A minimal AI chat application built with the Anuma SDK, Privy for authentication, and Next.js. The entire app fits in a single file — app/page.tsx — making it a good starting point for understanding how the pieces connect.

Sign in at dashboard.anuma.ai and create an app. This provisions the API account that powers AI responses.

git clone https://github.com/anuma-ai/starter-mini.git
cd starter-mini
pnpm install

Create a .env.local file in the project root:

NEXT_PUBLIC_API_URL=https://portal.anuma.ai
NEXT_PUBLIC_PRIVY_APP_ID=<your-privy-app-id>

pnpm dev

Open http://localhost:3000 to see the app. After signing in through Privy you'll get a chat interface where you can talk to an AI model.

Setup covers the Next.js configuration — wrapping your config with withAnuma to set up the API proxy the SDK needs.

Authentication explains the Privy integration: the provider, the auth gate that switches between login and chat, and how identity tokens are passed to the SDK.

Chat introduces the useChat hook from @anuma/sdk/react, which manages the connection to the Anuma backend, and the message state model.

Streaming describes how response text arrives in chunks through the onData callback and gets rendered as markdown in real time.

Sending Messages walks through the submit handler that formats messages for the API and sends the full conversation history with each request.