How to Configure LinkedIn Access Through the LinkedIn-MCP Server in Agent Reach

Configure LinkedIn access in Agent Reach by installing the linkedin-scraper-mcp Python wrapper, running the Node.js linkedin-mcp-server locally on port 3000, and registering the endpoint via mcporter config add linkedin http://localhost:3000/mcp, then verify with python -m agent_reach.cli doctor.

Agent Reach integrates with LinkedIn through a modular MCP (Micro-Content-Provider) architecture that delegates scraping tasks to a local Node.js server. To configure LinkedIn access through the linkedin-mcp server in Agent Reach, you must bridge the Python channel logic with the external MCP server using the mcporter CLI configuration tool.

Prerequisites and Installation

Install the Python Wrapper (linkedin-scraper-mcp)

The Agent Reach channel requires the linkedin-scraper-mcp package to communicate with the MCP backend.

pip install linkedin-scraper-mcp

Install the Node.js MCP Server and mcporter CLI

The actual LinkedIn scraping logic runs in a separate Node.js process. Install both the server and the configuration CLI globally.

npm install -g mcporter
npm install -g linkedin-mcp-server

Configuring the LinkedIn-MCP Server Connection

Start the Local MCP Server

Launch the linkedin-mcp-server locally. By default, it listens on port 3000.

linkedin-mcp-server &

# Or if running from source: node path/to/server.js

Register the Endpoint with mcporter

The LinkedInChannel class in agent_reach/channels/linkedin.py relies on the mcporter CLI to locate the MCP server. Register the local endpoint to establish the connection:

mcporter config add linkedin http://localhost:3000/mcp

This command creates the configuration entry that the health check probes when validating the backend.

Verification and Health Checks

Understanding the Channel Detection Logic

In agent_reach/channels/linkedin.py, the LinkedInChannel.can_handle() method (L18-L21) identifies LinkedIn URLs by pattern matching. Once a URL is detected, the check() method probes the backend status through the probe_command helper defined in agent_reach/probe.py.

The health check executes mcporter config list to verify the server is registered. If the output contains the word "linkedin", the channel sets active_backend = "linkedin-scraper-mcp" and marks the integration as active.

Running the Doctor Command

Execute the built-in diagnostics to confirm the integration is functional:

python -m agent_reach.cli doctor

Successful output shows LinkedIn: ok with availability notes indicating that profile, company, and job search functionality is fully available.

Fetching LinkedIn Data in Python

Once configured, the AgentReach core router in agent_reach/core.py automatically forwards LinkedIn URLs to the active MCP backend.

from agent_reach.core import AgentReach

ar = AgentReach()
profile = ar.read("https://www.linkedin.com/in/username/")
print(profile)  # Returns JSON with profile data

Troubleshooting Common Issues

If mcporter is missing from your system PATH, the LinkedInChannel reports "off" and prompts to install linkedin-scraper-mcp. If the binary exists but cannot execute (indicating a broken virtual environment), the health check returns a specific venv error hint. Always verify the Node.js server is running on the configured port before executing the doctor command.

Summary

• Install dependencies: linkedin-scraper-mcp (Python) and linkedin-mcp-server (Node.js).
• Configure mcporter: Run mcporter config add linkedin http://localhost:3000/mcp to point to your local MCP server.
• Verify health: Use python -m agent_reach.cli doctor to confirm the LinkedInChannel detects the backend in agent_reach/channels/linkedin.py.
• Access data: The AgentReach class in agent_reach/core.py routes LinkedIn URLs through the validated MCP connection.

Frequently Asked Questions

What is the difference between linkedin-scraper-mcp and linkedin-mcp-server?

The linkedin-scraper-mcp package is a Python wrapper that Agent Reach uses to communicate with the MCP protocol, while linkedin-mcp-server is the actual Node.js implementation that performs the LinkedIn scraping. Agent Reach requires both: the Python wrapper provides the channel interface, and the Node.js server provides the data extraction logic.

Why does the doctor command report LinkedIn as "off" even after installation?

This occurs when mcporter cannot find a configuration entry for LinkedIn. Run mcporter config list manually; if "linkedin" is missing, re-run mcporter config add linkedin http://localhost:3000/mcp. Also ensure the Node.js server is actually running on the specified port before executing the health check.

Can I run the MCP server on a port other than 3000?

Yes, but you must update the mcporter configuration accordingly. When starting the server on a custom port, use mcporter config add linkedin http://localhost:<custom_port>/mcp to ensure the LinkedInChannel.check() method in agent_reach/channels/linkedin.py can locate the endpoint during its probe routine.

How does Agent Reach determine if a URL should be handled by the LinkedIn channel?

The LinkedInChannel.can_handle() method in agent_reach/channels/linkedin.py inspects the URL pattern. If the domain matches LinkedIn, the channel claims responsibility, and the core router in agent_reach/core.py forwards the request to the active MCP backend configured via mcporter.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}