About DiskSpace AI
DiskSpace AI is a Mac disk analysis and cleanup app for people who want to identify reclaimable space with a seven-day recovery window. Through MCP, AI agents can inspect the disk and suggest cleanup, but must ask before cleaning.
Key features
- 01Explain the purpose of folders
- 02Classify items by whether they regenerate
- 03Keep files recoverable for seven days
- 04Find duplicate and similar photos
- 05Archive selected items as verified ZIP files
Use cases
- 01Find removable developer-tool caches
- 02Review space used by Xcode, Docker, and AI models
- 03Clean duplicates and old photos
MCP support
Claude Code, Claude Desktop, Codex, or Cursor can connect over MCP to analyze the disk and propose cleanup. The agent always asks before cleaning.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project README![]()
DiskSpace AI
Free up space on your Mac without fear.
It tells you what every folder is and only selects what recreates itself. Your files are kept for 7 days, in case you change your mind.
Website · Download the beta · Interactive demo · For AI agents (MCP) · Español

This repository is DiskSpace AI's public documentation: the catalog of folders the app knows, how to connect it to your AI agent, the release notes, and the place to report bugs or request features. The app itself is not open source.
What it does
- It knows what every folder is. uv, npm, pnpm and Homebrew caches, Xcode build data, node_modules from idle projects, Docker, AI models, simulators, installers you already used… The ones it doesn't know are explained by Apple's AI, on your Mac.
- Four colors, one rule. Green recreates itself and comes selected; amber is for review; gray is your data and is only shown; purple is unclassified.
- With a safety net. What recreates itself is deleted right away; your files are set aside for 7 days and can be restored.
- Duplicate and similar photos. In your Photos library and your folders: exact duplicates, near-identical shots (Apple's Vision picks the best one, on your Mac), old screenshots, big videos and bursts.
- Archive instead of delete. What you don't use but want to keep goes into a .zip that's fully verified before the original is set aside.
- An Explorer with sizes. A folder tree and a table sorted by size, with Quick Look and a filter: what Finder doesn't show you.
- Honest numbers. APFS clones and hard links are counted once, and Docker shows what it really takes up.
- For your AI agent. Claude Code, Claude Desktop, Codex or Cursor can analyze the disk and propose a cleanup; they always ask before cleaning. How to connect it
- Your files never leave your Mac. No analytics or tracking.

Download
DiskSpace AI 1.0.19 for Mac · macOS 15 or later · Apple silicon and Intel. During the beta, Pro is free until December 31, 2026. Install steps on the beta page.
Signed and notarized by Apple. Also with Homebrew:
brew install --cask diskspaceai/tap/diskspace-ai
The app updates itself. What changes in each version is in CHANGELOG.md.
In this repository
| folders/README.md | The folders the app knows: what each one is, whether you can delete it and how (also as JSON) |
| mcp/README.md | Connect DiskSpace AI to your AI agent (MCP server) |
| CHANGELOG.md | Release notes for every version |
| Issues | Report a bug or request a feature |
Help and contact
- FAQ and pricing: diskspaceai.com
- Support: [email protected]
- Bugs and suggestions: open an issue
The folder catalog and documentation are licensed under CC BY 4.0 (see LICENSE). DiskSpace AI is not affiliated with Apple. Mac and macOS are trademarks of Apple Inc.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.