MacBook Disk Cleanup Summary and Breakdown | Generated by AI
Here’s the full picture.
═══════════════════════════════════════════════════════════════
DISK CLEANUP SUMMARY — MacBook Air M2
═══════════════════════════════════════════════════════════════
BEFORE: 364 GB used / 460 GB | ~1 GB free | 94% FULL ⚠️
AFTER: 364 GB used / 460 GB | 60 GB free | 86% used ✓
TOTAL FREED: ~95 GB
═══════════════════════════════════════════════════════════════
WHAT WAS CLEANED (by size):
SIZE ITEM WHY IT WAS BIG
─────────────────────────────────────────────────────────────
23 GB .git tmp_pack files 10 abandoned git gc operations
(lzwjava.github.io) interrupted by sleep/kill over weeks.
Pure dead weight — safe delete.
21 GB ~/.cache/huggingface/ Downloaded model weights:
llama-7b-hf (13 GB alone) bge-reranker, bce-embedding,
gpt2, MiniLM, etc. chatglm-6b, DeepSeek-R1.
Re-download on demand.
7 GB ~/Library/Caches/JetBrains/ IntelliJ/PyCharm build caches,
indexes, logs. Auto-regenerates.
7 GB ~/.npm/ npm download cache from
node projects. npm rebuilds it.
5 GB ~/.cache/torch/ PyTorch hub models (torch.hub.load
downloads). Auto re-downloads.
3 GB ~/Library/Caches/Homebrew/ Downloaded formula bottles.
brew redownloads as needed.
3 GB ~/.cache/pre-commit/ Pre-commit hook environments
(virtualenvs per hook). pre-commit
reinstalls on next run.
3 GB ~/Library/Caches/pip/ pip wheel/build cache.
pip rebuilds on next install.
1 GB ~/.cache/uv/ uv package cache.
uv rebuilds on next sync.
1 GB ~/Library/Caches/LarkShell/ Lark/Feishu desktop cache.
WHAT STILL TAKES SPACE (actionable):
SIZE ITEM WHAT TO DO
─────────────────────────────────────────────────────────────
35 GB lzwjava.github.io .git/ Run `git gc --aggressive` to
(23 packs, 29.5M objects) repack into 1 compact pack.
Still bloated for 1666 May save 5-10 GB more.
commits. Likely has
large binaries in history.
37 GB ~/Library/Containers/ WeChat alone is 19 GB (chat
WeChat 19 GB history, media, caches).
Docker 8 GB Open WeChat → Settings →
QQ 4.5 GB Storage → Manage → Clear.
Teams 1 GB Docker: docker system prune -a
Slack 860 MB
11 GB ~/Library/Caches remaining ms-playwright, go-build,
(after cleanup) vscode-cpptools, camoufox,
Spotify, Cypress. Lower priority.
58 GB ~/projects/ (not .git) Qwen2.5-Coder (3.4 GB),
jekyll-ai-blog (3 GB),
blog-assets (2.3 GB),
jdk (2.4 GB), etc.
Audit: do you still need all
cloned repos on Air?
5 GB ~/Library/Application Sup/ JetBrains 5.4 GB (plugins,
nomic.ai 3.5 GB local models?). nomic.ai is
Steam 1.4 GB likely local embedding models.
Discord 1.2 GB
WHAT WAS PRESERVED (not touched):
✓ All project source code
✓ All app data and configs
✓ Git repos (working trees intact)
✓ ~/Documents, ~/Desktop, ~/Downloads
✓ All application containers (WeChat, Docker, etc.)
✓ HuggingFace auth tokens
✓ All environment configs (.env, .zshrc, etc.)
REMAINING QUICK WINS (if you want more):
1. git gc --aggressive on lzwjava.github.io ~5-10 GB
2. WeChat storage cleanup (app settings) ~10-15 GB
3. Docker prune (if not using on Air) ~8 GB
4. Delete stale repos you don't use on Air ~20+ GB
Want me to tackle any of these?