[damian123]

Summary

Production Postgres RLS hid workspace rows after Next.js dropped AsyncLocalStorage across await. That made a live Google Calendar connection look unavailable and caused availability saves to fail when updating the organizer timezone.

• Rebind signed tenant context with runWithDatabaseContext / reconstructed enterDatabaseAction scope on Google status, OAuth token persist, availability, onboarding, and related writes.
• Persist Google token refreshes under oauth_write, and update user timezone under account_write after the availability schedule write.
• Copy the Prisma query engine into the standalone web path so production signup/readiness can load.
• Log Prisma P#### codes on unhandled API errors without leaking tokens.

Verified on book.XXX.xyz: Integrations reports Google connected with complete calendar access, and PUT /api/availability returns 200.

Test plan

• Sign in on a production-path Postgres deploy and confirm Integrations shows Google Connected / full calendar access without reconnecting.
• Save weekly hours on /availability and confirm Availability saved.
• Complete onboarding if a new workspace is created.
• vitest run apps/web/src/server/db-context.test.ts apps/web/src/server/services/availability.test.ts apps/web/src/server/services/calendar.test.ts

[damian123]

Rebased check: this branch is already one commit ahead of current main (no additional rebase needed).

Local verification just re-ran and passed:

vitest run apps/web/src/server/db-context.test.ts \
  apps/web/src/server/services/availability.test.ts \
  apps/web/src/server/services/calendar.test.ts

39 passed / 0 failed.

Ready for review when you have time.