Skip to content

kzx commands

kzx is Kaizendex’s command-line tool: the app’s engine with no screen. You or a coding agent type kzx commands in a terminal. Every write goes through the same rules as a tap, and is logged and undoable.

Where kzx runs

Computer What kzx opens
With the desktop app the same database and folder as the desktop app
Without the desktop app (a server, Linux, a web-only user) its own engine: one more device of your account

Does it work in my setup?

Example: Codex runs kzx search "knee pain" on your laptop. This table says when that works.

Your setup or your data kzx and MCP on your computer A cloud MCP address (later)
Free, Cloud Sync on yes no: it needs Pro
Pro, Cloud Sync on yes yes
Cloud Sync off (desktop app only) yes no: it needs Cloud Sync
A Cloud Sync collection yes yes, on Pro
A Cloud Sync, end-to-end encrypted collection yes only with a key you give that connection
A This device only collection yes, on the device that holds it never
A No AI collection refused refused
Hidden entries refused refused
  • On your computer, kzx works the same in every setup: Cloud Sync on or off, end-to-end encryption on or off. Your own computer can already read your own data. End-to-end encrypted means only your devices hold the key, and your computer is one of them.
  • Your privacy choices still apply. “No AI” collections are never given to AI tools. “Ask me first” waits for your Approve. Locked files stay locked. Hidden stays hidden.
  • A cloud MCP address (later) lets cloud AI tools, such as Claude Code on the web, reach your data while your computer is off. It reads your data in Cloud Sync, so it needs Pro.

Commands

Command What it does Example
kzx status shows file edits not yet applied, and the problems list after an agent edits five files
kzx data where prints where every part of your data is, with its size: your folder, the database, the photo store, the backups, the start log, window settings, paired browsers, AI models and the note ~/.kaizendex/location.json kzx data where --json
kzx data move moves the database into a folder (--to) and, when they are still in %APPDATA%, the photo store and backups to ~/.kaizendex/app-data; each copy checked, the old files kept 30 days; a dry run unless --apply (it says how big the move is); close the app and every Claude Code session first kzx data move --to ~/Kaizendex --apply
kzx sync --wait waits until every file change is in the database, then returns edit a file, then query it
kzx search searches by words and meaning together kzx search "knee pain"
kzx entry add creates an entry a new Workouts entry
kzx import imports a CSV, a markdown file or a folder of them as one undoable step; --dry-run shows what would change; --collection names the collection kzx import runs.csv --collection Runs --dry-run
kzx go-back Go back in time: --list shows the backups; --to <backup> brings the account (or --collection one) back to it, as one undoable step that only adds and changes; --repair puts the newest good backup in place of a damaged data file kzx go-back --to hourly-20261107-09.sqlite --collection Runs --dry-run
kzx context returns an entry and its whole neighbourhood in one call see below
kzx tree prints the tree (entries inside entries, as under a collection’s row in the sidebar): titles and paths, to a chosen depth kzx tree
kzx log quick capture from the terminal log a run without opening the app
kzx run starts an automation that has a button or command trigger “Plan my week”
kzx automations list shows every automation; run <name> runs one now; on <name> and off <name> switch one, only when you ticked “AI can make and switch on automations” in Settings → Security → AI permissions kzx automations on "Done pile"
kzx repair-v1-keys finds card layouts from a v1 import that still name a property by its old v1 name (a workout’s set columns reps, weight) and fixes them; a dry run by default, --apply makes the fix one undoable step kzx repair-v1-keys, then kzx repair-v1-keys --apply

Run a bulk job as a dry run first: it shows the changes and writes nothing.

Emojis look wrong

kzx writes its answers as UTF-8, the usual text format for emojis. Your data is fine. Shown in a terminal window, the emojis look right. They break only when Windows PowerShell catches the answer, for example $x = kzx context Tasks or kzx context Tasks | Out-File notes.txt: it reads the answer with an old code page (an old table of letters), so ⏰ turns into odd letters.

Fix: type this once in that PowerShell window, then run the command again.

Terminal window
[Console]::OutputEncoding = [System.Text.Encoding]::UTF8

Git Bash and the MCP server read UTF-8 already.

kzx context

One call returns the entry, its parent chain, its children, its Relations both ways, its mentions and backlinks. Each comes as a title, its key fields and a short excerpt, with ids for a follow-up read.

Option What it does Example
<entry> the entry to start from "Projects/Launch"
--collection start from every entry in a collection --collection Goals
--depth how many links deep to go --depth 2
--fields which fields to include --fields status,progress,due
--max-tokens a size budget for the answer --max-tokens 4000

Example: “What is the status of my goals?”

Terminal window
kzx context --collection Goals --depth 2 --fields status,progress,due

Also offered, spelling not decided

kzx also offers these as commands; their command-line names are not decided: list skills, read a skill, structured search, recent entries, related entries, and read an entry. The MCP names are on MCP tools.

Any tool by name

kzx tools lists every tool. kzx tool <name> --args '{...}' runs one, with its arguments as JSON.

Example: move an empty collection to the Trash (needs “AI can move collections to the Trash” in Settings → Security → AI permissions):

Terminal window
kzx tool delete_collection --args '{"collection":"Old reading list"}'

With the checkbox off, kzx refuses and prints Open the permission: kaizendex://settings/security/ai-collections.

The MCP server

kzx also serves MCP, so agents without a shell can use the same tools. The AI in the app runs the same tools too, with the same rules. See MCP tools.

Privacy

kzx treats every caller as an AI, even you typing in a terminal: it cannot tell you from a coding agent that types the same command. So it obeys every collection’s AI setting. “No AI” refuses. “Ask me first” shows an Approve card in the app that names the program that asked: “A program using kzx wants to read Diary”. Hidden entries are refused. Cloud Sync and end-to-end encryption do not change this: the same rules hold in every setup.

Changes only you can make

kzx cannot make a change only you can make: where a collection lives, its AI choices, sharing, approving an automation, keys, devices, your AI safety settings or Delete for good. It can only suggest the change. The suggestion waits in the app, and nothing changes until you approve it on your device.

One exception is yours to allow: with “AI can make and switch on automations” ticked in Settings → Security → AI permissions, kzx automations on and off switch an automation. With it off, they change nothing and say where the checkbox is. The same holds for kzx tool delete_collection and “AI can move collections to the Trash”.

Example: Codex runs a command that would set your Diary to Any AI. The Diary stays as it was. The app shows the suggestion, and you decide.