LinkedIn MCP server: post to LinkedIn from Claude or ChatGPT
LinkedIn has no official MCP server. Which LinkedIn MCP servers break its terms, which use the official API, and how to post safely from Claude or ChatGPT.
William Lindberg, Customer Success
If you want Claude or ChatGPT to post to LinkedIn for you, you need a LinkedIn MCP server. The catch: LinkedIn doesn't make one. Every LinkedIn MCP server you can install today comes from someone else, and they fall into two very different camps. Some log in as you and drive LinkedIn like a robot, which LinkedIn's terms forbid. Others go through LinkedIn's official API, which is the route LinkedIn built for this.
This guide explains the difference, how to tell which kind you're looking at, and how to set up LinkedIn posting from your assistant without putting your account at risk.
What an MCP server is
MCP (Model Context Protocol) is an open standard that lets an AI assistant use outside tools. An MCP server is the tool's side of that connection: it tells the assistant which actions it can take, like "create a post," and runs them when you ask.
Does LinkedIn have an official MCP server?
No. As of October 2026, LinkedIn hasn't released or announced an MCP server of its own, and its developer documentation covers its regular APIs with no MCP option. Every LinkedIn MCP server out there is third-party.
What LinkedIn does have is an official API for posting. Its Posts API publishes text, images, documents, and more to a member's profile or a Company Page, after the member grants permission through LinkedIn's own login screen. Posting to a Company Page needs LinkedIn's Community Management API, and LinkedIn vets each app before granting full access. That vetting is likely one reason many "LinkedIn MCP" projects skip the official API altogether.
The three ways to connect LinkedIn to an assistant
1. Open-source servers that use your login session
Many free LinkedIn MCP servers on GitHub work by running a browser that's logged in as you, or by reusing your session cookie. They can do things the official API doesn't allow, like searching people or reading profiles. That's also the problem.
LinkedIn's User Agreement (section 8.2) says you agree not to:
"Use bots or other unauthorized automated methods to access the Services, add or download contacts, send or redirect messages, create, comment on, like, share, or re-share posts, or otherwise drive inauthentic engagement"
The same list also bans software, scripts, and browser plugins that "scrape or copy the Services." LinkedIn's help page on prohibited software and extensions spells out the consequence: accounts using such tools risk being restricted or shut down. In March 2026, LinkedIn's Tim Jurka also wrote that the company is working to "curb comment automation, including the use of third party software or browser extensions, which are not allowed on LinkedIn."
So a session-based server can work for weeks and then get your profile restricted. For a founder profile or a client's Company Page, that's a bad trade.
2. Servers that wrap LinkedIn's official API
Some open-source and hosted servers use LinkedIn's official API with proper login (OAuth). These stay inside LinkedIn's rules. The trade-off is setup: a self-hosted one usually needs your own LinkedIn developer app, and Company Page posting needs LinkedIn to approve that app. They also tend to post right away. LinkedIn's API has no "publish later" setting, so scheduling has to come from somewhere else.
3. A scheduler's MCP server
The third route is a social media scheduler that runs its own MCP server and already publishes through LinkedIn's official API. The assistant talks to the scheduler; the scheduler talks to LinkedIn. You get scheduling, drafts, and a calendar on top of the posting, and you don't need a LinkedIn developer app of your own.
That's how Lemonpost's AI posting works. Any MCP client (Claude, ChatGPT, Codex, Cursor, or anything else that supports MCP) can draft, schedule, queue, and publish posts. For LinkedIn, that means text, up to 20 images, PDF document posts, a first comment, and the link-preview toggle, to your profile or a Company Page, all through the official LinkedIn API. The same pre-publish checks and approval rules apply as in the app. The LinkedIn scheduler page shows every LinkedIn format it handles.
Official API vs session-based servers
| Official API (including schedulers) | Session or scraping-based | |
|---|---|---|
| How it connects | You grant access on LinkedIn's login screen | Your browser session or cookie |
| Allowed by LinkedIn's terms | Yes | No (User Agreement, section 8.2) |
| Risk to your account | Low | Restriction or shutdown |
| Post to your profile | Yes | Yes |
| Post to a Company Page | Yes, after LinkedIn approves the app | Depends on the project |
| Search people, read profiles | Very limited | Often, which is what breaks the rules |
| Breaks when LinkedIn changes its site | Rarely, the API is versioned | Often |
| Schedule for later | Only if the server adds it | Only if the server adds it |
If all you want is to post, you don't need the session-based kind. Posting is exactly what the official API is for.
Set up LinkedIn posting from your assistant with Lemonpost
The setup takes about 10 minutes. The steps are the same for Claude, ChatGPT, Codex, and Cursor; only the place you paste the server differs.
- Connect LinkedIn in Lemonpost. In the app, connect your LinkedIn profile and any Company Pages you manage. The login runs on LinkedIn's own site, so this step happens in the browser, not in the chat.
- Create a token in settings. This is the key your assistant uses to act as you. Treat it like a password and revoke it whenever you want.
- Add the Lemonpost MCP server to your client. Paste the token where your client adds MCP servers. Menu names change often, so the help center keeps the current steps.
- Test with a draft. Ask: "Draft a LinkedIn post about our new opening hours for my Company Page and save it as a draft." Open Lemonpost and check the calendar.
- Turn on approvals if others post. On Growth and Pro, set the brand's approval policy so posts wait for a sign-off.
For more on connecting each client and prompts for other networks, read how to schedule social media posts from Claude or ChatGPT.
5 LinkedIn prompts that work
Tell the assistant three things: which account (profile or Company Page), when, and whether it's a draft or scheduled.
- "Turn this blog post into 3 LinkedIn posts for my profile. Under 1,300 characters each, a strong first line, no hashtags. Put the article link in the first comment and save them as drafts for Tuesday, Wednesday, and Thursday at 8 am."
- "Schedule this PDF as a document post on our Company Page for Thursday at noon. Title it 'Q3 hiring report' and write a 3-line intro that tells people why it's worth a swipe."
- "Here are 6 photos from yesterday's event. Make one LinkedIn post for the Company Page with all 6 images, thank the speakers by name, and schedule it for tomorrow at 9 am."
- "What's scheduled on LinkedIn for the next two weeks? List the date, profile or Page, and the first line of each post."
- "Rewrite Tuesday's LinkedIn post in plainer language, cut it to 150 words, and turn off the link preview. Keep the same time."
Give it your voice first. Paste a short note on how you write, 3 posts you liked, and 3 things you'd never say. The drafts get much closer on the first try.
Guardrails worth keeping on
Drafts first. For the first couple of weeks, have the assistant save drafts only. Read them, fix them, then schedule. When the drafts stop needing fixes, let it schedule.
Approvals. On Growth and Pro, each brand has its own approval policy: off, contributors only, or every post. A post that needs sign-off sits in review and can't be scheduled until someone approves it. That holds even when an assistant created it. See approvals.
Validation. Every post is checked against LinkedIn's rules before it sends. The assistant can't post something the app would reject.
One token per person or assistant. If several people connect assistants, make a token each. Then you can cut one off without cutting off everyone.
Keep comments human. Let the assistant help you draft and schedule posts. Reply to comments yourself. LinkedIn has said it's working to make engagement pods ineffective and curb comment automation (more in how the LinkedIn algorithm works). Real replies build the relationships anyway.
FAQ
Is there an official LinkedIn MCP server? No. As of October 2026, LinkedIn hasn't released an MCP server. Every LinkedIn MCP server is third-party, either built on LinkedIn's official API or on your logged-in browser session.
Is it safe to use a LinkedIn MCP server? It depends on how it connects. Servers that use LinkedIn's official API with OAuth stay within LinkedIn's rules. Servers that drive your logged-in session break LinkedIn's User Agreement, and LinkedIn says accounts using such tools risk restriction.
Can Claude post to LinkedIn? Yes, through an MCP server. Connect Claude to a server that publishes through LinkedIn's official API, like Lemonpost, and ask it to draft, schedule, or publish a post to your profile or Company Page.
Can ChatGPT post to LinkedIn? Yes, the same way. ChatGPT can act as an MCP client, so once it's connected to a server like Lemonpost it can create LinkedIn drafts, schedule posts, and publish them through your account.
Can an MCP server post to a LinkedIn Company Page? Yes, if it uses LinkedIn's official API and LinkedIn has approved the app for Company Page posting. Lemonpost posts to profiles and to Company Pages you manage.
Can I schedule LinkedIn posts from an AI assistant? Yes, if the server supports scheduling. LinkedIn's API publishes immediately, so scheduling comes from the tool. Lemonpost's MCP server can draft, schedule, queue, or publish LinkedIn posts.
Which Lemonpost plan includes MCP posting? Every plan, starting at $29 a month on Creator (5 connected channels, unlimited seats). Approvals are on Growth ($99 a month) and Pro ($249 a month). Every plan has a 7-day free trial.
