docs(native-chat): correct why a slash command is inert in the answer row (#21111)

The previous note said running a command from the question card's free-text
row could only answer with command text or abandon the prompt. That is wrong
about skills, and silent on the real cause.

Verified against a live structured session: the typed answer is delivered
verbatim as the AskUserQuestion tool result, so it reaches the model but never
the command parser. A client-side command is therefore inert; a skill name can
still be acted on because the model simply reads it.
This commit is contained in:
Brennan Benson
2026-09-16 16:19:39 -07:00
committed by GitHub
parent 2e3a24c30f
commit 2fbdada551
@@ -192,10 +192,11 @@ export function NativeChatQuestionCard({
<span className="flex size-6 shrink-0 items-center justify-center rounded-md bg-muted text-muted-foreground">
<Pencil className="size-3.5" />
</span>
{/* Intentionally plain — no `/` or `@` grammar. This row answers the
question; a slash command addresses the session, so running one here
could only answer with command text or abandon the pending prompt.
That grammar belongs to the composer, which this card replaces. */}
{/* No `/` or `@` picker here — that autocomplete belongs to the composer,
which this card replaces. What you type is delivered verbatim as the
AskUserQuestion tool result: it reaches the model but never the command
parser, so `/compact` and friends are inert, while a skill name can
still be acted on. */}
<input
ref={answerInputRef}
disabled={isSubmitting}