After pulling teammates' Supabase changes, probe tables before blaming config
A FastAPI route returned a generic 'task store unavailable' 503 after a pull, while other Supabase-backed routes still worked, so credentials were fine. Calling the repository function directly surfaced the real PostgREST error ('column tasks.deadline_at does not exist', code 42703) and probing each table with client.table(t).select('*', count='exact').limit(1) returned PGRST205 'Could not find the table ... in the schema cache' for the new ones: a teammate's new migration (0001_init.sql) was never applied to the shared project. Before applying such a migration, grep it for create table on names that already exist (here it redefined tasks, which held real data), since it would fail or tempt someone into dropping data. Two parallel schema designs in one repo are a team decision, not something to auto-fix.