Contributing

temetro is open source — contributions are welcome. The repository lives at github.com/temetro/temetro.

Development setup

You need Node.js ≥ 20 and a PostgreSQL instance (Docker is the easy way to get one).

git clone https://github.com/temetro/temetro.git
cd temetro

# Backend
cd backend
npm install
cp .env.example .env          # point DATABASE_URL at your Postgres
npm run db:migrate            # apply migrations
npm run dev                   # API on :4000 (tsx watch)

# Frontend (second terminal)
cd ../frontend
npm install
npm run dev                   # app on :3000 (Turbopack)

frontend/ and backend/ are independent apps — each has its own package.json and node_modules; always run npm from inside the app you're working on.

Useful backend commands

After changing the auth configuration, run the three generation steps in order: auth:generate → db:generate → db:migrate.

Things to know before writing code

• Customized Next.js 16 — the frontend runs a customized build whose conventions differ from public docs in places (e.g. proxy.ts instead of middleware.ts). Read the docs bundled at frontend/node_modules/next/dist/docs/ before writing routing code.
• COSS / Base UI, not Radix — components compose via the render prop, not asChild. Use semantic Tailwind tokens (bg-muted, border-border), never hardcoded colors.
• RBAC lives in backend/src/lib/access.ts — if you add a resource, define its permission statements there and mirror them in frontend/lib/access.ts.

A good first contribution: the real AI chat

The chat UI is complete, but free-form messages get a mock reply generated in the frontend (components/chat/chat-panel.tsx). The planned next step is a /chat endpoint on the backend that:

1. receives the conversation and the active patient context,
2. calls an LLM with tool access to the patients API,
3. streams the answer back to the existing chat UI.

If that sounds like fun, open an issue on GitHub to coordinate before you start.