LinkedIn MCP server

The 13 tools

Five read, eight write. The names are what your assistant will show you when it asks permission to call one, so they are worth a glance before you connect — this is the entire surface area an agent has.

Scopes

An OAuth client only sees the tools its grant covers. A client authorized for analytics alone cannot list the content tools, let alone call them.

There is no publish scope, because there is nothing to grant it to.

Connect

Claude Code

claude mcp add --transport http blabigo https://studio.blabigo.ai/api/mcp/linkedin

Config-file clients

{
  "mcpServers": {
    "blabigo": {
      "url": "https://studio.blabigo.ai/api/mcp/linkedin"
    }
  }
}

The client runs OAuth against the endpoint and opens a browser window for you to approve. There is no key to create, copy or store.

Or call it directly over HTTP

curl -X POST https://studio.blabigo.ai/api/mcp/linkedin \
  -H "Authorization: Bearer mcp_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

API keys are created in the Blabigo dashboard under Settings → MCP. A key never expires and carries full access, so prefer OAuth unless your client cannot do it.

Frequently asked questions

What is the Blabigo LinkedIn MCP server?

It is a Model Context Protocol server that exposes your LinkedIn account to AI assistants as a set of 13 defined tools. Five read your posts and analytics; eight write, within fixed limits. An assistant that supports MCP can call them directly, so you can ask it to review last month's performance or queue Thursday's post without leaving the conversation.

Can an agent publish a LinkedIn post through the MCP server?

No. There is no publish tool and no publish scope. An agent can create a draft and queue it for a future time; Blabigo's own scheduler publishes it when that time arrives. The minimum lead time is 60 seconds, so queueing cannot be used as a disguised instant publish.

Can an agent delete my posts?

No. There is no delete tool and no delete endpoint. Cancelling a scheduled post reverts it to a draft, and archiving a draft hides it from the working list but keeps the record, reversibly. Whatever an agent withdraws, you can still see.

How do I authenticate?

Two ways. Most clients need only the endpoint URL: the server publishes standards-based OAuth discovery, so the client runs an authorization-code flow with PKCE and you approve the connection in your browser. For scripts, CI and self-hosted agents, you can instead create an API key in the Blabigo dashboard and send it as a bearer token.

Which protocol version and transport does it use?

JSON-RPC 2.0 over HTTP, using the streamable HTTP transport. Clients that cannot reach a remote server or do not implement OAuth can run the Blabigo CLI locally instead, which speaks MCP over stdio and forwards to the same endpoint.

Does the MCP server scrape LinkedIn?

No. It reads and writes through your own connected LinkedIn account using the official API and the permissions you granted when you connected it. Nothing automates the LinkedIn website, and nothing logs in as you.