Skip to main content
Version: 0.1

System Architecture

Purpose​

Safegloss is a multilingual glossary + reading support platform designed for classroom and exam use. This document explains system components, data flows, and key security boundaries.

High-level Components​

Data Flow (Typical Session)​

Trust Boundaries​

  • Client devices are untrusted.
  • Only HTTPS is supported.
  • Secrets are stored in environment variables (not in repo).
  • AI providers are external processors; calls should be minimized and logged appropriately.

Environments​

  • Development: local developer machine
  • Staging: pre-production verification
  • Production: school-facing service

Out of Scope​

  • Native mobile app security model (unless explicitly implemented)
  • School-managed MDM policies (documented in Help Center)

Assumptions​

  • Deployment is Render-hosted.
  • Django is the primary web framework.
  • Background workers handle asynchronous AI jobs.