pw-organize

name

pw-organize

description

Organizes a 1Password CSV export into multiple vault-specific CSVs. Session-based flow with slash commands for each phase: intake interview, auto-categorization, human review, and export. Parses 1Password CSV exports, applies deterministic rules (email matching, domain matching) and AI service knowledge to categorize each entry into user-defined vault categories, then generates per-vault CSV files importable back into 1Password. Use when: organizing passwords into vaults, splitting a 1Password export, separating personal from business passwords, cleaning up a password manager.

Password Organizer

Organize a 1Password CSV export into separate vault-specific CSVs. Walks through five phases via slash commands, producing importable CSVs for each vault.

This is password organization, not password management.

Session Lifecycle

/pw-organize:initialize ──▶ /pw-organize:intake ──▶ /pw-organize:categorize ──▶ /pw-organize:review ──▶ /pw-organize:export
       (SETUP)                 (INTERVIEW)            (CLASSIFY)                  (CONFIRM)              (OUTPUT)

Commands

Command	Phase	What It Does
/pw-organize:initialize [session]	0. Setup	Creates the session folder, source-files directory, and intake template.
/pw-organize:intake [session]	1. Intake	Conversational interview: who you are, your emails, your business, desired vault categories. Saves to {session}/working/intake.md.
/pw-organize:categorize [session]	2. Categorize	Parses the CSV, applies deterministic rules and AI service knowledge to categorize every entry. Flags ambiguous items.
/pw-organize:review [session]	3. Review	Walks through categorized entries vault by vault. Resolves flagged items. Finalizes assignments.
/pw-organize:export [session]	4. Export	Generates per-vault CSV files importable into 1Password, plus a summary report.

File Structure

{session}/
├── source-files/                      ← User drops 1Password CSV export(s) here
│   └── *.csv
│
├── working/                           ← All intermediate files
│   ├── intake.md                      ← Created by initialize command
│   ├── parsed_entries.json            ← Structured parse of CSV(s)
│   ├── rules_result.json             ← After deterministic rule pass
│   ├── categorized.json              ← After AI classification pass
│   └── reviewed.json                 ← After human review
│
└── output/                            ← Final deliverables
    ├── {vault-key}.csv               ← One CSV per vault category
    └── summary.md                    ← Report of what went where

Important Rules

• Interview first — Never categorize without understanding the user's emails, business, and vault preferences
• Preserve all fields — Output CSVs must contain every column from the original export, in the same order
• Don't guess on ambiguous items — Flag them for user review
• No passwords in logs — Never print or display password values during processing
• Source files are read-only — Never modify the original CSV exports
• Ask before overwriting — Check before writing to any existing file
• Dynamic vaults — Vault categories are defined by the user during intake, never hardcoded