skills/file-management/SKILL.md
Help users design, audit, and maintain a file management system — folder structures, naming conventions, storage strategies, and cleanup workflows. Use this skill whenever the user mentions organizing files, cleaning up folders, setting up a file structure, naming files, managing photos/albums, digital asset organization, cloud vs local storage, backups, or says things like "my files are a mess", "where should I put this", "help me organize my drive", "set up a folder structure", or asks about file naming.
npx skillsauth add duruii/scientific-skills file-managementInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
A practical framework for organizing digital files — born from studying two distinct approaches: a creator's asset-heavy workflow (Ami) and a knowledge worker's information architecture (Jeffrey). The core insight: organize by where you'll use it, not where you found it (David Allen).
Function over form. A folder structure that looks beautiful but requires discipline you don't have is worse than a slightly messy one you actually maintain. The best system is the one you stick to.
Organize by usage context. Meeting notes for Project A go inside Project A's folder — not in a global "Meeting Notes" folder. When you open the project, everything is there. When you share it, everything goes with it.
Name for searchability. The goal is: 6 months from now, you can find this file in under 30 seconds. That means consistent naming conventions, not perfect ones.
Don't over-optimize. Five levels of nested folders with a rigid numbering taxonomy will be abandoned in two weeks. Start simple, add structure only where friction actually appears.
Dual backup, always. Cloud + local. Never a single point of failure for anything you can't afford to lose.
Borrowed from Johnny Decimal, simplified. Use two-digit prefixes to control sort order and signal priority:
Root/
01-Active/ # What you're working on right now
02-Work/ # Work-related (projects, clients, deliverables)
03-Personal/ # Personal documents, finances, health
04-Reference/ # Things you look up regularly (templates, IDs, guides)
05-QuickShare/ # Temporary staging area for files to share with others
06-Assets/ # Reusable resources (fonts, icons, music, graphics)
07-Archive/ # Completed projects, cold storage staging
99-Archive/ # Deep archive — stuff you can't delete but rarely touch
Not every folder is required. Use only the categories your life actually needs. The key rules:
For active projects (especially creative/media work), use this template inside the project folder:
ProjectName/
01-Source/ # Raw footage, original files, source data
Camera-A/
Camera-B/
Screen-Recordings/
02-Working/ # Work-in-progress, drafts, iterations
03-Output/ # Final deliverables (video, PDF, slides)
04-Assets/ # Project-specific reusable elements
README.md # Brief project description, dates, status
This mirrors Ami's creator workflow: separate raw inputs from finished outputs so you can archive source material independently of deliverables.
For personal photo libraries:
Cull with reverse selection. Instead of marking favorites, delete everything first (move to trash), then recover the keepers from trash. Psychologically faster than heart-ing photos one by one.
Album naming: YYYY-MM Event
2025-03 Japan Trip2024-12 Mom BirthdayArchive cycle. Every 2-3 years, move older photos from phone/cloud to external drive. Keep the last 2 years on-device for quick access.
Dual storage. Cloud library (iCloud/Google Photos) for sync across devices + periodic backup to external drive.
Choose granularity based on how you'll search for it:
| Granularity | Format | Example | When to use |
|---|---|---|---|
| Year only | YYYY | 2025-Budget.xlsx | Annual docs, budgets |
| Year + Quarter | YYYY-QN | 2025-Q1-Review.pptx | Quarterly reports |
| Year + Month | YYYY-MM | 2025-05-Presentation.pptx | Monthly recurring |
| Full date | YYYY-MM-DD | 2025-05-05-Training.docx | One-off events |
Rule of thumb: the more granular the date, the more you need to know which parent
folder it lives in. Use full dates only when the parent folder provides context
(e.g., Jeff's Sharings/2025-05-05-Inbox-Zero-Training.docx).
Prefix or suffix files with a consistent type keyword. Pick 5-8 and stick to them:
doc- # Documents, reports
notes- # Meeting notes, personal notes
wk- # Working/draft files
brief- # Briefs, briefings
slides- # Presentations
form- # Forms, templates
ref- # Reference material
Combined: 2025-Q1 doc-Quarterly-Business-Review
90% findability rule: if you combine a project name + one type keyword, you should find the file in search. That's the bar.
- and _| Tier | Medium | What lives here | Access frequency | |---|---|---|---| | Hot | Local SSD / laptop | Active projects, today's work | Daily | | Warm | Cloud sync (iCloud, Google Drive) | Recent projects, shared docs | Weekly | | Cold | External drive / NAS | Completed projects, archives | Monthly or less |
3-2-1: 3 copies, 2 different media types, 1 offsite.
Practical version for most people:
When someone shares a file with you, decide immediately:
| Action | When | Why | |---|---|---| | Do nothing | One-off request, you'll edit once and forget | No clutter | | Make a copy | Template you'll reuse, or you need a frozen snapshot | Independent from source | | Add a shortcut | Ongoing document you need to find in YOUR system | Organize without moving |
For shortcuts: create a pointer in your own folder structure (e.g., inside 02-Work/ProjectX/)
that links to the original. You organize your way, the owner organizes theirs.
Starred / Flagged files: maximum 5 at any time. If everything is starred, nothing is. Criteria for starring:
For files you can't rename (shared docs, system files):
When helping a user clean up their files, run through this:
Then propose a structure that starts from their current state — don't prescribe a perfect system they'll never adopt. Migration should be gradual:
| Platform | Operator | Example |
|---|---|---|
| Google Drive | type:presentation | Find all Google Slides |
| Google Drive | from:[email protected] | Files shared by specific person |
| macOS Spotlight | kind:pdf | Find PDFs |
| macOS Spotlight | date:this week | Recent files |
| Alfred/Raycast | open <keyword> | Launch pinned files fast |
Things that look organized but create friction:
tools
Debugging methodology for LLM tool calls — trace from the model's side first, use runtime evidence over code inference, and follow live request chains for reasoning/thinking bugs.
testing
Research-grade single-paper analysis with evidence-grounded structured extraction and internal self-evaluation. Use when users ask to summarize or screen one academic paper from an arXiv link/ID or local PDF and need verifiable claims with citations, especially for Chinese-language output to students.
tools
Use browser MCP to access IEEE Xplore through university library proxy, preserve institutional session, run keyword/advanced/journal search, and optionally post-filter by CCF rank (for example CCF-A) with structured output.
testing
Fetch and organize course transcripts from DeepLearning.AI. Use this skill whenever the user mentions DeepLearning.AI courses, wants to download course transcripts, subtitles, or VTT files from a course, or asks to organize lesson transcripts from learn.deeplearning.ai. It does NOT trigger for general video subtitle downloading — only for DeepLearning.AI courses specifically.