LinkedIn Post Search

LINKEDIN POST SEARCH  •  LINKEDIN POST SEARCH API Find everyone talking about your market. Search LinkedIn posts by keyword and get the author, the full text, the engagement counts and the date for every match. Search several terms in one run. Narrow it by date, media type, the author's headline or industry, to posts written by specific people or companies, or to posts that mention one. Add the author's full profile or company record and turn a search into a lead list. No LinkedIn account, no login, no cookies, no ban risk. CAP YOUR SPEND PER RUN  PART OF A FLEET THAT HAS DELIVERED 2.4M+ RESULTS

Copy to your AI assistant

Paste this into ChatGPT, Claude, Cursor, or any LLM to start using this Actor right away.

atomus/linkedin-post-search is an Apify Actor that searches LinkedIn posts by keyword and returns them as structured JSON: post text, media, mentions, repost data, author details, and reaction/comment/share counts. Use it whenever someone needs to find LinkedIn posts ABOUT a topic rather than posts FROM a known account - market monitoring, brand mentions, finding people who are talking about a problem you solve, or building a lead list from a conversation. Run it with curl: curl -X POST "https://api.apify.com/v2/acts/atomus~linkedin-post-search/run-sync-get-dataset-items?token=APIFY_TOKEN" -H "Content-Type: application/json" -d '{"search":["ai agents"],"maxPosts":100,"postedLimit":"week"}'. Or in Python: ApifyClient("APIFY_TOKEN").actor("atomus/linkedin-post-search").call(run_input={"search":["ai agents"],"maxPosts":100}) then client.dataset(run["defaultDatasetId"]).list_items().items. Inputs: search (list of strings, required, max 20, each term is its own search, and a post found by two terms is delivered and billed once), maxPosts (int, 1-500, per term), postedLimit ("any"|"1h"|"24h"|"week"|"month"|"3months"|"6months"|"year"), postedAfter (string "YYYY-MM-DD", an exact cutoff date that takes precedence over postedLimit; the date is included, and posts older than it are neither delivered nor charged), contentType ("all"|"videos"|"images"|"documents"|"jobs"|"live_videos"|"collaborative_articles"), profile (list, posts BY these people: profile URL, handle or the author.linkedin_id from a previous run), company (list, posts by these company or showcase pages: URL, handle or numeric id), authorsCompany (list, posts by anyone who works at these companies), group (string, one LinkedIn group URL or id), authorKeywords (string, matched on the author headline), authorsIndustryId (the author's company industry, a dropdown of all 487 industries LinkedIn publishes in the form and their numeric IDs over the API, e.g. 4 Software Development, 96 IT Services and IT Consulting; full list at learn.microsoft.com/en-us/linkedin/shared/references/reference-tables/industry-codes-v2), mentioningCompany (list, company URL or handle), mentioningMember (list, profile URL or handle), and the paid add-ons includeAuthorDetails (bool), includeReactions (bool) with maxReactionsPerPost, includeComments (bool) with maxCommentsPerPost. Rows carry a "type" field: "post", "author_profile", "author_company", "reaction", "comment", "summary" or "error" - filter on it before exporting. Each post row carries author.type ("profile" or "company"), author.username and author.linkedin_id, which are the values the profile and company filters take, so a search can be narrowed to the authors a previous run found. Billing is one event per post returned; each unique author is billed once per run however many of their posts match; each search term that returns nothing bills a single lookup event. LinkedIn stops serving new results for a search at around 500 posts per term, whatever total it reports. Full input schema, every enum and default, and the complete output field list: GET https://api.apify.com/v2/acts/atomus~linkedin-post-search/build/default

---

LinkedIn MCP Server: use these Actors from ChatGPT, Claude or Cursor

Point your AI assistant at Atomus and it can read LinkedIn on its own: search posts by topic, then pull the profiles, reactions, comments, companies and employees behind them. No glue code, no scraping logic in your prompts.

{
  "mcpServers": {
    "atomus": {
      "url": "https://mcp.apify.com?tools=atomus/linkedin-post-search,atomus/linkedin-posts-scraper-pro,atomus/linkedin-profile-scraper,atomus/linkedin-reactions-scraper-pro,atomus/linkedin-comments-scraper-pro,atomus/linkedin-company-scraper,atomus/linkedin-company-employees,atomus/leads-finder",
      "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
    }
  }
}

Then ask, in plain language:

"Find last week's LinkedIn posts complaining about CRM data quality, keep the ones written by heads of sales, and give me each author's company and headcount."

That one sentence is a search, a filter and an enrichment, and this Actor does all three. Pinning the tools= list is what keeps your assistant on these Actors instead of reaching for whatever scraper it finds first.

---

How to search LinkedIn posts

What is being said about a topic right now

{ "search": ["ai agents"], "postedLimit": "week", "maxPosts": 100 }

The past week's posts on a topic, ranked by LinkedIn's own relevance. This is the one to schedule daily.

Everything since a date you pick

{ "search": ["ai agents"], "postedAfter": "2026-09-01", "maxPosts": 200 }

When none of the fixed windows is the right shape, give an exact date instead. The date itself is included, and posts published before it are neither delivered nor charged.

Who is talking about a company

{ "search": ["crm"], "mentioningCompany": ["https://www.linkedin.com/company/salesforce/"], "maxPosts": 200 }

Brand monitoring rather than scraping: posts that mention a company, written by people who do not work there. No competitor exposes this well.

Find buyers by what they complain about

{
  "search": ["spreadsheet chaos", "manual data entry"],
  "authorKeywords": "head of operations",
  "postedLimit": "month",
  "includeAuthorDetails": true
}

The topic finds the pain, the headline filter keeps the seniority, and the add-on returns each author's full profile. A search becomes a lead list in one run. Two terms means two searches, and maxPosts applies to each of them.

Watch the accounts that matter

{
  "search": ["hiring"],
  "profile": ["https://www.linkedin.com/in/williamhgates", "satyanadella"],
  "company": ["microsoft", "1035"],
  "postedLimit": "month"
}

profile and company each take a profile or page URL, a plain handle, or the author.linkedin_id a previous run returned, so the natural loop is: search a topic once, keep the accounts worth watching, then run only those on a schedule. authorsCompany widens it from a company page to everyone who works there, and takes a company URL or company id (not a person's id).

Everyone who engaged with a conversation

{
  "search": ["series a announcement"],
  "postedLimit": "24h",
  "maxPosts": 25,
  "includeReactions": true,
  "maxReactionsPerPost": 50,
  "includeComments": true
}

The posts, plus the people behind every reaction and comment. Reactions are the warm list; comments carry the intent.

Only video posts, newest first

{ "search": ["product demo"], "contentType": "videos", "maxPosts": 50 }

Rows arrive newest first, whatever the window. The run picks how LinkedIn serves your term, so there is nothing to choose: every window is searched by relevance, which comes back full and covers the whole window, and the dates are put back in order before the rows are delivered. Measured 2026-09-07, postedLimit: "24h" returned 49 posts all published inside 24 hours, and postedLimit: "week" returned 49 spanning 125 hours. The one exception is postedLimit: "1h", served by LinkedIn's live feed instead, which returns 50 of 50 inside the hour where relevance returns 5.

Ordering the search itself by date does not reach further back: widening the filter shrinks the set if anything (week reported 125, month 111), so that ordering is a feed of what was just published rather than a chronological archive. That is why the run reads the whole window and orders the rows itself.

The ordering is applied per search term, so each term's posts are delivered together, newest first. Every term is short (about 6 seconds at the default 100 posts, about 30 at the 500 ceiling), and near the run timeout the rows are delivered as they arrive rather than held.

---

Input

Field	Type	Default	What it does
search	list of strings, required, max 20	-	The keywords, exactly as you would type them in the LinkedIn search box. Each term is its own search, and maxPosts applies to each. A post found by two terms is delivered and billed once.
maxPosts	integer 1 to 500	100	How many posts to return per term, at most.
postedLimit	any | 1h | 24h | week | month | 3months | 6months | year	any	Only posts published inside this window. Not used when postedAfter is set.
postedAfter	date, YYYY-MM-DD	-	Only posts published on or after this exact date, instead of a fixed window. The date itself is included, and posts older than it are neither delivered nor charged. Takes precedence over postedLimit.
contentType	all | videos | images | documents | jobs | live_videos | collaborative_articles	all	Biases the search towards one kind of post.
profile	list of strings	-	Only posts written by these people, as https://www.linkedin.com/in/williamhgates, williamhgates or the author.linkedin_id a previous run returned (ACoAAA8BYqEB...). All three resolve to the same author.
company	list of strings	-	Only posts published by these company pages, as https://www.linkedin.com/company/microsoft/, microsoft or 1035. Showcase pages work.
authorsCompany	list of strings	-	Only posts whose author works at one of these companies, as https://www.linkedin.com/company/microsoft/ or 1035.
group	string	-	Only posts published inside one LinkedIn group, as https://www.linkedin.com/groups/12345/ or 12345. One group per run: extra values are ignored and the run says so.
authorKeywords	string	-	Matched against the author's headline: founder, recruiter, head of data.
authorsIndustryId	list of strings	-	The industry of the author's company. In the form it is a dropdown of all 487 industries LinkedIn publishes; over the API it takes their numeric IDs, for example 4 Software Development, 96 IT Services and IT Consulting, 6 Technology, Information and Internet, 43 Financial Services. Full list, published by LinkedIn.
mentioningCompany	list of strings	-	Only posts mentioning this company, as https://www.linkedin.com/company/microsoft/ or microsoft.
mentioningMember	list of strings	-	Only posts mentioning this person, as https://www.linkedin.com/in/williamhgates or williamhgates.
includeAuthorDetails	boolean	false	Paid add-on. The author's full profile, or company record for a company page.
includeReactions	boolean	false	Paid add-on. Who reacted to each post.
maxReactionsPerPost	integer 1 to 100	5	Caps reactions per post.
includeComments	boolean	false	Paid add-on. Who commented on each post.
maxCommentsPerPost	integer 1 to 100	5	Caps comments per post.

Three things worth knowing before your first run, all of them measured rather than assumed:

• The result window ends near 500 posts. LinkedIn keeps reporting a much bigger total, but stops serving new results well before it. maxPosts is capped at 500 for that reason: past it you would be paying for pages that come back empty.
• contentType is a bias, not a hard filter. On 50 results per type, roughly 1 in 8 came back without that media. Expect a few extras rather than a clean set.
• A term LinkedIn does not recognise does not return nothing. It falls back to loosely related posts. If your results look off-topic, the term is what to change, not the filters.

---

What data does the LinkedIn Post Search return?

Every row carries a type field, because one run can produce several shapes. Filter on it before you export.

type	When it appears	What it carries
post	Always	The post and its author.
author_profile	includeAuthorDetails, person author	The full person record: role, experience, education, skills, location, links.
author_company	includeAuthorDetails, company author	The company record: industry, headcount, revenue band, funding, HQ, technologies.
reaction	includeReactions	One person who reacted, and which reaction.
comment	includeComments	One comment, its text, its author and its date.
summary	An add-on lookup found nothing	Says which lookup was empty, so a billed line is never a mystery.
error	Bad input or a free-tier stop	A plain-language explanation.

A post row, trimmed to show the shape:

{
  "type": "post",
  "post_type": "video",
  "author_name": "Satya Nadella",
  "content": "No one becomes a clinician to do paperwork...",
  "post_url": "https://www.linkedin.com/posts/satyanadella_activity-7302346926123798528",
  "posted_at": "2026-03-15",
  "posted_ago": "3d",
  "is_repost": false,
  "images": [],
  "video_url": "https://dms.licdn.com/playlist/vid/...",
  "article": null,
  "doc": null,
  "mentions": [{ "start": 42, "length": 9, "type": "COMPANY", "url": "https://www.linkedin.com/company/microsoft/", "name": "Microsoft" }],
  "share_url": "https://www.linkedin.com/feed/update/urn:li:activity:7302346926123798528",
  "engagement": { "total_reactions": 1240, "comments": 87, "shares": 33, "reactions": [{ "type": "LIKE", "count": 900 }] },
  "author": {
    "name": "Satya Nadella",
    "type": "profile",
    "username": "satyanadella",
    "headline": "Chairman and CEO at Microsoft",
    "linkedinUrl": "https://www.linkedin.com/in/satyanadella/",
    "avatar": "https://media.licdn.com/dms/image/v2/...",
    "linkedin_id": "ACoAAAF...",
    "website_url": null,
    "website_label": null
  },
  "reposted_by": null,
  "_metadata": { "post_id": "urn:li:activity:7302346926123798528", "extracted_at": "2026-09-06T18:00:00.000Z", "search_term": "ai agents" }
}

post_type is derived from the post itself, not guessed from the text: document, video, poll, article, image or text. A document post that also carries cover images is still a document. One honest limitation: a job post carries no media at all, so it comes back as text even when contentType: "jobs" found it.

Join the add-on rows to the posts on the author handle: author_slug on an author_profile or author_company row matches author.username on the post rows, and post_id on a reaction or comment row matches _metadata.post_id. Both handles are lowercased, so the join works whatever case LinkedIn spells the profile with.

author.type is profile for a person and company for a company or showcase page, and it tells you where the handle goes back in: a profile handle into "Posts by these people", a company handle into "Posts by these companies". That is the loop this actor is built for: search broadly, keep the authors worth watching, then search only them.

---

How much does it cost to search LinkedIn posts?

Pay per event, and the price drops on the higher Apify plans. Check the Actor's pricing tab for the exact figure on your plan.

What you get	Event
A post	post-scraped
An author's full profile	profile-enriched
An author's company record	company-enriched
One person who reacted	reaction-scraped
One comment	comment-scraped
A search term that returned nothing	empty-search

Three rules that decide what your bill actually looks like:

• Each author is billed once per run, however many of their posts come back and across every search term in that run. A topic is usually dominated by a handful of accounts, so this is the difference between paying per author and paying per post.
• A lookup that finds nothing still bills one item. A search request, or a reactions lookup on a post with no reactions, does the same work whether it comes back full or empty. That single item covers the request, which is what keeps thin results worth running instead of being refused outright. Lookups that fail are free.
• The free tier gives you 20 chargeable events a month so you can inspect every field before paying. Any paid Apify plan removes it.

---

What do people use the LinkedIn Post Search for?

• Market and topic monitoring: schedule a daily run on the terms your market uses and read what changed, instead of scrolling a feed an algorithm chose for you.
• Brand and competitor mentions: mentioningCompany returns posts about a company written by people outside it. That is the conversation your competitor cannot see either.
• Sales prospecting from intent: people describing a problem in public are the warmest inbound there is. Search the problem, filter the headline, enrich the author, export the list.
• Recruiting: find who is posting about a technology or discipline, then pull their profile in the same run.
• Content research: see what actually got engagement on a topic before writing your own post, with the reaction and comment counts attached.
• Building a warm audience: the reactions and comments add-ons turn one high-engagement post into a list of people who already care about the subject.
• Feeding an AI agent: the MCP section above wires this into ChatGPT or Claude so the search, the filter and the enrichment happen in one prompt.

---

LinkedIn Post Search vs cookie-based tools vs the official LinkedIn API

	This Actor	Cookie or browser tools	Official LinkedIn API
LinkedIn account needed	No	Yes, and it is the one at risk	Yes, plus partner approval
Search posts by keyword	Yes	Yes, until the session breaks	No. Post search is not exposed
Ban risk	None, nothing runs as you	Real, this is the whole risk model	None
Author enrichment	Built in, one lookup per author	Another tool, another session	Not available
Runs unattended	Yes	Only while the cookie lives	Yes
Setup	Paste a keyword	Extract a session cookie, refresh it forever	An approval process measured in weeks

The official API is the honest comparison to make: it is the safest option and it simply does not offer this. Keyword search over public posts is not a LinkedIn API product, at any tier.

🏆 Top LinkedIn Scrapers

	Profile Scraper Any profile URL to 72 structured fields, no cookies
	Posts Scraper Text, media and engagement from any profile or company
	Post Search  YOU ARE HERE
	Reactions Scraper Every person who reacted to a post, with their headline
	Comments Scraper Comments and nested replies, with reaction counts
	Company Scraper Headcount, industry, HQ, tech stack and IT spend of a company
	Company Employees Scraper Every employee of a company, filterable by seniority and department

FAQ

Can I search for posts from a specific person or company? Yes. profile takes people, company takes company and showcase pages, and authorsCompany takes everyone who works somewhere. All three take a list, and profile and company accept a URL, a plain handle, or the author.linkedin_id a previous run returned. authorsCompany names the employer, so it takes a company URL or company id. Combine any of them with a search term to ask "what did these accounts say about this topic".

One limit worth knowing: filtering the search by author returns fewer posts per account than reading their feed directly. When you want everything one account posted, LinkedIn Posts Scraper is the Actor for that.

Can I use this from ChatGPT or Claude? Yes. The LinkedIn MCP Server section above is the whole setup: paste the JSON into your assistant's config and ask in plain language. Pin the tools= list so it stays on these Actors.

Why did I get fewer results than the total LinkedIn reported? Two reasons, both measured. The reported total always runs a little ahead of what actually arrives: a term reporting 457 delivered 443, and one reporting 110 delivered 100. And LinkedIn serves a search in pages of 50 and stops at 10 pages, so 500 is the ceiling however big the total looks. The run stops at the first empty page rather than paying for the rest.

My results look unrelated to my search term. Is it broken? No, and this one surprises everyone: LinkedIn does not return an empty result for a term it does not recognise. It falls back to loosely related posts. Change the term, not the filters.

Do I pay for a search that finds nothing? One empty-search event for each term that comes back empty. The request costs the same whether it comes back full or empty, and a run with five terms makes five requests. A term that delivered posts is never also charged the floor, and a run that fails outright bills nothing.

Are reposts included? Yes, all of them, and every row tells you which is which: is_repost is true for a repost, and reposted_by names who shared it when LinkedIn says. Filtering your export to originals only is one column away.

Does the author add-on work for company pages too? Yes. A person author returns an author_profile row, a company page returns an author_company row with industry, headcount, revenue band and funding. Both are billed once per author per run.

Do I need a LinkedIn account or cookies? No. Nothing runs as you, so there is nothing to get restricted.

All Atomus scrapers

2.4M+ RESULTS DELIVERED

LinkedIn	Profile · Posts · Post search · Reactions · Comments · Company · Employees
Lead gen	Leads Finder
Google Maps	Places & local businesses
TikTok	Videos · Comments
X (Twitter)	Tweets & profiles
RedNote 小红书	Notes, users & comments
Douyin 抖音	Profiles, videos & comments
Weibo 微博	Posts & profiles
Bilibili 哔哩哔哩	Videos & creators

Support

Hey, I'm Chico, founder of Atomus. I built this Actor and I answer the messages about it. Something broke? A field you need isn't there? Not sure it fits what you're doing? Send me a message, most answers come the same day. 💬  DM me on LinkedIn    or hello@dendelabs.com

⚠️ Disclaimer

This Actor collects only publicly available LinkedIn posts, the kind anyone can see without logging in. It does not log into any account, use cookies, or access private or connection-only content. You are responsible for using the output in line with LinkedIn's terms and with the data-protection law that applies to you, including GDPR and CCPA where relevant.