Feature tour

Everything is organised around projects. Each project has its own stack, its own data, and a management surface for the parts of Supabase you use every day.

How this tour is organised

Each area has its own deep page with the workflow, real screens, behaviour and the configuration names that apply. The tour follows the product's own navigation: a dashboard over projects, then the per-project workbench, then the manager-level administration surfaces.

Two areas are described but not given a deep page: edge functions and the Postgres integration panels. The code for both exists, but there is no clean, approved screen in the public screenshot set, and a page without a real screen would be padding. They are listed here so the tour stays honest about what the product contains.

One dashboard over every stack

Manager dashboard in dark theme showing project totals, system health cards for Docker, disk space and memory, quick actions, and a recent activity feed.
dashboard
The control room: project counts, system health, quick actions and recent activity.

Described but not given a deep page

Edge functions. The Functions page lets you pick a function by id or name, edit its code, deploy through the Supabase CLI, invoke it with a method, headers and a JSON body, manage secrets with masked values, and follow the edge-runtime log stream. The behaviour is documented from the source, but no approved screen shows the function studio, so there is no deep page and no image slot.

Integrations. The Integrations page routes to panels for cron jobs, queues, Vault and foreign-data wrappers, plus a GraphiQL panel. Again, the source is real and the panels are documented, but the approved screenshot set has no clean capture of them. They are named in the project settings page rather than given feature pages of their own.

What the tour leaves out

The public site describes the product's user-facing behaviour. It does not publish internal audit material, verification logs, scenario files or agent configuration. Claims on each page are tied to the source files that prove them; if a claim could not be tied to code or user documentation, it was removed rather than softened.