sql GPT

A natural-language to SQL query tool. Ask questions about your data in plain English and get instant results with visualizations.

Built with Next.js, Gemini AI, and Supabase.

Features

• Natural Language Queries — Type questions in plain English and get accurate SQL generated by Gemini 2.5 Flash.
• Live Query Execution — Generated SQL runs against a live Supabase PostgreSQL database with results displayed instantly.
• Smart Visualizations — Results are rendered as bar charts, line charts, or tables based on query intent (powered by Recharts).
• Multiple Schemas — Choose from pre-loaded database schemas: E-Commerce, SaaS Analytics, and HR & Operations.
• Dark / Light Mode — Toggle between themes with next-themes.
• Chat Interface — Conversational UI with message history and a "New Chat" reset.

Tech Stack

Layer	Technology
Framework	Next.js 14 (App Router)
AI	Gemini 2.5 Flash via Vercel AI SDK
Database	Supabase (PostgreSQL)
Charts	Recharts
UI	shadcn/ui, Radix, Tailwind CSS
Validation	Zod

Getting Started

Prerequisites

• Node.js 18+
• A Google AI Studio API key
• A Supabase project with seed data

Setup

1. Clone the repo

   git clone https://github.com/upendrapant/query-craft.git
   cd query-craft

2. Install dependencies

   npm install

3. Configure environment variables

   Create a .env.local file in the project root:

   GOOGLE_GENERATIVE_AI_API_KEY=your_gemini_api_key
   NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
   NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key
   SUPABASE_SERVICE_ROLE_KEY=your_supabase_service_role_key

4. Run the dev server

   npm run dev

   Open http://localhost:3000.

Usage

1. Select a database schema from the dropdown in the header.
2. Type a question or click a suggested query.
3. View the generated SQL, explanation, and results as a chart or table.

License

MIT