🚀 TubeLight — Your Desk Lamp for Ideas ✨

A calm, personal, multimodal AI workspace.
Chat with frontier models • Ground answers in your own documents • Private image library • Powered by InsForge

---

---

TubeLight is a beautiful, distraction-free web app that turns InsForge into your personal AI thinking companion.
Think of it as a soft tube light glowing over your late-night desk — focused, warm, and always ready to help you explore ideas.

⭐ Star this repo

---

✨ Features

Feature	Description
🔐 Secure Personal History	Private chats scoped only to you
🌐 Multi-Model Switching	GPT-4o mini, Claude 3.5 Sonnet, Gemini & more
📄 Document RAG	Ask smart questions grounded in your own PDFs, Markdown & text files
⚡ Fast & Beautiful UI	Modern, clean and responsive design with Tailwind CSS

Built for learners, builders, researchers, writers, and solo creators who want one peaceful place to think deeply with AI.

---

📸 Screenshots & Demo

🔐 Login / Sign Up

---

💬 Chat Interface

---

📄 Document Upload

---

🤖 Model Switcher

---

👤 Author & Connect

Md Fahim Imtiaz Khan

B.Eng in Software Engineering

• 🌐 Portfolio → mdfahimimtiazkhan.dev
• 💻 GitHub → @fahimimtiaz12

---

🚀 Quick Start (Local Development)

📋 Prerequisites

Make sure you have the following installed:

• Node.js 20+ (recommended)
• npm
• An InsForge project (free tier works great)

---

⚙️ Step-by-Step Setup

1️⃣ Link Your InsForge Project

npx @insforge/cli login
npx @insforge/cli link

---

2️⃣ Set Up Environment Variables

cp .env.example .env

---

3️⃣ Install Dependencies

npm install

---

4️⃣ Start Development Server

npm run dev

Open your browser and visit:

http://localhost:5173

---

💡 Important Tip

Add the following URL to Allowed Auth Redirects inside your InsForge dashboard:

http://localhost:5173

---

🗄️ Database & Storage

This uses 4 main tables automatically created by the migrations:

Table	Purpose
tubelight_chats	Stores chat sessions and metadata
tubelight_messages	Individual messages in each chat
tubelight_chunks	Document chunks with vector embeddings (powers RAG)
tubelight_document_sources	Metadata of uploaded documents (PDFs, etc.)

---

🔒 Security

• Row-Level Security (RLS) is enabled
• Every user can only access their own data
• All documents and images are stored in the private bucket:

tubelight-docs

Everything is private by default — only the signed-in user can access their data.

---

🌍 Deploy to Production

Build the project

npm run build

Deploy using InsForge CLI

npx @insforge/cli deployments deploy .

---

🛠️ Tech Stack

Layer	Technology
Frontend	Vite + React + TypeScript + Tailwind CSS
Backend	InsForge (Auth • Postgres • Vector Search • Storage • AI Gateway)

---

📜 License

MIT License — Free to use, modify, and build upon.

---

🤝 Contributing

We welcome contributions of all sizes!

Please read our Contributing Guide first before submitting pull requests or issues.

---

🌟 About TubeLight

TubeLight — crafted with care — 2026 🌟
Your calm corner of the internet for thinking with AI.

Made with ❤️ by Md Fahim Imtiaz Khan

---

💬 Final Note

Happy building & thinking!

If you have any questions, feel free to open an issue or reach out directly — I’d love to help you customize TubeLight for your workflow.

⭐ If you like this project, please star the repo!

It helps others discover TubeLight and supports the project.

---

🚀 TubeLight

Your Desk Lamp for Ideas ✨