A Chrome extension to control your browser with AI
Xeon Agent is a Chrome extension that allows you to control your browser using artificial intelligence. It's a secure and open-source alternative to OpenAI Operator, with a focus on privacy and local control.
Unlike other solutions, Xeon Agent:
- Works locally: no external server needed
- Respects your privacy: your data never leaves your browser (except for the AI API)
- Supports multiple AI models: compatible with Google Gemini and OpenAI
- Is fully customizable: adjust settings to your needs
See Xeon Agent in action:
Watch how Xeon Agent can help you automate browser tasks with simple natural language commands.
Hi! I'm a developer passionate about automation. This project was born from my curiosity for AI and web automation.
To be honest, I was mostly familiar with backend development before starting this project, and Chrome extensions were new to me. That's why the code isn't always elegant (let's face it, it's sometimes garbage 😅).
Despite this, I managed to create something that works! I continue to learn and improve Xeon Agent every day. If you have suggestions or want to contribute, don't hesitate!
- 🌐 Browser automation: performs complex actions on websites
- 🧠 Advanced AI: uses cutting-edge models like Gemini and GPT
- 🔍 Contextual analysis: understands page structure for precise interactions
- 🔄 Complex actions: navigation, form filling, clicks, and more
- 🔐 Security: your sensitive data stays on your device
- get the extension from the release and unzip it
- Open Chrome and go to
chrome://extensions/ - Enable "Developer mode"
- Click "Load unpacked extension"
- Select the project folder
- Configure your API key in settings
- Click on the extension icon in the toolbar
- Open settings
- Enter your API key (Gemini or OpenAI)
- Select your preferred AI model
- Adjust the temperature to control AI creativity
- Improve tab switching: Solve switch_tab issues by disconnecting/reconnecting the debugger
- Support more providers: Add support for Anthropic Claude, Mistral, and other models
- Enhanced bounding boxes: Make element visualization more precise and less intrusive
- Accessibility tree optimization: Improve conversion and representation of DOM structure
- Intelligent tab management: Automatically reconnect when a new tab is opened by the extension
- Improve complex tasks: Draw inspiration from Manus for better handling of long tasks
- Conversation history: Implement a system for storing and searching past interactions
- More intuitive user interface: Redesign the UI for a better user experience
Xeon Agent uses the Chrome Debugging Protocol to interact with web pages. It captures a snapshot of the page, sends it to an AI model with context, then interprets the instructions to perform actions on the page.
The agent can:
- Click on elements
- Fill out forms
- Use the keyboard
- And much more!
Contributions are welcome! Here's how you can help:
- Report bugs
- Suggest features
- Submit pull requests
- Improve documentation
This project is licensed under a custom proprietary license that requires attribution and prohibits creating competing projects without permission.
Key requirements:
- You must credit the original author (Louis, age 14)
- You cannot create competing browser automation tools
- Commercial use requires permission
- Personal and educational use is allowed
For the complete license terms, please see the LICENSE file.