Get paperr running
on your own machine.

Prerequisites

• Node.js 22.5+ — the server uses the built-in node:sqlite module, which needs 22.5 or later.
• Python 3 (optional) — lets npm install auto-provision paperr's own bundled AI server. Skipped safely if Python isn't found; add it later and re-run npm install in server/ to enable it.
• Ollama or LM Studio (optional) — external local LLM providers, if you'd rather point dotAi at one of those instead of the bundled server.

Quick start

Fetch paperr and start the dev servers. install:all creates server/.env from the example and auto-generates real random JWT_SECRET / JWT_REFRESH_SECRET values — there's no manual copy/paste step, and it never overwrites secrets you've already set.

$ npx paperr             # fetches paperr into ./paperr
$ cd paperr
$ npm run install:all   # root + server + client deps, writes .env
$ npm run dev           # API :3000 · client :5173

$ git clone https://github.com/biswasprateek/paperr.git
$ cd paperr
$ npm run install:all   # root + server + client deps, writes .env
$ npm run dev           # API :3000 · client :5173

# (re-)generate server/.env only — safe, never overwrites set secrets
$ npm run setup:env

Both options land in the same place — the npm package is just a thin fetcher. paperr is a self-hosted app, not a library, so there's nothing to npm install into an existing project.

Bundled AI server

npm run install:all (which runs npm install inside server/) automatically provisions paperr's own local AI model via litert-lm — a small, efficient AI model that runs on as little as CPU in under 2GB of RAM, uses your iGPU or GPU when available — scaling to the hardware you have — and auto-offloads from memory when idle. It looks for a Python interpreter to build a project-local virtual environment at server/ai/litert/venv:

Production

Recommended for everyday use. Build the client once, then let Express serve the API and app together on port 3000.

# one click — builds the client, then serves on :3000
> ./start-paperr.ps1
> ./start-paperr.ps1 -NoBuild   # skip rebuild if current

# one click on macOS / Linux
$ ./start-paperr.sh

$ npm run build   # bundles the React app
$ npm start       # Express serves API + app on :3000

Then open http://<your-machine-ip>:3000 from any device on your network.

Network access

paperr binds to 0.0.0.0, so it's reachable from every device on your LAN. Find your local IP (ipconfig on Windows, ifconfig / ip a on Linux/macOS) and open:

$ http://192.168.x.x:3000

Configuration

server/.env is created automatically by npm run install:all, with the JWT secrets pre-generated. Edit it directly to adjust anything else:

Demo data

Want paperr pre-populated to explore? The seed script creates a fully-populated demo space — "The Reyes Family", 4 members, realistic tasks, projects, notes and routines across every section.

$ node scripts/seed-demo.cjs   # log in with password: paperrdemo1

Log in as any member with the password paperrdemo1 (maya is admin). Re-running it wipes and recreates just that demo space.

Tech stack

Project layout

Back to the overview →