Connect LinkedIn to Gemini

The four steps

1. 1

   Create a Blabigo account and connect LinkedIn

   Sign up free, then connect your LinkedIn profile and any company Pages you manage from the dashboard. You authorize Blabigo against your own account through LinkedIn's normal consent screen. Gemini never handles your LinkedIn credentials, because it never sees them.

2. 2

   Add the server to Gemini CLI

   One command registers Blabigo as an MCP server over streamable HTTP. Add --scope project if you want it available only inside the current directory rather than for your user.

3. 3

   Authenticate

   Blabigo publishes OAuth discovery metadata, and Gemini CLI detects it: on the first 401 it finds the endpoints, registers itself dynamically and runs the flow. If it does not prompt you automatically, run /mcp auth blabigo to start it. No API key is created.

4. 4

   Check the tools are there

   Run /mcp list to confirm Blabigo is connected and see the tools it exposes, then ask Gemini to list your connected LinkedIn accounts. Getting your profile and Pages back exercises authentication, scope resolution and the LinkedIn API in a single call.

Or edit the config yourself

In ~/.gemini/settings.json for your user, or .gemini/settings.json inside a project:

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

httpUrl is streamable HTTP; url is SSE. Blabigo speaks the former, and using the wrong field is the usual reason a hand-written config does not connect.

Fallback: run it locally over stdio

If you would rather hold an API key than run a browser flow, create one in the dashboard and use the CLI as a local MCP server instead:

{
  "mcpServers": {
    "blabigo": {
      "command": "npx",
      "args": ["-y", "@blabigo/cli", "mcp"]
    }
  }
}

What to ask once it is connected

• “List my connected LinkedIn accounts.”
• “How did my posts perform over the last 30 days? Which one got the most impressions?”
• “Draft a post from these release notes and queue it for Thursday at 9am.”
• “Show me everything scheduled for next week, and flag anything that overlaps in topic.”

When it does not work

Frequently asked questions

How do I connect LinkedIn to Gemini?

Create a Blabigo account and connect your LinkedIn profile to it, then run `gemini mcp add --transport http blabigo` with Blabigo's MCP endpoint. Gemini CLI discovers the OAuth configuration automatically and walks you through signing in. After that the LinkedIn tools are available in your Gemini session.

Does Gemini support MCP servers?

Gemini CLI does. It supports stdio, SSE and streamable HTTP transports, configured either with the `gemini mcp add` command or by editing the mcpServers block in ~/.gemini/settings.json. It also supports OAuth 2.0 for remote servers, including discovering the configuration from the server rather than being told it.

Can Gemini post to LinkedIn?

It can draft a post and place it on your queue for a future time. It cannot publish — Blabigo's scheduler publishes at the scheduled time, and no tool in the integration reaches LinkedIn on demand. So Gemini can write and queue, while you keep the last look before anything goes out.

Where is the Gemini CLI settings file?

At ~/.gemini/settings.json for your user, or .gemini/settings.json inside a project for a project-scoped server. The `gemini mcp add` command writes to the user scope by default; pass --scope project to write the project one instead.

What is the difference between httpUrl and url in the config?

They select the transport. httpUrl is streamable HTTP and url is SSE. Blabigo's endpoint speaks streamable HTTP, so it belongs in httpUrl — this is the most common reason a hand-edited config fails to connect.

Does this work in the Gemini app as well as the CLI?

This guide covers Gemini CLI, which has documented MCP support and is the reliable path today. Support for custom MCP servers in Google's consumer and Workspace Gemini surfaces has been moving, so rather than describe a menu that may have changed, we would suggest checking Google's current documentation — the endpoint URL is the same wherever you end up adding it.