Multi-Tool MCP Server

Connect Claude AI to Google Sheets + SQLite database + PDF files simultaneously in a single conversation.

Fiverr Gig: "I will build a Claude AI assistant that reads your Sheets, database, and PDF files simultaneously"

---

What it does

A business owner can ask Claude:

• "How many active customers do we have?" → queries SQLite
• "Add a new lead to my spreadsheet" → writes to Google Sheets
• "Find the invoice for Alice" → searches PDFs
• "Give me a full business overview" → queries all 3 sources at once

---

Tools (9 total)

Tool	Data Source	Description
sheets_read	Google Sheets	Read any range as a table
sheets_write	Google Sheets	Append a row (dict or list)
db_query	SQLite	Run SELECT queries
db_insert	SQLite	Insert a new row
db_list_tables	SQLite	List all tables with columns
pdf_list	PDF Folder	List all PDF files
pdf_read	PDF Folder	Extract full text from a PDF
pdf_search	PDF Folder	Search all PDFs for a keyword
smart_query	All 3	Ask in plain English — auto-routes

---

File Structure

multi-tool-mcp-server/
├── server.py               # Entry point
├── mcp_instance.py         # Shared FastMCP instance
├── tools/
│   ├── sheets_tools.py     # sheets_read, sheets_write
│   ├── sqlite_tools.py     # db_query, db_insert, db_list_tables
│   ├── pdf_tools.py        # pdf_list, pdf_read, pdf_search
│   └── smart_query_tool.py # smart_query
├── utils/
│   └── sheets_auth.py      # Google service account auth
├── demo_data/              # Created by setup_demo.py
│   ├── sample.db           # 10 sample customers
│   └── sample.pdf          # Sample invoice
├── setup_demo.py           # Creates demo data
├── test_server.py          # Full test suite
├── requirements.txt
├── .env                    # Your credentials (gitignored)
└── .env.example

---

Setup

1. Create virtual environment

cd "e:\MCP SERVERS\multi-tool-mcp-server"
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

2. Configure .env

GOOGLE_SERVICE_ACCOUNT_FILE=C:\Users\yourname\service-account-key.json
DEFAULT_SPREADSHEET_ID=your_spreadsheet_id_here
DB_FILE_PATH=demo_data/sample.db
PDF_FOLDER_PATH=demo_data

3. Create demo data

python setup_demo.py

4. Run tests

python -m pytest test_server.py -v

5. Add to Claude Desktop

In %APPDATA%\Claude\claude_desktop_config.json, add under mcpServers:

"multi-tool": {
  "command": "e:\\MCP SERVERS\\multi-tool-mcp-server\\.venv\\Scripts\\python.exe",
  "args": ["e:\\MCP SERVERS\\multi-tool-mcp-server\\server.py"],
  "env": {
    "GOOGLE_SERVICE_ACCOUNT_FILE": "C:\\Users\\yourname\\service-account-key.json"
  }
}

---

Google Sheets Setup

1. Go to console.cloud.google.com
2. Enable Google Sheets API
3. Create a Service Account → download JSON key
4. Open your Google Sheet → Share it with the service account email
5. Copy the spreadsheet ID from the URL

---

Real-World Use Case

A freelance consultant uses this to manage their business:

Claude, give me a business overview.
→ smart_query("business overview")
   ✓ Shows all customers from SQLite (10 rows)
   ✓ Searches PDFs for recent invoices
   ✓ Shows latest leads from Google Sheets

Claude, search for any invoice mentioning Mumbai.
→ pdf_search("demo_data", "Mumbai")
   ✓ Returns matching lines from all PDFs

Claude, how many active customers?
→ db_query("SELECT COUNT(*) FROM customers WHERE status='Active'")
   ✓ Returns count instantly