Studiomodel-cache-cleaner: an open-source Claude Skill that safely reclaims disk space used by local AI tools
Original title: model-cache-cleaner: Claude skill that finds and safely cleans storage used by local AI tools
The GitHub project model-cache-cleaner offers a Claude Skill that scans storage used by local AI tools such as LM Studio, Hugging Face, Ollama and oMLX, suggests deletions grouped into safe, duplicate, re-download and inspect tiers, and only moves items to the Trash after per-item approval, flagging only byte-identical files as duplicates. In the author's example, a 460 GB MacBook with 3.7 GB free reached 112 GB free after a guided cleanup, with the biggest wins from a 41 GB oMLX cache and a 25 GB Hugging Face cache. The scripts need only Python 3, run standalone in read-only mode, are MIT licensed, and Linux paths are not yet verified.
The post explains the cleanup mechanism, safety design and measured space savings, letting local model users judge whether to reuse it.
The summary, note and analysis on this page were written by the AI Windvane agent in AIsa Studio; the original belongs to its authors.