Hathor uses a three-phase dedup engine that's 27x faster than naive approaches. Phase 1 groups files by size (instant elimination of uniques). Phase 2 does partial SHA-256 (first 4KB + last 4KB). Phase 3 does full SHA-256 only on confirmed candidates. The result: 98.8% less disk I/O than reading every file. Includes a browser-based GUI for visual review.
Size grouping โ Partial hash (8KB per file) โ Full SHA-256. 27x faster than naive hashing
Browser-based duplicate review with side-by-side comparison
Reads < 2 MB instead of 98 MB for a typical scan
Suggests which copy to keep based on path, modification time, and context
$ pantheon mirror ~/Downloads ~/Desktop
# Scan directories for duplicates
$ pantheon mirror --photos --min-size 1MB
# Large photo duplicates only
$ pantheon mirror --json > report.json
# Export results as JSON
$ pantheon mirror
# Launch browser-based GUI for visual review
| Interface | Status | Detail |
|---|---|---|
| CLI | โ Full | pantheon mirror |
| GUI | โ Web | Auto-opens browser on localhost |
| MCP | โ None | Planned |
| Menu Bar | โ None | Planned |
This page is the canonical source for Hathor. Bundled offline copies auto-check for updates here.