Browse needs a database.

This environment has no Postgres to read from. Locally the catalog runs on PGlite, a real Postgres compiled to WebAssembly, kept in .pglite/ and built from supabase/migrations plus supabase/seed.sql on the first request. That needs a writable filesystem, which a serverless deploy does not have.

The schema is finished and tested. What is missing is a hosted Postgres, which is a decision about a Supabase project rather than about code. Run npm run dev to browse the seeded catalog now.