Blog/Sales and GTM
5 min read

Job Postings API: Pull Hiring Signals for Any Company

Get job postings as data, for a role in a city or for one company, with the fields that come back and the traps in them. Billed per call from your agent.

Job Postings API: Pull Hiring Signals for Any Company

Find who is hiring for a role, and what a company is hiring for, in one prompt. Paste this into your agent:

Set up https://monid.ai/SKILL.md, and then use Monid to find 5 current data engineer job postings in Berlin, and give me a table with the title, company, posted date, workplace type, applicants and whether a salary is listed.

One prompt turns a job search into a table. 1 min setup · 10-24 hrs saved.

You want job postings as data: every open data engineer role in a city, or the roles one account is hiring for, with enough fields to filter and compare. A pattern of job ads says what a company is building before its website does. This guide covers how to pull postings from your agent, what the records contain (we ran the endpoints live on 2026-10-16), and the traps in them. It runs through Monid, the OpenRouter for agent tools: one key, one balance, every tool priced per call.

What is the easiest way to get job postings through an API?

Monid. Install it once, add a key, and your agent calls a job search endpoint with a title and a city and gets typed JSON back, with no board account to open and no scraper to run. Jooble, the aggregator people often search for here, offers its API to partners through an application, not as a key you sign up for and start calling.

  • Pay per call. A LinkedIn job search bills $0.001 per result plus a flat $0.001 per run (Verified 2026-10-16 on the jobs tools page). Buy calls as you need them: no monthly plan and no minimum top-up. Discovery and inspection are free; only a run bills.
  • One install, every API. The same key and balance call the job endpoints and every other tool on Monid, so adding firmographics or a contact for the same account is another call, not another vendor.
  • Lives inside your agent. Install the CLI, add your key, run the prompt above, and the tool is there whenever your agent needs it. Your agent reads SKILL.md and learns the rest, so setup is about a minute.

A five-row search cost $0.006 when we ran it (Verified 2026-10-16).

How do you get job postings for a role or a list of companies?

Run one search per role and city, or one lookup per company, and loop it over your list. There are two shapes, and they answer different questions: search the market ("who is hiring a data engineer in Berlin") or read one account ("what is Stripe hiring for").

RouteWhat you getCost shapeFits
Download a job postings datasetA frozen file, often from a past yearFree or one-offResearch on the past
Build your own scraperWhatever your parser coversYour hoursOne board you will maintain
Scraper marketplace actorRows from one boardPer run, per resultA single source
Monid job endpointsLive rows from several boards as typed JSONPay per call, buy flexiblyRole searches, account lists, agents

Monid fits because the work is the same call repeated, and you pay for the searches you run and nothing between them.

For agents

Grab an API key at app.monid.ai, then paste this to your agent and hand it the key:

set up https://monid.ai/SKILL.md

It learns the whole discover, inspect, run workflow itself. More in the agent quickstart.

For humans

npm install -g @monid-ai/cli
monid keys add -k <your-api-key> -l main

Search the market for a role

monid inspect -p apify -e /harvestapi/linkedin-job-search
monid run -p apify -e /harvestapi/linkedin-job-search \
  -i '{"jobTitles":["data engineer"],"locations":["Berlin"],"maxItems":25}'

The endpoint bills per query in jobTitles, so total results are roughly the number of titles times maxItems. Pass one title first and read what comes back before you add a second.

Read one company's openings

Resolve the company to a website first, because the jobs endpoint takes a website or an id and not a bare name. The resolver is free:

monid run -p akta -e /v1/company/search --query '{"query":"<company name>"}'

Then call /v1/company/jobs with the website as company. The next sections show what it returns and how it bills.

Or hand the whole job to an agent

Give this to your agent

$Set up https://monid.ai/SKILL.md, and then use Monid to find the latest data engineer job postings in Berlin, group them by company, and tell me which companies have more than one open role.

📖 See also which LinkedIn job API stays fresh, which tests how current the postings are.

Can you scrape job postings yourself with Python?

Yes, for one board and a few hundred rows it is a weekend. You fetch the search page, follow the pagination, and parse each posting into fields. The same method works for every board, and every board needs its own parser.

The cost is everything around the parser. Each site changes its markup, limits repeated requests and asks you to sign in for the full listing, so the fetch layer, the proxies and the repair work are the real job. Two boards mean two parsers and a pass to line their fields up. GitHub is full of job board scrapers for exactly this reason: most are written once and then break, and one feed per board is the shape of the problem.

The Monid route is the choice for anything you will run more than once. The fetch and the parsing are maintained for you, the output is JSON with the same field names every run, and a search is $0.001 a result (Verified 2026-10-16). A scraper you wrote yourself is the better fit when one board is all you need and you want to own it.

What does a job posting record look like, and what can you build with it?

A posting record is a long row about the role, the company and the application, and a search returns one per job. We ran a LinkedIn search for data engineer in Berlin on 2026-10-16 with maxItems set to 5, and every field below is from that run.

The response

Each of the five rows carried 32 fields. The ones that matter for signal work:

  • The role: title, postedDate, expireAt, jobState (LISTED on all five), employmentType, workplaceType, experienceLevel, descriptionText and descriptionHtml.
  • The demand: applicants (132, 25, 61, 25 and 146 across the five) and views (0 on all five).
  • The employer: a company object with name, website, employeeCount (124, 5751, 348366, 203 and 12), industries and a LinkedIn URL.
  • The stack and the contact: applicantTrackingSystem (Lever on one, LinkedIn on one, empty on three) and hiringTeam, which held one named recruiter with a position and a profile link and was empty on the other four.
  • The money: salary with text, min and max, all null on all five.

For a second shape we ran the Built In search for the same title and city. One call returned 25 postings with 11 flat fields each: id, title, url, company, company_url, company_logo, work_type, location, salary, experience_level and posted. It is a thinner record with no applicant count and no company size, and it comes back as a page of 25 with total_pages set to 3.

For one account, the Akta jobs endpoint returned a row for https://stripe.com with title, source (linkedin), url, date_posted, location, a short description, type, remote, salary_min_usd and salary_max_usd (138300 and 207500 on that row), experience_level and a key_skills list of 16 entries. The response also carried count, the total open roles for the company, which was 1284.

What you can build with it

  • A hiring watchlist. Search the roles that matter to you each week and keep the companies that appear. A first data engineer or a first security hire means a function is being started, which is when tooling gets bought. Shipping a LinkedIn job alert bot shows the loop that runs a search on a schedule.
  • A buying-intent signal on an account list. For each account, read the open-role count and the titles. A company with 1284 open roles is staffing up; one with none is not. Pair it with the company size and industry already in the row.
  • A technographic read. The description text names the tools a team hires for, and applicantTrackingSystem names the recruiting stack. Both arrive inside the same row, so you learn what a team runs from what it is hiring for.
  • A live role dataset. Collect searches over weeks into one table, keyed on the posting id, instead of a frozen Kaggle file.

What goes wrong when you pull job postings in bulk?

Six things show up when you read more than one row. All six are from the 2026-10-16 runs.

Why did a one-row request bill ten units?

Because the Akta jobs endpoint bills in blocks of ten results. We asked for one posting with limit set to 1, got one row back, and were billed ten units, $0.20 (Verified 2026-10-16). The listed price is that amount per ten results. Ask for ten at a time with limit 10 and use offset to page, so every unit you pay for comes back as a row.

Why does the LinkedIn search bill more than the row count suggests?

Because the price has two parts, a flat fee per run and an amount per result, and it applies per query. Our five-row run billed $0.006: five results at $0.001 plus the flat $0.001 (Verified 2026-10-16). Three titles and three cities multiply the result count, so run one query first.

Why can you not trust applicants as a smooth number?

Because it is not one on every row. Of five postings, two said exactly 25 and the others said 132, 61 and 146. A value like 25 on a posting made the same day is a floor reading, not a count of people. Read large values as demand, small repeated ones as "few", and never average the column.

Why is expireAt not a close date?

Because it sits at a fixed distance from the posted date on many rows. Three of five postings expired 30 or 31 days after postedDate; the other two ran 39 and 64 days. Read postedDate as the age of the role and do not read expireAt as the day the role closes.

Why do so many fields come back empty?

Because the platform does not hold them for most postings. salary was null on all five LinkedIn rows, experienceLevel was null on four, and workplaceType was null on two. On Built In, 21 of 25 rows had no salary, and the four that did carried text such as 100K-130K Annually with no currency code. Any model built on pay will have holes, so count the nulls before you chart anything.

Why does the same role show up twice?

Because boards repost. In the Built In page, 8 of 25 rows had a posted value starting with "Reposted", such as "Reposted 19 Days Ago". Another 11 had a location of "4 Locations" or "9 Locations" instead of a city. A daily pull counts a refreshed ad again and reads as accelerating hiring, so key your table on the posting id and keep the first date you saw. The title match is also loose: a mechanical engineer for data centres and a satellite data processing engineer came back for "data engineer", so filter titles on your side. The Indeed job search shows the same duplicate and filter work on another board.

A provider can also fail inside a call. We ran the Clay open-roles lookup for stripe.com; the provider answered "Not enough Clay credits", the run reported zero billed units, and no row came back. Your loop needs to handle the missing row.

Which endpoint should I use for which job?

EndpointWhat it doesInputBillingPrice (Verified 2026-10-16)
apify/harvestapi/linkedin-job-searchRole and city search on LinkedIn, 32 fields per rowjobTitles, locations, maxItemsPer result plus a flat fee$0.001
builtin/search_jobsRole search on Built In, 25 rows a pagequery, location, pagePer call$0.01
akta/v1/company/jobsPostings and open-role count for one companycompany, limit, offsetPer 10 results$0.20
apollo/organizations/{organization_id}/job_postingsActive postings for an Apollo organization idorganization_id, page, per_pagePer call$0.026
clay/enrichment/company-job-openingsOpen-role count and sample postings from a domainCompany DomainPer call$0.032
akta/v1/company/searchResolve a company name to a website and idqueryPer call$0.00

Prices Verified 2026-10-16 with monid inspect. The first row is the one for market searches and the third is the one for a named account. The Apollo endpoint needs an organization id, which comes from a separate Apollo company search, so it is two calls for one account; we inspected it and did not run it. The Clay endpoint returned a failure on our run, so we have not seen its rows. For the LinkedIn side of this table in more depth, see the LinkedIn data guide.

When is another route a better fit?

You are building a job board. That is a licensing conversation with the sources, not a data pull, and the endpoint is the smaller part of it.

You need every opening in a market. A search returns what the platform surfaces for your query, and the Built In search showed 3 pages for one title and city. For headcount analytics that must be complete, buy a dataset built for it.

You need salary data. The field is null on most rows we pulled, so compensation benchmarking needs a source built for pay.

You need history. These endpoints read what is live. A closed role is gone, so a trend means collecting on a schedule from today, not querying backwards.

You need one free look at one site. A browser and the board's own search do that with no key.

And the disclosure: we resell these endpoints. The habits in this post cost nothing whoever you buy from: key on the posting id, read postedDate over expireAt, and count the nulls.

Conclusion

The quickest way to get job postings as data is a search call per role and city, or a lookup per company, through Monid: $0.001 a result on the LinkedIn search (Verified 2026-10-16), one key for this and every other tool, about a minute to set up and ready inside your agent whenever you need it. A scraper of your own works for one board; Monid keeps the fetching and parsing maintained.

Two things matter more than the endpoint you pick. Deduplicate on the posting id, because reposts and refreshed ads otherwise read as accelerating hiring. And check how an endpoint bills before you loop it: a one-row company request billed ten units, and a multi-title search multiplies the result count.

How we estimated the time saved: a scraper for two boards, with the fetch layer and parsers, is roughly 8 to 16 hours, plus 2 to 8 hours for the deduplication and the first month of repairs. That is 10 to 24 hours. The 1 minute is the install, key and first call once you already have a Monid key.

Free next step: run monid discover for "job postings", then monid inspect -p apify -e /harvestapi/linkedin-job-search, and make one search with maxItems set to 5. Start at monid.ai.

FAQ

Is there a free job postings API?

Not free, but cheap for small jobs: through Monid a five-row LinkedIn search billed $0.006 when we ran it (Verified 2026-10-16), with no monthly plan and no minimum, and discovery and inspection cost nothing. Free job board APIs and public datasets exist, but they are frozen files or capped feeds, not live postings.

Is there a Jooble API?

Yes, but Jooble offers it to partners through an application, not as a self-serve developer key. If you want postings as data inside a pipeline, call a job search endpoint on Monid instead and get typed JSON for a role and a city.

How do I get the open jobs of one company?

Resolve the company to a website with akta /v1/company/search, which is free, then call akta /v1/company/jobs with that website. It returns the postings and a count of the company's open roles. Ask for ten at a time, because the call bills in blocks of ten results.

How do I avoid counting the same job twice?

Key your table on the posting id and keep the first date you saw it. Boards repost: 8 of 25 Built In rows we pulled were marked "Reposted", and a daily pull would count each refreshed ad again.

Why is the salary field usually empty?

Because most postings do not list one. It was null on all five LinkedIn rows we pulled and on 21 of 25 Built In rows. Where a salary was present it was a text range with no currency on Built In, and numeric minimum and maximum fields on the Akta row.

Published August 2026.

job postings apihiring signalsjooble apijob postings datajobs datagtm