"Runs on-device" is the easiest claim in software to make and one of the hardest to check. So here's the boring version: a list of everything SKI sends over the network, and everything it doesn't.
The short version:
Your voice never leaves the machine. Recognition and playback both run locally.
Your code never passes through us. SKI is a skill your agent loads, not a proxy in the middle.
Four things touch the network, and none of them carry audio, code, or transcripts.
What never leaves
Speech-to-text runs entirely on your machine. Your spoken input becomes text before anything else happens to it, there's no upload step where audio is sent somewhere to be transcribed and sent back.
Text-to-speech runs locally too, using one of four on-device neural voices. The reply you hear is synthesized on your own hardware.
Both halves of the loop work with no network connection at all. That's the practical test, and it's the one worth running yourself: pull the wifi and keep talking to your agent.
Why we're not in the middle of your code
This is the part people assume wrong, and it's the more important half. SKI doesn't sit between you and Claude Code's or Codex's model, it installs a shared skill your agent already knows how to read, teaching it that voice I/O exists...
Read the full article here: heyski.io/blog/is-voice-coding-private
Top comments (0)