Description
TaxyAI's browser extension leverages GPT-4 to automate browser interactions, enabling users to perform repetitive tasks efficiently. The extension is designed to execute actions based on user-defined instructions, streamlining workflows within the browser environment.
The core functionality involves simplifying the webpage's Document Object Model (DOM) to include only interactive elements. This simplified DOM, along with user instructions, is then sent to a selected Large Language Model (LLM), currently supporting GPT-3.5 and GPT-4. The LLM processes the information and generates actions, which the extension executes using the chrome.debugger API. The action cycle includes steps like clicking elements and entering text, with a maximum of 50 actions per task.
The extension is currently available through GitHub and will be released on the Chrome Web Store after adding features to increase its usability for a non-technical audience. Users can install the extension by cloning the repository, installing dependencies using yarn, and building the package. Once installed, the extension can be activated as a popup or a devtools panel. Users need to provide an OpenAI API key, which is stored securely within the browser.
Key features include the ability to automate tasks such as scheduling meetings, searching for content, and writing essays. The extension's architecture prioritizes safety, halting execution in case of unexpected responses. The project is open-source, ensuring transparency and community contributions. The tech stack includes Chrome Extension Manifest V3, React 17, Webpack 5, and TypeScript, among others. The project is actively maintained, with updates and improvements regularly added.
GitHub's Core Features
Automates browser actions using GPT-4
Supports ad-hoc instructions
Simplifies webpage DOM for LLM processing
Uses chrome.debugger API for action execution
Supports GPT-3.5 and GPT-4
Allows users to define custom workflows
Open-source and community-driven
Includes example demos for common tasks
Provides a devtools panel for debugging
Securely stores OpenAI API keys
Offers a waitlist for wider usage
Uses React 17 and TypeScript
Supports a maximum of 50 actions per task
Getting Started with GitHub
Clone: Clone the GitHub repository.
Install Dependencies: Run `yarn` to install the necessary dependencies.
Build: Run `yarn start` to build the package.
Load Extension: Navigate to `chrome://extensions/` and enable Developer mode.
Load Unpacked: Click 'Load unpacked' and select the build folder generated by `yarn start`.
Activate: Activate the extension via popup (cmd+shift+y on mac or ctrl+shift+y on windows/linux) or devtools panel.
Enter API Key: Provide your OpenAI API key in the provided box.
Experiment: Navigate to a webpage and start experimenting with TaxyAI.
GitHub's Use Cases
- Automated Web Tasks
- Content Creation
- Workflow Automation
- Web Application Testing
- Data Entry
- Information Gathering
- Task Scheduling
- Browser Automation







