LinkedIn MCP Server: Connect AI Coding Agents to LinkedIn Data

What LinkedIn MCP Servers Do

A LinkedIn MCP server connects AI coding agents (Claude Code, Cursor, Windsurf, Cline) to LinkedIn data through the Model Context Protocol. Instead of copying profile URLs into your chat, the agent queries LinkedIn directly: pull a prospect's work history, search for companies by industry, or look up job postings without leaving the terminal.

The important caveat: LinkedIn does not publish an official MCP server. Every implementation is community-built. Some use browser automation to interact with LinkedIn's frontend. Others use LinkedIn's API (which requires partner approval for most endpoints). A third category uses managed scraping platforms like Apify that handle the infrastructure.

Available Implementations

LinkedIn MCP servers fall into three categories: open-source browser automation, open-source API-based, and managed platforms. The right choice depends on whether you prioritize cost, reliability, or compliance.

stickerdaniel/linkedin-mcp-server (Browser Automation)

The most popular open-source option. Uses Patchright (a Playwright fork designed to evade bot detection) with persistent browser profiles. After LinkedIn cracked down on Playwright-based scraping in late 2025, this server migrated to Patchright to maintain functionality.

Supports profile lookup, company profiles, job search, and job detail retrieval. Available on PyPI as linkedin-scraper-mcp and as a Docker image. The tradeoff: browser automation is inherently fragile. LinkedIn frontend changes can break scraping, and aggressive usage triggers account restrictions.

felipfr/linkedin-mcpserver (API-Based)

A TypeScript server that uses LinkedIn's API directly. Supports profile search, profile retrieval, job search, messaging to connections, and network statistics. Built with TSyringe for dependency injection and Pino for logging.

The barrier: LinkedIn restricts API access. Most useful endpoints (People Search, Messaging) require LinkedIn partner approval or Community Management API access. Individual developers typically cannot get these credentials. If you have them, this is the cleanest implementation.

Managed Platforms (Apify, Composio, Zapier)

Apify runs LinkedIn scraping on their infrastructure. Six scraping modes: Profiles, Companies, Jobs, Posts, Search, and Profile Complete. No LinkedIn credentials needed, just an Apify API token. $5/month in free credits covers light usage.

Composio handles OAuth authentication and provides a hosted MCP endpoint. Supports posting, deleting posts, retrieving organization data, and fetching profiles. Zapier takes a no-code approach, consuming 2 Zapier tasks per MCP tool call.

Configuration

stickerdaniel/linkedin-mcp-server (Claude Code)

Install prerequisites, then add the server to your MCP configuration. The first run requires an interactive login to create a browser session.

# Install uv (Python package manager)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install Patchright browser
uvx patchright install chromium

# First-time login (opens browser for LinkedIn auth)
uvx linkedin-scraper-mcp --login

{
  "mcpServers": {
    "linkedin": {
      "command": "uvx",
      "args": ["linkedin-scraper-mcp"]
    }
  }
}

felipfr/linkedin-mcpserver (Cursor / Claude Desktop)

Requires Node.js 20+ and valid LinkedIn API credentials. Clone the repo, build, then point your MCP config at the built output.

git clone https://github.com/felipfr/linkedin-mcpserver.git
cd linkedin-mcpserver
npm install
npm run build

{
  "mcpServers": {
    "linkedin": {
      "command": "node",
      "args": ["/path/to/linkedin-mcpserver/dist/index.js"],
      "env": {
        "LINKEDIN_ACCESS_TOKEN": "your-access-token"
      }
    }
  }
}

Apify LinkedIn MCP (Any Client)

{
  "mcpServers": {
    "apify-linkedin": {
      "command": "npx",
      "args": [
        "-y", "@apify/actors-mcp-server",
        "--actors", "apify/linkedin-scraper"
      ],
      "env": {
        "APIFY_TOKEN": "your-apify-token"
      }
    }
  }
}

Available Tools

Tool availability varies by implementation. Browser automation servers expose fewer tools but require no API credentials. API-based servers offer more functionality but need approved credentials.

Typical Workflow

In practice, most developers use LinkedIn MCP servers for point lookups, not bulk operations. A common pattern: you're writing outreach code and need to verify a prospect's current role before generating an email. The agent calls get_person_profile with a LinkedIn URL, gets structured data back (name, title, company, experience), and uses it to personalize the message.

User: "Look up the CTO of Acme Corp on LinkedIn and
draft a personalized intro email"

Agent calls: get_person_profile("linkedin.com/in/jane-doe")
  -> Returns: { name, title, company, experience[], skills[] }

Agent calls: get_company_profile("linkedin.com/company/acme-corp")
  -> Returns: { industry, size, description, recent_posts[] }

Agent drafts email using both data sources.

API Limitations

LinkedIn has the most restrictive API among major social platforms. Understanding these constraints is essential before investing time in a LinkedIn MCP setup.

The practical impact: most individual developers can only access basic profile data through the official API. Anything beyond that requires either partner approval (months-long process, typically reserved for companies) or browser automation (faster but riskier).

Use Cases for Developers

Combining LinkedIn MCP with Codebase Tools

A LinkedIn MCP server handles external context: who works where, what companies do, what jobs are available. For the internal half of the equation, a codebase search MCP like WarpGrep gives the same agent deep understanding of your code. The combination lets an agent research a prospect on LinkedIn, then find the relevant outreach template in your codebase, and draft a personalized email. Two MCP servers, one workflow.

Compliance and TOS Considerations

The legal landscape around LinkedIn scraping is evolving. The hiQ Labs v. LinkedIn Supreme Court case (2022) established that scraping publicly available data is not necessarily a CFAA violation, but LinkedIn's terms of service still prohibit it, and they enforce those terms through technical measures (account restrictions, legal action against commercial scrapers).

Risk Mitigation

• Use API-based servers when possible. If you have LinkedIn API credentials, use felipfr/linkedin-mcpserver or similar. This is the only fully compliant approach.
• Keep volumes low. If using browser automation, limit to a handful of lookups per day. Do not run bulk scraping operations.
• Use a dedicated account. Do not run browser automation against your primary LinkedIn account. If the account gets restricted, it should not affect your professional network.
• Consider managed platforms. Apify and Composio assume the compliance risk on their side. Check their terms for your use case.
• Never automate messaging at scale. Automated outreach via LinkedIn messaging is the fastest path to account suspension.

Frequently Asked Questions

Is there an official LinkedIn MCP server?

No. LinkedIn does not publish an official MCP server. All implementations are community-built using either LinkedIn's restricted API, browser automation (Patchright/Playwright), or managed scraping platforms like Apify. LinkedIn has shown no indication of releasing an official MCP server.

Do I need LinkedIn API credentials?

Depends on the implementation. Browser automation servers (stickerdaniel) use your LinkedIn login session. API-based servers (felipfr) need OAuth 2.0 credentials from LinkedIn's Developer Portal. Managed platforms (Apify) only need their own API token. For most individual developers, browser automation or Apify is the practical path since LinkedIn API access is heavily restricted.

Will LinkedIn ban my account?

Browser automation carries real risk. LinkedIn detects automated patterns and can restrict accounts temporarily or permanently. Keep request volumes low (under 20-30 lookups/day), add delays between requests, and consider using a secondary account. API-based approaches with proper credentials do not carry this risk.

Can LinkedIn MCP servers send messages?

Some implementations support it. The felipfr server includes messaging via the LinkedIn API. Browser automation servers can technically interact with the messaging UI. However, automated messaging violates LinkedIn's terms and is the most likely trigger for account restrictions. Use this capability sparingly, if at all.

Which LinkedIn MCP server should I use with Claude Code?

For zero-cost setup with minimal configuration: stickerdaniel/linkedin-mcp-server via uvx. For reliability without managing browser sessions: Apify's LinkedIn MCP server ($5/month free credits). For API-based access with proper credentials: felipfr/linkedin-mcpserver. Most developers start with the stickerdaniel server and move to Apify if they need consistent results.

Related Resources

• Claude Code MCP Servers Guide
• Claude Code vs Cursor
• WarpGrep: Semantic Codebase Search MCP
• CLAUDE.md Configuration Guide