GCC Jobs Aggregator

Searches public Bayt, GulfTalent, and NaukriGulf pages, or extracts one/multiple supported job URLs directly. Records are detail-verified, deduplicated, and normalized with title, company, GCC location, description, salary, posting date, skills, seniority, source-published labels/sections, application evidence, and direct-access provenance.

Modes

• search: generate searches from keywords, locations, and sources, discover job links, then enrich detail pages.
• single: extract exactly one supported public job URL.
• multiple: extract up to 100 supported public job URLs sequentially.

Use maxItems to bound the requested dataset. Requests use fixed ordinary direct HTTPS settings with a bounded timeout, polite spacing, no retries, no proxy, and no fingerprint or identity changes. includeArabic retains optional Arabic labels. Only HTTPS URLs on Bayt, GulfTalent, and NaukriGulf hosts are accepted.

Each source is tracked independently. An HTTP 401/403/429/451 or visible verification/access-denial challenge stops that source and records it as SKIPPED; the actor does not retry, switch routes, or use a proxy. HTTP 407 and unverified transport failures are recorded as DEFERRED. Other available sources continue, and the run summary preserves each source's access evidence.

The dataset contains jobs only. Source failures and no-data explanations are stored in the OUTPUT run summary, preventing diagnostics from being mistaken for jobs. Public search layouts can change or withhold server-rendered results; direct URL modes are the most deterministic fallback.

Run status and result status

The summary's status is the Apify platform outcome and is limited to SUCCEEDED or FAILED. resultStatus separately describes data completeness: COMPLETE, LIMITED, SKIPPED, DEFERRED, NO_DATA, or FAILED. records and emittedRecords count only complete job records whose Dataset push resolved successfully. A 401/403/429/451 response or visible source-authored challenge is SKIPPED only when its HTTPS receipt's requested and final URLs both match the same requested first-party source; then requests stop for that source, its unpersisted rows are discarded, and other requested sources may continue. Redirects are not followed. Cross-source responses, generic HTTP errors, proxy-authentication responses, uncertain transport, parser ambiguity, and unexplained empty results remain DEFERRED; this Actor has no affirmative empty-search detector and does not infer NO_DATA. Diagnostics and failed/block request receipts stay in key-value artifacts; successful per-job response receipts may accompany the corresponding real job row. Diagnostics are never written as Dataset rows. A fatal zero-row failure uses the Actor failure lifecycle; later non-block failures after successful Dataset writes preserve those rows as SUCCEEDED / LIMITED.

{ "mode": "search", "keywords": ["software engineer"], "locations": ["Dubai"], "sources": ["gulftalent"], "maxItems": 20 }

Local QA

npm test
npm run lint
npx apify validate-schema

An Actor run or dataset validator can read/write local storage. Before using either, follow the isolated-storage and preservation guidance in AGENTS.md; this lifecycle audit did not run the Actor or inspect local storage.