# Changelog of LinkedIn Jobs Search Scraper (`jobsapi/linkedin-jobs-search-scraper`) Actor

- **URL**: https://apify.com/jobsapi/linkedin-jobs-search-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/jobsapi/linkedin-jobs-search-scraper.md

## Changelog

### 2.2.0 - 2026-09-29

#### Changed

- Added a terminal `SKIPPED` access policy based solely on the existing root `AGENTS.md` record of LinkedIn's public-search sign-in modal. No source request or Dataset write is made under this policy.
- Separated platform run `status` (`SUCCEEDED`/`FAILED`) from data `resultStatus`; successful stored rows remain `SUCCEEDED`/`LIMITED` after later non-block failures, and fatal zero-row failures exit non-zero.
- Added offline lifecycle regression tests, strict summary validation, and documentation of the still-unqualified source comparison.

### 2.1.0 - 2026-08-31

#### Changed

- Replaced the unused Playwright container with the smaller Node 24 Actor image.
- Added Apify-safe production logging and truthful local/cloud execution metadata.
- Updated output schemas, validation, and marketplace documentation for cloud deployment.

### 2.0.0 - 2026-08-30

#### Changed

- Replaced the legacy browser/proxy flow with bounded native HTTPS requests to LinkedIn's official public guest search and detail HTML endpoints.
- Added explicit `search`, `searchMultiple`, `single`, `multiple`, and `startUrls` modes with strict official URL and numeric posting-ID validation.
- Added complete-record buffering, detail/canonical identity checks, rich description parsing, source-backed application-link handling, deduplication, retries, request budgets, and local KVS run receipts.
- Tightened input, dataset, output, and key-value-store schemas; added parser tests and a recursive local dataset validator.

#### Safety and limitations

- No proxy, fingerprinting, login, challenge bypass, cloud call, or cloud push is used.
- LinkedIn-published optional values are omitted when absent; no application URL is fabricated.
- Access walls, HTTP errors, and incomplete public details fail closed with structured KVS diagnostics and no diagnostic dataset rows.

### 1.1.0 - 2025-05-21

- Legacy release retained for historical context. It used browser-based crawling and is superseded by 2.0.0.