Updated
What Is a LinkedIn MCP Server?
A LinkedIn MCP server is a program that gives an AI assistant such as Claude, Cursor, or Claude Code a set of tools for LinkedIn data through the Model Context Protocol (MCP). You connect it once, then ask questions in plain language, such as "which of this person's posts did best this year?", and the assistant calls the server's tools to fetch the answer. LinkedIn publishes no MCP server of its own, so every one you can use is built by a third party. Flux (fluxgraph.app) runs a hosted one.
Updated November 2026.
What you'll learn
- What MCP is and what a "LinkedIn MCP server" adds to it
- The kinds of tools these servers expose
- The three connection designs and what each can read
- What no LinkedIn MCP server can see
MCP in one paragraph
The Model Context Protocol is an open standard, introduced by Anthropic in November 2024, that lets an AI client discover and call tools from a server. A server lists its tools with names, descriptions, and input schemas; the client picks a tool, sends arguments, and gets a result back. The assistant never needs to know how the data was fetched. A LinkedIn MCP server is simply an MCP server whose tools return LinkedIn data.
What the tools do
Tool sets vary by server, but they cluster into a few jobs.
- Read a profile or company page. Headline, follower count, recent posts, and engagement counts.
- Analyze posting patterns. Which days, hours, formats, and lengths beat an account's own baseline. Flux returns the sample size behind each finding.
- Search posts. Find posts by keyword or topic across accounts you track.
- Map engagement. Who reacts and comments, aggregated across posts, by company and industry.
- Act on LinkedIn. Some servers can post, like, comment, connect, or message. Flux cannot; it is read-only.
Flux's server exposes more than 60 tools across these jobs. The full list, with inputs and return keys, is in the MCP tool reference.
The three ways a LinkedIn MCP server can connect
The connection design decides what a server can read and what it risks, so it is the first thing to check.
| Design | How it gets data | Typical use |
|---|---|---|
| Logged-in session | Automates a browser or reuses your LinkedIn cookie; runs as you | Open-source servers for reading whatever you can see when logged in |
| Official LinkedIn API | Your own developer app with OAuth | Posting to and managing your own profile or company page |
| Hosted, read-only | A remote endpoint; the provider reads public data on its own systems | Researching any public profile or company page without logging in |
Flux is the third kind. It reads public LinkedIn data through a third-party data provider, not through an official LinkedIn API, and it is not affiliated with or endorsed by LinkedIn. You never give it a LinkedIn password or cookie, and nothing runs under your account.
What no LinkedIn MCP server can see
Some data is private to the account owner and visible only inside LinkedIn: impressions, profile viewers, dwell time, and anything behind a login such as messages. No third-party server can read those for someone else's account. Reactors and commenters, by contrast, are public and named, and they aggregate across posts. That second half is where most of the research value is.
Trying one
Flux's LinkedIn MCP server is hosted at https://www.fluxgraph.app/mcp. A free preview covers up to 50 recent posts from a public profile or company page, within a daily budget. Full post history, weekly refresh, trends over time, and the engagement graph are on the Edge plan at $15 per month as of November 2026. Setup for Claude, Claude Code, and Cursor is in connecting Claude and other AI clients.
Related questions
- Does LinkedIn have an official MCP server?
- Is there a Claude MCP for LinkedIn?
- Can Claude scrape LinkedIn?
FAQ
Is a LinkedIn MCP server the same as the LinkedIn API?
No. The LinkedIn API is LinkedIn's own interface for approved developer apps, limited to your own content and ad accounts. An MCP server is a tool layer for AI clients. Some MCP servers are built on the official API; most are not, and read data another way.
Do I need to install anything to use one?
It depends on the server. Open-source servers run on your machine and need Python or Node plus your LinkedIn login. Hosted servers such as Flux are a URL you add to your client; in Claude you add a connector and sign in, and in developer clients you paste a config.
Can a LinkedIn MCP server post for me?
Some can. Servers built on the official API or on your logged-in session can publish posts, and session-based ones may also like, comment, and message. Flux is read-only on LinkedIn: it can help you draft and score a post from your own history, but you publish it yourself.
Is using a LinkedIn MCP server allowed?
LinkedIn's User Agreement restricts automated access and data collection, so no third-party tool can promise LinkedIn's approval. The risk sits with servers that automate your own account. Flux reads only public data on its own systems and takes no actions on LinkedIn. This is not legal advice.
Does LinkedIn Have an Official MCP Server? (2026)
No. As of November 2026 LinkedIn publishes no MCP server and exposes none of its APIs as one. Every LinkedIn MCP server is third-party. Here is what exists, how each kind connects, and what to check before you install one.
Is There a Claude MCP for LinkedIn? (2026)
Yes, but not from Anthropic or LinkedIn. Claude connects to LinkedIn data through third-party MCP servers. How to add one in Claude, Claude Code, and Cursor, which kind to pick, and what Claude can and cannot do with LinkedIn once connected.