Skip to main content
LinkedIn pages are read with a purpose-built site rule instead of a browser. You don’t turn anything on: send a LinkedIn URL as usual and the rule is used automatically. A page costs 1 credit, and pages that can’t be delivered are free. Spidra reads LinkedIn the way a visitor who isn’t signed in sees it. It never signs in, and it doesn’t use cookies you pass for LinkedIn pages. The rule steps aside when a request needs something only a browser can do: a screenshot, browser actions, forEach or an AI instruction. Those URLs use the normal browser flow.

​
Page types

Links work as you copy them. Tracking parameters are removed, country subdomains such as uk.linkedin.com work, a search link with a job selected (currentJobId=) opens that job, and a company’s /about/ link reads the company page, which shows the same details.

​
Job searches

Put your search in the URL the same way LinkedIn does: keywords, location, and filters such as f_TPR=r86400 (posted in the past 24 hours). The first page returns up to 60 jobs. For more, add start: start=60, then start=70, and so on in steps of 10. Each response includes nextStart, the value to use for the next page. LinkedIn shows at most 1,000 results per search to visitors who aren’t signed in. A request with start above 990 fails straight away with a message and costs nothing. To cover more jobs, split the search by location, keyword or date.

​
Fields you get

Every field is optional on LinkedIn’s side, so a missing value comes back as null or an empty list, never a guess. A few notes on coverage:
  • salary appears only when the employer shows pay. In testing, about 1 in 5 job postings did.
  • Public profiles show current roles, education and a limited part of the experience history. Since November 2025, LinkedIn shows a profile’s full history only to signed-in members. Some entries are masked with asterisks by LinkedIn itself; those come back with "masked": true.
  • employeesOnLinkedIn is the number of members who list the company, not the company’s own headcount.

​
What comes back

  • output: "json" with no prompt or schema: the structured data 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.
  • With a prompt or schema: the AI reads a compact copy of the structured data instead of the whole page. A company page with a five-field schema cost 2 credits in testing.

​
Limits you should know about

  • Signed-in pages are refused. People search, a company’s People tab, a profile’s full details, the feed, messaging and Sales Navigator need a signed-in member. Those requests fail straight away with a message and are not charged.
  • School pages are often refused. LinkedIn turns away most visitors who aren’t signed in from school pages. When it does, the request fails with a message and is not charged.
  • A profile LinkedIn won’t show fails with a message and is not charged. LinkedIn answers a missing profile and a refused request the same way, so the message can’t say which it was.
  • 1,000 results per job search, as above.
  • A page that doesn’t exist, such as a mistyped company slug, fails with a message and is not charged.
Check LinkedIn’s User Agreement before you scrape it, and the data protection laws that apply to you. Profiles contain personal data, which laws such as GDPR protect even when it is public. You are responsible for how you use the data you collect. Spidra is not affiliated with LinkedIn.