az11k-dev/jarvis-app — reverse-engineered prompt

Build me a mobile app called JARVIS, a lightweight phone based voice assistant that feels like a smart companion. I want it to work on both iPhone and Android, listen to spoken commands, send them to OpenAI for understanding, and reply out loud with a natural voice. Let me switch between available system voices, and give it a clean, simple interface that works well on different screen sizes.

It should ask for Face ID, Touch ID, or the device passcode before opening. I also want practical voice commands like opening apps such as YouTube or Telegram, handling YouTube searches from speech, getting current info from the internet, remembering recent conversation context during a session, and setting reminders or notifications. If possible, include the developer style GitHub voice commands mentioned in the project, like creating a repo or showing recent commits.

Please wire up the OpenAI key through environment settings, make it runnable in Expo, and leave the code organized so it is easy to extend later. Look up current docs online if you need to.