LinkedIn MCP Server: How to Set One Up in 2026 (Open-Source vs Hosted)

Search “LinkedIn MCP” and you'll mostly find one of two things: an open-source server you install and run yourself, or a hosted connector you just plug in. They sound similar, but they're built for very different people - and one of them comes with an account-safety warning that's easy to miss.

This guide explains what a LinkedIn MCP server is, how to set up the popular open-source one, what it can and can't do, and how the hosted approach compares - so you can pick the right one before you spend an afternoon in a terminal.

What Is a LinkedIn MCP Server?

The Model Context Protocol (MCP) is an open standard that lets AI assistants securely connect to outside tools and data. A LinkedIn MCP server is simply an MCP connector for LinkedIn: it gives an assistant like Claude, ChatGPT, Gemini, or Poke permission to work with LinkedIn on your behalf.

What “work with LinkedIn” means depends on the server. Some are built to read data - looking up profiles, companies, or job listings. Others, like Postbeam's, are built for the content workflow: drafting posts in your voice, pulling your analytics, and scheduling. That difference matters more than the setup instructions, so keep it in mind as we go.

The Two Kinds of LinkedIn MCP Server

Option 1: The Open-Source LinkedIn MCP Server

The best-known one is linkedin-mcp-server - an independent, community project (not affiliated with LinkedIn, supported by Unipile) that lets AI assistants read LinkedIn data through your own logged-in browser session.

It's genuinely useful if you want an assistant to pull profile, company, or job data. Here's the gist of setting it up.

How to Set Up the Open-Source Server (Step-by-Step)

Before you start, you'll need:

• •A LinkedIn account you're logged into (the server acts through your own session)
• •uv installed (it ships with uvx, the recommended runner)
• •An MCP client like Claude Desktop - the server runs locally and connects to it

First, install uv (macOS / Linux):

curl -LsSf https://astral.sh/uv/install.sh | sh

Then add the server to your MCP client's config (e.g. Claude Desktop's claude_desktop_config.json):

{
  "mcpServers": {
    "linkedin": {
      "command": "uvx",
      "args": ["mcp-server-linkedin@latest"],
      "env": { "UV_HTTP_TIMEOUT": "300" }
    }
  }
}

Restart your client. On the first tool call, the server opens a LinkedIn login window and stores a local browser session - then you're connected. The full step-by-step:

The Catch: Account Safety

Because the open-source server acts through your own logged-in LinkedIn session, it lives in the same gray area as any browser-automation tool. The project is upfront about this, and so should you be:

“Use at your own risk; there is no guarantee of account safety.”- from the linkedin-mcp-server documentation

Why the warning? LinkedIn's User Agreement discourages automated access that isn't through its official APIs, and accounts that trip its automation detection can be restricted or suspended. Browser-session tools operate in that gray area by design - it's the trade-off for being free and unofficial.

That's not a knock on the project - it's an honest disclaimer. But if your LinkedIn account is tied to your pipeline or your job, it's worth weighing before you automate your own session. A hosted option that works through supported channels and OAuth (rather than driving your login) sidesteps that question entirely.

Option 2: The Hosted LinkedIn MCP (Postbeam)

If you don't want a terminal, a JSON config, or that account-safety question hanging over you, the hosted route is simpler. Postbeam's LinkedIn MCP is a connector someone else runs for you. You add one URL and authorize with OAuth - no install, nothing to maintain.

It's also built for a different job. Instead of just reading LinkedIn, it runs your content workflow from chat: draft posts in your voice from your week's Slack, email, and meetings; ask what's working in your analytics; and schedule posts - in Claude, ChatGPT, Gemini, or Poke.

• ✓Connects through secure OAuth with your Postbeam account, revocable anytime
• ✓Publishes through compliant, supported channels - not by automating your login
• ✓Works with Claude, ChatGPT, Gemini & Poke using the same server URL
• ✓No code, no terminal - setup takes under two minutes

Open-Source vs Hosted: Side by Side

Which One Should You Choose?

Go open-source if you're a developer who wants raw LinkedIn data reads inside your own MCP client, you're comfortable in a terminal, and you accept the account-safety trade-off of automating your session.

Go hosted with Postbeam if you want to actually run your LinkedIn presence from chat - drafting in your voice, checking analytics, and scheduling - without code and without putting your account on the line.

Frequently Asked Questions

Related Articles