altic-dev/FluidVoice — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a polished macOS dictation app called FluidVoice. I want it to live in the menu bar and let me press a global hotkey, speak, and have the words typed into whatever app or text box I’m using.

It should work mostly offline and feel very fast, with a small overlay that shows live transcription while I’m talking. Add a settings area where I can choose different speech models, use Apple Speech as a built in option, and support Whisper for broader Mac compatibility. It should ask for microphone and accessibility permissions clearly.

Please include a write mode for dictating or rewriting text anywhere, plus a command mode where I can say actions for the Mac. Add a history screen with usage stats so I can see past dictations. Also include optional AI cleanup using providers like OpenAI, Groq, or a custom provider, with keys stored securely in Keychain.

Make it feel native, simple, and reliable on macOS. Look up current Apple docs if needed.

Want more depth? Deep Reverse