organize（organize 搜索）

Organization Skill

Version: 1.2 Source: L1/L2/L3 Cognitive-Phase Framework

Folder structure should match how your brain navigates, not how a database stores records.

North Star

Goal: You always know where you are and what you're looking for.

Definition: Any person (or AI) should be able to find a file in 3 decisions or fewer, without guessing which folder it's in.

Smell: If you open a folder and feel lost, the structure has failed.

Quick Start

Already know what you need? Start here:

For the full framework explanation, read on.

The L1/L2/L3 Framework

Folder navigation is three cognitive phases. Each phase answers a different question, and folder names should match the type of thinking you're doing at that level.

L1: Context (Nouns / Arenas)

Question: "Where am I operating?"

Brain mode: Wayfinding. You're orienting yourself to a domain before doing any work.

Folder names are typically nouns — broad arenas of activity. Mindset-based verbs (e.g., Fulfill, Maintain, Grow) are valid when they trigger the same wayfinding shift:

Drive/
├── Finance/
├── Marketing/
├── Operations/
├── Production/
├── Legal/
└── Personal/

Test: You're still in L1 as long as you're narrowing the domain. You might click through multiple folders and still be in L1 if each click is answering "which world am I in?"

Success condition: Your brain naturally shifts from "where am I?" to "what am I trying to do?"

L2: Execution (Verbs / Outcomes / Subject Categories)

Question: "What am I trying to accomplish?" — or in creative/domain-heavy projects — "What type of thing am I working on?"

Brain mode: Targeting. You've picked an arena, now you're selecting an objective or subject focus.

Folder names are verbs, outcomes, or subject categories — what work gets done here:

Marketing/                    # Business example — verbs/outcomes
├── Launch/
├── Grow/
├── Campaign/
├── Brand/
└── Analyze/

Arctic/                       # Creative example — subject categories
├── Locations/                # "I'm working on PLACES"
├── Residents/                # "I'm working on CHARACTERS"
└── Lore/                     # "I'm working on STORY"

L2 doesn't require verb names. In creative, world-building, or domain-heavy projects, subject-matter categories (Locations, Characters, Assets, Environments) function as L2 when they trigger the brain shift from wayfinding to intent. The test is whether your brain changes modes — not whether the folder name is a verb.

Test: You're in L2 as long as you're narrowing the objective, work mode, or subject focus. L2 completes when you start hunting for a specific artifact.

Success condition: Your brain shifts from "what am I doing?" to "which thing do I need?"

L3: Retrieval (Objects / Artifacts)

Question: "What exact thing do I need?"

Brain mode: Grabbing. You've picked an outcome, now you grab the thing.

Folder names are specific objects — the artifacts themselves:

Campaign/
├── Q1-2025-Product-Launch/
│   ├── Brief.docx
│   ├── Timeline.xlsx
│   ├── Assets/
│   └── Results.pdf
└── Q2-2025-Brand-Refresh/
    └── ...

Test: You're in L3 when you're scanning for a specific file or artifact. You know what you want — you're just finding it.

Key Principle: Layers Are Cognitive Phases

The layers are NOT literal folder counts. They describe your mental state as you navigate.

The test isn't "how many folders?" but "has my brain shifted modes?"

Diagnostic Guard

Before diagnosing a structure as broken or a layer as missing, apply these checks:

The framework serves navigation, not the other way around. If someone navigates smoothly, the structure is working regardless of whether folder names are verbs, nouns, or subject categories. Diagnose problems people actually experience, not theoretical label mismatches.

Scaffolding Evaluation

Empty folders aren't always clutter. Pre-built folder structures serve a real purpose in creative and content-heavy projects — they communicate "this is a slot for future content" and ensure consistency across parallel branches. Before flagging empty structure as a problem, determine whether it's intentional scaffolding or scattered clutter.

Intentional Scaffolding (Acknowledge It)

Scaffolding is working when it passes these checks:

When you encounter scaffolding that passes these checks, acknowledge it as useful structure rather than flagging it for removal. Note where content has filled in and where it hasn't, but don't recommend tearing down the framework.

Example — good scaffolding in a world-building project:

VR World/
├── Arctic/
│   ├── Locations/
│   │   ├── Commercial/
│   │   │   ├── _Ideas/
│   │   │   └── Winter Lodge Restaurant/    ← named, filling in
│   │   ├── Natural/
│   │   │   └── _Ideas/
│   │   └── Residential/
│   │       └── Dream House/                ← built out with rooms
│   └── Residents/
│       └── Alexander Borealis/             ← named character
├── Forest/                                  ← same template, different fill state
│   └── Residents/
│       ├── Big Bear/                        ← fully developed
│       └── Zebra Fox/
└── Desert/                                  ← same template, less filled
    └── ...

The template is consistent, reflects the world-building domain (biomes × content types), and is progressively filling with real content. The empty slots are a content roadmap, not noise.

Example — good scaffolding in a content pipeline:

character-types/
├── bears/images/
│   ├── head/        (31 files)              ← filling in
│   ├── body/        (15 files)
│   └── tail/        (0 files)               ← empty but expected
├── dogs/images/
│   ├── head/        (26 files)
│   └── legs-and-feet/ (28 files)
└── foxes/images/                            ← mostly empty, work not started yet

Same body-part template across 25 species. Bears and dogs are well-populated; foxes haven't started yet. The empty folders are task slots in a production pipeline.

Scattered Clutter (Flag It)

Scaffolding has become clutter when:

What To Do With Each

The Key Test

Would deleting this empty folder lose information? If the folder name + location tells you "this is a slot for [specific type of content]," it's scaffolding. If deleting it wouldn't confuse anyone about what goes where, it's clutter.

Progressive Disclosure

The framework uses progressive disclosure — each level reveals only what you need for your current decision:

Step 1: See 5-8 arenas          (L1 — which world?)
Step 2: See 3-6 outcomes         (L2 — which goal?)
Step 3: See the artifacts         (L3 — grab the thing)

Ideal Counts per Level

If any level exceeds its max, it's a signal to split or restructure.

Naming Conventions

L1 Names: Broad, Stable, Recognizable

L2 Names: Action-Oriented, Scannable

L3 Names: Specific, Descriptive, Findable

Universal Naming Rules

Context-Sensitive Naming Rules

Don't apply code conventions to personal drives, or personal conventions to codebases. Match the rules to the environment.

File Name Semantics

File names should communicate what the file contains at a glance. A good file name answers "what is this?" without opening it.

Non-Semantic Names (Flag These)

What Makes a Good File Name

When to Flag vs. When to Ignore

Suggesting Better Names

When flagging non-semantic names, suggest replacements based on:

1. The folder context — a file in Arctic/Locations/Natural/_Ideas/ is probably a nature/landscape reference
2. The file's visible content — if you can read the image, describe what you see
3. Sibling files — if named files exist alongside unnamed ones, match the pattern
4. Don't require perfection — snowy-village.jpg is better than 241076.jpg, even if it's not a complete description

Decision Guide: Where Does This File Go?

Active vs Reference: The Intent Test

The test: "Am I opening this folder to *do work* or to *look something up*?"

• If both → split into two L2 folders. The active folder holds the work and links to the reference folder.
• Active folders archive when work is done. Reference folders persist.

Overlap Resolution

The active folder owns the work. The reference folder owns the knowledge. They link to each other — never duplicate.

Never duplicate. Two copies means one gets stale.

System Folders

Infrastructure folders that support the structure, not content. Use a leading underscore.

Rules for System Folders

Transit Folders

Folders holding content that is actively being moved to a known destination. Use a leading tilde (~). Distinct from _Unsorted/ (destination unknown).

Key rules: always tilde-prefixed, include a date suffix (~Transfer 2026-02/), include _STATUS.md for multi-session transfers, delete when empty.

For full details (naming conventions, lifecycle, staleness signals, analysis scope), see references/transit-folders.md.

File Lifecycle

Created → Active → Completed → Archived → (Deleted)

Files move through stages based on project status. Archive at L2 level (_Archive/ within the outcome folder). Default to keeping — storage is cheap, regret is expensive.

For full lifecycle rules, archive strategy, and maintenance cadence (weekly/monthly/quarterly), see references/lifecycle.md.

Adapting by Scale

Red Flags

Stop and reorganize when you see:

Anti-Patterns

Exceptions: When to Break the Rules

Every rule in this framework has legitimate exceptions. The key: break rules consciously, not by accident.

Date-First at L1

Test: Is the time period genuinely the primary organizing principle, not just a habit?

More Than 12 L1 Arenas

Test: Would merging any two arenas force you to constantly context-switch within a single folder?

Flat L3 With Many Files

Test: Can you still find what you need without subgrouping? If search or date-sorting is sufficient, don't force artificial categories.

General Rule for Exceptions

If you're breaking a rule, write down *why* in a _README.md at that level. If you can't articulate the reason, you shouldn't break the rule.

Quality Checks

Before considering a drive "organized," verify these rules hold.

Structure Checks

Naming Checks

Boundary Checks

Applying the Framework

Step 1: Audit Current State

List your top-level folders. For each one, ask:

• Is this clearly an L1 arena (a domain of activity)?
• Or is it something else? (date, project, catch-all)

Step 2: Design L1 Arenas

Write down the 5-8 major domains of your life or work. These become your L1 folders. They should feel natural and stable.

Step 3: Design L2 Outcomes Per Arena

For each arena, list the 3-6 things you actually do there. These are your L2 folders.

Step 4: Let L3 Emerge

Don't pre-create L3 structure. Let it emerge as you actually produce artifacts. Structure follows work.

Step 5: Migrate Gradually

Move files in batches. Start with the arena you use most. Don't try to reorganize everything at once. For detailed migration steps, see references/migration-guide.md.

References

• references/naming-patterns.md — Detailed naming examples and conventions for each layer
• references/migration-guide.md — Step-by-step guide for reorganizing existing messy drives
• references/transit-folders.md — Transit folder conventions, lifecycle, and staleness signals
• references/lifecycle.md — File lifecycle rules, archive strategy, and maintenance cadence

Assets

• assets/home-drive-template.md — Starter structure for a personal home drive
• assets/work-drive-template.md — Starter structure for a professional work drive

Scripts

• scripts/audit_structure.py — Scan a folder tree and report L1/L2/L3 issues (catch-alls, depth, naming)