TL;DR

• MCP Server connects AI assistants (like Claude or OpenAI agents) to GoHighLevel data
• Setup uses Private Integration Tokens (PITs) and secure HTTP headers
• Supports tools like contacts, conversations, calendars, pipelines, payments, workflows
• Works with any HTTP-capable client—no SDK needed
• Enables automating follow-ups, tagging, messaging, reporting and multi-tool orchestration

What Is the HighLevel MCP Server?

MCP stands for Model Context Protocol—an open standard that lets AI agents query and update your CRM using real-time context. GoHighLevel's hosted MCP server lets you:

• Read and modify contacts, conversations, pipelines, calendar events, payments, and more
• Trigger workflows or automations via natural language commands
• Connect external tools (e.g. Slack, ClickUp, n8n) through unified AI orchestrations

Why Use the MCP Server?

• Fast AI integration—no SDK or code required, just HTTP requests
• Secure access via scoped PIT tokens only granting minimal permissions
• Scalable design, built for future tool expansion up to 250+ functionalities
• Compatible with clients like Cursor, Windsurf, Claude Desktop, OpenAI Playground

Step-by-Step Setup

a. Generate Your Private Integration Token (PIT)

• In GoHighLevel go to Settings > Private Integrations
• Create new integration, select required scopes (contacts, conversations, payments, etc.) and copy the token

Add the MCP server config:

{
  "mcpServers": {
    "ghl": {
      "url": "https://services.leadconnectorhq.com/mcp/",
      "headers": {
        "Authorization": "Bearer YOUR_PIT_TOKEN",
        "locationId": "YOUR_LOCATION_ID"
      }
    }
  }
}

c. Start Calling Tools

Use calls like:

• contacts_get-contact
• conversations_send-a-new-message
• opportunities_update-opportunity
• payments_list-transactions
• locations_get-location
  … and more—all available tools are listed in the Help Center docs

Real-World Use Cases

Troubleshooting Guide

What You Need to Get Started

• GoHighLevel account with admin permissions
• PIT (Private Integration Token) with tool access scopes
• AI client supporting HTTP MCP (e.g. Claude, Cursor)
• Location ID from your sub-account headers or setup prompts

Sample AI Workflow

1. Agent pulls unread contacts
2. Tags “New Lead” and sends follow-up SMS
3. Updates pipeline stage
4. Creates task for sales rep
5. Sends alert to Slack automatically