PreCode

PreCode is an AI-powered educational platform designed to cultivate better programming habits by encouraging students to plan and specify their logic before writing a single line of code. It enforces a structured workflow that emphasizes deep thinking, requirement gathering, and self-reflection.

Core Features

• Specification Phase: Students must define the behavior, inputs, outputs, and edge cases of their logic. An AI tutor provides feedback and asks clarifying questions without giving away solutions.
• Coding Phase: Once the specification is approved, students implement their solution using an integrated Monaco Editor. Real-time AI assistance is available for syntax help, but the tutor remains strict about debugging.
• Reflection Phase: After submission, the AI compares the code against the initial specification, identifies mismatches, and prompts the student for a final reflection on their learning process.
• Persistence: All sessions, including specifications, code, and reflections, are securely stored for future review.

Tech Stack

• Framework: Next.js 16 (App Router)
• Editor: Monaco Editor (@monaco-editor/react)
• AI Engine: Groq SDK (Llama 3.3 70B)
• Database: Neon (Serverless PostgreSQL)
• Styling: Tailwind CSS
• Language: TypeScript

Getting Started

1. Prerequisites

Ensure you have Node.js installed on your machine. You will also need:

• A Groq API Key
• A Neon Database Connection String

2. Environment Setup

Create a .env file in the root directory and add the following variables:

GROQ_API_KEY=your_groq_api_key
DATABASE_URL=your_neon_database_url

3. Database Initialization

Run the setup script to create the necessary tables in your Neon database:

node setup-db.mjs

4. Installation

Install the project dependencies:

npm install

5. Running the Application

Start the development server:

npm run dev

The application will be available at http://localhost:3000.

Educational Philosophy

PreCode is built on the principle that programming is 80% thinking and 20% typing. By forcing a "Think-First" approach, it prevents the common pitfall of "guess-and-check" coding. The AI tutor is specifically tuned to be a facilitator rather than a solver, ensuring that the cognitive heavy lifting stays with the student.