One command.
A finished Next.js app.

npm create lacspace-app@latest — no config files, no setup wizard.

Nine templates. Choose static (frontend only) or --fullstack — frontend plus a real Node/Express/MongoDB/Redis backend. Then layer on any of 18 add-ons with --with ui,dataviz,ai-chat,rag,content,search,auth-pages,analytics,payments,email,i18n,quality,uploads,push,captcha,notify,realtime,consent — AI, content, search, auth, analytics, payments, email, i18n, quality gates, uploads, web push, keyless CAPTCHA, in-app toasts, live SSE updates and GDPR cookie consent, all keyless and free-first. The backend ones auto-upgrade a static project to full-stack.

cd in, npm run dev, and you’re live — SEO, headers and (optionally) a working AI already wired.

create-next-app gives you a blank starter. create-lacspace-app gives you a finished-looking app — a polished template already wired with SEO, security headers, a sitemap, robots.txt, a dynamic OG image and a working contact form — and then lets you layer on real features with one flag, including a streaming AI chat or chat-with-your-docs RAG that runs free and keyless on local Ollama.

Yes — free and open under the Lacspace Free Licence. The app you generate is entirely yours, including for commercial use.

Next.js 15 with the App Router, Tailwind CSS v4 and TypeScript. Every generated app builds cleanly out of the box.

Yes. Add --fullstack (alias --dynamic) and you get an npm-workspaces monorepo: a frontend/ Next.js app wired to the API with a typed client and /login, /register and a protected /account page; a backend/ Node · Express · MongoDB · Redis · TypeScript API that boots as a working app — JWT auth (register, login, me) and an example per-user CRUD resource, built on zero-dep @lacspace packages (password hashing, JWT, request validation, typed env, rate limiting); and a shared types/ package imported by both, so the API contract can't drift. One npm install, one npm run dev runs the API on :4000 and the frontend on :3000 together. The default stays static — a single frontend-only app, byte-for-byte unchanged.

No. Redis is optional — with no REDIS_URL the API falls back to an in-memory cache automatically, so it runs with zero infrastructure. MongoDB can be a local Docker container (a root docker-compose.yml with Mongo + Redis is included) or a free Atlas cluster. It's keyless and free-first, and every generated file carries plain "how this works" teaching comments.

Yes. The blog template turns content/posts/*.md into statically-generated pages with Article JSON-LD and sitemap entries (via @lacspace/markdown); the docs template is a full Markdown docs site with an auto-generated sidebar, on-this-page TOC and prev/next.

Yes — npx create-lacspace-app add pricing faq testimonials drops prewired, themed sections into any page, and add ai-chat or rag drops a whole feature (routes, UI and wiring) into an existing project. The UI kit installs automatically if it’s missing.

Eighteen, each requested with --with <key>, in the interactive picker, or add <key> later: ui (the 96-component @lacspace/components library wired into your template — a working settings page, a <UIProvider>, and a theme bridge that maps your template’s accent, radius and font onto the kit so the components match your brand and follow the dark-mode toggle you already have), dataviz (a working /insights page built from @lacspace/charts, @lacspace/table and @lacspace/date — stat cards, an SVG line chart, a date-range filter and a sortable, searchable, CSV-exportable table), ai-chat (a streaming AI chat on local Ollama), rag (chat with your Markdown docs, grounded with sources), content (a /updates Markdown section + RSS + llms.txt), search (instant keyless BM25 full-text search over your Markdown — no key, no service), auth-pages (profile editing, password change and TOTP 2FA on the built-in login), analytics (privacy-first, cookieless web analytics with a MongoDB collector and dashboard), payments (checkout wired to eSewa & Khalti with integer-safe money), email (transactional email with ready templates that logs to the console until you add SMTP), i18n (a dependency-free t() translator and language switcher with English + Nepali locales), quality (one-command quality gates — bundle-size budget, dependency audit and fake fixtures — plus a GitHub Actions CI workflow), uploads (authenticated file uploads stored in MongoDB, served via signed, expiring URLs — no S3), push (web push notifications with zero vendors — VAPID + RFC 8291 over pure Web Crypto, a subscribe endpoint and a service worker, no Firebase/FCM), captcha (a keyless, privacy-friendly proof-of-work CAPTCHA — stateless HMAC-signed challenges verified with one server secret, no Google or Cloudflare), notify (beautiful in-app toast notifications — a <Toaster/> and useToast with promise toasts, actions and positions), realtime (live updates over Server-Sent Events — a channel/room hub, a Web-standard stream endpoint and a typed browser client with a useSSE hook, no WebSocket server or vendor) and consent (GDPR-friendly cookie consent — a framework-agnostic store, a drop-in banner and a React <ConsentBanner/> + useConsent, cookie-persisted with whenConsent() script-gating). Everything is keyless and free-first; the backend ones — auth-pages, analytics, payments, email, uploads, push, captcha and realtime — contribute real Express routes and models via a route manifest and auto-upgrade a static project to full-stack, while ui, dataviz, i18n, quality, notify and consent are frontend/tooling add-ons.

Yes — a recipe bundles a template, full-stack mode and a set of add-ons into one command. npx create-lacspace-app my-app --recipe ai-saas gives you a SaaS with accounts, payments, a streaming AI chat and analytics; other recipes are store (e-commerce + eSewa/Khalti checkout + email + analytics), blog (content + search), docs-ai (RAG + search) and internal-tool (dashboard + auth + analytics + email). Explicit --template, --with and --fullstack flags still merge on top.

No. The ai-chat and rag add-ons default to local Ollama — free, keyless and private (nothing leaves your machine). Install Ollama, pull a model, and it just works; or set LACSPACE_AI_* to point at any hosted OpenAI-compatible model instead. It’s built on the zero-dependency @lacspace AI packages, so the code is yours to read and own.