Supabase work crosses project configuration, data, functions, and the operational question that prompted the request. Strawberry can list organisations and projects, inspect tables and rows, describe schema, retrieve project and Edge Function details, list secrets and API keys, and execute supported row, secret, and SQL operations within the connected Supabase scope.
01
Understand the project surface before changing the table.
Data work goes wrong when a requested update arrives before anyone has checked the project, schema, or current rows. Strawberry can list Supabase projects and tables, describe a table, and select data so the change is grounded in the actual structure rather than a guessed column name.
02
Make every row operation visible before it becomes real.
Supabase’s native operations can insert, update, or delete rows, and those verbs need an explicit target set. Strawberry can select the relevant rows first, prepare the proposed values, and keep the identifiers and criteria visible so a reviewer can tell exactly what a write will affect.
03
Treat SQL and secrets as production work.
The available Supabase operations include running SQL, setting or deleting secrets, and retrieving project configuration, API keys, and Edge Function details. Those capabilities are powerful for a tightly defined maintenance task, but the query, project, and secret name should be inspected before permission is granted for a consequential action.
04
Put safe checks on a schedule, not schema changes.
Projects benefit from recurring checks for missing records, unexpected schema changes, or Edge Function inventory, but not from unattended mutation. Save the read-only check as a Supabase skill and use a routine to report defined exceptions without inserting, updating, deleting, or running unreviewed SQL.
Supabase Strawberry