Trust Model

Data And Safety

The project uses Postgres for core records, explicit retention cleanup for generated history, and safety-first wording for medical output.

Key Points

Core records live in Neon/Postgres.

Generated/history rows can be cleaned automatically by scheduled jobs.

AI guidance is framed as support, not diagnosis.

1

Stored Data

The backend stores users, health profiles, budget plans, exercise plans, food items, and clinic records in Postgres.

User auth uses backend-issued access and refresh tokens.

Frontend sessions are stored in HTTP-only cookies.

Clinic seed data is now database-backed with CSV fallback.

2

Retention

A cleanup script can remove old generated/history rows after configured retention windows while keeping each user's latest record by default.

Exercise plans default to 30 days.

Budget plans default to 180 days.

Health profiles default to 365 days.

3

Medical Safety

The product is decision support. It should guide users toward qualified care for emergencies or concerning patterns and avoid diagnosis or medication claims.