AI textarea with a toolbar that includes a send button and optional attachments and voice input buttons.
Set showVoiceInputButton to add a microphone to the toolbar. Tap it and speak: dictation shows as grey preview text and commits to the form control when you confirm. Cancel discards the whole session.
The transcript is shown as the browser reports it. Thirty seconds without new speech ends the session and commits what you said; you can also cancel or confirm at any time.
The microphone only renders in Chrome (or Chromium) where the Web Speech API is available, so it does not appear in Firefox, Safari, or Edge.
This field starts with a value to show how dictation joins existing text at the caret.
Speech lands at the caret rather than always at the end, so click into the middle of the sentence before dictating to splice it in there.
False = static submit button (no expand/sweep or label fade-in).
Per-file upload progress; matched by `File` reference. Requires `showAttachmentsButton`.
Only used when `showAttachmentsButton` is true.
Replaces submit content with a spinner and disables the button.
Height in px of the submit button.
Horizontal padding in px of the static (`animateButton: false`) submit button.
Submit expands to a labelled pill once `minCharacters` is reached.
Textarea font size (px) on tablet and up.
Textarea font size (px) on mobile.
Removes the outer glow only; gradient border stays.
Disables submit only; textarea and other controls stay interactive.
Prevents submit from stealing focus so the virtual keyboard stays open.
Scrollable line count on desktop/tablet.
Scrollable line count on mobile (>= 400px).
Scrollable line count on small mobile (<400px).
Character threshold for submit expand (requires `buttonText`).
Lines shown at rest before auto-grow, including the stacked toolbar row. Set to 1 for a single-line field with the button inline.
False = label-only submit in every state.
Toolbar text button before submit; pair with `onSkip`.
Rounded-rectangle submit instead of round/pill.
BCP-47 tag for voice recognition. English-only for V1.