Skip to content

Enhancement: Integrated Browser View – AI can use browser directly within OpenGUI #56

@touch2be

Description

@touch2be

Feature Request: Integrated Browser Panel

Summary

Currently, when an AI agent needs to interact with a website or check results in a browser, it has to spin up an isolated headless browser (e.g. Playwright, Puppeteer, Chromium Headless) in the background — completely detached from the OpenGUI interface. The user cannot see what the AI is doing in the browser, and the AI cannot interact with a visible, shared browser session.

This creates friction: users often have to write system prompts explicitly forbidding the AI from using their real browser and instructing it to use headless-only tools.


Proposed Solution

Integrate a built-in browser panel directly into OpenGUI so that:

  1. The AI can navigate, click, fill forms, and read pages using the embedded browser — no external headless setup required.
  2. Results are visible to the user in real-time inside the OpenGUI interface.
  3. The AI and user share the same browser session — the user can observe and intervene if needed.
  4. No more workarounds like manually configuring Playwright or Chromium Headless just to give the AI basic web access.

Motivation

Today, a typical setup looks like this (from a user's system prompt):

"Du darfst nicht meinen sichtbaren Browser übernehmen, öffnen oder fernsteuern. Verwende ausschließlich einen isolierten Headless-Browser im Hintergrund, z. B. Playwright, Puppeteer oder Chromium Headless."
("You must not take over, open or remote-control my visible browser. Use only an isolated headless browser in the background, e.g. Playwright, Puppeteer or Chromium Headless.")

This is entirely unnecessary if OpenGUI provides a sandboxed, integrated browser that the AI is allowed to use freely — without touching the user's system browser.


Possible Implementation Approaches

  • Embed a Chromium/WebView panel inside the OpenGUI window (e.g. via Electron's webview tag, Tauri's WebView, or CEF).
  • Expose a CDP (Chrome DevTools Protocol) endpoint that the AI backend can connect to for automation.
  • Show a live preview of the browser state alongside the chat, so users always see what the AI is doing.
  • Optionally: allow the user to take over control of the embedded browser at any time.

Benefits

Today With integrated browser
AI uses headless browser, user sees nothing AI uses embedded browser, user sees everything
Complex system prompts needed to restrict browser access No restrictions needed — browser is sandboxed by design
External tools (Playwright etc.) must be installed Browser is built into OpenGUI
No shared state between AI and user AI and user share the same browser session

Related

  • Similar concept: Cursor with browser preview, Devin with integrated browser agent view
  • Could pair nicely with tool-use / function-calling features in OpenGUI

This would be a significant UX improvement and would make OpenGUI much more powerful as an agentic AI interface.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions