disk-coldstore 0.1.0

Tiered laptop storage reclaimer: find junk, LZX-compress cool folders, 7z-archive cold ones — all reversible.

ColdStore

Tiered laptop storage reclaimer for Windows. Point it at a folder, get an honest, categorized plan of what can be freed, then apply the plan — every action is journaled and reversible.

What it does

Three tiers, safest first:

Tier	Trigger	Action	Reversible via
JUNK	Regenerable folders (node_modules, .venv, __pycache__, browser caches, Temp) or app caches	Send to Recycle Bin	Restore via Shell.Application COM
COOL	Not accessed in 30+ days, non-media content	NTFS/LZX compress in place with compact.exe	compact /u /s /a /i /exe
COLD	Not accessed in 180+ days	7z archive; original folder replaced with a .coldstore stub file	coldstore restore <stub>

Bonus: bit-exact JPEG -> JPEG XL lossless transcode using libjxl — reversible to the byte-identical original JPEG.

Install

pip install coldstore

Optional external tools that unlock extra performance / features:

• 7-Zip — if 7z.exe is on PATH, cold archiving uses the native binary (much faster on large folders). Otherwise the pure-Python py7zr fallback is used automatically.
• libjxl — cjxl.exe / djxl.exe on PATH enable the bit-exact JPEG -> JXL media tier.

Usage

coldstore scan     <path>                         # dry-run report of reclaimable space
coldstore analyze  <path> [--tree]                # WizTree-style disk analysis
coldstore apply    <path> --junk --cool --cold    # do the work
coldstore restore  <stub.coldstore>               # restore an archived folder
coldstore undo     --last N | --entry K           # undo previous actions
coldstore journal  [--limit N]                    # timeline of everything done

coldstore gui                                     # graphical interface
coldstore register [--uninstall]                  # register .coldstore file association
coldstore schedule --target <path> --cadence weekly

coldstore media scan|apply|restore  <path>        # bit-exact JPEG <-> JXL

Recommended first run — read-only, no changes:

coldstore analyze  C:\Users\%USERNAME% --tree --tree-depth 4 --tree-min-share 1
coldstore scan     C:\Users\%USERNAME% --min-size-mb 100

Then, when you're ready to actually reclaim space:

coldstore apply    C:\Users\%USERNAME% --min-size-mb 100 --junk --cool --cold

Every action is written to %USERPROFILE%\.coldstore\actions.jsonl. To undo anything:

coldstore journal --limit 20
coldstore undo --entry 14

How the tiers avoid overlap

If a folder qualifies as COOL/COLD but contains a JUNK sub-folder, the container's projection subtracts the JUNK bytes before estimating savings. The report therefore never double-counts across tiers.

Safety

• Every action is journaled with before/after bytes.
• JUNK goes to the Recycle Bin (never hard-deleted).
• COLD archives are integrity-verified with 7z t (or py7zr's testzip) before the original folder is removed.
• OneDrive cloud-placeholder files are detected via GetFileAttributesW (real attribute check, not a path heuristic) and excluded by default.
• Never touches C:\Windows, Program Files, ProgramData, or the Recycle Bin.

Requirements

• Windows 10 1607+ (LZX compression) or Windows 11
• Python 3.10+
• NTFS filesystem for the COOL tier

License

MIT