# HireHive Jobs API > Query jobs from HireHive boards through one API with ?source=hirehive — plus how HireHive's cross-tenant aggregator really behaves. Source: https://hyperjobs.io/ats/hirehive HireHive is recruiting software for small and mid-sized employers and the agencies that staff them, with boards at `.hirehive.com`. It's a small platform, and it does one thing almost no ATS does. ## It has a public cross-tenant aggregator Most ATS give you a per-tenant endpoint and nothing else, which means the board list is your problem and the board list is the hard part. HireHive inverts that. There is **no per-tenant API at all** (`/v1/{slug}/jobboard` returns 404), and instead there is one public endpoint that returns everyone: ```bash curl "https://api.hirehive.com/v1/hirehive/jobboard?top=2000" ``` **1,562 jobs across 256 tenants, in one unauthenticated call.** The `hirehive` in that path is not a tenant slug — it's a public aggregator key. Discovery and listing are the same request, which is the same trick [WeRecruit](/ats/werecruit)'s sitemap pulls and one of only a handful of cross-tenant routes anywhere in this pool. So this platform reverses the usual work: there's no discovery to do, and the whole difficulty is in the two ways the aggregator lies to you. ## `skip` is silently ignored The endpoint takes `top`. It appears to take `skip`. It does not. Proven with `top=10`: `skip=0` and `skip=10` return the **identical 10 ids**. So do `offset`, `page`, `limit` and `$skip`. Every one of them returns HTTP 200 with a perfectly plausible page of results, which happens to be page one again. An adapter that loops on `skip` fetches page one **fifty-two times** and stores **100 of 1,562** jobs. Nothing errors. Nothing 400s. Every request is a 200, the run completes, and you are missing 94% of the platform with no signal that anything went wrong. `data.pages` doesn't save you either: it's computed client-side from the `top` **you** sent, so it will happily report a page count for a thing that only ever serves one page. Don't loop on it. The fix is to stop paginating. Ask for `top=2000`, take the whole platform in one call, and filter per board in memory. ## The detail's date is a rolling SEO date The aggregator rows carry no body, so the description comes from the job's own page (`hosted_url`), which does carry schema.org JSON-LD. Read the whole JSON-LD and you inherit a quiet corruption: **`datePosted` is synthetic**. A job whose list `published_date` is 2026-01-15 reports `datePosted` of 2026-07-01. Six months out. Every old job on the platform reports that same rolling date, because it's tuned for search engines rather than for you. A feed built on it says the entire back catalogue was posted this week. The list's `published_date` is the truth. It's unix **seconds**, not milliseconds. So `postedAt` comes from the list and only the description comes from the detail — the JSON-LD is right about the body and wrong about the date, and you have to want fields from it selectively rather than trust it whole. ## Two smaller ones **`/rss` and `/feed` return 200 with the board's HTML**, not a feed. That's an HTTP 200 and a `Content-Type` you can't lean on, so a feed parser gets a document and finds no items. **The aggregator is a near-complete subset, not a mirror.** `patagonia` shows 18 jobs through the aggregator and 19 on its own board. Small, consistent, and worth knowing before you reconcile counts against a customer's careers page. Where the aggregator is generous: each row carries `company_name` — the real employer, not a slug — plus a logo, a structured `country` that the free-text location label often omits, an experience band and a category. Descriptions came back at 100% on the boards tested (`cornmarket` 13/13, `patagonia` 18/18). ## Where this stands `?source=hirehive` is served today. At 1,562 jobs, HireHive is a small source and it will never be a big one. It's covered because the cost of covering it is one request, and because a platform that hands you every tenant for free deserves to be taken up on it. The keys of [`/v1/meta`](/docs/api-reference/meta)`.sources` are the live vocabulary — the values `?source=` actually accepts, with counts, and the only thing worth coding against. [Sources & coverage](/docs/reference/sources) says the same thing in prose. See [all platforms](/ats) · [quickstart](/docs/quickstart).