StudyBuddy Studio

An internal authoring and ingestion pipeline for the StudyBuddy AI QBank — a browser-based studio for scaffolding medical exam questions and a validated batch ingester that pushes them (and their media) into Supabase.

What it does

• Studio (studio.html / studio.ts) — a browser UI for authoring structured QBank questions against a defined subject/domain/topic taxonomy.
• Scaffolder (scaffoldQuestion.ts) — generates a correctly-shaped question template so every question conforms to the schema from the start.
• Ingester (ingestQuestions.ts) — validates each question JSON (external ID format, difficulty, reasoning order, five options, correct answer, explanation, teaching point), verifies and uploads associated media to the qbank-media storage bucket with license metadata, writes to the database, and logs any failures for re-runs.

Question schema (enforced at ingest)

Each question carries: external_id (e.g. CV-PATH-015), subject, domain, topic, difficulty (Easy/Medium/Hard), reasoning order (1st/2nd/3rd), competency, stem, five options, correct option, explanation, teaching point, and optional licensed media.

Tech stack

TypeScript · Node.js · Supabase JS · dotenv

Running

npm install
# create .env with SUPABASE_URL and SUPABASE_SERVICE_ROLE_KEY
npx tsx scripts/src/ingestQuestions.ts

The ingester uses the Supabase service_role key. It is loaded from a gitignored .env and must never be committed or exposed client-side.