Export LinkedIn connection requests and new connections with an AI agent: the profiles of everyone who added you (2026)

Overview

When people find you on LinkedIn, after a post, a talk or a job ad, they send you connection requests. Founders, recruiters, salespeople and creators want to know who these people are: which ones are buyers, candidates or peers, which ones to accept, and which ones to follow up with. To decide, they need each person's profile, and they need the list in a sheet or a CRM.

"People who added me" means two lists on LinkedIn: the connection requests you received and did not answer yet, and the people who connected with you recently. This page covers both.

Why it's hard

• The Received tab has no export. LinkedIn shows your pending requests 10 at a time, with a name and a short line. To see a person's role, company and location, you must open each profile.
• The data archive is partial. The Invitations file (Settings & Privacy, Data privacy, Get a copy of your data) has names, profile URLs, the date and the message, but no headline, company or location. The Connections file has company and position, but LinkedIn lists it in the group that can take up to 48 hours. Both are snapshots: new requests need a new download.
• There is no API for you. Microsoft's documentation says the LinkedIn Connections API is restricted to developers approved by LinkedIn. We found no public API for received requests.
• The automations stopped or need your cookie. Zapier's LinkedIn app has no "new connection" trigger: Zapier staff wrote in October 2023 that it was removed. The tools that export received requests run on their own servers with your LinkedIn session cookie, or are Chrome extensions that you use one list at a time.
• Many requests are noise. Requests from bots, fake profiles and sales pitches arrive with the real ones, so each profile needs a check before you accept it.

How Reduck does it

Your AI agent runs Reduck's LinkedIn scripts in your own Chrome, where you are already signed in. It needs no API key and sends no cookie to a server.

1. It lists the two lists. get_received_invitations reads your Received tab: each request with the name, the profile link, the type (person, page, event, newsletter) and the id that accept_invitation takes. get_connections reads your connections, newest first, with each person's headline, profile link and the "connected on" date.
2. It opens each profile. get_profile returns the headline, location, current company, About text, experience and education. It does not record a profile view.
3. It applies your rule. You write the rule in your own words, for example: "accept founders and marketers in B2B software, ignore anyone whose headline says sales or recruiting, and flag profiles with no photo and no experience". The agent marks each person accept, ignore or check.
4. It acts only when you say so. accept_invitation and ignore_invitation change your account. The agent shows you the list first, and accepts or ignores the ones you approve.

What you get: one sheet or CSV with every person who added you in the period you choose, with name, headline, current company, location and profile link, the "connected on" date for new connections, and a suggested decision for each pending request. Ask again next week to get the new ones.

How to start

1. Create a free account at reduck.ai.
2. Install the Reduck extension in Chrome and pair it.
3. Connect Reduck to your agent: docs.reduck.ai.
4. Ask: "List everyone who sent me a LinkedIn connection request or connected with me in the last {number of days} days. Open each profile and put them in a sheet with their name, headline, current company, location and profile link. Mark the requests I should accept, using this rule: {your rule}. Do not accept or ignore anything until I say so."

Tools to export LinkedIn connection requests, compared

As each vendor's own page describes it, read on October 29, 2026. "Not documented" means we did not find it on their page.

Where the others are stronger: Outscore keeps a history of every request as you browse, with no prompt. Apify and PhantomBuster run on a schedule with your computer off, and Linked API is built for a product that needs an API. Reduck runs when your agent asks, and your Chrome must be open.