  "maxCandidates": 12,
  "maxPages": 1,
  "maxRequests": 24
}

Stable detail URLs look like:

https://www.civilservicejobs.service.gov.uk/csr/jobs.cgi?jcode=2010702

The Apply now control on current adverts is an official login form rather than a final external destination. The output records that source-backed form action and explicitly marks finalDestinationPublished: false; it never invents a destination.