---
title: Create an AI Agent App
description: Create an AI agent app with Next.js, AI CLI workflow docs, MCP, skills, Better Auth, Drizzle, SQLite, Tailwind CSS, and shadcn/ui.
updated: 2026-05-21
translationStatus: pending
category: Packs
tags:
  - ai-agent
  - nextjs
  - mcp
  - ai-cli
  - skills
keywords:
  - create ai agent app
  - build ai agent app
  - nextjs ai agent starter
  - ai cli starter
  - mcp app starter
---

Use the AI Agent App pack when you want the application and the agent workflow to start together. It prepares a Next.js project plus AI CLI-oriented documentation and local agent tooling.

<StarterPackCta id="ai-agent-app" />

## Generate the AI agent starter

Open the [AI Agent App pack in the Stack Builder](/new?preset=ai-cli-agent-workbench&view=command), or run:

```bash
bun create better-fullstack@latest my-ai-agent-app \
  --ecosystem typescript \
  --frontend next \
  --backend self \
  --runtime none \
  --database sqlite \
  --orm drizzle \
  --auth better-auth \
  --api trpc \
  --ai ai-cli \
  --part codeQuality:universal:biome \
  --part aiTooling:universal:ruler \
  --part workspaceRunner:universal:turborepo \
  --part aiTooling:universal:mcp \
  --part aiTooling:universal:skills \
  --ai-docs claude-md agents-md cursorrules
```

## What the pack includes

- Next.js for the product surface.
- SQLite and Drizzle for local-first persistence.
- Better Auth for accounts and sessions.
- AI CLI setup, MCP support, skills, and generated agent docs.
- Biome and Ruler so agent instructions and code quality live with the repo.

## First things to customize

1. Define the user-facing agent job: research, coding, support, automation, or analysis.
2. Add the app data model the agent is allowed to read and write.
3. Tighten MCP/tool instructions before connecting external services.
4. Move from SQLite to a managed database when collaboration or production hosting needs it.

## When this is the right pack

Choose this for AI-assisted developer tools, internal automation, agent dashboards, workflow copilots, or experiments where app structure and agent instructions should evolve together.

For the stack-level walkthrough, read [Next.js AI CLI Agent Workbench](/guides/ai/nextjs-ai-cli-agent-workbench/).