miragecoa/AI-Cubby 是一个托管在 GitHub 上的开源项目，主要使用 Vue 编写，采用 GNU Affero General Public License v3.0 许可证，目前获得 383 个 Star、7 次 Fork。
仓库全名
miragecoa/AI-Cubby
README 摘要
中文 | English AI Cubby Everything you've ever opened — always findable. 🌐 aicubby.app · Download · Discord Why does this exist? Scenario 1: The desktop landfill Your desktop is buried under screenshots, installers, and random files you saved "just for now." Finding anything takes forever, and the mess itself is draining. You tell yourself you'll organize it — spend an afternoon sorting into folders — and three days…
Your desktop is buried under screenshots, installers, and random files you saved “just for now.” Finding anything takes forever, and the mess itself is draining. You tell yourself you’ll organize it — spend an afternoon sorting into folders — and three days later it’s chaos again. So you give up, and next time you still have to dig.
With AI Cubby:
Create one folder on your desktop, drag everything into it — desktop instantly clean, deeply satisfying
Drag that folder into AI Cubby — one click, everything’s indexed
Next time you need something, just search — it appears instantly
No categories. No folder hierarchies. Desktop stays clean, nothing is lost.
Scenario 2: Tool amnesia
You found a great tool that fixed your corrupted MP4. Used it once, problem solved. Six months later the same issue comes up — what was that tool called? Where did you install it? No idea.
That’s me with WizTree — the perfect disk analyzer I use when my C: drive fills up. Can never remember the name. Every single time I have to search for it all over again.
AI Cubby silently logs every program and file you open during normal computer use. Next time you need it, one search is all it takes.
The compound effect: it grows on you
At first I thought manual tagging was a non-starter — nobody bothers. But after talking to real users, the picture was more nuanced: people won’t tag random files, but they absolutely will tag the handful of tools and files they care about most.
So before AI fully takes over, AI Cubby’s core identity is a usage history recorder — the kind of tool that gets more useful the longer you use it.
You finally track down that one specific cleanup utility, and this time you slap a tag on it. The app also silently records that you opened it. Next time you need it — one second. Every find is a deposit. Over time, your “digital memory” comes alive.
Core Features
1. Automatic recording, zero configuration
Runs silently in the background and logs everything you use:
File monitoring — images, videos, documents you open are recorded automatically
Desktop shortcuts — games and apps on your desktop are detected and indexed
Process detection — every program you launch is captured
No manual imports needed. Just use your computer normally.
You can also drag files or folders directly into the window for instant import.
2. Usage tracking
Not just what you used — but how: open count, total runtime, live timer, last opened time.
That game from three months ago, that batch of photos from last week — open AI Cubby and see it all at a glance.
3. AI features (coming soon)
Natural language search — describe what you remember, not the filename:
“the video editor I used last year”
“files I worked on yesterday”
“the spreadsheet my boss sent me to fill in”
“the presentation we used at the trade show last week”
AI auto-categorization — use it, let AI organize it:
Photos — tag one “Canada, 2018, Niagara Falls” and AI labels the rest of the batch
Apps & Games — drop in a folder of exe files, AI identifies which are games and which are tools
How it compares
AI Cubby
Everything
Playnite
Auto-records usage history
Yes
No
No
Zero configuration
Works out of the box
Filename search only
Requires plugin setup
Runtime tracking
Built-in
None
Plugin required
All resource types
Games / apps / images / video / docs
Files only
Games only
Data privacy
Fully local
Fully local
Fully local
More Features
Search & recall — full-text search across titles and notes, find anything you’ve used
Pin & ignore — pin important resources to the top, ignore unwanted ones (5-second undo)
Tag system — manual tags with smart suggestions based on resource type
Multiple profiles — separate work and personal libraries, each with its own database
Folder as resource — import an entire folder as a single entry, great for mod packs and project directories
System tray — closing the window minimizes to tray, stays running quietly
Auto-start — set once, always running
Roadmap
.nfo metadata — read Kodi / Jellyfin .nfo files to import existing media libraries automatically
Screenshots
Drag & Drop Import
Usage Tracking
Search & Filter
Themes & Customization
Download
Visit aicubby.app for the latest release, changelog, and screenshots.
Portable (no installer): AI-Cubby-x.x.x-portable-win-x64.zip — unzip and run
Requires Windows 10 or later
Build from source
git clone https://github.com/miragecoa/AI-Resource-Manager.git
cd AI-Resource-Manager/app
# Install dependencies
npm install --ignore-scripts
# Compile native module (better-sqlite3)
npm run rebuild
# Development mode
npm run dev
# Production build + package
npm run package
No. All data is stored 100% locally on your machine — nothing is uploaded to any server. AI Cubby only records programs and files you actively open; it does not capture screenshots or read file contents. You can delete any record at any time from the ignored list.
Will antivirus software flag it?
No. AI Cubby uses the Windows native filesystem watch API (fs.watch) to detect recently opened files — a standard system API. Process detection uses WMI event subscriptions, not process list polling.
Where is my data stored?
All data is stored in a local SQLite database. The exact path is shown on the Settings page. Cover image thumbnails are cached in a covers/ folder in the same directory.
Is macOS / Linux supported?
Windows is the current focus. macOS and Linux support is on the roadmap.
Does it track every process?
System service processes (svchost, services, etc.) and helper processes (updater, crashpad, etc.) are automatically filtered out. Only programs you actively use are recorded.
How much CPU / memory does it use?
Almost none. File monitoring uses OS-native event push (no polling). Process detection uses WMI event subscriptions (no polling). Typical memory usage is 80–120 MB; CPU usage is near 0%.
Community
Contributions are welcome — feel free to fork and open a pull request.
评论