This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Hello Realtime is an OpenAI Realtime API demo application supporting WebRTC users. It's built on Val Town's Deno runtime with TypeScript and uses the Hono framework for routing.

• Run locally: Deploy to Val Town platform - no local dev server
• Test deployment: Visit https://hello-realtime.val.run for WebRTC
• Format Run deno fmt

1. Main Entry Point (main.tsx)

   • Hono app with route mounting
   • Routes: /rtc, /sip, /observer
   • Serves frontend from /frontend/index.html

2. WebRTC Route (routes/rtc.ts)

   • Creates OpenAI Realtime calls via POST to /rtc
   • Handles SDP offer/answer exchange
   • Triggers observer WebSocket connection

3. Observer Route (routes/observer.ts)

   • Establishes WebSocket connection to OpenAI
   • Monitors call events and transcripts
   • Auto-sends response.create after connection

4. Utilities (routes/utils.ts)

   • makeHeaders(): Adds OpenAI API authentication
   • makeSession(): Configures Realtime session with model, instructions, and voice settings

• Single-page vanilla JavaScript application (frontend/index.html)
• WebRTC peer connection management
• Real-time status indicators and logging
• No external dependencies or build process

Required:

• OPENAI_API_KEY: For all OpenAI API calls

• Uses Val Town's file reading utilities from https://esm.town/v/std/utils@85-main/index.ts
• Deno runtime with TypeScript support
• No local development server - deploy directly to Val Town
• Import npm packages via npm: prefix (e.g., npm:hono)

• OpenAI Realtime model: gpt-realtime
• Voice: marin (configurable in utils.ts)
• Audio configuration includes noise reduction for near-field devices
• WebSocket messages are logged except for audio transcript deltas
• Observer automatically triggers 250ms after connection