TLDR:
I'm a software engineer.The only app I have open is Google Chrome.
My code editor, my terminal, and remote work tools are all inside Chrome, managed using my browser extension.
It is easy to forget I am using macOS under the hood as it is just an undifferentiated (sometimes intrusive) layer beneath the web platform.
How it started: limitations with macOS window management
macOS likes to look pretty, even if it means having animations you can't disable 🙂.
The transition when switching between full-screen macOS apps was the slowest daily part of my workflow. Even with the "Reduce Motion" accessibility setting on, keyboard focus takes half a second to switch between macOS Spaces.
I even wrote a custom macOS window manager in hopes of avoiding this issue. However, macOS is not optimized for custom window management, so the experience proved slow and unstable.
This is just one example of a pattern. A system you can't customize to your needs is a system you don't own.
I then realized that Chrome's tabs suit my workflow much better - no unskippable animations, no overlapping windows, and plenty of keyboard shortcuts. This led me to try to fit every app I use into Google Chrome.
The first steps: mail and meetings in the browser
Gmail, Google Calendar, Google Meet, and Google Docs were born on the web and are an awesome experience to use.
Microsoft Teams and Outlook are much less so. The web apps have papercuts, bugs, and uncustomizable behavior. Yet, I need to use them daily.
Here comes the cool part - you can customize the experience of these web apps by writing your own browser extension. The browser extension can adjust the styling, fix small bugs, or even add new features. Nowadays, coding agents like Claude can vibe-code your wishes with just a few prompts - the barrier is lower than ever.
For example, for Microsoft Teams, I added notification batching and protection against accidental hand-raising during meetings. For Outlook, I reduced favicon updates to eliminate distractions and customized the calendar to resemble Google Calendar a bit more.
Small improvements for tools you use daily make a big impact.
The big step: VS Code in the browser
The reason I love VS Code is that it is just a web app. It runs on HTML/CSS/JavaScript, so you can customize it extensively. This same tech stack means it can be opened inside Google Chrome, without needing a separate window.
VS Code is built on a server/client architecture. By default, starting VS Code opens an Electron window, but it doesn't have to - you can open VS Code in the browser too. Start the server using the code serve-web command, then open the browser at the URL it prints. See the code serve-web tutorial.
It is easy to forget VS Code is running inside Chrome as we get a completely functional VS Code with all the same capabilities. VS Code also includes a capable terminal emulator. All terminal commands, file writes, and Copilot requests are processed by the same server used by the standalone VS Code window.
The only constraint is keyboard shortcut conflicts with the browser UI and browser extensions. Fortunately, both VS Code and browser extensions permit rebinding the shortcuts, so conflicts can be resolved. Even better, when in macOS full-screen mode (Cmd+Shift+F), Google Chrome passes nearly all keyboard shortcuts to the website - even Cmd+W and Ctrl+Tab are handled by VS Code.
An additional perk: I don't have separate Microsoft Teams and VS Code Chromium instances eating my RAM - they all share a single Google Chrome process.
The window management
The Google Chrome Extensions API includes a very capable tabs API. An extension can jump you between tabs, move tabs around, or even swap tabs between windows.
With Microsoft Teams, Outlook, VS Code, GitHub, and other web apps being in the browser, the browser is the only window I have open - one window per monitor.
Inside that window, my Web Commander extension provides keyboard shortcuts for fast navigation. Key shortcuts I defined:
- Jump to the nth VS Code tab
- Move the current tab left/right
- Move the current tab to a neighboring screen
- Cycle between the last N used tabs
- Rebalance the VS Code tabs across monitors - useful when working with 3 monitors in the office and then continuing work with only 2 monitors at home.
In addition, I defined a Cmd+D commander UI for opening commonly used pages with just a few keystrokes. For example, typing gecia navigates to "Github.com / Esri / Calcite-design-system / Issues / Authored by me".
Looking back: Project Fugu
Moving everything into browser tabs only works when those tabs can replace desktop apps. Efforts like Project Fugu have helped close that gap, giving web apps deeper integration:
- VS Code uses the File System Access API
- Photopea uses the File Handling API to open local files
- Google Meet uses Picture-in-Picture, tab presenting, and WebCodecs
- Microsoft Teams applies client-side webcam blur effects
- Gmail handles
mailto:links - Figma gets a listing of locally installed fonts
- Twitter and Pinterest support the Web Share API
The beauty of the web
What I love about the web is that it is the most hackable platform - I don't have to accept apps exactly as they are shipped. When something gets in my way, DevTools lets me investigate, and browser extensions let me change it.
I was over the moon when someone made GTA 5 playable on the web using WebGPU and WebAssembly - even AAA games are now within reach. Let's push the limits of the web further.
I started by trying to escape a window-switching animation. Now my editor, terminal, meetings, and mail are all browser tabs.
The web is my operating system.
P.S.: the cat is out of the bag on macOS, but Apple doesn't want the same to happen on iOS. That is why iOS does not permit a competitive browser engine.
Next steps
While the Chrome Extensions API surface is extensive, there are a few limitations:
- The extension can't move the mouse to another screen when moving focus to that screen.
- The browser extension can't handle keyboard shortcuts when focus is in the Chrome URL bar, DevTools, or on
chrome://pages. - There is no keyboard shortcut to dismiss a notification.
To handle these cases, I'm looking to write a tiny Swift companion that would communicate over a WebSocket with my browser extension.
Top comments (1)
code serve-web is a neat way to get VS Code into a tab. Small extensions to fix papercuts in web apps you use daily is a low-effort idea more people should try. The Swift companion over WebSocket for the URL bar and DevTools gaps sounds like a fun next step.