Jharu -A disk cleaner that can think.
Free disk cleaner for your AI model and dev caches
Start new thread

Jharu -A disk cleaner that can think. - Free disk cleaner for your AI model and dev caches

by•
Jharu is a free, open-source disk cleaner for macOS and Windows that understands AI and developer clutter. Reclaim gigabytes from Hugging Face, Ollama, and PyTorch model caches, npm and pip caches, and app leftovers. A free CleanMyMac alternative — no telemetry, nothing permanently deleted.

Add a comment

Replies

Best
Maker
📌
Jharu is the free, open-source disk cleaner that actually understands developer and AI clutter — on macOS and Windows. 🧹 Generic cleaners show you folder sizes. Jharu knows what the data is, and whether you've ever used it. What it does: Per-model ML cleanup — lists every Hugging Face, Ollama, PyTorch and Whisper model individually and flags which ones you actually loaded vs. only downloaded. Reclaim the 6 GB model you forgot about without touching the ones you use. Nothing else does this. Your whole disk as a treemap — the biggest space-eater is the biggest block. Click to drill in. Deep dev-cache scan — npm, pip, conda, Cargo, Docker, Xcode and 20+ more, each rated safe-to-clear or review-first. Python environment analyzer — finds every virtualenv and shows what's duplicated across them (the "torch installed 11 times" problem). ❌ Clean uninstaller — removes an app and the leftovers it scatters across your system. Why you can trust it: ✅ Free forever, open source (Apache 2.0) ✅ Zero telemetry, zero tracking ✅ Nothing is ever permanently deleted — everything goes to the Trash ✅ No registry "cleaner" snake oil, and it tells you what it can't read instead of hiding it Free download for macOS (Apple Silicon + Intel) and Windows → Try a scan and drop a comment with how much you reclaimed 👇

Reclaimed a surprising amount of space from Hugging Face and pip caches I didn't even know were bloating my drive. Love that nothing gets deleted without permission and there's zero telemetry.

Maker

 Love to hear it — the pip caches sneak up on everyone. Thanks for giving it a run. If you like its please share and support us.

The idea of targeting AI caches like Hugging Face and Ollama is genuinely useful, not something I've seen elsewhere. One thing I'd love is a dry-run preview before cleanup so I can see exactly what's flagged and skip anything I'm still using, like an active virtualenv or a model mid-download.

Maker

 Good news — that already exists. Nothing is deleted without you confirming, and there's a preview of exactly what's flagged before anything moves. Active virtualenvs and in-use models are marked "review first," never auto-selected.

Does Jharu let you preview exactly what will be removed from AI model caches (like Hugging Face and Ollama) before cleaning, and can you easily restore anything if you accidentally remove it?

How does Jharu decide which cache files are safe to remove?