# linkedin-mcp-server [Health: Active]

**Category:** 🌐 Social Media  
**Repository:** https://github.com/stickerdaniel/linkedin-mcp-server  
**GitHub Stars:** 3961  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/linkedin-mcp-server-3

## Description
Open-source MCP server for LinkedIn. Give Claude and any MCP-compatible AI agent access to profiles, companies, jobs, and messages.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `uvx` (confidence: high):

```json
"mcpServers": {
  "linkedin-mcp-server": {
    "command": "uvx",
    "args": ["mcp-server-linkedin"]
  }
}
```

## Documentation

## What linkedin-mcp-server MCP server does

linkedin-mcp-server MCP server gives MCP-compatible agents access to LinkedIn through a user-controlled browser session. It is intended for research and account workflows involving people, companies, jobs, posts, messages, and connection requests. The project is open source, independent of LinkedIn and Microsoft, and belongs in the 🌐 Social Media category.

Profile tools can read selected sections such as experience, education, skills, projects, and posts. Agents can retrieve the signed-in user's own profile, search for people by keyword, location, connection degree, or current company, and inspect recommended profiles. Company tools provide company information, posts, open jobs, employee searches, and keyword-based company discovery.

Job tools search by keyword and location, list saved jobs, return job details, and identify whether an application uses LinkedIn Easy Apply or an employer's application page. Feed and post tools read the home feed, retrieve company posts, and search posts with optional recent-time limits.

## How it works

The agent communicates with the MCP server, while the server uses the user's browser session to access LinkedIn. This approach uses the existing account context rather than describing a separate LinkedIn API integration in the provided material. The server also exposes a session-closing tool; closing the shared browser session does not delete the saved login.

Conversation tools cover several stages of inbox work. An agent can list recent conversations, open a thread by LinkedIn username or thread ID, search messages by keyword, and send a message after confirmation. Profile targets may open a new direct message. The connection tool can send a request with an optional note or accept an incoming request.

## Setup and configuration

The repository documents several delivery options: installation with `uvx`, a Docker setup, and client-specific packages or plugins for Claude Desktop, Codex, Claude Code, and Cursor. The recommended installation method shown in the documentation uses `uvx`, which requires `uv` to be installed first. The project also publishes a Python package named `mcp-server-linkedin`.

Configuration is added to an MCP client's `mcpServers` settings when using the `uvx` method. Because access is provided through the user's browser session, the agent should be used with the intended LinkedIn account and session. The available material does not specify additional environment variables or a separate API key configuration.

## Tools and capabilities

The documented tools include:

- Read personal and other people's profiles, selected profile sections, and recommended profiles.
- Search people, companies, jobs, and posts using the supported filters.
- Read company profiles, company posts, employees, home-feed posts, job descriptions, and application URLs.
- List inbox conversations, inspect threads, and search conversation messages.
- Send confirmed messages and create or accept connection requests.
- List saved jobs and close the shared browser session.

## Limitations and notes

linkedin-mcp-server MCP server is an independent project and is not affiliated with LinkedIn or Microsoft. Its documented access model depends on a browser session, so its behavior is tied to the account and session available to that browser. Sending messages requires confirmation according to the tool documentation. The repository lists links and identifiers for many results, but the provided material does not define broader rate limits, LinkedIn plan support, or reliability guarantees.

_Full upstream README: https://allmcps.com/mcp/linkedin-mcp-server-3/readme_