Customer Support MCP Server
README.md
# 🎫 Customer Support MCP Server
<div align="center">
## 🤖 AI-Powered Customer Support MCP Server
**Connect Cursor AI to a real SQLite database through Model Context Protocol (MCP).**
</div>
---
## 🌟 Project Overview
The **Customer Support MCP Server** is a Python-based MCP server that allows an AI application such as **Cursor** to interact with a real **SQLite database**.
Instead of allowing the AI to directly execute SQL queries, the server exposes safe and well-defined **business tools** for customer support operations.
### 🎯 Main Goal
Demonstrate how an AI application can use **MCP (Model Context Protocol)** to:
- 🎫 Create customer support tickets
- 🔍 Retrieve existing tickets
- 🔎 Search tickets
- ✏️ Update ticket information
- 💾 Store ticket data in SQLite
- 🤖 Allow Cursor AI to perform operations using natural language
---
# 🏗️ Architecture
```text
👤 USER
│
│ Natural Language
▼
┌───────────────┐
│ Cursor │
│ AI Agent │
└───────┬───────┘
│
│ MCP
▼
┌────────────────────────┐
│ Customer Support MCP │
│ Server │
├────────────────────────┤
│ 🎫 create_ticket │
│ 🔍 get_ticket │
│ 🔎 search_tickets │
│ ✏️ update_ticket │
└────────────┬───────────┘
│
│ SQL
▼
┌─────────────────┐
│ SQLite │
│ support.db │
└─────────────────┘
===================
🚀 Features
Feature Description
🎫 Create Ticket Create a new customer support ticket
🔍 Get Ticket Retrieve a ticket using its ID
🔎 Search Tickets Search using keyword, email, status, or priority
✏️ Update Ticket Update status, priority, or description
💾 SQLite Persistent local database
🤖 Cursor AI client for interacting with MCP
🧪 MCP Inspector Test and debug MCP tools
⚡ uv Fast Python package/environment management
=======================
🧰 Technology Stack
Technology Purpose
🐍 Python 3.12 Application development
🔌 MCP AI ↔ Tool communication
🗄️ SQLite Database
🤖 Cursor AI/MCP client
⚡ uv Python environment & dependency management
🧪 MCP Inspector MCP server testing
=====================
📁 Project Structure
customer_support_mcp/
│
├── 📁 .cursor/
│ └── 📄 mcp.json
│
├── 📁 data/
│ └── 🗄️ support.db
│
├── 📁 .venv/
│
├── 🐍 database.py
├── 🐍 server.py
├── 📄 pyproject.toml
├── 🔒 uv.lock
├── 📄 README.md
└── 📄 .gitignore
==================
🗄️ Database Schema
The SQLite database contains a tickets table.
┌─────────────────────────────────────────┐
│ tickets │
├─────────────────────────────────────────┤
│ id INTEGER PRIMARY KEY │
│ customer_name TEXT │
│ customer_email TEXT │
│ subject TEXT │
│ description TEXT │
│ status TEXT │
│ priority TEXT │
│ created_at TIMESTAMP │
│ updated_at TIMESTAMP │
└─────────────────────────────────────────┘
============
🧪 Test Using MCP Inspector
The MCP Inspector allows you to test the tools before connecting Cursor.
Run:
uv run mcp dev server.py
Open the Inspector URL displayed in the terminal.
You should see:
Customer Support MCP Server
Tools
✓ create_ticket
✓ get_ticket
✓ search_tickets
✓ update_ticket
=======================
🧑💻 Connect MCP Server to Cursor
Create:
.cursor/mcp.json
=======================
🔄 Restart Cursor
After saving .cursor/mcp.json:
Cursor
↓
Restart Cursor
↓
Open customer_support_mcp
↓
Settings
↓
Tools & MCP
-------------
You should see:
🟢 customer-support
Tools:
✓ create_ticket
✓ get_ticket
✓ search_tickets
✓ update_ticket
===============
🤖 Using Cursor AI
Once the MCP server is connected, you can use natural language.
🎫 Create Ticket
Ask Cursor:
Create a ticket for John Smith.
Email:
john@example.com
Issue:
Customer cannot login.
Priority:
high
Cursor can call:
create_ticket()
🔍 Retrieve Ticket
Ask:
Show me ticket number 1.
Cursor can call:
get_ticket(ticket_id=1)
🔎 Search Tickets
Ask:
Find all open urgent tickets.
Cursor can call:
search_tickets(
status="open",
priority="urgent"
)
✏️ Update Ticket
Ask:
Change ticket 1 to resolved.
Cursor can call:
update_ticket(
ticket_id=1,
status="resolved"
)
===================
🔥 Example Workflow
👤 User
│
│ "Create a ticket for John"
▼
🤖 Cursor AI
│
│ MCP Tool Call
▼
🔌 create_ticket()
│
│ SQL INSERT
▼
🗄️ SQLite
│
│ Ticket #1
▼
🔌 MCP Server
│
▼
🤖 Cursor
│
▼
👤 User
==================
📊 Example or Sample database of Tickets Output
After creating tickets, the database might contain:
┌────┬──────────────┬─────────────────────┬──────────────┬──────────┐
│ ID │ Customer │ Subject │ Status │ Priority │
├────┼──────────────┼─────────────────────┼──────────────┼──────────┤
│ 1 │ John Smith │ Unable to login │ open │ high │
│ 2 │ Sarah Jones │ Payment failed │ open │ urgent │
│ 3 │ Mike Brown │ Password reset │ resolved │ medium │
└────┴──────────────┴─────────────────────┴──────────────┴──────────┘
=================
🧠 MCP Concept
The most important concept in this project is:
AI APPLICATION
│
▼
┌───────────┐
│ MCP │
└─────┬─────┘
│
▼
BUSINESS TOOLS
│
┌────────┼────────┐
▼ ▼ ▼
CREATE SEARCH UPDATE
│ │ │
└────────┼────────┘
▼
DATABASE
The AI does not directly manipulate the SQLite database.
---------
Instead:
AI
↓
MCP Tool
↓
Python Function
↓
SQLite
↓
Result
↓
AI
===============
✅ MCP Business Tools
AI
↓
MCP Tool
↓
Validated Python Function
↓
SQLite
===============
🔍 Verify SQLite Database
The database will be created here:
data/support.db
You can verify the database using Python:
sqlite3 "C:\Users\your name\path\folder name\data\support.db"
============
🎓 Learning Outcomes
After completing this project, you will understand:
🔌 MCP
Model Context Protocol
↓
MCP Server
↓
MCP Tools
↓
MCP Client
---
🗄️ Database Integration
Python
↓
sqlite3
↓
SQLite
---
🤖 AI Integration
Cursor AI
↓
MCP
↓
Python Tools
↓
SQLite
---
💼 Business Operations
CREATE
↓
READ
↓
SEARCH
↓
UPDATE
============
🏆 Project Summary
Component Technology
🤖 AI Client Cursor
🔌 Protocol MCP
🐍 Backend Python
🗄️ Database SQLite
⚡ Environment uv
🧪 Testing MCP Inspector
🎫 Business Domain Customer Support
=============
🌟 Final Architecture
╔════════════════════╗
║ CURSOR ║
║ 🤖 AI AGENT ║
╚═════════╤══════════╝
│
│ MCP
▼
╔══════════════════════╗
║ CUSTOMER SUPPORT MCP ║
║ SERVER ║
╠══════════════════════╣
║ 🎫 create_ticket ║
║ 🔍 get_ticket ║
║ 🔎 search_tickets ║
║ ✏️ update_ticket ║
╚══════════╤═══════════╝
│
│ SQL
▼
╔══════════════════════╗
║ SQLite ║
║ support.db ║
╚══════════════════════╝
==============
🚀 Complete Command Flow
# 1. Go to project
cd C:\Users\path
# 2. Create environment
uv venv --python 3.12
# 3. Activate environment
.venv\Scripts\Activate.ps1
# 4. Install MCP
uv add "mcp[cli]"
# 5. Verify SQLite
uv run python -c "import sqlite3; print('SQLite OK')"
# 6. Verify MCP
uv run python -c "from mcp.server.fastmcp import FastMCP; print('MCP OK')"
# 7. Run MCP Inspector
uv run mcp dev server.py
# 8. Run MCP server directly
uv run python server.py
==============
❤️ Final Message
<div align="center">
🎫 Customer Support MCP Server
Python 🐍 + MCP 🔌 + SQLite 🗄️ + Cursor 🤖
AI + MCP + Database = Intelligent Business Operations
TDQS
A4.2/5.0
Scored across 4 tools
Disambiguation5/5
Each tool has a distinct purpose: create, retrieve by ID, search with filters, and update. No overlapping functionality, making it easy for an agent to select the correct tool.
Naming Consistency5/5
All tool names follow a consistent verb_noun pattern: create_ticket, get_ticket, search_tickets, update_ticket. Minor pluralization in search_tickets is acceptable and consistent with typical search APIs.
Tool Count5/5
Four tools cover the essential ticket operations without bloat. The count is well-scoped for a customer support server, fitting the ideal range of 3-15 tools.
Completeness5/5
The server provides full CRUD-like coverage (create, read, search, update) for tickets. Closing a ticket is achievable via update_ticket's status parameter, so no critical lifecycle gaps exist.
Maintenance
ActivitySlowing
ResponsivenessNo issues