Keep client data apart
Create one project per client app. example-project is only a fictional slug. A real client gets its own slug and its own database. The Data API will not read another project’s tables from that host. That is a sharper boundary than a shared database with a client_id column.
The Console lists the projects your developer account owns. CubiBase does not currently provide an invite API, so “the team” is not an automatic shared workspace. Plan account ownership with that limit in mind.
Hand off a host, not a bespoke server
Each app calls its own https://{slug}.api.cubibase.com/rest/v1/{table}. You can keep the schema work in the Console, expose only the operations that client needs, and rotate the publishable key without touching the other clients.
Questions
Can a client log in to the Console and edit their own tables?
Only if they have a CubiBase developer account that owns the project. There is no separate “client viewer” role in the current product, and there is no invite flow to add one.
Can I delete a client project when the engagement ends?
The project owner can delete a project through the Console’s control API. Confirm export and backup behavior before you promise a client a retention period. Those periods are not published.