# Changelog of Hiring Signals Monitor: Buying Signals from Job Postings (`humble-echidna/hiring-signals`) Actor - **URL**: https://apify.com/humble-echidna/hiring-signals/changelog.md - **Full Actor documentation**: https://apify.com/humble-echidna/hiring-signals.md ## Changelog Versions follow MAJOR.MINOR.PATCH (`src/version.py`); Apify shows MAJOR.MINOR from `.actor/actor.json`. Every run logs its version and records it in the `RUN_STATS` key-value record. > **About future failures:** this actor reads each company's public job-board feed and home page. Job-board vendors > change their feeds without notice. If a company that used to work starts failing, suspect a format change first: > the run log names the company and says what went wrong, and one broken company never affects the others. ### 1.0.1 (2026-09-26) - **Re-check each home page every (days)** (`stackCheckEveryDays`, default 7): with monitoring on, a home page checked less than that many days ago isn't fetched again; its last stack is reported with `techStackCheckedAt`, and a change shows up at the next check. The home-page check is most of a company's cost, and stacks change slowly, so this keeps daily monitors cheap. Set 1 to check every run. - New field `techStackCheckedAt`. ### 1.0.0 (2026-09-26) First release (deployed for its confirming run, not published). - **Input:** a list of companies (a name, a careers page or a job-board URL, as in Company Career Page Jobs Scraper), each optionally followed by ` | `; signal keywords (skills, tools or role words, matched as whole words in the title and description, or the title only); optional departments and locations; "only signals since my last run" (monitoring); thresholds for hiring surges; an optional list of technologies that count as stack signals. - **One result per company with a signal**, with `signalTypes` (`new_roles_matching`, `hiring_surge`, `stack_added`, `stack_removed`), a one-line `summary`, the new matching roles (title, link, location, department, posting date, which keywords matched and where), open and matching role counts now and at the last run, the home page's tech stack and what it added or dropped. Companies with nothing new are listed in the run's `OUTPUT` record and are not results, so they're never charged. - **Monitoring:** one memory per search (keywords, where they're matched, departments, locations) and per saved task, in the `hiring-signals-memory` store in the user's account. A company's entry is only updated once its result is in the dataset (or it was quiet), so a signal cut by the run's limits comes back next run. Guards against false signals: a failed board isn't evaluated; a role that disappears is remembered for 30 days; a board that suddenly lists nothing or loses half its roles is held until the next run confirms it; a home page that fails, shows a bot check or suddenly shows nothing never reports technologies removed. - Up to 500 companies per run. Charged per company with a signal through Apify's standard `apify-default-dataset-item` event; Max results per run and the maximum cost per run stop the run cleanly (the companies not checked are listed in `OUTPUT.notChecked`). - Shares its code with Company Career Page Jobs Scraper (`mms_ats`) and Website Technology Detector (`mms_stack`), so a fix to a job board or a fingerprint reaches all of them. #### Sources and terms (checked before building; re-read 2026-09-26) Same use as Company Career Page Jobs Scraper, whose terms audit (2026-09-24) cleared these six platforms: each company's public job feed, fetched on the user's behalf for the companies the user lists; no index is built or resold. Re-read for this actor: - **Greenhouse** — docs (docs.greenhouse.io/job-board.html): "Job Board data is publicly available, so authentication is not required for any GET endpoints." greenhouse.com/legal has no terms for website visitors or feed readers; the only automated-access clause found (my.greenhouse.io user agreement) binds job-seeker accounts, not the feed. - **Lever** — docs (github.com/lever/postings-api): "all job postings in the `published` state are publicly viewable. These jobs may be scraped by third parties." Terms (lever.co/legal/terms-of-service): no scraping or automated-access clause; access limits name only Lever's direct competitors. - **Ashby** — docs (developers.ashbyhq.com/docs/public-job-posting-api): "This API allows you to get data for all currently published Job Postings for your organization", no key. Terms (ashbyhq.com/resources/terms) bind customers only; nothing forbids reading the public feed. - **Recruitee** — docs (docs.recruitee.com/reference/intro-to-careers-site-api): "The Recruitee Careers Site API allows to view company's jobs ... from candidate perspective." Terms (recruitee.com/terms) bind the Subscriber; "An End-User that is not the Subscriber does not derive any rights from these Terms"; no scraping clause. - **Personio** — docs (developer.personio.de/docs/retrieving-open-job-positions): "Current open job postings can be retrieved in XML format under myaccount.jobs.personio.de/xml". Terms: cleared in the 2026-09-24 audit; the re-read on 2026-09-26 got HTTP 429 from personio.com (rate limit), so that audit stands. Re-check when the site answers. - **Teamtailor** — docs (support.teamtailor.com RSS guide): the `.rss` feed serves "metadata that can be shared elsewhere". Terms (teamtailor.com/en/terms-and-conditions) bind the customer and its users, not feed readers. - **Company home pages** — one page per company per run, as Website Technology Detector reads them (cleared for it): a normal logged-out visit, robots.txt followed, no protection bypassed, cookie values never output. The domain is one the user typed or the company itself publishes (careers page, job links, career-site host); it is never guessed from a name. All requests go through `mms_common` (honest User-Agent `HumbleEchidnaApify`, robots.txt per RFC 9309, private-network guard, ports 80/443 only). The platforms refused by the ats-jobs audits (Workday, Workable, SmartRecruiters, Breezy, Rippling and 18 more) are refused here with the same reasons.