Stickystein

Next.js + Bun + PostgreSQL/pgvector app for asking questions over indexed PDFs with citations.

Setup

1. Install dependencies: bun install
2. Copy envs: cp .env.example .env
3. Run migrations: bun run db:migrate

Bun Commands

• bun run dev Start the app in development mode.
• bun run build Build the app for production.
• bun run start Start the built app.
• bun run worker Start the background worker.
• bun run db:migrate Apply database migrations.
• bun run db:migrate:status Show migration status.
• bun run index:corpus Discover PDFs from PDF_DATA_DIR, chunk them, embed them, and store them in Postgres.
• bun run smoke:llm Run a simple LLM smoke test using the current LLM_* env vars.
• bun run test Run all tests.
• bun run test:unit Run unit tests.
• bun run test:integration Run integration tests.
• bun run test:e2e Run render/e2e-style tests.
• bun run check Run lint, typecheck, and tests.
• bun run scrape Run the scraping script.

Notes

• DATABASE_URL is required for migrations and runtime DB access.
• PDF_DATA_DIR controls which PDF folder bun run index:corpus scans.
• EMBEDDING_* env vars control indexing and query embedding.
• LLM_* env vars control answer generation.