Why is the kit free and the generator paid?
The kit is code you download once, so we give it away to anyone with an account. The generator holds most of the knowledge in this product and it keeps moving — running it as a service means a fix reaches you the day it is made, not at the next release. What you pay for is new modules; what you receive is ordinary source that is entirely yours. Both cover unlimited projects, your own and your clients’ — the one thing you may not do is resell the kit itself.
Can I try it first, and what if it does not fit?
Yes, on your own schema: get the kit free, then press “Get my free generator key” in your portal — 10 modules over 30 days, renewed every 30 days while the beta runs, generated straight into your project, and they stay yours whether or not you subscribe. If you subscribe and have not generated a module yet, write to us within 14 days of the first payment and we refund it, no reason needed. Cancel any time from your portal; what you generated stays yours.
Is my database sent to your server?
Its shape, not its contents. Sent: table and column names, types, indexes, foreign keys, check constraints, column comments and a row count per table — plus the files the generator edits or must leave alone: the module’s existing files, the shared files it wires into (routes, seeder, sidebar, search) and your translated labels. Never sent: a single row of data, your .env or database credentials, or any other file. What collects it is two small classes inside your own project, under 500 lines together — open them and read them.
If the service goes away, what happens to my code?
It keeps running, forever. A generated module is ordinary source in your repository — zero runtime dependency on our server. Only generating new or changed modules needs an active generator key, free or paid.
What if I need to change a generated module?
Change the schema and regenerate — refining a module is free within your current period. Files that are genuinely yours, such as translated labels, are never overwritten. For changes the schema cannot express, the code is entirely yours; edit it like any other code.
Why this when Filament and Laravel’s React starter kit are free?
Both are excellent, and sometimes they are the right choice. Filament is the better fit if your team writes Livewire: it builds admin screens from your columns and ships multi-factor sign-in, multi-tenancy and imports at no cost. Laravel’s React starter kit gives you sign-in and two-factor on the same React 19 + shadcn/ui stack this kit uses — a clean foundation, with no admin layer on top.
This kit is for React teams, and for what comes after sign-in: roles with a permission matrix, an audit trail with field-level diffs, passkeys and Google/GitHub sign-in handled inside your own app rather than by a hosted auth service, exports, and a production Docker stack. On top of that, the generator writes ordinary Laravel for every module — controllers, form requests, a REST API with its OpenAPI document, and 50 or more Pest tests. What it does not have is multi-tenancy: if you need tenants on day one, Filament is ahead of it there.
When is an AI assistant better?
For one unusual screen that follows no pattern — a bespoke dashboard, a one-off report — an assistant will get you there faster, and we would rather say so than have you find out after paying. This is built for the thirty screens that do follow a pattern, and for keeping them identical when five people build them.
Which databases are supported?
PostgreSQL, MySQL 8.4 and MariaDB 11. PostgreSQL is the engine the kit is built on and the one where every feature exists. MySQL and MariaDB run the product — install, migrations, CRUD, filters, exports, roles, the audit trail, two-factor, the REST API, and the generator itself: make:crud reads a MySQL schema as well as a PostgreSQL one — with four things that are simply absent there: array columns and the filter that reads them, the inet and macaddr types, trigram search (MySQL gets a plain LIKE: fine on a small table, a full scan on a large one), and reusing the email address of a soft-deleted account. Measured, not assumed: before a release the bundle is installed from nothing in a clean virtual machine four ways — Docker and native, each on PostgreSQL and on MySQL/MariaDB — and the suite is run on each engine there. For 2.9.1: 2,441 passing on PostgreSQL and 2,348 on MariaDB, and 971 on MySQL after the example modules are removed — none failing on any.
Does it work with our company identity provider?
Today: Google and GitHub, plus two-factor and passkeys. That is consumer SSO — it is not SAML or OIDC against Okta, Entra ID (Azure AD) or Keycloak, and there is no SCIM provisioning. If your client’s IT department hands you a metadata URL, this kit does not read it yet. Need one of those? Tell us which provider — it is buildable on the sign-in flow that is already there, and knowing which provider you actually use is what decides whether we build it.