For Data Sources
New job boards, ATS feeds, careers pages, and bulk dataset operators can plug into the Job Pool ecosystem through three documented paths. Every contribution is reviewed, attributed, and routed through the canonical schema. The origin stays the system of record — the pool just gives your data a structured cache and a documented contract.
Why join
The pool is a cooperative cache. Your records show up in the canonical API, the dataset, and downstream Job Pool surfaces alongside attribution back to your origin.
Your listings get normalized into job-listing.schema.json and served through /v1/jobs and reviewed CSV snapshots.
Every record carries url, apply_link, and source_business_url back to your origin. Consumers are pointed at your apply flow, not a clone.
Your source is added to /v1/sources with repository, DVC pointer, schema, and freshness signals. Downstream consumers know what to trust.
Responsible aggregators are pointed at the cache first, reducing duplicate crawls against your origin pages and apply flows.
Joining the pool grants no exclusive rights to your data, your audience, or your crawler traffic. You can withdraw the source at any time and the cache stops refreshing.
Submissions land in a staged review queue on jobpool.live/docs/submissions before they show up in the public dataset and API.
Integration Paths
Pick the path that matches how your data is shaped today. You can mix paths over time — for example, expose a feed for incremental updates and ship an initial CSV for backfill.
Best when your job board already exposes a stable, machine-readable feed (RSS, Atom, JSON Feed, or a documented JSON listing endpoint).
https://example.com/jobs.json) that returns the listings you want represented in the pool.JobPoolBot robots product token in your robots.txt and reference the feed in a comment or sitemap.Link header on listing pages: Link: <https://example.com/jobs.json>; rel="alternate"; type="application/json"./v1/jobs.Best when there is no first-party feed and the data lives in HTML pages that need a small adapter to normalize into the canonical schema.
job-listing.schema.json, including stable url, apply_link, and source_business_url attribution./v1/sources.Best when you have a periodic bulk export (CSV, Parquet) you can hand over for review and inclusion in the canonical dataset.
job-listing.schema.json, with one job per row and consistent column order./openapi.json if you also want to mirror the API shape./datasets/latest.csv./v1/sources.What records look like
The pool serves a single normalized record shape. Every integration path eventually produces records that match the same schema, regardless of whether they came from a feed, a scraper, or a CSV upload.
| Field | Type | Notes |
|---|---|---|
job_title | string | Public-facing role title as it appears on the origin. |
url | string | Canonical detail URL on your origin. Must resolve. |
apply_link | string | Application URL, preserved verbatim. Carries forward attribution. |
source_business_url | string | Origin business or board URL the listing belongs to. |
country_code | string | 2-letter ISO style code, for example US. |
industries | array | Best-effort classification. Free-form strings, deduped. |
{
"job_title": "Backend Engineer",
"url": "https://example.com/jobs/backend-engineer",
"apply_link": "https://example.com/apply/backend-engineer",
"source_business_url": "https://example.com",
"country_code": "US",
"industries": ["Software", "Engineering"],
"ingestion_date": "2026-05-09"
}Reference Agent
JobPoolBot is the canonical data-source agent for the pool. It is the same code path used by reviewed scrapers and the same crawler that picks up first-party feeds. Treat it as the reference implementation for what an automated data source should look like.
JobPoolBot/1.0 (+https://jobpool.live/jobpoolbot)JobPoolBot/v1/sourcesrobots.txt, including Crawl-delay hints when supplied.Link headers and HTML metadata before deep-crawling listing pages.If-Modified-Since / ETag revalidation where available.# /robots.txt
User-agent: JobPoolBot
Allow: /jobs
Allow: /careers
Crawl-delay: 5
# Preferred structured feed:
# https://example.com/jobs.json
Sitemap: https://example.com/sitemap.xmlSubmission Lifecycle
Every integration path follows the same staged review flow. There is no private, undocumented bypass.
You file a submission on jobpool.live/docs/submissions with feed URL, scraper, or dataset, and source metadata.
Submissions are reviewed for correctness, politeness, schema fit, and attribution. Feedback is posted on the submission record.
Approved submissions land in a staging dataset where records are validated end-to-end and dedup keys are checked.
Promoted sources show up in /v1/sources and their records flow into /v1/jobs and the next reviewed dataset snapshot.
/datasets/latest.csv; consumers can read freshness from /v1/sources at any time.FAQ
No. If your data only lives in HTML, the reviewed-scraper path on jobpool.live/docs/scrapers is the supported alternative. Feeds are encouraged but not required.
No. The pool stores normalized records; apply_link is preserved verbatim. Consumers are pointed at your apply URL, not a copy.
Open a withdrawal on the same submissions UI. The catalog entry is updated, the scraper or feed pull is paused, and the cache stops refreshing on the next cycle.
JobPoolBot is the canonical, documented agent. Other consumers may still crawl your origin under their own UAs — the pool just gives them a cache to prefer first via /v1/jobs.
Crawl-delay hints in robots.txt are respected. Origin-specific pacing is set per source during review and tuned if you flag issues on jobpool.live/docs/status.
JobPoolBot itself is the reference example, listed as a product on /products/. Its operator page is jobpool.live/jobpoolbot.
Next Steps
The submission UI on jobpool.live captures everything reviewers need: feed URL, scraper, dataset, refresh cadence, license posture, and contact info.