LinkedIn MCP Server: The Complete Guide (2026)

The short answer: LinkedIn MCP servers come in two incompatible kinds. Scrapers read LinkedIn data through your logged-in session — profiles, companies, jobs — and serve prospecting workflows. Content servers never touch your account; they give an AI assistant the research and voice context to help you write. Check the tool list before installing: get_person_profile means scraper, generate_content_ideas means content.

Last updated: 1 August 2026.

Search for a LinkedIn MCP server and you'll find a dozen options within a minute. What nobody tells you is that they're doing two completely different jobs, and picking the wrong kind wastes a week.

We analysed the live top ten Google results for "linkedin mcp server" in August 2026: eight of the ten document a scraper, and only two frame it as a content tool. The term draws roughly 480 US searches a month at a Google competition index of 15 — real demand, very little commercial contest.

Most LinkedIn MCP servers are scrapers. They log into LinkedIn as you and read data out of it — profiles, companies, job listings, sometimes messages. They exist to feed prospecting and recruiting workflows.

A much smaller group are content servers. They don't read LinkedIn at all. They give your AI assistant the research and context it needs to help you write for LinkedIn.

Both are legitimate. They are not substitutes. This guide covers what a LinkedIn MCP server is, the tools each kind exposes, how to install and configure one, and how to decide which you actually need.

What is a LinkedIn MCP server?

An MCP server is a small program that gives an AI assistant a set of callable tools; a LinkedIn MCP server is one whose tools relate to LinkedIn. The Model Context Protocol is an open standard, introduced by Anthropic in November 2024, for connecting AI assistants to external tools and data. An MCP server is a small program that advertises a set of tools to AI agents; an MCP client — Claude Desktop, Claude Code, Cursor — connects to it and can then call those tools during a conversation.

The practical effect: instead of copying data into a chat window, your assistant fetches it. Instead of you describing your LinkedIn profile, the model reads it. The protocol handles discovery, schemas and transport; you just add a server to a config file.

A LinkedIn MCP server is simply an MCP server whose tools relate to LinkedIn. What those tools do is where the fork happens.

What are the two kinds of LinkedIn MCP server?

Scraper and automation servers

This is the majority of what's available — eight of the ten pages ranking for the term in August 2026. The same named tools recur almost verbatim across the GitHub, Docker Hub and LobeHub listings:

Notable implementations include stickerdaniel's open-source server on GitHub, the Apify LinkedIn MCP server, Bright Data's enterprise scraping infrastructure, and Zapier's MCP bridge. Several are catalogued on mcpservers.org, LobeHub and the Docker MCP Catalog.

How they authenticate matters. Most drive a real browser session using your li_at session cookie, or your email and password directly. That is worth understanding before you install one: you are handing a local process your LinkedIn session. Read the source, prefer open implementations, and be aware that LinkedIn's user agreement restricts automated data collection. Accounts have been restricted for aggressive automation.

Content and research servers

A much smaller group — including ContentIn's glossary entry, Postiv and Viralz — treat LinkedIn as the destination rather than the data source. They expose research, audience context and drafting tools, and never touch your account credentials.

The distinction shows up immediately in the tool list. There is no get_person_profile here. Instead you get things like business context, ICP definitions, creator benchmarking and content generation.

The gap is measurable. Of the ten pages ranking for "linkedin mcp server", five link to github.com, four to linkedin.com and three to anthropic.com — a developer-documentation SERP, not a marketing one. Anything published here has to read like documentation to belong.

It may not stay a niche question for long. Search Engine Land reported on 31 July 2026 that OpenAI appears to be testing an ad format where the click opens a business-specific ChatGPT conversation rather than a landing page — with the business agent fed by product feeds, lead forms and MCP tools. OpenAI has not confirmed this and the report is explicitly hedged, so treat it as a signal rather than a fact. But the direction is worth noting: MCP is drifting from a developer convenience toward customer-facing infrastructure.

Which tools does a LinkedIn content MCP expose?

Viralz is a content server. It runs alongside your assistant and gives it the context needed to write posts that sound like you and target your actual buyers. Here are the main tools:

The workflow this enables: ask Claude for three post ideas, have it pull your ICP and vocabulary automatically, draft in your voice, then run simulate_draft before publishing. No tab switching, no copy-paste, no credentials handed over.

How do you install a LinkedIn MCP server in Claude?

Configuration is the same shape across every MCP client. You add an entry to a JSON config file, restart the client, and the tools appear.

Claude Desktop

Edit claude_desktop_config.json — on macOS it lives at ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "viralz": {
      "command": "npx",
      "args": ["-y", "@viralz/mcp-server"],
      "env": {
        "VIRALZ_API_KEY": "your-api-key-here"
      }
    }
  }
}

Restart Claude Desktop. The tools appear in the tool menu.

Cursor

Add the same block to .cursor/mcp.json in your project, or to the global Cursor MCP settings. The schema is identical.

A scraper server, for comparison

An open-source scraper server needs rather more setup, because it drives a browser:

git clone https://github.com/stickerdaniel/linkedin-mcp-server
cd linkedin-mcp-server
uv sync
export LINKEDIN_COOKIE="your-li_at-cookie-value"

{
  "mcpServers": {
    "linkedin": {
      "command": "uv",
      "args": ["--directory", "/path/to/linkedin-mcp-server", "run", "main.py"],
      "env": {
        "LINKEDIN_COOKIE": "your-li_at-cookie-value"
      }
    }
  }
}

You'll also need ChromeDriver on your path. Getting the li_at cookie means opening LinkedIn in a browser, going to developer tools, and copying it out of application storage — a step worth pausing on, since that cookie is your session.

Why isn't my LinkedIn MCP server working?

Tools don't appear after restart. Almost always malformed JSON. Run your config through a validator — a trailing comma will silently break the whole file.

Server starts but every call fails. Check the API key or credentials in the env block. Most clients surface server logs; read them before guessing.

Scraper server returns empty results. The session cookie has expired. They typically last weeks, not months, and re-authentication is manual.

npx command not found. Node isn't on the PATH that your MCP client sees, which is often different from your shell's. Use an absolute path to the binary.

Rate limiting or account warnings. Stop. Reduce request volume immediately. This is the failure mode that costs people their LinkedIn account, and no data run is worth that.

Which LinkedIn MCP server should you use?

Ask what you're trying to produce.

If you need data out of LinkedIn — prospect lists, company research, job market analysis — you need a scraper server. Accept the credential handling and the terms-of-service exposure, or use a hosted service like Apify or Bright Data that carries some of that operational risk for you.

If you need to publish into LinkedIn — posts, carousels, a consistent habit — a scraper gives you nothing useful. Profile data is not the bottleneck. Knowing what to say, in your own voice, to people who might buy, is the bottleneck. That's a content server.

If you're a founder doing your own content, the second is almost certainly what you actually want, even though the first is what search results push you toward. The scraper-heavy results are an artefact of who built first, not of what most people need.

One caveat we'd rather state than hide: a content MCP is only worth installing if you already work in an MCP client daily. If Claude or Cursor isn't already open on your machine, a browser-based tool will serve you better, and the honest recommendation is to use one.

Best fit for a content MCP: you publish on LinkedIn yourself, you already work in Claude or Cursor, and you want research and drafting without handing over account credentials.

Not a fit: you need prospect lists or company data extracted, you're doing recruiting research, you don't use an MCP-compatible client, or you want scheduling and publishing rather than drafting.

Frequently asked questions

What is a LinkedIn MCP server? An MCP server whose tools relate to LinkedIn. Two distinct kinds exist: scrapers that read LinkedIn data via your session, and content servers that help you write for LinkedIn without touching your account.

Is it safe to give an MCP server my LinkedIn cookie? It's a real risk. That cookie is your logged-in session. If you do it, use open-source code you can read, run it locally rather than on someone else's infrastructure, and never paste the cookie into a hosted service you don't trust. Content servers avoid the question entirely by not needing account access.

Will using a LinkedIn MCP server get my account restricted? Reading your own data at low volume is low risk. Automated connection requests, bulk messaging and high-volume scraping are what trigger enforcement. LinkedIn has removed access from tools that crossed that line.

Does it work with Cursor as well as Claude? Yes. MCP is client-agnostic. The same server works in Claude Desktop, Claude Code, Cursor and any other MCP-compatible client, with the same config shape.

Can I use more than one at once? Yes — add multiple entries under mcpServers. Running a scraper for research and a content server for drafting is a perfectly sensible combination.

Do I need to write code? No. Hosted servers need an API key in a config file. Self-hosted open-source servers need a clone, a dependency install and an environment variable — closer to following a README than programming.

Where can I find available LinkedIn MCP servers? mcpservers.org, the Docker MCP Catalog and LobeHub all maintain directories. Read the tool list before installing — it tells you immediately which kind you're looking at.

Do you need a LinkedIn MCP server at all?

The LinkedIn MCP server category is dominated by scrapers because scrapers were easier to build first. If your job is prospecting or recruiting, that's good news.

If your job is publishing — if the thing standing between you and pipeline is a blank page rather than missing profile data — then a scraper solves a problem you don't have. Check the tool list before you install anything. get_person_profile means data extraction. generate_content_ideas means the other kind. The names tell you everything.

Viralz runs as a content MCP server for Claude and Cursor, from €30/month. See the tool list and setup guide.