Three-Layer Tool Architecture

Profy provides three layers of interaction tools, covering the full spectrum from pixel-level desktop control to DOM-aware browser automation. All layers share a unified tool(action=...) calling convention.

Architecture

Layer Comparison

Layer	Tool	Scope	Awareness	Backend
computer	`computer(action=...)`	Full screen/window	Pixel screenshots + coordinates	Cloud: noVNC / Desktop: CUA
browser	`browser(action=...)`	Controlled browser	DOM + Accessibility Tree	Cloud: CDP / Desktop: Electron
chrome	`chrome(action=...)`	User’s Chrome	DOM + Accessibility Tree	Chrome Extension CDP

When to Use Each

Use `computer` for:

Native desktop applications (Finder, Terminal, IDE)
Game automation
Any non-browser GUI application
Screen-level visual verification

Use `browser` for:

Browsing public web pages
Testing in-development sites (localhost)
Web scraping
Form automation (no auth required)
3D/WebGL visual verification

Use `chrome` for:

Operations requiring user’s login state
Accessing authenticated internal systems
Performing actions on logged-in websites
Secure credential management

Unified Calling Convention

All three layers share the same call pattern:

# Screenshot
computer(action="screenshot")
browser(action="screenshot", url="http://localhost:5173")
chrome(action="screenshot")

# Click
computer(action="click", x=100, y=200)
browser(action="click", ref="button[0]")
chrome(action="click", ref="button[0]")

# Type
computer(action="type", text="Hello")
browser(action="type", ref="input[0]", text="Hello")
chrome(action="type", ref="input[0]", text="Hello")

Backend Polymorphism

The same tool name routes to different backends depending on execution environment:

Cloud Sandbox: computer → noVNC pixel stream, browser → CDP protocol
Desktop: computer → macOS CUA daemon, browser → Electron webview
Chrome Extension: chrome → Native Messaging → Extension CDP

Tool authors don’t need to worry about backend differences — just pick the right layer.

Confirmation Policy

Tool	Policy	Reason
`computer`	Depends on action	May affect system
`browser`	Not required	Fully sandboxed
`chrome`	4-level confirmation	Operating user’s real browser

Computer Use

Pixel-level desktop control

Browser

DOM-aware browser control

Chrome Extension

User Chrome browser control

Rendering Surfaces

Eight display channels

A2UI

Declarative interactive UI components

Video Production

AI video production pipeline & timeline editing

3D & Visualization

Data viz, 3D scenes & game dev

Knowledge Connectors Rendering Surfaces

Get Started

Core Concepts

Chat with Experts

Creation & Automation

Plugin Tools

Memory & Personalization

Marketplace & Content

IM Channels

Account & Billing

Team

How It Works

Help

Plugin Tools Overview

Three-Layer Tool Architecture

Architecture

Layer Comparison

When to Use Each

Use `computer` for:

Use `browser` for:

Use `chrome` for:

Unified Calling Convention

Backend Polymorphism

Confirmation Policy

Computer Use

Browser

Chrome Extension

Rendering Surfaces

A2UI

Video Production

3D & Visualization

​Three-Layer Tool Architecture

​Architecture

​Layer Comparison

​When to Use Each

​Use computer for:

​Use browser for:

​Use chrome for:

​Unified Calling Convention

​Backend Polymorphism

​Confirmation Policy

​Related Documentation

Computer Use

Browser

Chrome Extension

Rendering Surfaces

A2UI

Video Production

3D & Visualization

Three-Layer Tool Architecture

Architecture

Layer Comparison

When to Use Each

Use `computer` for:

Use `browser` for:

Use `chrome` for:

Unified Calling Convention

Backend Polymorphism

Confirmation Policy

Related Documentation