LinkedIn MCP Server

One conversation. Every LinkedIn workflow.

Research competitors, draft posts, analyze engagement, schedule content, and manage your professional presence—all from a single AI-powered interface.

The Power of Unified AI Workflows

Imagine this: You ask Claude to research a competitor's LinkedIn presence. Minutes later, you have insights on their content strategy, top-performing posts, and audience engagement patterns. You draft a response post, Claude optimizes it for engagement, and you schedule it for peak hours—all without leaving the conversation.

But it doesn't stop there.

When you combine this MCP with others, the possibilities multiply:

LinkedIn Research → Blog Post → WordPress Publish → LinkedIn Promotion

1. Pull insights from a LinkedIn thought leader's post
2. Ask Claude to write a blog article expanding on those ideas
3. Publish it to your WordPress site (via another MCP)
4. Create a LinkedIn post promoting the article with the link

This is the future of professional workflows—one AI assistant orchestrating your entire digital presence across platforms. No more context switching. No more copy-pasting between tabs. Just seamless, intelligent automation.

Who Is This For?

Content Creators & Marketers

• Create and schedule LinkedIn posts with AI-powered optimization
• Manage drafts, analyze content performance, and find optimal posting times
• Build a consistent content calendar without leaving your AI assistant

Sales & Business Development

• Research prospects and companies before outreach
• Find similar profiles to identify new leads
• Look up companies by domain to qualify opportunities

Recruiters & HR Professionals

• Research candidate profiles and backgrounds
• Discover professional interests and expertise areas
• Find similar professionals based on target profiles

Analysts & Researchers

• Gather company and professional data at scale
• Analyze engagement patterns and content performance
• Generate comprehensive reports on LinkedIn presence

Features

Content Creation

Content Planning

Profile Research

Company Research

Analytics & Engagement

Quick Start

Prerequisites

• Python 3.11+
• uv (recommended) or pip
• LinkedIn account
• LinkedIn Developer App (free, 5-minute setup)

Installation

git clone https://github.com/southleft/linkedin-mcp.git
cd linkedin-mcp

# Install with uv (recommended)
uv venv && source .venv/bin/activate
uv pip install -e .

# Install browser automation (optional, for enhanced features)
playwright install chromium

LinkedIn Developer Setup

1. Go to LinkedIn Developer Portal
2. Click "Create app"
3. Fill in details:
   • App name: e.g., "My LinkedIn MCP"
   • LinkedIn Page: Select or create a company page
   • Privacy policy URL: Your website or LinkedIn profile
4. Go to Products → Request "Share on LinkedIn" (instant approval)
5. Go to Auth tab:
   • Copy Client ID and Client Secret
   • Add redirect URL: http://localhost:8765/callback

Configuration

Create a .env file in the project root:

# LinkedIn OAuth (Required for posting)
LINKEDIN_CLIENT_ID=your_client_id
LINKEDIN_CLIENT_SECRET=your_client_secret
LINKEDIN_API_ENABLED=true

# Enhanced Data API (Recommended for reliable profile/company lookups)
# Powers the top 2 fallback sources: Professional Network Data API + Fresh LinkedIn Data API
THIRDPARTY_RAPIDAPI_KEY=your_api_key

Note: The RapidAPI key enables the most reliable data sources in the fallback chain. Without it, the server falls back to cookie-based methods which are prone to LinkedIn's bot detection.

Authentication

# Authenticate with LinkedIn (opens browser)
linkedin-mcp-auth oauth

# Optional: Extract cookies for additional features
linkedin-mcp-auth extract-cookies --browser chrome

# Verify setup
linkedin-mcp-auth status

Connect to Claude Desktop

Add the following to your Claude Desktop config file:

Complete configuration example:

{
  "mcpServers": {
    "linkedin": {
      "command": "/path/to/linkedin-mcp/.venv/bin/python",
      "args": ["-m", "linkedin_mcp.main"],
      "cwd": "/path/to/linkedin-mcp",
      "env": {
        "LINKEDIN_CLIENT_ID": "your_client_id",
        "LINKEDIN_CLIENT_SECRET": "your_client_secret",
        "LINKEDIN_API_ENABLED": "true",
        "THIRDPARTY_RAPIDAPI_KEY": "your_api_key",
        "PYTHONPATH": "/path/to/linkedin-mcp/src",
        "LOG_LEVEL": "INFO"
      }
    }
  }
}

Critical: Claude Desktop does not read .env files—all environment variables must be explicitly passed in the config.

After updating the config:

1. Quit Claude Desktop completely
2. Reopen Claude Desktop
3. The LinkedIn MCP server will connect automatically

Connect to Claude Code

Add to ~/.claude.json:

{
  "mcpServers": {
    "linkedin-mcp": {
      "type": "stdio",
      "command": "/path/to/linkedin-mcp/.venv/bin/python",
      "args": ["-m", "linkedin_mcp.main"],
      "cwd": "/path/to/linkedin-mcp",
      "env": {
        "LINKEDIN_CLIENT_ID": "your_client_id",
        "LINKEDIN_CLIENT_SECRET": "your_client_secret",
        "LINKEDIN_API_ENABLED": "true",
        "THIRDPARTY_RAPIDAPI_KEY": "your_api_key",
        "PYTHONPATH": "/path/to/linkedin-mcp/src"
      }
    }
  }
}

Run /mcp in Claude Code to connect.

Usage Examples

Content Creator Workflow

You: "Create a LinkedIn post about the importance of continuous learning in tech"

Claude: Here's a draft optimized for engagement:

"In tech, the moment you stop learning is the moment you start falling behind.

Every new framework, language, or methodology is an opportunity to expand your toolkit and thinking.

The best engineers I know aren't just skilled—they're perpetually curious.

What are you learning this week?

#ContinuousLearning #TechCareers #GrowthMindset"

Should I post this now, schedule it, or save as a draft?

Content Planning

You: "Save this as a draft for my product launch and schedule it for Tuesday 9am"

Claude: Done! I've:
1. Saved your draft titled "Product Launch"
2. Scheduled it for Tuesday, 9:00 AM (your timezone)

You can view your drafts anytime or adjust the schedule.

Profile Research

You: "Look up the CEO of Microsoft on LinkedIn"

Claude: Found Satya Nadella's profile:
- Chairman and CEO at Microsoft
- Focus on cloud computing and AI transformation
- Author and thought leader in technology

Would you like me to find similar profiles or see their recent articles?

Company Research

You: "Look up information about anthropic.com"

Claude: Found Anthropic:
- AI safety company
- Headquarters in San Francisco
- Focus on AI research and safety

Would you like more details or to search for employees?

Available Tools

Content Creation

Draft Management

Scheduling

Profile Research

Company Research

Analytics

CLI Commands

# Authentication
linkedin-mcp-auth status              # Check auth status
linkedin-mcp-auth oauth               # Authenticate via OAuth
linkedin-mcp-auth oauth --force       # Force re-authentication
linkedin-mcp-auth extract-cookies     # Extract session cookies
linkedin-mcp-auth logout --all        # Clear all credentials

# Server
linkedin-mcp                          # Start MCP server
python -m linkedin_mcp                # Alternative server start

Configuration Reference

Limitations

Troubleshooting

"Profile lookup returns error"

Ensure THIRDPARTY_RAPIDAPI_KEY is set in your Claude Desktop config:

"env": {
  "THIRDPARTY_RAPIDAPI_KEY": "your_api_key",
  ...
}

Important: Claude Desktop does not read .env files. You must add the key directly to the config's env section.

"LinkedIn client not initialized"

linkedin-mcp-auth status    # Check configuration
linkedin-mcp-auth oauth     # Re-authenticate

"Application context not initialized"

Reconnect the MCP server:

• Claude Desktop: Quit and reopen the app
• Claude Code: Run /mcp to reconnect

Posts not appearing

• Verify OAuth: linkedin-mcp-auth status
• Ensure "Share on LinkedIn" product is enabled in Developer Portal

Comments failing with permission error

• Comments require the "Community Management API" product
• Apply for access in your LinkedIn Developer Portal

Search returns empty results

• Search relies on session cookies which expire after ~24 hours
• Refresh cookies: linkedin-mcp-auth extract-cookies --browser chrome
• Note: Profile and company lookups use the RapidAPI fallback and are unaffected

Security

• OAuth tokens: Stored securely in system keychain
• Session cookies: Encrypted local storage
• Credentials: Environment variables only
• No data collection: Everything runs locally

Development

# Install dev dependencies
uv pip install -e ".[dev]"

# Run tests
pytest tests/ -v

# Linting
ruff check src/ && ruff format src/

# Type checking
mypy src/linkedin_mcp

Architecture

Intelligent API Fallback

Under the hood, the server uses a multi-source architecture that automatically falls back through APIs ordered by reliability:

This ensures high availability, automatic recovery from failures, and best-effort data retrieval.

License

MIT License - see LICENSE for details.

Acknowledgments

• FastMCP - MCP SDK for Python
• Model Context Protocol - AI tool protocol
• Playwright - Browser automation

Built for AI-powered professionals