How to Use Claude to Find What's Actually Eating Space on Your Computer

The Rundown In this guide, you will learn how to use Claude as a storage-cleanup copilot so you can figure out what is actually eating space on your Mac before you start deleting random folders. This works especially well if you use AI coding tools and have a machine full of old projects, oversized caches, and developer clutter that the built-in storage menu does not explain very well. Who This Is Useful For Operators, founders, and freelancers  who use AI tools on their main computer and keep running into storage warnings Developers and vibe coders  with old repos, large  node_modules  folders, Homebrew clutter, or app caches they do not fully understand Mac users  who want to free up space without blindly deleting files and breaking a working setup What You Will Build You will use Claude to investigate the biggest storage offenders on your machine, rank which ones are low-risk to clean up, and work through a few common categories like  node_modules , Homebrew cache, and Claude desktop cache. Then you will turn the process into a reusable monthly checklist. What You Need to Get Started A Mac with low storage or a cluttered dev environment Access to  Claude ,  Claude Code , or Claude Cowork Terminal access on your Mac Step 1: Investigate Before You Delete Anything Start a new Claude session and make the safety rule explicit: you want help investigating first, not deleting first. The point is to have Claude explain what is large, what it likely contains, and what is usually safe to remove before any destructive command gets suggested. Use a prompt like this: Help me investigate what's taking up storage on this Mac. Start by identifying the biggest directories in my home folder. For each large folder, explain what it likely contains, whether it's usually safe to delete, and what questions I should answer before removing anything. Do not suggest destructive commands until we've reviewed the findings. Once Claude gives you commands, run them in Terminal and paste the results back in. This gives Claude real local information to work from instead of vague guesses. Pro tip:  The built-in Mac storage screen is useful for broad categories, but it often misses the real story. The problem is usually not one huge file. It is a pile of medium-size directories, old dependencies, and app caches that add up fast. Step 2: Find Old  node_modules  and Other Reproducible Bloat If you do any kind of web development,  node_modules  is usually one of the first places to check. Old experiments, abandoned repos, and half-finished client work can each carry hundreds of megabytes or several gigabytes of dependencies. Ask Claude to scan your development folders and sort these by size and age: Scan my dev folders for large node_modules directories. Show me the largest ones, where they live, and when they were last modified. Separate them into: 1. likely safe to remove because the project looks old or inactive 2. review manually because the project may still be active Explain your reasoning for each one. This is one of the easiest cleanup wins because you are usually deleting installed dependencies, not your source code. If a project is old and you need it later, you can reinstall everything with  npm install . Pro tip:  If you do not know which projects are still active, ask Claude to interview you. It can walk through the list one folder at a time and help you decide what is stale. Step 3: Clean Up Homebrew and Other Tooling Cache Carefully After  node_modules , move on to package-manager and tooling clutter. Homebrew is a common offender because it accumulates old downloads, outdated package versions, and dependencies you no longer need. Docker can do the same thing with images, containers, and build cache. Try this prompt for Homebrew: Explain how to safely clean up Homebrew on this Mac. Tell me what old versions, caches, or unused dependencies are taking space, what the normal cleanup commands would do, and which ones are low-risk versus which ones need manual review first. If you use Docker, run the same playbook there: Help me find Docker files, images, and caches that are taking significant space on this Mac. Show me what is low-risk to remove, what needs review first, and how much space each cleanup might save. The habit that matters is asking Claude to preview the impact first. Do not run cleanup commands just because they look standard. Step 4: Clear Claude Cache Without Deleting the Wrong Thing App cache is another easy place to reclaim space, and the Claude cache can get surprisingly large. During setup for this guide, my cache had grown to around 15 GB, much of it tied to Claude Cowork usage. Open the Claude desktop app and click  Help  >  Troubleshooting  >  Clear Cache and Restart . You can also click  Delete Cowork VM Bundle and Restart . That just deletes virtual coding sandboxes that Claude Cowork has used. Do not click  Delete Cowork VM Sessions and Restart  as that will delete all your Claude Cowork threads. Pro tip:  This same logic applies to other local AI apps too. If you use ChatGPT desktop, Codex, or similar tools heavily, their local caches may be much larger than you think. Step 5: Turn This Into a Reusable Monthly Cleanup Checklist The bigger win here is not just getting space back once. It is creating a repeatable cleanup workflow so you do not end up rage-deleting files the next time your disk fills up. When you finish the audit, ask Claude to turn your process into a reusable Claude skill: You can also ask Claude to create a checklist in a markdown skill file so you can manually track the same process every month. Going Further Here's some other non-obvious storage eaters that you probably have if you code with AI. Python environments  like  .venv  folders and old conda environments Xcode simulators  and derived data if you do any Apple-device development Downloads  filled with installers, ZIP archives, models, and exports Screen recordings and generated media  from Loom, CleanShot, or AI video tools Package-manager caches  from  npm ,  pnpm ,  yarn , or  pip

• Duration

Save 15% on Thoughtly by joining The Runway University

Through our vast network of over 100 partnerships with major AI companies, we also offer University members free access or major discounts to the many popular AI tools along with AI certified courses, daily guides on AI tools and many more.

Join AI University

Learn more