stickerdaniel/linkedin-mcp-server

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

3,034 ★518 forksPythonUpdated 22h ago

What you need to know

MCP server that lets AI assistants like Claude read LinkedIn through your own logged-in browser session — profiles, companies, people/job search, messaging, feed, and posts (built with FastMCP and Patchright).

Install

uvx mcp-server-linkedin@latest (recommended, universal; keep @latest and uvx exactly — pinned versions rot)
Claude Desktop: download the .mcpb bundle from GitHub releases and click to install
Docker: docker run --rm -i -v ~/.linkedin-mcp:/home/pwuser/.linkedin-mcp stickerdaniel/linkedin-mcp-server:latest
Local dev: uv sync && uv run -m linkedin_mcp_server

Usage

  • One-time session setup: uvx mcp-server-linkedin@latest --login (opens a browser; uses ~/.linkedin-mcp/profile/)
  • Or --import-from-browser [chrome|chromium|brave|edge|arc|vivaldi|helium|yandex|whale|auto] to reuse an existing logged-in browser session
  • Tools: get_person_profile / get_my_profile, search_people, search_companies, get_company_profile, get_company_posts, get_company_employees, search_jobs, get_job_details, get_saved_jobs, get_feed, search_posts, get_inbox, get_conversation, search_conversations, send_message, connect_with_person, get_sidebar_profiles, close_session
  • HTTP mode for web clients: --transport streamable-http --host 127.0.0.1 --port 8080 --path /mcp

Key features

  • ~19 tools covering profiles (with explicit section selection), people/company/job search, company posts and employees, inbox messaging, feed, and post search
  • Session import from your everyday browser (Chrome, Brave, Edge, Arc, Vivaldi, etc.) or manual --login
  • stdio (default) and streamable-http transports; Docker deployment with a profile mounted from the host
  • Tool calls serialized to protect the shared browser session; shared-browser handoff between multiple clients
  • Proxy support (--proxy-server) with credentials via env vars

Best for

Personal LinkedIn research, outreach, and job hunting driven through an AI assistant using your own logged-in session.

Caveats

  • LinkedIn's User Agreement prohibits automated access — accounts using automated tools can be restricted or banned; use at your own risk (explicit FAQ)
  • Requires a real LinkedIn login; captcha challenges and mobile-app login confirmation may occur; only one active session at a time
  • Docker runs headless — you must create the profile on the host first with --login; Docker creates a fresh session each startup and sessions may expire
  • macOS/Linux enforce that the browser be at least as new as the one that last opened the profile
  • Use of a proxy is discouraged for most users (triggers LinkedIn checkpoints); must be set up before creating the session
  • Unaffiliated with LinkedIn/Microsoft; personal-use only, community project
Platforms: macOS · Linux · WindowsClients: Claude Desktop · Claude Code · Cursor · Any MCP client
Documentation ↗

Reviewed 2026-08-07

Topics

ai-agentsanthropicchatgptchatgpt-desktopclaudeclaude-aiclaude-codeclaude-desktopdesktop-extensiondxtlinkedinlinkedin-apilinkedin-mcplinkedin-profile-scraperlinkedin-scrapermcpmcp-servermodel-context-protocolpython
View on GitHub ↗
Stars
3,034★
Forks
518
Language
Python
License
Apache-2.0
Created
2025-04-13
Last push
2026-08-06