LinkedIn MCP Server

<!-- mcp-name: io.github.prakharagarwal-dev/linkedin-mcp-server -->

A local MCP server that lets AI assistants search LinkedIn and perform supported LinkedIn actions through the visible website.

Like the project? Star the repository to help other MCP users find it.

If LinkedIn MCP saves you time, sponsor its continued development.

[!IMPORTANT] This is an unofficial project and is not affiliated with or endorsed by LinkedIn. LinkedIn interface changes can temporarily break features or prompt security checkpoints. Use it only with accounts and activity you are authorized to operate.

What you can do

• Search current jobs with LinkedIn's visible filters and read complete job descriptions.
• Search people and companies, then read exact visible profiles and company details.
• Search and read posts, comments, replies, reactions, and media.
• Create personal posts, comment, reply, and react after confirmation.
• List and search connections, and send, accept, or ignore invitations.
• Search messages, read one-to-one conversations, and send text or attachments.
• Use it with Codex, Claude, Cursor, VS Code, Gemini, or any MCP client.

See the capability matrix for every supported filter, input, output, and visible postcondition.

Getting started

Requirements

• macOS, Linux, or Windows
• Python 3.12 or 3.13
• uv
• a local MCP client and a LinkedIn account

Claude Desktop users can use the packaged .mcpb extension instead of the standard uvx setup below.

Standard configuration

This configuration works in most MCP clients:

{
  "mcpServers": {
    "linkedin-mcp": {
      "command": "uvx",
      "args": [
        "--from",
        "linkedin-mcp-local",
        "linkedin-mcp",
        "serve",
        "--transport",
        "stdio"
      ]
    }
  }
}

Every current capability is available after installation. Before the server posts, messages, connects, comments, or reacts, your MCP client should show the exact action and ask you to confirm. You can optionally restrict capabilities in Configuration.

One-click installs

[![Add to Kiro](https://ki