Skip to main content
Most pages are scraped with a real browser. For a few sites that is slow, costly or unreliable, so Spidra reads them with a purpose-built site rule instead. You do not turn anything on: send the URL as usual and the rule is used automatically. A rule only applies when the request needs nothing a browser alone can do. If you ask for a screenshot, browser actions, forEach or an AI instruction, Spidra uses the normal browser flow for that URL.

​
Indeed

Indeed pages are read directly rather than through a browser. A typical search page returns in a few seconds and costs 1 credit.

​
Every country site

All of Indeed’s country sites work the same way: uk.indeed.com, ca.indeed.com, de.indeed.com, ng.indeed.com, jp.indeed.com, br.indeed.com and the rest. In testing, 59 of the 60 country sites returned results; the exception was ae.indeed.com, which answers 404 for its search page. Older country-code domains such as indeed.co.uk, indeed.jp and indeed.com.br that redirect to those sites are understood too, and plain http:// links are upgraded to https://. A search that Indeed answers with no matching jobs, for example an English query in a country with no such listings, comes back as a normal result with zero jobs and costs 1 credit.

​
Fields you get

Fields are read from the data Indeed puts on the page. Every field is optional on Indeed’s side, so a missing value comes back as null or an empty list, never a guess. Coverage in our tests is shown where it varies. Coverage to expect:
  • requirements (skills and certifications, each marked required or preferred) appears on roughly 10% to 50% of listings, depending on the role.
  • benefits and schedule appear on most single jobs, but only when the employer lists them.
  • address and coordinates appear on fewer than half of jobs, because Indeed shows them only when the employer publishes a street address.
  • salaryMin and salaryMax appear when the posting includes pay.
  • employees and revenue on company profiles are ranges, for example “10,000+” and “over $10B”.
Links copied from a results page carry long tracking tokens (ad=, sjdu=, xkcb=, tk= and more). You do not need to clean them. Spidra reduces a job link to the job key and opens that job, so this works as pasted:

​
What comes back

Nothing is filtered out. For a search page you get every job on the page, including sponsored ones, which are marked "sponsored": true, and jobs older than your fromage window, marked "outsideWindow": true. Filter on those flags, or describe what you want in a prompt.
  • No prompt or schema, output: "json": the structured data comes back as is. No AI is used, so the cost is the base credit only.
  • output: "markdown": the whole page as markdown, followed by the structured data for every job.
  • With a prompt or schema: the AI reads the whole page and all of the structured data, so the token cost is higher than for a normal page. A full search page of about 40 jobs cost 13 credits in testing. If you only need the data, use JSON output without a prompt.

​
Limits you should know about

  • Only the first page of a search. Indeed asks visitors who are not signed in to sign in before showing page 2 and beyond. A request with start=10 or higher fails straight away with a message and costs nothing. Use narrower searches instead: add a location, try title variants, or set fromage=1 to get only new postings. If you pass your own session cookies, Spidra will try deeper pages with your session.
  • Very generic searches can fill the first page. A broad title in a large city can have more new jobs than one page holds. Split it by city or by a more specific title.
  • A job must still be listed. If a job has expired or been removed, the request fails with a clear message and costs nothing.
  • Click-tracking links are not pages. Ad-click and tracking URLs (/pagead/clk, /uie/clk, /applystart and similar) would register a click that an advertiser pays for, so Spidra refuses them immediately with a message and does not charge. Job links that carry such tokens are fine: Spidra reduces them to the job key and opens the job.
  • Signed-in areas are refused. Sign-in pages, account and employer dashboards, résumé search and the application flow need a login, so there is nothing to read. They fail immediately and are not charged.
  • Pages that do not exist. If Indeed answers a page with an error such as 404, the request fails with that status and is not charged.
Check Indeed’s terms of service before you scrape it. You are responsible for how you use the data you collect.

​
Tracking parameters are removed automatically

For every site, Spidra removes well-known marketing and click-tracking parameters from your URL before it runs the job, for example utm_source, gclid, fbclid and msclkid. They never change what a page shows. Any parameter Spidra does not recognise is left exactly as you sent it, because the same name can be a real filter on another site.

​
Pages that cannot be delivered are free

If a page comes back as an HTTP error such as 400, 401, 403, 429 or a server error, or a CAPTCHA was solved and the page is still blank, the URL is reported as failed and you are not charged for it, including the CAPTCHA. See How credits work.