Folder Organizer — Development skill for Claude Code

folder-organizer

**A Claude skill that organizes any messy folder by reading what's inside — then dedupes, classifies, and renames, without ever hard-deleting.**

一个先读懂内容再动手的文件整理 skill · [中文说明](#中文)

Most "organizers" sort by filename. That's the problem — filenames lie. A file called `final` is often a draft; an `invoice.html` is often a login page; three `report_v2.xlsx` copies are often *not* the same file. `folder-organizer` opens each file, figures out what it actually is, and only then decides where it goes. Nothing is deleted — removals go to a staging trash you empty yourself.

---

What it catches that a rename tool can't

• Exact duplicates by content, not name — the macOS … 2.png copies, the (1) re-downloads, the zip that's just a packed copy of files already sitting next to it.
• Version residue — draft / v2 / final / 副本 / 20250807 chains — grouped for you to prune, never auto-deleted.
• Mis-filed files — a spreadsheet whose sheet names reveal it's a duplicate of another; an .html that's actually a marketing page, not a report.
• Sensitive files loose in a work folder — IDs, contracts, bank details — flagged to move out, not filed away.
• The real space hogs — one 500 MB export hiding among 200 documents.

How it works — 5 phases, each ending at a checkpoint you approve

Phase	Does	Script
0 · Profile	Size hotspots, type mix, biggest files, date span	scan.py
1 · Dedupe	md5 content dedup → A exact dups · B version residue	find_dupes.py
2 · Read	A content signature for every file, gracefully degrading	signatures.py
3 · Propose	Ask the organizing axis, propose a structure + mapping	—
4 · Execute	Move / categorize; optional content-driven rename	safe_stage.sh

In action

**Phase 0 — know the terrain before touching it:**

$ python3 scripts/scan.py ~/Downloads
# Folder profile  ~/Downloads
1,204 files ·