MCP Server Prototypes

Overview

This project investigates the Model Context Protocol (MCP) and demonstrates how AI assistants can interact with external data sources through MCP servers.

Three MCP server prototypes were developed:

• Excel MCP Server
• PDF MCP Server
• SQL MCP Server

These servers expose tools that allow AI assistants such as Claude Desktop to read, create, update, and analyze local resources through a standardized protocol.

---

MCP Architecture

MCP follows a client-server architecture.

1. The user interacts with an AI client.
2. The AI client communicates with MCP servers using the MCP protocol.
3. MCP servers expose tools.
4. Tools perform operations on external resources.
5. Results are returned to the AI client and presented to the user.

Architecture Diagram

User
  ↓
Claude Desktop (MCP Client)
  ↓
MCP Protocol
  ↓
MCP Servers
 ├── Excel Server
 ├── PDF Server
 └── SQL Server

---

Technical Stack

Component	Technology
Language	Python 3.x
MCP Framework	MCP Python SDK (FastMCP)
Excel Processing	openpyxl
PDF Processing	pdfplumber
Database	SQLite
Testing Client	Claude Desktop
Development Environment	VS Code

---

Project Structure

mcp-prototypes/
│
├── README.md
├── requirements.txt
│
├── excel_mcp/
│   ├── server.py
│   └── sample.xlsx
│
├── pdf_mcp/
│   ├── server.py
│   ├── sample1.pdf
│   └── sample2.pdf
│
├── sql_mcp/
│   ├── server.py
│   └── database.db
│
└── .gitignore

---

Setup

Create Virtual Environment

python3 -m venv venv

Activate Environment

macOS/Linux

source venv/bin/activate

Windows

venv\Scripts\activate

Install Dependencies

pip install -r requirements.txt

SQLite is included with Python and does not require additional installation.

---

Claude Desktop Configuration

Example MCP configuration:

{
  "mcpServers": {
    "excel-mcp": {
      "command": "/path/to/venv/bin/python",
      "args": [
        "/path/to/excel_mcp/server.py"
      ]
    },

    "pdf-mcp": {
      "command": "/path/to/venv/bin/python",
      "args": [
        "/path/to/pdf_mcp/server.py"
      ]
    },

    "sql-mcp": {
      "command": "/path/to/venv/bin/python",
      "args": [
        "/path/to/sql_mcp/server.py"
      ]
    }
  }
}

Restart Claude Desktop after modifying the configuration.

---

Excel MCP Server

Tools

Tool	Description
server_directory	Show server location
list_excel_files	List available Excel files
create_excel	Create workbook
read_excel	Read active sheet
read_range	Read cell range
read_sheet	Read specific sheet
write_row	Append row
update_cell	Update cell

Example Prompts

• List available Excel files
• Create sales.xlsx with columns Customer, Product, Quantity
• Add a new row for Alice
• Read sales.xlsx
• Update Quantity to 500
• Read Sheet2 from sample.xlsx

---

PDF MCP Server

Tools

Tool	Description
server_directory	Show server location
list_pdf_files	List available PDFs
read_pdf	Extract text
read_pdf_tables	Extract tables
pdf_metadata	Retrieve metadata

Example Prompts

• List available PDF files
• Summarize sample1.pdf
• Extract tables from sample2.pdf
• Show metadata for sample2.pdf
• What are the key findings in sample2.pdf?

---

SQL MCP Server

Tools

Tool	Description
server_directory	Show server location
list_database_files	List SQLite databases
list_tables	Show tables
describe_table	Show schema
run_query	Execute SELECT queries
run_write	Execute INSERT/UPDATE/DELETE queries
seed_demo_data	Create demo data

Example Prompts

• Create demo data
• Show all orders
• Describe the orders table
• Add a new order for Sofia
• Update Sofia's quantity to 3
• Which customer placed the most orders?

---

Key Findings

Traditional AI assistants are limited to information available within the conversation context.

The Model Context Protocol extends these capabilities by allowing AI clients to discover and invoke external tools exposed by MCP servers.

In this project:

• The Excel MCP server provided access to spreadsheets.
• The PDF MCP server provided access to document contents.
• The SQL MCP server provided access to database records.

These tools enabled Claude Desktop to perform real file and database operations through a standardized protocol.

---

Limitations

These prototypes were designed for demonstration and learning purposes.

Current limitations include:

• No authentication
• Limited error handling
• Direct SQL execution
• Files restricted to server directories
• No role-based access control