LinkedIn MCP on GitHub: the Open-Source Servers Compared (2026)

Search "linkedin mcp" on GitHub and you get a wall of repos with near-identical names: linkedin-mcp-server, mcp-linkedin, linkedin-mcp. They are not the same. Some scrape your profile through a headless browser, some hit an unofficial API, some only read jobs, and a few barely work. Picking the wrong one is how people get their LinkedIn account flagged.

This is the map. Below are the open-source LinkedIn MCP servers worth knowing, what each actually does, how it connects, and the risk that comes with it. Then the honest part: when an open-source repo is the right call, and when a hosted, official MCP is the safer one.

What a LinkedIn MCP server does

An MCP (Model Context Protocol) server is a bridge that lets an AI assistant like Claude read data and take actions in a tool. A LinkedIn MCP exposes LinkedIn as tools the AI can call: read a profile, search jobs, fetch a feed, or, for the content-focused ones, draft and publish a post.

The critical question for any LinkedIn MCP is how it connects. There are three methods, and they carry very different risk:

• Browser scraping: the server drives a logged-in browser session to read pages. Powerful, but fragile and against LinkedIn's terms, which is what gets accounts restricted.
• Unofficial API: the server calls private LinkedIn endpoints with your cookies. Lighter than scraping, still unofficial, still risky.
• Official API on your account: the server acts through a sanctioned product integration. Safe and stable, but limited to what that product supports.

Keep that lens as you read the repos.

The open-source LinkedIn MCP servers on GitHub

stickerdaniel/linkedin-mcp-server

The most-starred of the bunch. It scrapes LinkedIn through browser automation to pull profiles, companies, and jobs, and it is genuinely capable for read and research tasks. It is also the clearest example of the scraping trade-off: it needs your session and works until LinkedIn changes a selector or flags the activity.

southleft/linkedin-mcp

An AI-focused LinkedIn MCP aimed at richer profile and content data. Similar scraping-based approach, similar caveats. Worth a look if stickerdaniel's does not fit your data shape.

adhikasp/mcp-linkedin

This one uses an unofficial LinkedIn API to read your feed and search jobs. It is lighter to run than a full browser scraper, but the unofficial-API route is fragile: it breaks when LinkedIn rotates endpoints, and it still touches your account in a way LinkedIn does not sanction.

The long tail

There are many more (fredericbarthelet/linkedin-mcp-server, Dishant27/linkedin-mcp-server, felipfr/linkedin-mcpserver, alinaqi/mcp-linkedin-server, rugvedp/linkedin-mcp, eliasbiondo/linkedin-mcp-server). They cluster around the same patterns: a browser scraper or an unofficial-API client, packaged as an MCP, with varying levels of maintenance. Check the last commit date and the open-issues count before you trust one in a workflow.

How they compare

The catch with open-source LinkedIn MCPs

Open-source LinkedIn MCPs are great for one thing: pulling data when you are a developer who can run and babysit them. But for actually running your LinkedIn presence, three problems show up fast:

1. Account risk. Scraping and unofficial APIs both violate LinkedIn's terms. The more you automate, the higher the chance of a restriction. Your main account is not a good place to test that.
2. Maintenance. These repos break when LinkedIn ships a change. You become the on-call engineer for your own posting workflow.
3. They read, they do not ship. Most are built to extract data, not to draft in your voice, schedule, and publish. That is the part a marketer or creator actually needs.

The hosted, official alternative: the Taplio LinkedIn MCP

If your goal is content and personal brand rather than scraping data, the Taplio LinkedIn MCP solves all three problems. It runs on your own Taplio account through the official path, so there is no scraping, no browser automation, and no ban risk. It is built for the job: search the viral-post inspiration index, draft a post in your voice, schedule or publish it, and read your analytics, with nothing going live until you approve it. And there is nothing to maintain, because it is hosted.

It is not a code repo you clone. You add it as a connector and it works, which leads to the setup.

How to add a LinkedIn MCP (two ways)

As a custom connector (Claude web or ChatGPT, no terminal): open Settings, go to Connectors, click Add custom connector, paste the MCP server URL, and authenticate. For Taplio that URL is https://mcp.taplio.com. This is the path most people should use.

In the terminal (Claude Code, Cursor):

claude mcp add --transport http taplio https://mcp.taplio.com

Self-hosted open-source servers follow their own README: you clone the repo, install dependencies, provide your session or cookies, and run the server locally before pointing your client at it.

How to choose

• You are a developer who needs to extract LinkedIn data for a custom build, and you accept the risk and upkeep: start with stickerdaniel/linkedin-mcp-server.
• You want to run your LinkedIn content (research, draft, schedule, publish, analyze) safely and with zero maintenance: use the Taplio LinkedIn MCP. For the full landscape of options, see our roundup of the best LinkedIn MCP servers, and the step-by-step setup with Claude.

FAQ

Is there an official LinkedIn MCP server?

LinkedIn does not publish its own MCP server. The closest to "official" is a hosted MCP that acts through a sanctioned product integration on your account, like the Taplio LinkedIn MCP, rather than scraping LinkedIn. The GitHub repos are community projects, most of which scrape or use unofficial APIs.

Are open-source LinkedIn MCP servers safe to use?

They carry real risk. Most rely on browser scraping or unofficial APIs, both of which violate LinkedIn's terms and can get an account restricted. If you use one, do it on a test account, not your main profile, and keep the automation light.

What is the best LinkedIn MCP server on GitHub?

For data extraction, stickerdaniel/linkedin-mcp-server is the most capable and maintained. But for running your LinkedIn content safely, a hosted official MCP like Taplio's is the better choice, because it does not scrape and there is nothing to maintain.

Do I need to self-host a LinkedIn MCP?

Only for the open-source repos, which you clone and run yourself. A hosted MCP like Taplio's needs no install: you add it as a custom connector in Claude or ChatGPT and authenticate with your account.