quiper 0.1.0

A macOS overlay for AI web applications, providing a floating window with a web view and hotkey support.

Quiper

A versatile macOS application that provides a system-wide overlay for multiple AI services, including Grok, Gemini ChatGPT and everything other engine that provides a web application. This tool allows you to quickly access your favorite AI assistants with a global hotkey, quickly switch between engines and manage multiple chat instances for each service.

Features

• Multi-Service Support: Seamlessly switch between Grok and Gemini.
• Multi-Instance Chats: Manage up to 10 simultaneous chat sessions for each service.
• Global Hotkeys:
  • Option+Space (customizable) to show/hide the application window.
  • Command+Control+<digit-n> to switch to nth engine.
  • Cmd+0 through Cmd+9 to switch between chat instances.
• Customizable: Easily change the global hotkey to your preferred combination.

Installation

You can install the application using pip:

python3 -m pip install quiper

Or you can clone the repository and run it directly:

git clone https://github.com/sassanh/quiper.git
cd quiper
uv run quiper

To build the application into a standalone macOS app, run:

uv run poe build-dmg

The dmg file will be created in the dist directory. You can then drag the app to your Applications folder.

To have the application launch automatically at login, run:

/Applications/quiper.app/Contents/MacOS/quiper --install

How It Works

This application is built with PyObjC, which allows Python to interact with Apple's native Objective-C frameworks. It creates a borderless, always-on-top window that contains a WKWebView for each chat instance. The global hotkeys are registered using the quickmachotkey library, and the application state is managed in the AppController class.

Contributing

This is an open-source project, and contributions are welcome. If you have ideas for new features or improvements, please open an issue or submit a pull request on the GitHub repository.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Credits

It is greatly inspired by the macos-grok-overlay project by tchlux, which was originally designed for the Grok AI service. The base of this codebase is mostly copied from that project, mostly the boilerplate code for setting up the macOS application.

Most of the code has been written with gemini-cli and grok. That's why it is so unorganized and messy :D If people find it useful, I will refactor it and make it cleaner, maybe even rewrite it in Swift.