Scraper Workers

LinkedIn prospector

The Ultron LinkedIn Prospector takes a set of search filters and returns matching profiles with their public fields: name, headline, location, current company, current title, profile URL. It pages through LinkedIn's People search using the caller's li_at cookie behind a residential proxy, follows redirects up to three hops, and dedupes by canonical profile id.

Updated today

Overview

At a glance

Worker: ultron-linkedin
Runtime: Node 20, raw https requests through HttpsProxyAgent
Source: apify-actors/linkedin/
Auth: li_at cookie required
Redirect handling: Up to 3 hops
Dedupe key: Canonical profile id from the final URL
Downstream: decision-maker-prospector, vc-prospector, hiring-manager-prospector

The prospector is the lightest of the LinkedIn scrapers because it does not need a full browser. The People search endpoint returns enough useful HTML in the initial response that a plain GET plus a small parser produces the fields the platform needs. Skipping Playwright saves seconds per request and removes the headache of Chromium memory inside the worker container.

Authentication

The li_at cookie is passed in headers on every request. Cookie missing means abort.

apify-actors/linkedin/src/main.jsjs

1if (!input.li_at) {

2 log.error('li_at cookie missing - refusing to run')

3 await Actor.fail('li_at cookie required')

4 return

5}

7const headers = {

8 'cookie': `li_at=${input.li_at}`,

9 'user-agent': ua,

10 'accept': 'text/html,application/xhtml+xml',

11 'accept-language': 'en-US,en;q=0.9',

12}

Input schema

Search filters + cookie + proxy controls.

apify-actors/linkedin/.actor/actor.jsonjson

1{

2 "queries": ["AI angel investor", "operator angel"],

3 "locations": ["United States", "United Kingdom"],

4 "current_company": [],

5 "title_keywords": ["investor", "partner"],

6 "industry": [],

7 "max_per_query": 100,

8 "li_at": "<required>",

9 "proxy_url": "http://session-<id>:<pass>@proxy.apify.com:8000",

10 "webhook_url": "https://app.51ultron.com/api/apify-webhook?source=linkedin",

11 "webhook_secret": "<shared>",

12 "user_id": "<uuid>",

13 "conversation_id": "<uuid>"

14}

Crawl flow

One pass per query × location. Each pass walks paged results.

apify-actors/linkedin/src/main.jsjs

1for (const query of input.queries) {

2 for (const location of input.locations) {

3 let page = 0

4 while (true) {

5 const url = buildPeopleSearchUrl({ query, location, page })

6 const res = await httpsGet(url, headers, proxyAgent)

7 const profiles = parseProfileCards(res.body)

8 if (!profiles.length) break

10 for (const p of profiles) {

11 if (seen.has(p.profile_id)) continue

12 seen.add(p.profile_id)

13 await Actor.pushData({ ...p, query, location, scraped_at: now() })

14 }

15 page += 1

16 if (page * pageSize >= input.max_per_query) break

17 }

18 }

19}

Proxy and redirects

Residential proxy with a hop-bounded redirect chaser.

apify-actors/linkedin/src/main.jsjs

1async function httpsGet(url, headers, agent, timeout = 20000, _depth = 0) {

2 const res = await httpsGetOnce(url, headers, agent, timeout)

3 if (

4 (res.status === 301 || res.status === 302 || res.status === 307 || res.status === 308) &&

5 res.headers.location &&

6 _depth < 3

7 ) {

8 const next = new URL(res.headers.location, url).toString()

9 return httpsGet(next, headers, agent, timeout, _depth + 1)

10 }

11 return res

12}

Note

LinkedIn frequently redirects unauthenticated and auth-edge cases through a few hops before settling on either the result page or the login wall. The chaser caps at three hops so it cannot loop forever. Any response after three hops that is still a redirect is treated as a soft failure for that page.

Output shape

One row per unique profile.

dataset-row.jsonjson

1{

2 "profile_id": "AC123abc",

3 "profile_url": "https://www.linkedin.com/in/maria-schultz",

4 "name": "Maria Schultz",

5 "headline": "Head of Sales at Acme | Angel @ ...",

6 "location": "Berlin, Germany",

7 "current_company": "Acme",

8 "current_title": "Head of Sales",

9 "query": "AI angel investor",

10 "scraped_at": "2026-05-11T18:42:00Z"

11}

How to trigger it

Same shape as the other Ultron Workers: HTTP POST against the Apify run endpoint with the platform-injected metadata.

invocationtext

1POST https://api.apify.com/v2/acts/ultron-linkedin/runs?token=<APIFY_TOKEN>

2Content-Type: application/json

4{

5 "queries": ["AI angel investor"],

6 "locations": ["United States"],

7 "max_per_query": 100,

8 "li_at": "<from the connected LinkedIn integration>"

9}

File map

apify-actors/linkedin

package.json

src

main.jsauth + crawl loop + parser + dedupe + push