TabOrg

TabOrg is a Chrome extension for granular, incremental tab management. It is designed for asynchronous workflows, allowing users to create tabs and windows freely while providing AI-assisted suggestions for grouping and organization.

This workflow distinguishes TabOrg from fully automated solutions that make decisions for you, and from manual-only approaches that require tedious organization.

Features

Semi-Automated Tab Organization

TabOrg provides AI assistance for tab management while retaining user control:

1. Granular Selection: Select specific tabs using checkboxes in the tree view, or quickly select all ungrouped tabs.
2. AI Suggestions: Click "Organize Tabs" to have Google Gemini or Chrome Built-in AI analyze selected tabs and suggest groupings based on content.
3. Review & Apply: Preview suggested groups before applying them. Accept all suggestions or select specific groupings.
4. Manual Refinement: Reassign tabs to different groups or create groups manually using the dropdown menu.

Core Features

• Hierarchical Vertical Tab View: View all your open windows and tabs in a tree within the sidepanel.
• Tab Management: Close, switch to, and reorder tabs with clicks or drag-and-drop.
• Group Management: Create, rename, and colorize tab groups. Move tabs between groups or windows.
• AI Window Naming: Automatically generate names for windows based on their contained tabs and groups using Google Gemini or Chrome Built-in AI.
• Theme Support: Adapts to system light or dark mode, including dynamic extension icons.

MCP Server

TabOrg exposes a local Model Context Protocol (MCP) server that allows external LLMs (like Claude Desktop or MCP Inspector) to interact with your browser tabs.

Capabilities

• List Tabs: Query your open tabs and groups.
• Manipulate: Group, move, and close tabs programmatically.

Setup

1. Start the MCP Bridge Server: The extension requires a local bridge server to communicate with the MCP client.

   yarn workspace @taborg/server start

2. Enable in Extension:

   • Open TabOrg Settings.
   • Toggle Enable MCP Server to ON.
   • Verify the status shows "Connected".

3. Connect your MCP Client:

   Note: Claude Desktop requires an SSE-to-Stdio bridge to connect to local HTTP servers.

   {
     "mcpServers": {
       "taborg": {
         "command": "npx",
         "args": ["-y", "mcp-proxy", "http://localhost:3000/mcp"]
       }
     }
   }

   • MCP Inspector:

     npx @modelcontextprotocol/inspector http://localhost:3000/default/mcp

   • Gemini CLI:

     Option 1: Using the CLI (Recommended)

     # Connect to the default instance
     gemini mcp add taborg http://localhost:3000/default/mcp
     
     # OR connect to a specific profile (e.g., your email)
     gemini mcp add taborg-work http://localhost:3000/your.email@example.com/mcp

     Option 2: Manual Configuration Add to your ~/.gemini/settings.json:

     {
       "mcpServers": {
         "taborg": {
           "httpUrl": "http://localhost:3000/default/mcp"
         }
       }
     }

     Note: Replace default with your Instance ID (found in Extension Settings) if you are using specific profiles or multiple instances. By default, it uses your Chrome profile email.

Tech Stack

• Framework: WXT (Web Extension Framework)
• UI Library: Shoelace & Lit
• Language: TypeScript
• Package Manager: Yarn

Development

Prerequisites

• Node.js (v18+)
• Yarn (v4+)

Setup

1. Clone the repository.
2. Enable Corepack to use the correct Yarn version:

   corepack enable

3. Install dependencies:

   yarn install

Running Locally

To start the development server with hot module replacement:

yarn dev

This will open a new Chrome instance with the extension loaded.

Building

To build the extension for production:

yarn build

The output will be in the .output/ directory.

License

MIT