Skip to content

ixruby/ixr-agent

IXR Agent

IXR Agent

License: MIT TypeScript

The GUI Agent for IXR Systems. Control any web interface with natural language — no Selenium, no headless browser, no screenshots.

Built on top of page-agent (MIT License) by Alibaba, rebranded and extended for the IXR ecosystem.


✨ Features

  • 🎯 Zero-dependency integration — Just inject a script. Works in any webpage.
  • 📖 Text-based DOM manipulation — No OCR. No multi-modal LLMs needed.
  • 🧠 Bring your own LLMs — Connect OpenAI, Gemini, Qwen, local Ollama, or any OpenAI-compatible API.
  • 🎨 IXR-branded UI — Human-in-the-loop panel styled for IXR products.
  • 🔗 IXR system integration — Connects natively to the IXR automation stack.

💡 IXR Use Cases

  • CRM Copilot — Automate data entry in any web CRM with a sentence
  • Lead Scraping — Control PropertyFinder/Bayut in-browser to extract villa owner contacts
  • Client Onboarding — Auto-fill forms, generate contracts, trigger workflows
  • Dashboard Automation — Control internal dashboards without API access

🚀 Quick Start

One-line integration (demo mode)

<script src="https://cdn.jsdelivr.net/npm/ixr-agent/dist/iife/ixr-agent.demo.js" crossorigin="true"></script>

NPM

npm install ixr-agent
import { IXRAgent } from 'ixr-agent'

const agent = new IXRAgent({
    model: 'gpt-4o-mini',
    baseURL: 'https://api.openai.com/v1',
    apiKey: 'YOUR_API_KEY',
    language: 'en-US',
})

await agent.execute('Click the Submit button')
await agent.execute('Fill in the contact form with: Name=John, Email=john@ixr.com')
await agent.execute('Extract all villa prices from the current page')

🏗️ Architecture

IXR Agent (browser layer)
    ↕ natural language commands
IXR Controller (DOM interaction)
    ↕ structured actions
IXR LLMs (model abstraction)
    ↕ OpenAI-compatible API
Your LLM (Qwen / GPT / Gemini / Local)

🔧 Configuration

const agent = new IXRAgent({
    model: 'qwen-plus',
    baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1',
    apiKey: process.env.DASHSCOPE_API_KEY,
    language: 'en-US',
    // Optional: IXR gateway integration
    ixrGatewayUrl: 'ws://bore.pub:PORT',
    ixrGatewayToken: 'minimax-agent',
})

📦 Packages

Package Description
ixr-agent Main entry point — IXRAgent class
@ixr/core Core DOM agent logic
@ixr/ixr-controller Page controller & action executor
@ixr/llms LLM abstraction layer
@ixr/ui IXR-branded panel UI
@ixr/ext Chrome extension for multi-tab control

📄 License

MIT — See LICENSE

Based on page-agent by Alibaba Group (MIT License).

About

IXR Agent - GUI agent for controlling web applications with natural language.

Topics

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors