Databases
Create an isolated Postgres or Redis instance tied to your project — once it's ready, the connection info is automatically injected into your next deploy.
How to create one
Open your project in the dashboard and switch to the Databases tab. Choose the engine (Postgres or Redis), enter a name, and click Create. The request returns immediately; the database is provisioned in the background — the status badge moves through Provisioning → Ready (or Failed if something goes wrong). The dashboard refreshes automatically until it's ready, so you don't need to reload the page yourself.
Isolation
Each database is a separate instance in the project's own namespace — not a schema on a shared cluster. You can't reach another user's or another project's database at the network level.
Automatic injection
Once a database reaches the Ready state, a connection URL is automatically generated per engine: Postgres → DATABASE_URL, Redis → REDIS_URL. These two keys are reserved, so you can't set them by hand — they're injected automatically on your next deploy.
If you create more than one database for the same engine (only possible on paid plans), only the first database created for that engine is used — which one connects doesn't change from deploy to deploy.
Getting the connection URL
In the database list, click Show connection info on a record in the Ready state — the full URL, including the password, appears; use Copy to copy it to your clipboard (for example, to connect from your local development environment). Because this URL contains a password, keep it hidden while screen-sharing or taking notes.
Free tier limit
On the free tier you can create at most 1 database across your whole account (not per project — across all your projects combined). Once the limit is reached, new database creation requests are rejected.