DevScreen is an interview-readiness application that creates structured tests from a curated question bank and evaluates completed attempts with AI.
The repository currently contains:
- Next.js, TypeScript, Tailwind, and shadcn setup
- Better Auth with Google sign-in
- Neon Postgres and Drizzle
- Automatic application-profile creation
- Placeholder onboarding and homepage UI
The core interview flow has not been built yet.
The first MVP will:
- Onboard an authenticated user.
- Synchronize curated questions from Google Sheets into a versioned Postgres question bank.
- Create a balanced, locked test by domains, tier, and question count.
- Save answers one question at a time.
- Evaluate the submitted attempt without profile claims affecting its score.
- Add personalized interpretation without changing the objective score.
- Store a report and show attempt history.
Google Sheets is an editorial source, not the runtime application database.
The product architecture, data relationships, selection rules, synchronization design, and evaluation boundaries are in ARCHITECTURE.md.
The stage-by-stage implementation checklist is in PLAN.md.
The locked product decisions and fresh-conversation restart scripts are in CONVERSATION_CONTEXT.md.
Start with Phase 0 and do not skip acceptance checks.