Which LinkedIn MCP Servers Can Actually Publish a Post?

“LinkedIn MCP server” sounds like one category of thing. It is not. Almost every result you find reads LinkedIn; a much smaller number can write to it. If your goal is to draft a post in Claude or ChatGPT and have it appear on your profile, that single difference decides whether a tool is useful to you at all.

So this is a capability audit, not a ranking. One question, asked of each option: can it create a post?

Which LinkedIn MCP servers can actually publish a post?

MCP — the Model Context Protocol, an open standard for connecting AI assistants to outside tools — lets anyone build a LinkedIn connector. Here is what the main options do.

Where a cell says “not verified here”, that means exactly what it says: it was not checked directly, so it is not being claimed either way. Prices and plans change, and a guess in a table looks identical to a fact.

What do these tools say about themselves?

Two of these are worth quoting directly, because the tools themselves are clearer than most comparison articles.

The top Google result for “linkedin mcp” is the open-source repository stickerdaniel/linkedin-mcp-server. Its own README states that it “does not publish or create posts on your behalf”, and warns that accounts “can be restricted or banned”. That is an unusually straight piece of documentation and it should be taken at face value. The project is a well-made profile and job scraper. It is not a publishing tool, and it never claimed to be.

Taplio is on the other side of the line — it can post — but its own comparison page says its MCP is “text-only... you cannot upload images, video, or PDF carousels”. Again, credit for saying so plainly. If your posts are mostly plain text, that limit may not matter to you at all.

Why can most LinkedIn MCP servers not post?

This is the part that explains the whole table. Reading and writing are not equally hard on LinkedIn, and the gap is not about engineering skill.

Reading can be scraped. A profile page is HTML. If a tool has your logged-in browser session or a copied cookie, it can fetch pages and pull text out of them. No permission from LinkedIn is involved, which is why scrapers are quick to build and why there are so many of them. It is also why their documentation warns about account restrictions.

Publishing needs permission. To create a post through LinkedIn’s sanctioned route, an app needs the w_member_social permission, granted through the “Share on LinkedIn” product. That is not a code trick. It requires:

• A registered LinkedIn developer application, with the “Share on LinkedIn” product requested and approved.
• A real OAuth flow (OAuth is the “sign in with...” handshake where LinkedIn, not the app, collects your approval). The app never sees your password.
• A server that holds tokens securely and can be reached by the AI app. A tool running only on your laptop, with no registered app behind it, has nothing to post with.

So a scraper does not skip publishing because it forgot. It skips it because the only way a scraper could post is by driving your logged-in session to write — exactly the behaviour that gets accounts restricted, and exactly what the leading repository declines to do.

What about Apify, Bright Data and Zapier?

These three often appear in the same search results, so it is worth being precise about what they are for.

Apify and Bright Data offer LinkedIn MCP endpoints aimed at data extraction — pulling profiles, companies and job listings at scale. They are serious infrastructure for a real job, and that job is not publishing. If you came looking for a way to post, they are not the answer; if you came looking for bulk data, they are much closer to it than any of the posting tools.

Zapier’s LinkedIn MCP exposes only around two actions. Zapier’s strength has always been chaining apps together rather than being a deep interface to any one of them, and the MCP surface reflects that. Check which two actions are exposed for your account before assuming it covers what you need.

What separates the tools that can post?

Once you are down to the short list that can genuinely publish, the differences are about what a post is allowed to contain, and who controls the moment it goes out.

• Media. Plain text is the baseline everywhere. Beyond that: a single image, a swipeable carousel of 2–9 images, or a PDF document carousel. Taplio’s MCP is text-only by its own description. Kalovio supports all four, with a 3000-character limit on the text, which is LinkedIn’s own hard limit. One caveat worth carrying: multi-image and document posts go through LinkedIn’s newer versioned interface, which may not be switched on for every self-serve app. If yours is not, Kalovio returns a clear message rather than failing silently, and single-image posting still works.
• Approval. This matters more than it sounds. In Kalovio, publishing and deleting are two-step: the first call returns a preview plus a signed token covering the exact content, and nothing happens until you confirm. That is what stops a stray instruction hidden in a web page from posting for you.
• Scheduling. Posting now is easy. Posting at 08:30 next Tuesday needs a server that is still awake when you are not.
• Reconnecting. LinkedIn member tokens expire in roughly 60 days, and self-serve apps get no refresh token. Every honest tool in this category will ask you to reconnect periodically. If one claims permanent access, ask how.

How do you check a tool’s claim yourself?

None of the four ranking SaaS pages in this market names a single OAuth scope or API endpoint. “Official, no scraping” is written on all of them and proven on none of them. You can test it in under a minute:

• Ask which scope it requests. For posting the answer is w_member_social. A tool that cannot name it is not built on the official API.
• Ask which endpoint it posts to. POST /v2/ugcPosts for text, single images and article shares; LinkedIn’s versioned /rest/posts for multi-image and document posts.
• Watch the sign-in. A real OAuth flow sends you to linkedin.com, shows what you are granting, and is revocable from your settings. A box asking you to paste a cookie or a password is not that.

So which one should you use?

Match the tool to the job, and ignore the category name:

• Reading profiles, companies or jobs? The scraper repositories and the data platforms do this well. Read their own warnings about account restrictions first — they wrote them for a reason.
• Writing and publishing your own posts, text only? Several tools handle this, including Taplio’s MCP.
• Publishing with images, carousels or PDF documents, and scheduling them? That is a much shorter list. Kalovio is on it, is free while in beta, and is honest about what it cannot do: it cannot read posts it did not create, cannot send messages or connection requests, cannot scrape leads, and has link-click tracking rather than real LinkedIn engagement analytics.

The market is noisy mostly because one word — “LinkedIn MCP server” — covers two completely different products. Decide whether you want to read or to write, and the list gets short very quickly.