---
title: "App CLI: Scaffold Production-Ready AI Applications"
slug: "app-cli-frontend-scaffolding"
date: "2025-03-10"
tag: "Application"
summary: "Interactive CLI for generating Next.js 15 applications with chat interfaces, authentication, and full ibl.ai SDK integration. Supports AI-assisted customization."
author: "ibl.ai Engineering"
repo: "iblai/vibe"
---

The **ibl.ai App CLI** is now available for scaffolding production-ready frontend applications. Run a single command to generate a complete Next.js 15 app with:

- **Chat interface** — Pre-built conversational UI components with streaming support
- **Authentication** — Login, registration, and session management out of the box
- **ibl.ai SDK integration** — Full platform SDK wired up and configured
- **AI-assisted customization** — Use Anthropic or OpenAI to customize the generated app via natural language

### Quick Start

```bash
npx iblai-app-cli create my-app
cd my-app
npm run dev
```

The CLI walks you through configuration: platform URL, authentication method, default mentor, theme, and more. The output is a standard Next.js 15 + React 19 + TypeScript project — no vendor lock-in on the frontend.

### Tech Stack

- Next.js 15 with App Router
- React 19
- TypeScript
- Tailwind CSS
- ibl.ai Platform SDK

### Why scaffold instead of embed

An embedded widget leaves the interface on someone else's server: you configure what the
vendor exposes and you ship what the vendor ships. The CLI takes the opposite approach — it
writes a complete application into your repository. The generated project is ordinary
Next.js source code, so it builds with your toolchain, deploys to your infrastructure, and
goes through your review process like any other frontend you maintain.

That matters because the frontend is where institutional identity lives. Branding, copy,
navigation, accessibility requirements, and the accessibility and privacy reviews that come
with them are all decisions you own outright, rather than tickets filed against a vendor
roadmap. Nothing in the generated app is obfuscated or licensed separately.

### Customizing with AI

Point the CLI at an Anthropic or OpenAI key and you can describe changes in plain language
instead of hand-editing components — restyle the chat surface, adjust the onboarding flow,
or wire in an extra page. The model edits the generated source directly, so every change
lands as a normal diff you can read, revise, or revert.

The scaffolded app talks to the platform through the ibl.ai SDK, which means it works the
same way against a hosted deployment, a VPC installation, or a fully air-gapped one. Point
it at a different platform URL and the same frontend runs there unchanged.

The same scaffold is now reachable without a terminal at all: [Code Mode](/updates/code-mode)
is the coding agent built into the Agentic OS chat box, and it materializes this template,
installs dependencies, and runs the app from a single prompt.

Once the app exists, shipping it is documented end to end in
[Deploy Your AI App: Web, Your Server, and the Stores](/updates/deploy-ai-app-web-server-app-stores)
— platform-hosted deployment with only your API key, the same app as a container on your own
infrastructure, and a generated Makefile that submits the Tauri build to the App Store and
Google Play.