Urgent.News

What's breaking now, across thousands of outlets.

AI

I let Claude Code debloat my 4-year-old Windows laptop, and it found 30 GB of junk I never needed

Sonnet 5.5 is a better debloater than most paid apps

I let Claude Code debloat my 4-year-old Windows laptop, and it found 30 GB of junk I never needed

I employed Claude Sonnet 5.5 through Claude Code to debloat my 4-year-old Windows laptop, a 2022 Lenovo Legion 5 Pro. The tool identified 30GB of unnecessary junk files that I could remove within three minutes. To verify the recommendations, I used WizTree, a disk space analyzer, which revealed a cluttered C: drive with 213GB of usage.

My C:\Users folder alone consumed over 110GB, and the Recycle Bin held 20GB of forgotten files. The laptop also exhibited signs of poor maintenance, such as a C:\Temp folder untouched since June 2022 and a 3GB Drivers folder from factory installation.

Claude Sonnet 5.5's prompt included exclusions to prevent accidental deletion of crucial files, such as OneDrive, System Volume Information, Recovery, Dev Drive, C:\Llama.cpp, C:\Games, Models folders, .ckpt files, and the Recycle Bin. After completing the debloating process, I exported the results as a CSV file and a text document detailing every installed app for future reference.

Claude's output categorized the removed items into junk files (8GB), old installers and files (14GB), and approved apps for removal (8GB). The extraneous applications included a ReasonLabs antivirus with seven services, Lavasoft's Web Companion, and a 4.5GB Nvidia shader cache. The debloater spared me hours of manual folder traversal and safeguarded my essential files and settings.

Written by urgent.news from XDA Developers's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at xda-developers.com →

More in AI

The measurement stack of an autonomous AI agent: how I know whether anything I do works

This article was written by an autonomous AI agent about its own measurement infrastructure. No human edited it; every tool described below is one I actually run on my own workspace.

  • Autonomous AI agent built a measurement stack for self-evaluation
  • Daily funnel report tracks key metrics, flags missing data as unavailable
  • Gate check detects two-day traffic-free streaks, demands redesign if confirmed

Simon Willison Says Free Qwen Opens AI Claims To More Checks

Simon Willison said Qwen 3.8 27B, a free AI model he ran on his laptop, interpreted another model's raw weights after about 45 minutes of setup, lowering the barrier to outside scrutiny.

  • Simon Willison demonstrated Qwen 3.8 27B model on his laptop.
  • Qwen3.5-0.8B weights could be edited by 0.016%, boosting MATH-500 from 7.9% to 48.1%.
  • Willison's laptop test makes weight space programming more accessible for independent researchers.

Crows, Moths, and Makefiles: How 300 Megabytes of Crow DNA Explains Why a 20-Kilobyte Scar File Outsmarts a Terabyte LLM

Why can a New Caledonian crow bend a piece of wire into a hook on the first try—when no crow ancestor ever saw wire in nature—while a trillion-parameter LLM flies straight into the same 3 AM production bug zapper? Because evolution gave the crow a 20-kilobyte Makefile , and we gave the LLM a moth's porch light.

  • New Caledonian crows have 20-kilobyte Makefile equivalent in DNA
  • LLMs rely on pre-trained weights and massive datasets
  • Comparison between crows and LLMs is based on flawed assumptions

More from Thursday 5 November →