1. Install
You can also run it without installing, with npx crescodb. The global install is worth it because everything after this is a cresco command.
2. Create a project
schema.cresco is the whole thing: your models, your fields, and who is allowed to read them. It is JSON, it lives in your repo, and it is the only place you describe your data.
3. Start the server
Every model now has full CRUD. GET, POST, PUT, PATCH and DELETE on the collection and on /:id, plus bulk insert, filtering, ordering and pagination. You did not write a route.
4. Add auth
Those are real files in your project that you own and can edit. Not a dependency you have to guess at. If you want to change how login works, you open the file.
5. Check it before it ships
This answers "who can read this table" from the schema in your repo, offline. It is worth running before you deploy anything, and cresco deploy refuses to ship on a critical finding.
Where to go next
The full command reference, the SDK, storage, realtime and deployment all live in the docs.