Tabman: Brave Tab Manager

A command-line tool designed to streamline your browsing experience by intelligently organizing your open tabs from the Brave browser (Works for any chromium-based browser). tabman helps you categorize, tag, and save your browsing sessions, making it easier to find and revisit websites later.

Features

• Tab Retrieval: Fetches all open tabs from Brave using its remote debugging interface.
• Intelligent Categorization: Uses AI-powered language models to categorize your tabs:
  • Google Gemini: Employs the Gemini 2.0 Flash model for advanced and fast categorization.
  • Mistral AI: Leverages Mistral's large language models as an alternative to Gemini.
  • Ollama: Enables local categorization using Ollama, supporting a variety of LLMs (e.g., llama 3.2, Mistral).
• Dynamic Tagging: Generates a list of concise and relevant keywords to describe the content of each tab.
• Flexible Saving:
  • Saves the tab data (title, URL, main category, and tags) as a JSON file, organized in date-based subfolders within the data/ directory.
  • Creates a Markdown version of each tab session, also stored in the date-based subfolders within the data/ directory.
  • Maintains a central all_tabs.md file in the data/ directory, where all tab data is appended for easy access.
• Secure API Key Management: Utilizes a .env file to securely store API keys, which can be set using command-line arguments.

Installation

1. Clone the repository:

   git clone https://github.com/Anshulgada/brave-tab-manager.git

2. Navigate to the project directory:

   cd Brave Tab Manager

   This assumes that your project folder is named Brave Tab Manager.

3. Install pipx (if you don't have it):

   python -m pip install --user pipx
   python -m pipx ensurepath

   This ensures that pipx is installed globally and its executables are present in your system's PATH. You may need to restart your terminal for these changes to take effect.

4. Install tabman using pipx:

   pipx install -e .

   This command installs your tool into an isolated environment and makes the tabman command accessible from anywhere. Make sure you run this command from the root of the project directory, i.e. Brave Tab Manager.

Usage

1. Enable Remote Debugging in Brave: To allow tabman to connect with your Brave browser, you must enable remote debugging. Here's how:

   • Windows:
     1. Right-click on the Brave shortcut in your Start Menu or Taskbar and select "Properties."
     2. In the "Target" field, add --remote-debugging-port=9222 (or your preferred port number) to the end of the existing path, ensuring there's a space between the path and the flag. For example:

        "C:\Program Files\BraveSoftware\Brave-Browser\Application\brave.exe" --remote-debugging-port=9222
        

     3. Click "Apply" and "OK."
   • macOS/Linux: Modify the Brave shortcut (or application launcher) to include the --remote-debugging-port=9222 flag in the command line that launches Brave. Consult your system's documentation for the specific method.
   • After enabling remote debugging, make sure that you have restarted your Brave browser.

2. Run the tabman tool:

   tabman [OPTIONS]

   You can use the following options:

   • -h, --help: Displays the help or info menu.
   • -c, --categorize: Categorize your open tabs using the selected LLM model, and save the data. This is not necessary when setting the api keys.
   • -v, --version: Display the version of tabman and exits.
   • -m, --model <MODEL-NAME>: Specify the LLM model for categorization (gemini, mistral, or ollama).
   • --save-keys: Save API keys to the .env file. Use this in combination with -gk or -mk to set the keys.
   • -mk, --mistral-key <API-KEY>: Provide your Mistral AI API key.
   • -gk, --gemini-key <API-KEY>: Provide your Google Gemini API key.
   • -om, --ollama-model <MODEL-NAME>: Specify the Ollama model name (e.g., llama3.2, mistral). Default is llama3.2.
   • -o, --output-dir <PATH>: Specify the directory to store the output JSON and Markdown files, as well as the central all_tabs.md file (default: data). You can specify any directory with absolute path as well, but do note that it will remove the directory if the test fails, or any new files with same name are created in that directory.

Examples:

• Categorize tabs using the default Gemini model:

  tabman -c

• Categorize tabs using Mistral:

  tabman -c -m mistral

• Categorize tabs using Ollama with a specific model:

  tabman -c -m ollama -om mistral

• Save API keys to the .env file:

  tabman --save-keys -gk YOUR_GEMINI_API_KEY -mk YOUR_MISTRAL_API_KEY

  Replace YOUR_GEMINI_API_KEY and YOUR_MISTRAL_API_KEY with your actual API keys.

• Specify a custom output directory:

  tabman -c -o custom_output

  This will save the output into a directory named custom_output, which will be created in the root of your project.

• Specify an absolute path as an output directory:

   tabman -c -o "C:\\Users\\<user-name>\\Desktop\\My_Tabs"

• Show help message:

   tabman

Configuration

• API Keys: API keys for Gemini and Mistral can be provided either directly via command-line arguments or saved to a .env file for later use. The tool will read keys from environment variables, and then the command line arguments and the .env file when using the tool.
• Ollama: If you select Ollama model, then you will have to install and run ollama on your system with your desired models.

Development and Testing

• To avoid re-installing the package again and again, during local development, you should run the tool inside your virtual environment. For this you have to activate your environment and then run pip install -e . from the root directory. Now you can use the tool with the activated virtual environment. Whenever you want to update the global version then you have to uninstall the current version using pipx uninstall tabman and install again using pipx install ..

• Running tests:

  • The tests are located in the tabman/tests directory. You can run the tests using pytest rom the root of your project i.e. where setup.py exists.

  pytest -v -s tabman/tests

Dependencies

• openai
• ollama
• markdown
• requests
• tenacity
• playwright
• python-dotenv
• beautifulsoup4
• google-generativeai
• google-api-python-client

License

This project is licensed under the MIT License.

This project is tested with BrowserStack