voice-shelllisted

Let the user send prompts by voice. Start a resident process that keeps listening to the microphone, take what the user says through Monitor, and treat it as an instruction. Use it when the user says "voice mode", "talk to me", "hands-free", "dictate my prompts", or "speak instead of typing", in whatever language they say it. To stop, "stop voice mode". If the user asks to set voice-shell up ("set up voice-shell"), follow SETUP.md, work out which environment this is (macOS 26 or newer, or not), and guide them from there.
ykuwai/voice-shell · ★ 2 · AI & Automation · score 74

Install: claude install-skill ykuwai/voice-shell

# Voice prompt mode Take what the user speaks and receive it as a prompt, without the keyboard. The part that listens to the microphone appends one line to a JSONL file every time an utterance is finalized. Follow that log with Monitor and treat each line that arrives as an instruction from the user. **There are three ways of recognizing speech.** These are all you can pick from, there is no other. | Name | What it is | Where the audio goes | |---|---|---| | `browser` (default) | Chrome's Web Speech API. Works with nothing installed | **Google's servers** | | `apple` | On-device recognition that ships with macOS 26. Light | Only inside this machine | | `whisper` | faster-whisper. Strong on proper nouns | Only inside this machine | The argument is `$ARGUMENTS` (`start` / `stop` / `status` / `setup`. `start` when omitted) ## When it is not set up yet If `start` fails with `No Python it can run was found`, or if the user says "set it up", walk them through [SETUP.md](SETUP.md). On macOS 26 or newer there is nothing to install, anywhere else it means installing Whisper. Check first, then confirm which way to go. Do not install everything on your own. ## Starting **When in doubt, just run `start`.** It remembers which way of recognizing was picked last time (`~/.config/voice-shell/config.json`), and the first time it is "This browser" (Chrome's Web Speech API). Nothing to install, so there is no wait. 1. Start it. ```bash ${CLAUDE_SKILL_DIR}/scripts/voice-shell.s