Space Wizard: Architecture & Technical Reference

Space Wizard is a cross-platform Electron desktop utility for visualizing storage usage and clearing duplicate files. It has no server component, no account system, and makes no network calls as part of normal operation. This reference describes its scan architecture, duplicate-detection method, and license-gated capabilities precisely, so a reviewer can evaluate the product against its actual behavior rather than marketing language.

The application ships as a desktop Electron app with a single-window UI organized into Files, Duplicates, and Usage tabs. There is no dual-pane browser and no archive (ZIP) handling built in — scope is deliberately narrower than a general file manager, focused on diagnosing and clearing storage.

On macOS and Linux, a scan shells out to the native `du` command (with size-in-kilobytes flags) for speed on large directory trees. On Windows, or if the native command is unavailable or fails, the scan falls back to a Node.js breadth-first directory walk.

Cloud-sync folders — OneDrive, Dropbox, iCloud Drive, Google Drive, Box, pCloud, and MEGA — are excluded from scans by default, since these often contain placeholder files that don't reflect actual local disk usage. This exclusion is a UI-togglable default, not a hard restriction.

Scan results render as a bar chart or a donut/pie chart, toggleable, showing the largest categories by size plus an aggregate 'other' bucket. This is a category breakdown, not a hierarchical, drillable treemap — there is no nested-rectangle, folder-by-folder zoom view in the current product.

Duplicate detection runs a three-pass algorithm designed to minimize expensive full-file hashing:

• Pass 1 — group files by exact byte size; files with a unique size are immediately excluded
• Pass 2 — compute a partial hash (first 4,096 bytes, SHA-1) for remaining candidates to cheaply eliminate near-misses
• Pass 3 — compute a full SHA-1 hash only for files that survived pass 2, confirming true duplicates

For the avoidance of doubt: there is no fuzzy or filename-based duplicate matching (matching is by content hash only), no interactive treemap, no point-in-time snapshot comparison between scans, and no live/real-time filesystem monitor — scanning is on-demand only. There is no built-in ZIP/archive tool and no dual-pane file browser.

Deletions — whether a single file or a bulk duplicate clear — are routed through the operating system's trash or recycle bin, not a raw filesystem unlink. A deletion is recoverable through the normal OS mechanism unless the user separately empties the trash.

Connected drives are inspected for filesystem type to detect network shares (SMB/AFP/NFS/CIFS on macOS/Linux, WMI drive-type on Windows) and are labeled as such in the UI. This is detection, not a dedicated network-browsing protocol implementation.

License tiers are enforced at the code level on two axes: scan-size cap (Free: 100 MB per scan; Pro: uncapped) and drive scope (Free: the local home drive only; Pro: any detected drive, including network shares). Team-tier features beyond Pro's gates (volume licensing, centralized preference management) are an account/licensing-layer distinction rather than a difference in the scanning engine itself.

No file content, metadata, or scan result is transmitted off the machine running Space Wizard. There is no telemetry collection and no account requirement for the Free tier. Everything described above runs as a local process against local (or locally-mounted network) storage.