What Digital Clutter Roaster is

Digital Clutter Roaster is a free, local-first storage analyser for macOS. It reads folder metadata directly in the browser using the File System Access API, identifies which files are quietly consuming disk space, and produces a cleanup plan you can export as Markdown or JSON. Because analysis runs entirely on-device, no file name or file content is ever uploaded to a server.

My role: Sole designer and developer — architecture, UI and heuristics.

The problem it solves

Mac storage fills up invisibly. Downloads, Desktop and developer caches accumulate installers, screenshots and build artefacts that nobody deliberately keeps. Existing cleanup tools solve this by asking for full-disk access and uploading an index of your filesystem — a level of trust that is wildly disproportionate to the problem of finding a few forgotten DMG files.

How it works

The whole tool is built around a constraint: the server must never see your data. The browser-native File System Access API lets a user grant access to one specific folder, and the app reads only metadata — names, sizes, timestamps — from that folder. Classification runs against macOS-specific heuristics (screenshot naming patterns, installer extensions, common developer cache directories) rather than any generic size ranking, so the output flags things you can actually delete instead of just listing your largest files. State is held in Zustand and the build is a static Vite bundle, which means there is no backend to trust in the first place.

The outcome

A shipped, publicly available tool that returns a scan and a prioritised cleanup plan in seconds, with exportable reports. It is the clearest example of a design principle that runs through my backend work too: choose the architecture that removes the need for trust rather than the one that asks for it.

Technology used

  • React
  • TypeScript
  • Tailwind CSS
  • Zustand
  • Vite

Engineering highlights

  • Folder-by-folder scanning using the browser-native File System Access API
  • Intelligent detection of macOS-specific clutter patterns (screenshots, DMGs, etc.)
  • AI-powered "Roast Report" that archetypes storage habits with a playful tone
  • Privacy-first architecture: all metadata analysis happens locally on your device
  • Actionable cleanup plans with exportable reports in Markdown and JSON formats