🧠 GHL MCP Server – Anthropic LangGraph Agent Integration

🧭 Purpose

This project exposes GoHighLevel (GHL) sub-account tools to an Anthropic-powered LangGraph agent using the Model Connection Protocol (MCP). It acts as a secure intermediary layer between the LangChain ecosystem and GHL APIs—designed for clean modular separation and zero-handoff latency.

This repo is built to support:

• 🔹 Claude-based LangGraph Agents as MCP clients
• 🔹 FastAPI MCP Server exposing GHL sub-account tools (contacts, pipelines, automations)
• 🔹 Secure credential loading via .env
• 🔹 Dynamic tool registration using LangChain MCP adapters

---

🧱 Architecture

 ┌────────────────────────────────────┐
 │          LangGraph Agent           │
 │  (Claude, Anthropic, LangChain)    │
 └────────────────────────────────────┘
                │
         MCP Client (HTTP)
                │
 ┌────────────────────────────────────┐
 │         GHL MCP Server (This)      │
 │   FastAPI + LangChain Tools        │
 └────────────────────────────────────┘
                │
        REST API → GHL Sub-Account

• MCP Server = FastAPI server exposing LangChain-compatible tools from GHL
• LangGraph Agent = Claude-powered LangChain graph calling MCPClient.load_tools()
• Secure Comm = All tool usage routed over HTTP via LangChain's MCP standard

---

📦 Tool Endpoints to Expose

Each tool is implemented using @mcp_tool and automatically discoverable by any LangChain-compatible MCP client.

Tool Name	GHL Endpoint	Description
get_contact_info	/contacts/{id}	Fetch contact details
list_opportunities	/opportunities/	List all open opps
trigger_webhook	Custom workflow trigger	Hit custom workflow webhook
get_pipeline_info	/funnels/	Retrieve funnel/pipeline structure
create_note	/contacts/{id}/notes	Create a note on contact

---

🔐 ENV VARS (.env)

Set the following variables in .env at the root of ghl_mcp_server/:

GHL_API_BASE_URL=https://rest.gohighlevel.com/v1
GHL_API_KEY=sk_your_primary_api_key_here
GHL_SUB_ACCOUNT_ID=abc123456789
ALLOWED_ORIGINS=*

---

🕵️ GHL SCAVENGER HUNT CHECKLIST

To go live, collect these credentials:

🧩 What	📍 Where to Find It
GHL_API_KEY	GHL → Settings → API Keys (create or use existing)
GHL_SUB_ACCOUNT_ID	URL bar while inside the sub-account dashboard
Webhook Trigger URLs	Automations → Webhook → Copy full URL
Funnel / Pipeline IDs	Funnels → Click → Copy from URL
Contact ID (for testing)	Contacts → Click any → Copy ID from URL

---

🚀 Dev Startup

# 1. Create .env file from template
cp env.template .env
# Then edit .env with your real GHL credentials

# 2. Install deps
pip install -r requirements.txt

# 3. Run server
uvicorn mcp_server:app --host 0.0.0.0 --port 8000
# OR use the startup script:
python scripts/start_server.py

# 4. Test the server
python examples/simple_client.py

# 5. Try interactive mode
python examples/simple_client.py interactive

---

🧰 Stack

• Python 3.11+
• FastAPI
• langchain-mcp-adapters
• MCPServer + Claude LangGraph agent
• Optional: NGROK or Azure App Service for public MCP endpoint

---

🎯 Goal

By running this repo, your Claude-powered agent can query GHL data and trigger workflows through MCP with zero coupling. Extend this to include new GHL tools as needed.

"Modular tools. Global reach. Absolute power."

⚡ Ready to connect your Claude agent to GoHighLevel!