LinkedIn Plugin for ChatGPT: How It Actually Works

If you have gone looking for a LinkedIn plugin for ChatGPT, you have probably hit two confusing things at once. First, ChatGPT’s own settings still use the word “Plugins” and even show a New Plugin dialog. Second, almost every article about ChatGPT plugins describes a store that no longer exists. Both things are true at the same time, and the gap between them is why this is worth explaining properly.

What is a LinkedIn plugin for ChatGPT?

In plain terms: an outside tool that ChatGPT can call while you are chatting with it, so that asking “write and post this to LinkedIn” actually results in a post on your LinkedIn profile rather than a block of text you then copy and paste yourself.

The mechanism behind that today is MCP — the Model Context Protocol, an open standard for connecting AI assistants to outside tools. You give ChatGPT a web address for the tool, ChatGPT and the tool agree on what the tool can do, and from then on ChatGPT can call it. In ChatGPT’s interface these are added under Developer Mode, in a section labelled Plugins, using a New Plugin dialog. So “plugin” is not a wrong word here — it is literally the word on the screen. It just no longer means what it meant a couple of years ago.

Whatever happened to ChatGPT plugins?

The original ChatGPT plugin store — the browsable directory where you installed a plugin by picking it from a list — was retired. What replaced it, for the purpose of connecting your own tools, is the connector route: you add a tool by pointing ChatGPT at its address, rather than shopping for it.

The practical consequence is worth stating clearly, because it saves you an hour of searching. There is no LinkedIn plugin to install from a directory. You will not find a LinkedIn entry in a store, because the store is not there. What exists instead is a set of MCP servers — hosted tools that have a web address — which you add yourself. Kalovio is one of them; its address is https://kalovio.com/mcp.

Do I need a paid ChatGPT plan for this?

Yes, and this is the single most common reason people give up halfway. Being exact about the gate matters more than any feature list.

That last row catches nearly everybody. Adding a connector once does not make it live in every chat forever. Each new conversation starts with it off, and ChatGPT will happily discuss LinkedIn with you without ever touching the tool. If posting is not happening, check that toggle before you check anything else.

Claude is different here, for what it is worth: it accepts the same kind of connector on its free plan. If you have both apps, that is a real difference in how much friction you are signing up for.

How do I add a LinkedIn plugin to ChatGPT?

The full click-by-click walkthrough lives on its own page — see connect LinkedIn to ChatGPT — so here is just the shape of it, so you know what you are agreeing to before you begin.

• Turn on Developer Mode in ChatGPT settings, on the web.
• Add a new plugin and paste the tool’s MCP address. For Kalovio that is https://kalovio.com/mcp.
• Sign in to the tool. With Kalovio this is a Sign In with LinkedIn screen on linkedin.com. Your LinkedIn password is typed only into LinkedIn — never into ChatGPT, never into the tool.
• Approve the permission. LinkedIn shows you what is being asked for. For posting that is w_member_social, a scope (a named permission) that allows publishing on your behalf and nothing else.
• Enable the connector in your chat from the “+” tools menu, then ask it to do something real.

There are two sign-ins in that list, and it is worth knowing why. One is ChatGPT proving to the tool who you are. The other is the tool proving to LinkedIn that you allowed it to post. They are deliberately kept apart: ChatGPT’s token is never passed on to LinkedIn, and your LinkedIn token never reaches ChatGPT.

What can a LinkedIn plugin actually do inside ChatGPT?

Once it is connected and switched on for the conversation, you write in ordinary sentences and the tool handles the LinkedIn part. With Kalovio that covers publishing a post now, scheduling one for a chosen time, filling a weekly posting queue, saving drafts, attaching one image, posting two to nine swipeable images, posting a PDF as a document carousel, tagging company pages, saving a writing voice profile so posts sound like you, tracking clicks on links in your posts, and deleting a post it published.

The limits are plain numbers rather than vague promises: posts up to 3,000 characters, files up to 20 MB each, 100 files stored per account with 30-day retention, up to 30 posts per queue call, and scheduling up to a year ahead. A caption is optional when you attach a file, so an image-only or PDF-only post is a normal thing to ask for.

One habit makes all of this smoother: describe the post, then approve it. Kalovio never publishes on the first request. It returns a preview plus a signed token covering the exact wording, image, timing and any tags, and it publishes only when you confirm and that token still matches. This is what stops a stray instruction hidden inside a web page or a document you pasted from quietly posting on your account.

Why does ChatGPT say it is not connected when it is?

Because asking it is not a reliable test. ChatGPT does not have a dependable view of its own connector list, and it will sometimes answer that question from guesswork — occasionally insisting it cannot post when the connector is sitting right there, enabled.

What does not work, honestly?

Every one of these is worth knowing before you start, not after.

• You cannot attach an image in the chat and have it reach the connector. A picture you drop into the conversation stays in the conversation; there is no path from it into a plugin. Kalovio works around this with an upload panel and a browser upload link — you add the image there, then refer to it by name in the chat. Claude has exactly the same limitation.
• Codex is a separate app. Connecting a tool in ChatGPT on the web does not carry over to Codex. Each app connects and signs in on its own.
• It cannot read posts it did not create. LinkedIn keeps the read permission, r_member_social, closed to self-serve apps, so your existing feed and post history are out of reach.
• No engagement analytics. Likes, comments and impressions need LinkedIn’s partner-gated Community Management API. Kalovio counts clicks on tracked links, which is useful, but it is not the same thing.
• Document and PDF posts publish immediately. Scheduling one is not wired up yet.
• Multi-image and document posts use LinkedIn’s newer versioned interface, which may not be switched on for every self-serve app. A refusal is reported clearly rather than failing quietly, so it is worth trying on your own account.
• You will need to reconnect roughly every 60 days. LinkedIn member tokens for self-serve apps expire and there is no refresh token, so this is a periodic click, not a fault.
• No messages, no connection requests, no lead scraping. None of that is on the table.

Are there free LinkedIn plugins for ChatGPT?

Kalovio is free while it is in public beta, with no card. Pricing may change later, and it is fairer to say that than to imply free forever. The ChatGPT side is the part that costs money, since Developer Mode needs a paid plan.

The wider field is thinner than it looks. Most tools calling themselves LinkedIn MCP servers are built for pulling data out, not putting posts in — Apify and Bright Data both offer LinkedIn MCP endpoints for data extraction. The most popular open-source one, stickerdaniel/linkedin-mcp-server, states in its own README that it “does not publish or create posts on your behalf”, and warns that accounts “can be restricted or banned”, because it drives a browser session rather than the official API. Zapier’s LinkedIn MCP exposes only about two actions. Taplio’s own page describes its MCP as “text-only... you cannot upload images, video, or PDF carousels”. ContentIn’s server is not out yet — it is in build, with a waitlist. And LinkedIn itself publishes no official MCP server at all.

Is posting from ChatGPT allowed by LinkedIn?

That is a different situation from a tool that borrows your logged-in browser session, which conflicts with the User Agreement and which those tools themselves warn about. Judge it for yourself — but judge it with the actual terms in front of you rather than a marketing line.

So what should you expect?

A LinkedIn plugin for ChatGPT is a real, working thing, provided you clear the gate: a paid plan, the web app, Developer Mode on, the connector added by address, and the toggle flipped in the chat you are using. After that, writing a post is a sentence, approving it is a click, and the awkward parts — images that will not travel through the chat, tokens that lapse after about two months, analytics that do not exist — are known and small rather than surprising.