mac assistant: AI intelligence for macOS devices to automate desktop actions

General Introduction

mac assistant is a project of AI intelligences designed for macOS, aiming to simplify user operations by combining native software and web features. The project currently supports OpenAI and GEMINI API, and plans to support Ollama Runs on a native large language model. mac_assistant is able to efficiently perform simple tasks such as clicking buttons on web pages and in applications, thanks to its deep integration with the DOM tree and operating system helper functions. However, it currently has limitations in handling operations such as Google Docs and PDF files. Future updates will further enhance its performance in these areas. The project was developed by browser-use and macOS-use drivers that will allow AI to utilize more tools to improve its performance in the future.

Function List

• Support for OpenAI and GEMINI APIs
• Plan to support the Ollama Native Large Language Model
• Efficient execution of button clicking tasks in web pages and applications
• Deeply integrated DOM tree and OS helper functions
• Provides an easy-to-use interface to interact with intelligences and APIs

Using Help

Installation process

1. Installation of required libraries: First, install the necessary libraries by running the following command:

   pip install PyQt5 pyinstaller langchain-openai browser-use mlx-use python-dotenv

2. Creating .env files: Create a.envfile and add the following:

   OPENAI_API_KEY=你的OpenAI密钥
GEMINI_API_KEY=你的GEMINI密钥

3. Running the application: Run the following command in the terminal to start the application:

   python agent_ui.py

Guidelines for use

1. Introduction to the interface: Upon launching the application, you will be presented with a user interface that allows you to interact with intelligences and APIs. The interface is designed to be simple and intuitive for user-friendly operation.
2. operate: Through the interface, you can enter commands for the Intelligence to perform specific tasks, such as clicking buttons on a web page or performing actions in an application. The Intelligence will automatically perform these tasks according to the instructions.
3. Future Updates: As the project continues to evolve, support for more file types (e.g., Google Docs and PDF) will be added in the future, and the overall performance of the intelligences will be improved.