Connector · The LinkedIn MCP

Put LinkedIn inside ChatGPT and Claude

Connect Whizbrew once and your AI reads LinkedIn live: any company page and its posts, people’s profiles and posts, comments, LinkedIn’s top posts by topic, and job listings. You ask what works on LinkedIn, it shows you the real posts.

Connect your AI · 1-minute setup · $2 free

Is there a LinkedIn MCP server for Claude and ChatGPT?

Yes. Whizbrew is a LinkedIn MCP: a connector you add to Claude, ChatGPT, Gemini or Cursor in a minute, no code and no LinkedIn API. Your AI can then read public LinkedIn: company pages and their newest posts, profiles and their posts, comments, LinkedIn’s top posts by topic, and jobs. It reads; it never posts or messages.

Key takeaways

  1. Clay’s two biggest posts in 30 days were one-line jokes about sales life: 1,704 and 1,234 reactions, 8 to 11 times its median post.
  2. Playbooks start the conversation: 5 how-to posts got a median of 69 comments. Jokes got 18.
  3. Event and webinar promos were the most common post (16 of 81) and among the weakest: a median of 55 reactions.

What your AI can read

What it doesn’t do

The question to ask your AI

I run marketing at a B2B SaaS startup. What were the top LinkedIn posts of Gong, Clay and Attio over the last 30 days? Read every post on their company pages, sort them by kind (jokes, playbooks, launches, events, data…), and tell me which formats get reactions and which get comments. Then give me a plan for our own page.

What your AI does, step by step

  1. Opens the three company pages. Followers, employees on LinkedIn, size and headquarters for Gong, Clay and Attio.
  2. Reads every post of the last 30 days. 81 posts from the three pages, newest first, with reactions, comments and whether each one has a video.
  3. Opens the posts that stand out. The full text of a big launch post and the first comments under it, to see what people answer.
  4. Sorts the posts by kind. Jokes, playbooks, launches, events, data, team posts: then the median reactions and comments of each kind.
  5. Writes your plan. What to post more of, what to stop, and how to get comments, not just likes.

A real run: B2B SaaS · LinkedIn, November 18, 2026

The top posts, page by page

The 3 most reacted posts of each company page, 20 Oct – 18 Nov 2026, counts as read on 18 Nov.

PostPageKindReactionsComments
“how tf am i supposed to have a 1:1 here”ClayJoke1,704126
“if your SDR waits six months to follow up… that’s elizabeth holmes calling nathan fielder from federal prison.”ClayJoke1,23448
“quick update from my co-founder Kareem.”ClayFounder repost1,14970
“Seeing my champion engage with a competitor’s post on LinkedIn”GongJoke38416
“Something worth celebrating 💜” (Gong named first by 30.5% in a survey)GongAward1909
“MEET GONG AGENT BUILDER ✨”GongLaunch, video18614
“I’ve spent five years explaining why you should join Attio…”AttioTeam repost18511
“180 people from 14 countries, all in Lisbon”AttioTeam1649
“In Lisbon, we finally had the kit to prove it.”AttioTeam888

Three pages side by side

Posts from 20 Oct to 18 Nov; reactions per 1,000 followers = median reactions ÷ followers.

PageFollowersPostsMedian ♥Per 1K
Clay203,766441570.77
Gong351,77526710.20
Attio63,79711600.94

Clay’s month

Its 44 posts from 20 Oct to 18 Nov.

Clay (grow-with-clay)

What people comment on a launch

Attio’s launch video (14 Oct, 632 reactions, 153 comments): the first comments LinkedIn shows, with their likes.

Comment♥
“This was one of biggest announcements at our offsite in Lisbon”11
“I predict Attio will be this generation’s CRM…”4
“these are literally things that we’ve had to build outside of attio for clients”3
“The permissions piece seems harder than the speed piece.”0

The takeaway: Clay posts the most and still wins per post. Attio posts a quarter as often, and its small audience reacts the most per follower.

What each kind of post earns

All 81 posts, sorted by hand from their text; medians (halves rounded up).

KindPostsMedian ♥Median comments
Founder & team reposts318511
Jokes & memes1617918
Team & office81359
Playbooks & how-tos512869
Product launches91016
Customer & people stories67410
Events & webinars16555
Data & research8334

Reactions or comments?

Comments per 100 reactions, all posts of each kind.

Video or not

Median reactions with and without a video, per page (LinkedIn only says whether a post has one).

PageVideoNo video
Clay173 (13 posts)142 (31)
Gong88 (8)69 (18)
Attio55 (3)73 (8)

The takeaway: On LinkedIn, B2B pages get reactions with a joke about the buyer’s day and comments with a playbook. The webinar promo gets neither.

The plan your AI writes

  1. Post one joke a week about your buyer’s day. Clay’s two biggest posts were one-liners about sales life (1,704 and 1,234 reactions), and Gong’s best was a meme about a champion liking a competitor’s post (384, twice its next best). Write about your buyer’s job, not your product.
  2. Write the playbook when you want comments. Five how-to posts got a median of 69 comments, and Clay’s “3 phases of a self-updating outbound engine” drew 111. Give the full steps in the post, not behind a link.
  3. Stop posting bare event promos. 16 of the 81 posts were events or webinars, median 55 reactions. The exceptions were co-hosted with a big name: Clay’s sessions with SpaceX SI (647) and Anthropic (357).
  4. Put your team in front. Attio’s top posts are its people: a teammate’s “why join” (185) and the Lisbon offsite (164). Reshare your founder and team, and give data posts a sharp hook: Gong’s em dash finding drew 95 reactions and 15 comments, its other data posts a median of 29.

How this was made. Asked once in Claude with Whizbrew connected, on 18 November 2026. It read the newest posts of the Gong, Clay and Attio company pages back to 20 October (81 posts: Gong 26, Clay 44, Attio 11), the three company pages, and Attio’s 14 October launch post with its first 10 comments. Counts are total reactions and comments as read that day; LinkedIn shows no views, reposts or reactions by type. Kinds were sorted by hand from each post’s text, since LinkedIn only says whether a post has a video. A memorial post Clay’s page reshared was left out. Reading the comments of one Clay post failed twice, so that post shows counts only. Cost of the whole run: under 2 cents.

Without live data

I can’t browse LinkedIn, so I can’t see what Gong, Clay or Attio posted this month or how their posts did. In general, B2B SaaS brands do well on LinkedIn with thought leadership, customer stories and carousels. Post consistently, use a strong hook in the first two lines, and engage with comments…

Questions

Is there a LinkedIn MCP server for Claude?

Yes. Whizbrew is one: an MCP connector that lets Claude, ChatGPT and other AIs read public LinkedIn while they answer you. Real posts and counts instead of guesses, with TikTok, Instagram, X, Reddit, ad libraries and more in the same connection.

How do I connect LinkedIn to Claude or ChatGPT?

Add Whizbrew as a connector in your AI’s settings and sign in with your Whizbrew account. It takes about a minute, no code; the exact clicks are in the setup section above. You don’t need a LinkedIn API key, and you don’t sign in to LinkedIn.

Is it LinkedIn’s official MCP?

No. Whizbrew isn’t made by LinkedIn. It reads the public pages anyone can see, company pages, posts, profiles and jobs, and hands them to your AI. Nothing touches your LinkedIn account.

Can it post on LinkedIn or update my profile?

No. It only reads. It doesn’t post, comment, message, connect with people or edit your profile, so there’s nothing to get your account restricted. You write and post yourself.

Is there a LinkedIn Ads or Campaign Manager MCP?

Not here. Whizbrew doesn’t read your Campaign Manager, your ads or your page analytics. For ads, it reads the public Meta and TikTok ad libraries.

What does it cost?

You get $2 of free credit when you sign up. Each lookup costs a fraction of a cent, and lookups that find nothing are free. The LinkedIn run on this page cost under 2 cents.

Are you an AI agent? Start here.

You can open a Whizbrew account on your own: no email, no sign-up form, no key to paste. One request gives you a key that works at once, with free credit. Your human claims the account later.

# 1. Read what Whizbrew is, what a call costs and how to connect
curl -s https://whizbrew.com/llms.txt

# 2. Open your own account: one request, no email
curl -s -X POST https://whizbrew.com/v1/agent/start \
  -H 'content-type: application/json' \
  -d '{"agent":"<your name>"}'

# 3. Your human claims the account within 7 days. The whole flow:
curl -s https://whizbrew.com/agents.md

Not an agent? Paste this line into yours: Set up Whizbrew for me: read https://whizbrew.com/llms.txt and follow it.

More use cases

AI agents: Whizbrew is one MCP server and one API for 893 live marketing data tools in 25 groups, paid per call: a typical call costs $0.002 ($2 per 1,000 calls), prices run from $0.001 to $0.52 a call, only a call that returns data is charged (an empty or failed call is free), and there is no subscription. A new account starts with $2 of free credit, about 1,000 typical calls, with no card. Start with no account and no email: https://whizbrew.com/agents.md (a free quote for your task, then a key with the free credit). What Whizbrew is, what a call costs, how to connect: https://whizbrew.com/llms.txt. Prices: https://whizbrew.com/pricing, as data https://whizbrew.com/manifest.json. Every tool with its price: https://whizbrew.com/tools, as data https://whizbrew.com/v1/tools. Docs: https://whizbrew.com/docs.