sqlite_manager MCP server

Context-Driven - (MCP) SQLite Server Manager

An enterprise-grade Model Context Protocol (MCP) Server v2 implementation designed to manage local SQLite databases dynamically through LLM chat sessions. This server provides AI agents (such as Claude Desktop, Cursor, and VS Code) with deep schema inspection, secure read/write capabilities, multi-statement ACID transactions, and structural database optimization tools natively via STDIO.

🚀 Key Features

• Zero-Config Stateless Sessions: Pass the target database file path directly within your chat session text—no hardcoded environment variables required.
• Multi-Platform Target Compatibility: Natively parses and handles Windows absolute path strings (C:\data\db.sqlite) as well as macOS and Linux paths (/var/db/app.db) without configuration changes.
• Deep Schema Inspection: Structural tools to dynamically extract schemas, tables, columns, indexes, and active foreign key constraints.
• Strict Security Boundaries: Built-in read-only query paths enforcement using system-level database structures to prevent unauthorized modifications.
• Transactional Pipeline Execution: Execute sequential modification loops within a secure rollback transaction boundary.
• Built-in Maintenance Hooks: Native optimization commands supporting automated database defragmentation (VACUUM), data planner analytics (ANALYZE), and database validation checks (integrity_check).

---

🛠️ System Requirements & Prerequisites

• Operating System: Windows 10 / 11, macOS, or Linux.
• Python Environment: Python 3.10 or higher.
• Supported Clients: Any Model Context Protocol host client (Claude Desktop App, Cursor, Zed, etc.).

---

📦 Installation & Setup Guide

1. Clone & Initialize the Project Workspace

Open your preferred terminal window (such as PowerShell or Command Prompt on Windows) and run:

# Clone the repository workspace
git clone https://github.com/KikeVen/sqlite-manager.git
cd sqlite-manager

# Build an isolated Python virtual environment
python -m venv venv

2. Activate Environment & Install Dependencies

Activate the virtual environment according to your operating system platform:

On Windows (PowerShell):

.\sqlite-manager\Scripts\Activate.ps1

On Windows (Command Prompt):

.\sqlite-manager\Scripts\activate.bat

On macOS / Linux:

source sqlite-manager/bin/activate

Install the production dependencies utilizing the provided tracking file:

pip install -r requirements.txt

---

⚙️ LLM Integration Configuration

To integrate this utility with your preferred AI client application, link the executable environment directly via your system config path.

Claude Desktop Integration

Navigate to your active system app settings path:

• Windows Configuration Path: %APPDATA%\Claude\claude_desktop_config.json
• macOS Configuration Path: ~/Library/Application Support/Claude/claude_desktop_config.json

Add or merge the following block inside your configuration schema:

{
  "mcpServers": {
    "sqlite-manager": {
      "command": "C:\\absolute\\path\\to\\mcp-sqlite-env\\Scripts\\python.exe",
      "args": ["C:\\absolute\\path\\to\\server.py"]
    }
  }
}

⚠️ Important: Ensure you replace the absolute file path tokens with the true location pointers on your disk. On Windows systems, escape all folder separators using double backslashes (\\).

---

💡 Conversational Chat Usage

Once integrated, your LLM agent will dynamically recognize the tools and extract the file paths out of your conversational phrasing. There is no separate setup script required.

Example Prompts to Try

• "Using sqlite-manager check my SQLite database structure at C:/Data/app.db and output a full list of all available tables."
• "Using sqlite-manager run a structural analysis of the users table inside /Users/username/Projects/analytics.db to show its columns and keys."
• "Using sqlite-manager execute a SELECT query over the items table in C:\Cache\local.db to extract records where the total is greater than 500."
• "Using sqlite-manager optimize the database file page size located at /var/lib/store.db using a vacuum operation."

---

🛠️ Core Tool Specification Sheet

This server exposes the following functional tool endpoints to your LLM agent:

Tool Method Identifier	Parameter Arguments	Core Operational Focus
list_tables	db_path	Returns an array string containing all tracking tables and active database views.
describe_table	db_path, table_name	Pulls table layout structures including columns, constraints, and indexes.
execute_read_query	db_path, sql_query	Securely runs read-only data evaluation lookups using PRAGMA query_only.
execute_write_query	db_path, sql_query, params	Handles mutation cycles like INSERT, UPDATE, or structural ALTER TABLE changes.
execute_transaction	db_path, sql_queries	Loops multiple operational updates safely within an atomic ACID rollback block.
optimize_and_maintain	db_path, operation	Performs low-level runtime engine processes: vacuum, analyze, or integrity_check.

---

📄 License

Distributed under the MIT License. See LICENSE file for more details.