SQL Pilot

A LangGraph-based AI agent that connects with a custom MCP server to provide natural language database interactions. SQL Pilot enables users to query, manipulate, and manage SQLite databases through conversational AI.

Features

• Natural Language Queries: Ask questions about your database in plain English
• Database Operations: Read, insert, update, delete, and create tables
• Schema Management: View table structures and relationships
• Authentication & Security: Built-in user authentication with consent management
• Web Interface: Clean Gradio-based UI for easy interaction
• Session Memory: Maintains conversation context and database schema awareness

Capabilities

Database Operations

• List all tables in the database
• Read data with filtering, limits, and ordering
• Insert new records with validation
• Update existing records with WHERE clauses
• Delete records with safety constraints
• Create new tables with custom schemas
• Drop tables (with consent)

AI-Powered Features

• Intelligent query parsing and operation planning
• Context-aware follow-up questions
• Automatic parameter mapping and validation
• Error handling and fallback mechanisms
• Session-based memory and learning

Quick Start

Prerequisites

• Python 3.8+
• OpenAI API key (set in environment variables)

Installation

1. Clone the repository

   git clone https://github.com/chandraprvkvsh/SQL-Pilot.git
   cd SQL-Pilot

2. Install dependencies

   pip install -r requirements.txt
   cd Query-MCP
   pip install -r requirements.txt

3. Set environment variables

   export OPENAI_API_KEY="your-openai-api-key"

Running the System

1. Start the MCP Server

   cd Query-MCP (github.com/chandraprvkvsh/Query-MCP)
   python server.py

2. Start the Web Interface (in a new terminal)

   python ui.py

3. Access the Interface

   • Open http://localhost:7860 in your browser
   • Upload a SQLite database or create a sample database
   • Start chatting with your database!

Usage Examples

"Show me all tables in the database"
"List all users with their email addresses"
"Add a new user named 'Alice' with email 'alice@example.com'"
"Update the user with id 1 to have name 'Bob'"
"Create a new table called 'products' with columns id, name, and price"
"Show me the schema of the users table"
"What was the last operation I performed?"

Architecture

• LangGraph Agent: Orchestrates the conversation flow and decision making
• Query-MCP Server: Custom MCP server handling database operations
• FastMCP Framework: Enables tool calling and resource management
• Gradio UI: Provides the web interface for user interaction
• SQLite Database: Stores and manages your data
• Persistent Agent Memory: Stores user context, session history, and schema awareness for intelligent multi-turn conversations

Security Features

• Consent-based destructive operations
• Parameterized queries preventing SQL injection
• Input validation and sanitization
• Session timeout and cleanup