Why voice coding
Two common reasons: RSI/wrist pain forcing time off the keyboard, and raw output speed. Experienced voice coders hit 90+ WPM dictating code, which beats most keyboard typists on prose-heavy code (comments, tests, docs).
What works today
In VSCode, Cursor, and Xcode: dictating identifiers in camelCase or snake_case, navigating ("go to line 42"), refactoring ("rename this to currentUser"), running tests ("save and run tests"), and writing comments + docstrings naturally.
What doesn't work yet
Pure symbol-heavy code (regex, complex generics) is still faster typed. Most voice coders use a mixed mode: voice for prose, identifiers, and navigation; keyboard for symbols.
Setup
Install Jarvis. Grant Accessibility permission. Open VSCode. Press hotkey, dictate. That's it. No per-language grammar files, no command vocabulary to memorize.