explaingit

xiaofenggan01/disk-cleaner-skills

21PythonAudience · generalComplexity · 1/5Setup · easy

TLDR

A Claude Code skill that scans your Mac or Windows computer for safe-to-delete files, generates an interactive HTML report with color-coded cleanup suggestions, and lets you approve deletions in a browser.

Mindmap

mindmap
  root((disk-cleaner))
    What it does
      Scan for junk files
      Interactive HTML report
      Browser-based approval
    Safety
      200 plus protected apps
      Whitelist tiers
      No path traversal
    Coverage
      Messaging apps
      Browsers and dev tools
      Creative software
    Tech Stack
      Pure Python
      No pip install needed
      Mac and Windows
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Tell Claude Code your disk is full and receive an HTML report listing which files are safe to delete, color-coded by risk level.

USE CASE 2

Review cleanup cards for apps like Slack, browsers, and developer tools in a local web interface, then copy the deletion command for each item.

USE CASE 3

Safely remove caches and logs from more than twenty apps on macOS or Windows without accidentally deleting protected system files.

Tech stack

Python

Getting it running

Difficulty · easy Time to first run · 5min

No pip install required, uses only the Python standard library. Requires Claude Code to be installed.

In plain English

disk-cleaner is a skill for Claude Code, Anthropic's coding assistant, that scans your Mac or Windows computer for files that can safely be deleted and helps you reclaim disk space. You trigger it by telling Claude something like "my disk is full" or "I need to free up space," and it runs a scan script that reads your file system without deleting anything. Claude then analyzes the scan results, classifies each found item by how safe it is to remove, and generates an interactive HTML report you can open in a browser. The report shows a disk usage bar, color-coded cards for each finding, a ranking of the five largest space consumers, and suggested cleanup actions. You can expand each card to see details and copy the cleanup command for that item. A local web server can be started to let you approve or reject deletions directly in the browser, with each operation logged to a file on disk. Safety is a central focus. The skill combines path-checking rules drawn from an open-source Mac cleanup project called Mole, a list of more than 200 protected applications that must not be touched, and a tiered whitelist system that separates items safe for permanent deletion from items that should only go to the Trash. The scanner rejects relative paths, path-traversal attempts, symbolic link tricks, and any path inside system directories. Trash failures do not fall back to permanent deletion. The scan covers more than twenty applications across categories including messaging apps, browsers, developer tools, creative software, and system folders. The project has no external dependencies and runs on pure Python from the standard library, so nothing needs to be installed with pip. It automatically detects whether it is running on macOS or Windows.

Copy-paste prompts

Prompt 1
I have the disk-cleaner-skills Claude Code skill installed and typed 'my disk is full' to Claude. Walk me through what happens next and how to read the HTML report it generates.
Prompt 2
The disk-cleaner skill flagged my Xcode derived data folder. How do I tell from the HTML report whether it is safe to delete permanently or should go to Trash only?
Prompt 3
I want to add a new app to the disk-cleaner-skills scanner on macOS. Where in the Python code do I add its cache directory paths and which whitelist tier should I assign?
Open on GitHub → Explain another repo

← xiaofenggan01 on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.