Support

First: did you get a 200? Invalid parameters and enum values return a 400 naming the parameter, so a typo can't hide. On a genuine 200 with zero rows, the usual suspects:

• skill commas mean AND. skill=python,aws needs both — it's the one filter where comma isn't OR.
• A displayed skill fed back as a filter. skill=Machine Learning → 0 rows. It's machinelearning.
• An over-tight combination. Every parameter still ANDs with the others; drop them one at a time and watch total (or use /v1/jobs/count).

The filtering guide has the full procedure.

Omitted by default — the key is absent, not null. Add description_format=text or description_format=html. It's opt-in because descriptions are multi-kilobyte and most callers don't want 200 of them per page. See the Job object.

If it's from browser JavaScript, it's the CORS preflight: an Authorization header makes the request non-simple, and there's no OPTIONS handler. Use ?api_key= from browsers — but see why you shouldn't call the API from a browser at all.

Otherwise: the key may be revoked, or you may be sending a key from a different account. 401 covers missing, malformed, unknown, and revoked without distinguishing.

Expected. Postings are reaped when they disappear from the employer's board, and /v1/jobs excludes anything expired. It may also have been marked as a duplicate of an equivalent posting from another source. See duplicates.

If you keep a local mirror, treat a 404 on a previously-seen id as a delete. Syncing the feed covers this.

Both are correct. /v1/health counts every row including duplicates and expired postings; /v1/meta counts only live, non-duplicate ones. /v1/meta is the number you want. See coverage.

Usually description=, which scans every description in the dataset. Pair it with a narrower filter. Also check you're not fetching descriptions you don't read — see coverage on query cost.

Pace the client rather than retry-looping. Limits are per account, not per key, so multiple workers share one budget and have to divide it. Rate limits has a token-bucket example.

Also check you're using limit=200: fifty pages of 20 costs ten times what five pages of 200 costs, for identical data.

Not through the API — there's no bulk jobs export. For flat-file delivery, talk to us. For keeping a live mirror, sync the feed.

No. Sync is pull-based via /v1/jobs/feed — one request returns everything since your cutoff. See syncing the feed.

It's scraped from employers' own ATS boards, so apply_url points at the real application form and postings track the source board within a few hours. Enriched fields — skills, seniority, taxonomies, salary — are derived by classifiers and are not perfect. Coverage varies by field; see coverage.

If you find data that's wrong, send the job id. That's the most useful bug report we get.