Tab Bankruptcy: Turn Your Chrome Tab Chaos into a Local AI-Powered Vault

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

I built Tab Bankruptcy, a local AI-powered Chrome tab manager for a friend who had a habit of keeping way too many browser tabs open across multiple Chrome windows.

The problem was simple: after collecting dozens or even hundreds of tabs, finding one specific tab became difficult, and closing the browser meant potentially losing that entire context.

Tab Bankruptcy lets them:

• Capture tabs across multiple Chrome windows
• Save complete browsing sessions locally
• Automatically categorize tabs using AI
• Generate short AI summaries
• Search tabs by title, URL, category, or summary
• Filter tabs by category
• Restore saved sessions back into Chrome
• Preserve window grouping, tab order, and pinned tabs where possible

Instead of treating all those tabs as clutter, Tab Bankruptcy turns them into a searchable personal vault.

Demo

🎥 Demo Video: https://youtu.be/0Kyy7g3F-cs

The demo shows the complete workflow:

Chrome tab chaos → Capture → AI categorization → AI summaries → Search → Restore

Code

💻 GitHub: https://github.com/AdityaAlibade/Tab-Bankruptcy.git

The project is organized into three main parts:

• Chrome Extension — captures and restores browser tabs
• FastAPI Backend — handles local APIs and persistence
• React Dashboard — provides the interface for searching and managing sessions

How I Built It

The project is built around local open-source AI, rather than a cloud AI API.

Tech Stack

• Chrome Extension — TypeScript + Manifest V3
• Frontend — React + TypeScript + Vite
• Backend — Python + FastAPI
• Database — SQLite
• Local AI Runtime — Ollama
• AI Model — Qwen2.5-Coder:7B

The architecture is:

Chrome Extension
       ↓
    FastAPI
       ↓
    SQLite
       ↓
React Dashboard
       ↓
    Ollama
       ↓
Qwen2.5-Coder:7B

The extension captures the user's tabs and sends the session to the local FastAPI backend. SQLite stores the session and tab information.

When AI functionality is needed, the backend sends the relevant tab title and URL information to the locally running Ollama model. The model produces categories and short summaries, which are then stored locally.

The dashboard provides local search and filtering without requiring another AI request.

The restore flow works in reverse: the dashboard creates a restore request, and the Chrome extension uses Chrome's Tabs API to recreate the saved session.

There is no cloud AI API required.

Why Does Open Innovation Matter?

Open innovation made it possible to build this project around local AI instead of depending on a closed AI API.

For Tab Bankruptcy, this matters because browser tabs can contain sensitive or personal information. Running the AI locally means the project does not need to send tab information to an external AI provider just to categorize or summarize it.

Using Ollama and an open-weight model also gives the project flexibility. The AI model can potentially be replaced with another compatible local model without redesigning the entire application.

More importantly, it let me experiment with AI as part of a real application rather than treating it as a separate cloud service.

Final Thought

Tab Bankruptcy started with a very ordinary problem: "I have too many tabs and I can't find anything."

The goal wasn't to build another complicated AI application.

It was to build something small, useful, and personal — and use open-source AI where it actually improves the experience.

Your tabs don't have to become a bankruptcy.