TabMind - Your Local Browser Agent 🧠

Stop sending your browsing data to the cloud. Bring the AI to your browser.

TabMind is a privacy-first browser extension that gives your browser a brain of its own. It connects directly to your local Large Language Models (like LM Studio or Ollama) to analyze web pages, summarize articles, and answer questions—all without your data ever leaving your machine.

---

🚀 Why TabMind?

• 🔒 Private by Design: Your data stays on localhost. Zero cloud leakage.
• ⚡ Local Power: Run 7B, 13B, or even 70B models on your own hardware.
• 🧠 Context Aware: It reads what you read. Select text to focus on specific paragraphs, or let it digest the whole page.

✨ Features

• Local LLM Integration: Plug & play with LM Studio, Ollama, or any OpenAI-compatible local server.

• Smart Context Extraction:

  • Full Page Mode: Digests the entire article, blog post, or documentation page.
  
  • Precision Mode: Highlight specific text, and TabMind instantly focuses on just that snippet.
  

• Interactive Sidebar:

  • Chat: Have a conversation with the page or any highlighted snippet.
  • Dynamic Suggestions: Suggestions swap automatically between full-page and selection mode.
  • One-Click Copy: Hover any AI response to copy the full answer instantly.
  • Customizable: Dark mode, resizable width, optional page context pane, and toggleable system messages.

  

  

  

  

• Model Management: Auto-detects your local models. Swap between Mistral, Llama 3, and DeepSeek in two clicks. 

• Power User Settings:

  • Custom API endpoints.
  • Configurable timeouts (because sometimes local 70B models need a minute to think).
  • Fine-grained visibility controls for system messages and page context.
  

🛠️ Installation

1. Clone or Download this repo.
2. Build it:

   npm install
   npm run build

3. Load it (Firefox only — this add-on relies on Firefox-specific APIs):
   • Go to about:debugging > This Firefox.
   • Click Load Temporary Add-on… and select manifest.json.

🎮 Usage Guide

1. Fire Up Your Model 🔥

Make sure your local server is running:

• LM Studio: Start server on port 1234. (Turn on CORS!)
• Ollama: ollama serve (usually port 11434).

2. Connect TabMind 🔗

• Open the sidebar (click the icon).
• Hit the Gear Icon ⚙️.
• Enter your address (e.g., http://localhost:1234/v1).
• Click Detect Models 🔄.
• Pick your weapon of choice from the dropdown.

3. Chat Away 💬

• Analyze the Page: "What is the main argument of this article?"
• Analyze a Selection: Highlight a complex paragraph and ask, "Explain this in simple terms."
• Use Chips: Click "Summarize" for a quick TL;DR (or swap to selection-specific actions).

🔌 Supported Backends

If it speaks "OpenAI API", TabMind understands it.

• LM Studio (Highly Recommended for UI)
• Ollama (Great for CLI lovers)
• LocalAI
• vLLM

👨‍💻 Development

Want to hack on it?

• src/background.js: The brain. Handles API calls, timeouts, and LangGraph logic.
• src/content.js: The eyes. Reads the DOM, extracts text, and watches for selections.
• src/sidebar.js: The face. Manages the chat UI, settings, and those snappy suggestion chips.

📄 License

MIT. Go wild. build cool stuff.