How Do I Get Started with the LinkedIn MCP Server?

If you write LinkedIn posts with Claude or Cursor, you know the gap: you draft in the AI, then leave it to copy-paste into LinkedIn, fix the formatting, and schedule by hand. It's the same disconnect we dug into in Why LinkedIn's Scheduler Is Holding Back Your Content Strategy: creation and publishing living in two separate places. The LinkedIn MCP server closes that gap: it connects LinkedIn to your AI client so the assistant can publish to your feed without you leaving the conversation.

One thing up front, because it shapes everything below: the LinkedIn MCP server is currently pre-launch. You can get early access by joining the waitlist (https://contentin.io/linkedin-mcp-server/), and the exact setup steps and supported features will land in the official documentation when it ships. What follows is what's worth understanding now, based on how MCP and LinkedIn's API actually work, not guesses about the unreleased build.

What the Model Context Protocol actually is

Model Context Protocol (MCP) is an open standard, introduced by Anthropic in late 2024, that lets AI clients talk to external tools through a shared language. Instead of every AI app building its own custom LinkedIn integration, someone writes one MCP server for LinkedIn, and then any MCP-compatible client (Claude, Cursor, and others) can use it.

That portability is the real point. The same server works across any client that supports the protocol, so you're not locked into one assistant or waiting for each app to build LinkedIn support of its own.

What the LinkedIn MCP server does

At its core, the server bridges your AI client and LinkedIn's API, authenticated through your own LinkedIn account. The anchor capability is publishing: LinkedIn lets approved apps post to a member's own feed through its self-serve "Share on LinkedIn" permission (w_member_social). That's what makes "draft this post and publish it" possible from inside your AI conversation.

It's worth being realistic about reading data, because LinkedIn's API is tightly gated. Posting to your own feed is self-serve, but reading your own posts and pulling analytics sit behind stricter access — the scope for reading a member's own feed is currently paused for new apps, and richer analytics generally require LinkedIn Partner-program approval. So whether features like analytics or scheduling are available will depend on the access ContentIn has secured, and the specifics will be confirmed in the launch docs. Treat publishing as the reliable capability and the rest as "check at launch."

If you want the wider landscape of what's possible beyond MCP, our Ultimate Guide to LinkedIn Automation Tools covers the broader toolkit this fits into.

What you'll need

• An MCP-compatible AI client (Claude Desktop is the most common starting point; Cursor and others also support MCP).

• A standard LinkedIn account, no developer account required to authorize an app to post on your behalf.

• Any runtime prerequisites (for example, some MCP servers require Node.js) will be listed in ContentIn's official requirements at launch.

How MCP setup works (the general pattern)

Every MCP server is added to a client the same way, so you can know the shape of this before the LinkedIn server is public:

1. Open your client's MCP configuration file. In Claude Desktop it lives at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows.

2. Add an entry for the server under the mcpServers key — a label plus the command the client uses to launch it.

3. Save and fully restart the client. The server's tools then appear in the client's tools/integrations menu.

The exact package name and command for the LinkedIn server will come from ContentIn's official setup guide, I'm deliberately not inventing it here, because a wrong command is worse than none. The structure above is the same for any MCP server, so the launch instructions will slot straight into it.

Authenticating with LinkedIn (OAuth)

Connecting your account uses OAuth, in plain terms, "log in with LinkedIn and approve some permissions." You'll be sent to LinkedIn, asked to approve what the app can do, and returned to your client with an access token.

Two facts worth getting right, since outdated guides get them wrong: posting on your behalf uses the w_member_social permission, and basic sign-in/profile for any new app now uses OpenID Connect scopes (openid, profile, email) — the old r_liteprofile scope was deprecated in August 2023 and no longer works for new apps. When you authorize, review what's requested: it should only be what's needed to post (and any read access the product has been granted). Be cautious if you're asked for anything unrelated.

What to expect right now

The server is in early access via the waitlist. The payoff it's built for is a tighter loop (draft and publish in the same conversation, without the copy-paste tab-switching that breaks your momentum) and because it's MCP-based, that loop is portable across any compatible client.

If you'd rather wire up an automated posting flow today while the server is still pre-launch, we walk through a do-it-yourself route in How to Build a Free LinkedIn Post Scheduler with n8n. And when you're ready for the native integration, join the waitlist (https://contentin.io/linkedin-mcp-server/) and you'll get the full setup instructions as soon as access opens.