engenty wizards

Make a wish. Get your wizard.
It guides you step by step.
Let AI do the work.

For personal use and small businesses
Turn a task you repeat into a wizard, or pick a ready one from the gallery. Use it yourself or share it with colleagues and clients: the wizard guides you step by step, you confirm its drafts, and you get the result.

For developers
Wizards are skills on steroids: workflows with their own interface, built and run on your machine with Codex, Claude Code or your own API key. They connect to other systems over MCP and HTTP. And you can put them into other places, such as Claude and other AI chats, or your own website with an embed code.

For enterprises
Wizards make AI dependable. Your domain experts define the business processes themselves, without waiting for developers. Host it in the cloud or inside your own walls, with any model provider. Get in touch.

What it does

Open format

A wizard is one JSON document. Anyone may read, write and run wizards in their own tools. The wizard format and the runner that shows a wizard to people are under the MIT licence, and so are four libraries in packages/: the decision model, the search index, search on libSQL and Goose, the copilot. The studio, the runtime and the rest are source available under FSL-1.1-MIT. Each release of them becomes MIT two years after it is published. License says which file is under which.

How it looks

The studio: the conversation that changes a wizard, the preview of each page as people see it, and its steps as a flow with its branches.

A wizard in use: it asks on one page, then on the next, and ends with the result to download or share.

Get started

Create an account at engenty.ai: it is in a closed beta, so “Get started” takes your email for an invitation, or your invitation code. Signed in, the studio asks who your team is, reads your website into your space, takes first documents, and you build your first wizard in the browser. Links run day and night on the team’s credits.

Install it on your computer

The same account on your own machine: it runs on your Claude, ChatGPT, Google AI or Cursor subscription, reads your local files and browses for you, and what you publish keeps running in the cloud behind the same links. On macOS or Linux, in a terminal:

curl -fsSL https://engenty.ai/install.sh | bash

On Windows, in PowerShell: irm https://engenty.ai/install.ps1 | iex. With Node.js 20 or newer on the machine, npx wizards does the same.

It needs nothing but curl and tar and asks for no password. It brings its own Node.js (a Node you already have is left alone), installs engenty wizards into ~/.engenty/wizards and then walks you through the rest:

• Your engenty account. “Sign in or create your account now?”: a link and a code, and the browser for the sign-in, which also creates a new account (with your invitation code); what you publish on this computer then runs in the cloud too. Esc while it waits, or say no, and wizards account sign-in does it any time. --local leaves the account out: nothing leaves the computer, links answer on it only (local only).
• An AI client. It looks for Codex, Claude Code, Antigravity and Cursor Agent and offers to install the missing ones – first the one whose app (Claude, ChatGPT, Antigravity, Cursor) is on your Mac, as its subscription is most likely yours already. The first page of the studio installs them too. Or use your account’s credits, an API key, Ollama or, on a Mac with Apple silicon and macOS 26, Apple Intelligence.
• A browser and ffmpeg (optional): Chrome, Edge or Chromium (on a Mac also Brave) for PDF and PNG exports and for steps that use a browser, ffmpeg for MP4 videos of animated widgets.
• Start at login (optional): engenty wizards starts in the background when you log in (a LaunchAgent on a Mac, a systemd user service on Linux, a scheduled task on Windows). Links and AI clients then always reach it. wizards autostart on or off switches it later.
• Your AI apps (optional): adds your wizards to the AI apps it finds, such as Claude Desktop, Cursor or Codex. wizards connect and wizards disconnect do it later.

It starts nothing. Afterwards:

wizards start

starts it and opens the studio; it runs while that terminal is open, Ctrl-C stops it. wizards start -s starts it in the background instead. wizards opens a menu with all of it, wizards status says what is installed and what runs, wizards setup runs the guided setup again, and wizards --help lists the rest.

Install it as an app. In Chrome or Edge the studio installs as an app: its own window, its own icon in the Dock or taskbar. The studio offers it in a banner at the top; or click the install icon at the right of the address bar.

From source

You need Node.js 24.11 or newer and pnpm 10 (corepack enable; Node 25 and later ship without corepack, so install it first with npm install -g corepack).

git clone https://github.com/engenty/wizards.git
cd wizards
pnpm install
pnpm build
pnpm wizards

pnpm may list “Ignored build scripts” – that’s expected, nothing to do.

Code steps (shell, Node, Python) run in a sandbox. wizards sandbox add installs one (about 480 MB on a Mac, 820 MB on Linux, no Docker needed); a checkout also uses Docker when the sandbox image is on the machine.

First start

The setup shows your account at the top and asks how the AI should run. Pick an AI client that is installed on your computer – engenty uses your existing subscription and signs it in for you in a terminal right on the page – your account’s credits, Apple Intelligence on a Mac that has it, or your own keys. It stays until a test call works, then the studio opens.

Change it later under Settings → Models.

Using it

The user guide shows how to build a wizard, test and share it, connect services and use it from your AI apps.

Your data

Everything lives in ~/.engenty/wizards/data: the databases and the files of your wizards and results. Keys you enter are kept in the Keychain on a Mac, elsewhere encrypted in that folder. Back up the folder; delete it to start over. Settings go into ~/.engenty/wizards/.env (see .env.example); ENGENTY_HOME moves the whole install.

pnpm start in a checkout keeps its data in the data folder next to the code instead.

Update

engenty wizards updates itself: it looks for a new version at start and once a day, downloads it in the background and switches to it at the next start, never during a run. wizards autoupdate off turns that off. To update now:

wizards update

This runs the installer again without its questions. The studio says in a banner at the top when a newer release is out and can run the same update from there or from its About dialog. From source: git pull, pnpm install, pnpm build. Your data is brought up to date at the next start.

To remove engenty wizards, run wizards uninstall (or npx wizards uninstall if the command no longer runs). It shows what it will remove first: your AI apps' entries, the login item, every running process, ~/.engenty/wizards and ~/.local/bin/wizards. Your data in ~/.engenty/wizards/data is kept unless you choose to delete it, and a reinstall picks it up again. --dry-run only shows the plan, --yes runs it without questions and keeps your data, and --yes --delete-data deletes your data too.

Run it on a server

With Docker, behind an https proxy (Caddy, Traefik, Coolify):

git clone https://github.com/engenty/wizards.git
cd wizards
cp .env.example .env    # set APP_URL to your https address and add a model key
docker compose up -d --build
docker compose logs     # shows the link to open once

Point the proxy at port 8891. On a server, engenty uses your keys or Ollama; the AI clients’ subscriptions live on your own computer. The code sandbox is off in this image.

Troubleshooting

More settings: .env.example.

For developers

Development server, checks, how it is built, the managed mode: docs/development.md.

Plugins add tools for steps, routes with their own tables and pages in the studio, without changing the runtime: docs/content/dev/plugins.

The documentation – a user guide and the developer docs – is in docs/content and is a site of its own: pnpm dev:docs.

License

The wizard format, the runner and four libraries are MIT. The rest of the code is FSL-1.1-MIT.

FSL-1.1-MIT is source available: use it for yourself and your company, run it, change it – just don’t offer it to others as a competing product or service. Each release becomes MIT two years after it is published. No licence grants rights to the name or the figures as a brand.

COPYRIGHTS.md lists the exact files under each licence. It also names the owner, the marks, what is not ours and the third-party software with its licences. In the docs: Licensing.