I work on a MacBook Pro with 512 GB of storage. For most people that's plenty. For me, it was full every couple of weeks.
I spend my days spinning up AI proofs of concept and MVPs and doing model research. Every experiment leaves something behind: a fresh `node_modules`, a Python virtual environment, a build folder, a few gigabytes of model weights pulled from Hugging Face or Ollama, Docker images I used once. Multiply that by dozens of projects and the disk disappears quietly, until macOS tells you there's no room left to save a file.
Traditional cleaners weren't cutting it. They're built to clear browser caches, images, docs and log files. They don't know what a `target/` folder is, that DerivedData can be rebuilt, or which of my 40 `node_modules` folders belong to projects I haven't touched in months.
So I cleaned by hand. I'd open Ghostty and run the same ritual over and over: `du -sh * | sort -h` to find the culprits, then `rm -rf` on node_modules, `brew cleanup`, `docker system prune`, digging through `~/.cache/huggingface`. It worked, but it was slow, I had to remember every tool's hiding spot, and one wrong `rm -rf` could take something I needed with it.
Why I built it
I work on a MacBook Pro with 512 GB of storage. For most people that's plenty. For me, it was full every couple of weeks.
I spend my days spinning up AI proofs of concept and MVPs and doing model research. Every experiment leaves something behind: a fresh `node_modules`, a Python virtual environment, a build folder, a few gigabytes of model weights pulled from Hugging Face or Ollama, Docker images I used once. Multiply that by dozens of projects and the disk disappears quietly, until macOS tells you there's no room left to save a file.
Traditional cleaners weren't cutting it. They're built to clear browser caches, images, docs and log files. They don't know what a `target/` folder is, that DerivedData can be rebuilt, or which of my 40 `node_modules` folders belong to projects I haven't touched in months.
So I cleaned by hand. I'd open Ghostty and run the same ritual over and over: `du -sh * | sort -h` to find the culprits, then `rm -rf` on node_modules, `brew cleanup`, `docker system prune`, digging through `~/.cache/huggingface`. It worked, but it was slow, I had to remember every tool's hiding spot, and one wrong `rm -rf` could take something I needed with it.
Yokoi is the tool I wanted instead. It knows where developer clutter lives, from node_modules and build artifacts to package caches, AI models, Docker images and simulators. It shows me what it found and why, then cleans it with a safety net: everything goes to the Trash first, every cleanup can be put back, and anything I touched recently or still have open is left alone.
Now cleaning my Mac takes one scan and one click, not an evening in the terminal.