SQLite Server

Query, explore, and manage SQLite databases through the Model Context Protocol. Connect any MCP-compatible AI client to your databases — Claude Desktop, Cursor, VS Code, and more.

License: MIT Python 3.10+

---

✨ Features

• 🔍 3 Tools: query, execute, list_tables — full database interaction
• 📄 2+ Resources: sqlite://tables (full schema), sqlite://{table}/schema (per-table)
• 💬 1 Prompt: database_analyst(question) — guided analysis workflow
• 🔒 Security-First: Read-only by default, explicit --allow-write flag for mutations
• 📊 WAL Mode: Write-Ahead Logging for better concurrent access
• 🖥️ Dual Transport: stdio (local) and Streamable HTTP (remote)
• 🎲 Sample Data: Seed script included for instant testing

---

🚀 Quick Start

# 1. Setup
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# 2. Create sample database
python seed_data.py --db-path /tmp/sample.db

# 3. Start server (read-only mode)
python sqlite_server.py --db-path /tmp/sample.db

---

🛠️ Tools

query(sql: str) -> str

Execute a SELECT query. Only SELECT and PRAGMA statements are allowed.

Example: query("SELECT name, price FROM products WHERE category = 'Electronics'")

execute(sql: str) -> str

Run INSERT, UPDATE, or DELETE. Requires --allow-write flag.

Example (with --allow-write):
  execute("INSERT INTO products (name, price, category, stock) VALUES ('Webcam 4K', 129.99, 'Electronics', 50)")

list_tables() -> str

List all tables with row counts.

---

🔒 Security Model

Feature	Default	With --allow-write
SELECT/PRAGMA queries	✅ Allowed	✅ Allowed
INSERT/UPDATE/DELETE	❌ Blocked	✅ Allowed
DROP/ALTER/CREATE	❌ Blocked	❌ Blocked
Row factory protection	✅ Enabled	✅ Enabled

⚠️ Production tip: Always use read-only mode for public-facing or shared databases. Only enable --allow-write in trusted environments.

---

📄 Resources

sqlite://tables

Returns complete database schema with column types, nullability, defaults, and primary keys.

sqlite://{table}/schema

Returns schema for a specific table.

---

💬 Prompts

database_analyst(query_question: str)

Creates a structured prompt for analyzing database data.

---

🔌 Connecting to Clients

Claude Desktop

{
  "mcpServers": {
    "sqlite": {
      "command": "python",
      "args": ["/ABSOLUTE/PATH/mcp-sqlite-server/sqlite_server.py", "--db-path", "/tmp/sample.db"]
    }
  }
}

Cursor

Settings → Features → MCP → Add Server:

• Name: sqlite
• Type: command
• Command: python /ABSOLUTE/PATH/mcp-sqlite-server/sqlite_server.py --db-path /tmp/sample.db

---

📁 Project Structure

mcp-sqlite-server/
├── sqlite_server.py      # Main server (FastMCP)
├── seed_data.py          # Sample database creator
├── requirements.txt
├── setup.sh
├── README.md
└── .gitignore

---

📜 License

MIT