Unibot from Belweave

A multi-provider LLM chat interface built with Next.js 16 that connects to AWS Bedrock, OpenAI, and OpenRouter. The application features real-time streaming responses, configurable model selection, and a modern, curve-focused UI design following shadcn/ui best practices.

Tech Stack

• Framework: Next.js 16 (App Router)
• Database: PostgreSQL with Prisma ORM
• UI Components: shadcn/ui
• Styling: Tailwind CSS 4
• AI Backends: AWS Bedrock, OpenAI, OpenRouter
• Language: TypeScript
• Icons: Lucide React
• Design System: Custom design tokens with comprehensive documentation

Features

• Google OAuth Authentication - Secure sign-in with Google accounts
• Real-time streaming responses from multiple LLM providers
• Multi-provider support - AWS Bedrock, OpenAI, and OpenRouter
• Configurable model selection - Switch between providers and models via settings
• Image support - Send images with vision-capable models (Claude 3.5 Sonnet, GPT-4o, etc.)
• Modern UI - Curved, friendly interface with smooth animations
• Dark/Light mode - System-aware theme with manual toggle
• Design system - Comprehensive design tokens for consistent styling

Design Documentation

This project includes a comprehensive design system:

• 📖 Design System Guide - Complete design philosophy, tokens, and patterns
• 🚀 Quick Reference - Common patterns and code snippets
• 📝 UI Modernization - Recent UI updates and migration guide

Prerequisites

• Bun 1.3.5 or higher
• Docker and Docker Compose (for local PostgreSQL)
• AWS Account with Bedrock access (if using Bedrock)
• AWS credentials configured (see AWS Configuration below)
• Google Cloud Platform account for OAuth (see Authentication Setup below)
• OpenAI API key (if using OpenAI)
• OpenRouter API key (if using OpenRouter)

Database Setup

1. Start PostgreSQL with Docker

docker compose up -d

This starts a PostgreSQL 17 container with:

• User: unibot
• Password: unibot_secret
• Database: unibot
• Port: 5432

2. Configure Environment Variables

Add the following to your .env.local file:

DATABASE_URL="postgresql://unibot:unibot_secret@localhost:5432/unibot?schema=public"

3. Run Database Migrations

# Generate Prisma client
bun run db:generate

# Push schema to database (development)
bun run db:push

# Or create a migration (production)
bun run db:migrate

Available Database Scripts

Command	Description
bun run db:generate	Generate Prisma client
bun run db:push	Push schema to database
bun run db:migrate	Create and run migration
bun run db:migrate:prod	Deploy migrations in production
bun run db:studio	Open Prisma Studio (database GUI)

Database Schema

The database includes the following models:

• User - User profile (email, name, Google ID, timestamps)
• Account - OAuth provider accounts (Google, etc.)
• Session - User sessions for authentication
• VerificationToken - Email verification tokens