Find people on LinkedIn

Runs a people search in the LinkedIn app as the signed-in account and reads each result card: name, headline, location, and the current role or school snippet. LinkedIn's terms and its defences make this one of the hardest targets for a scraper, which is why third-party ones go stale. Here it is simply the app, signed in as you, reading what is on the screen.

Nothing is sent. No connection requests, no messages, no profile opens.

Inputs

Run it

Open the results by link: typing into LinkedIn's search box is the least reliable step in the app, and the link lands on the People tab directly. URL-encode the query in the link (spaces as %20).

run_task({
  "phases": [{
    "launch": "com.linkedin.android",
    "openUrl": "https://www.linkedin.com/search/results/people/?keywords=<query>",
    "goal": "The LinkedIn People search results for '<query>' are open. Keep scrolling the people results. Do not type, do not use the search box, do not open All filters, do not tap any person's name or open a profile, and do not tap Connect, Invite, Message or Follow.",
    "maxSteps": 16,
    "collect": {
      "record": "a person in the People search results",
      "fields": { "name": "the person's full name only, without Verified or degree text",
                  "headline": "their headline, the line under the name",
                  "location": "their location, if shown",
                  "degree": "the connection degree such as 1st, 2nd or 3rd+",
                  "action": "the label of the button on the right of the card, such as 'Invite <name> to connect', 'Send a message to <name>', 'Pending' or 'Follow'" },
      "where": "people search results for '<query>' are showing",
      "count": <count>
    }
  }]
})

With degree 2nd, put a phase before the read that taps the chip by its full label, and drop launch and openUrl from the read so it stays on the filtered page:

{ "launch": "com.linkedin.android",
  "openUrl": "https://www.linkedin.com/search/results/people/?keywords=<query>",
  "goal": "In the row of filter chips near the top, tap the chip whose label is 'Filter by 2nd connections', once, so the results reload. Do nothing else: do not open All filters, do not tap any other chip, do not scroll, do not tap any card.",
  "require": { "label": "Filter by 2nd connections" },
  "steps": ["the 'Filter by 2nd connections' chip is selected and the results have reloaded"],
  "maxSteps": 6 }

What comes back

result.collected holds one record per person, from the result cards alone. degree often arrives as the whole name line (Chu Li Premium • 2nd): read the • 2nd / • 3rd+ token from it. When name comes back empty, the action label still carries it (Invite <name> to connect). Profile URLs and contact details are not on this screen; getting those means opening each profile, which is a separate task per person.

Notes

• The button on each card says what you can do, with no judgment needed: Invite <name> to connect is a 2nd-degree person one tap can invite; Send a message to <name> on a Premium account is InMail to a 3rd+ person, not an existing connection; Pending is already invited; Follow is creator mode. degree and action together classify a card.
• Name the chip by its full label. "2nd" alone was ambiguous on a measured screen; the full label is one tap. require ends the phase in one look if the chip row is not there.
• Do not open All filters: it is the most common reason a run sends or reads nothing.
• Measured on 2026-10-06: by link, 6 of 6 people in 2 steps and 6 credits, every card's button label read; typing into the search box had stalled in five of six earlier attempts.
• To act on the list, see linkedin-invite-from-results (one tap per card) or linkedin-connect-with-note (one person, with a note). LinkedIn caps invitations near 100 a week and throttles quietly past that; check linkedin-sent-invitations for what really went out.