ui: the chat comeback — side-based authorship, restored quotes, tucked whys

Blockquotes: the callout override had displaced Streamdown's own styled
blockquote component, so plain quotes rendered as bare unstyled text
since that commit. The pass-through branch now restores Streamdown's
verbatim classes (border-l-4, muted italic).

Explanations tucked for ALL questions (owner): the prediction's Why /
Hide-why toggle is universal — every quiz answered card sits its
explanation behind the same collapsed button. Same pattern, one law.

The thinking row's rhythm: my-1 -> my-3. The 4px margin hugged the
options on non-graded answered cards (owner report); 12px matches the
card's mt-2/mt-3 spacing grammar everywhere the row renders.

Side-based authorship (owner round, 2026-10-07): the document shape
blurred user into agent — three left-rail idioms (user text, user
thinking, quoted prose) collided once quotes rendered. User rows go
RIGHT: 'You' messages and practice thinking cards hug the right edge at
85% width with the rail mirrored (border-r), text still reading
left-to-right. In-card thinking (inside the answered card, the agent's
context) stays left. Transcript thinking rows arrive EXPANDED — the
thinking was clearly sent. Cmd+T (toggle-user-thinking, chat scope,
cheat-sheet row) sets every user-thinking row to one state via a module
store: explicit sets stomp individual toggles, late mounts honor the
last set (the state initializer reads the store, so static render
reflects it too).

components/thinking-block.tsx

@@ -1,35 +1,86 @@
 "use client";
 
-import { useState } from "react";
+import { useEffect, useRef, useState } from "react";
 import { ChevronDownIcon } from "lucide-react";
 import { cn } from "@/lib/utils";
 
 /** The unified user-thinking row (owner round, 2026-10-05): ONE visual
  *  language for every surface where the learner's own thinking renders —
  *  practice thinking traces, the reading channel lifted above the answered
- *  card, the quiz answer's thinking inside its container. It arrives
- *  COLLAPSED as a trigger row in the exact register of the agent's
- *  reasoning rows (muted label + chevron + hover affordance); expanding
- *  opens the quote block — rail, 50% opacity, italics. The italics + the
- *  half-opacity text are the inner-voice signature; the rail marks it as
- *  learner material. No shell, no border: document material, full width.
+ *  card, the quiz answer's thinking inside its container. It arrives as
+ *  a trigger row in the exact register of the agent's reasoning rows
+ *  (muted label + chevron + hover affordance); expanding opens the quote
+ *  block — rail, 50% opacity, italics. The italics + the half-opacity
+ *  text are the inner-voice signature; the rail marks it as learner
+ *  material. No shell, no border: document material, full width.
  *
  *  The glance-level law (owner): user thinking = rail + 50% italic;
- *  agent thinking = no rail. */
+ *  agent thinking = no rail.
+ *
+ *  SIDE-BASED AUTHORSHIP (owner round, 2026-10-07): transcript-level rows
+ *  go RIGHT — mirrored rail (border-right), block hugging the right edge
+ *  at ~85% width, text still reading left-to-right. In-card rows (inside
+ *  the answered card, the agent's context) stay left. `mirror` selects
+ *  the side; `defaultOpen` marks the transcript rows, which arrive
+ *  EXPANDED (owner: clearly conveyed that the thinking was sent).
+ *
+ *  ⌘T (toggle-user-thinking) sets EVERY row to one state via the module
+ *  store below — one source of truth, no provider tree: rows subscribe on
+ *  mount; explicit sets (and only those) stomp individual toggles. */
+
+// ── The ⌘T store (module-scope, one per page load) ─────────────────────────
+// null = "rows run their own defaults" (the pre-toggle state); the first ⌘T
+// resolves from `true` (transcript rows default open) and then alternates.
+let allUserThinkingOpen: boolean | null = null;
+const userThinkingListeners = new Set<(open: boolean) => void>();
+
+export function currentUserThinkingOpen(): boolean {
+	return allUserThinkingOpen ?? true;
+}
+
+/** The row's starting state: an explicit ⌘T set wins over the row's own
+ *  default (late mounts honor the last set — same path as the live sync,
+ *  readable at render time so static SSR reflects the store too). */
+function resolveUserThinkingOpen(defaultOpen: boolean): boolean {
+	return allUserThinkingOpen ?? defaultOpen;
+}
+
+export function setAllUserThinkingOpen(open: boolean) {
+	allUserThinkingOpen = open;
+	for (const l of userThinkingListeners) l(open);
+}
+
+function subscribeUserThinking(fn: (open: boolean) => void): () => void {
+	userThinkingListeners.add(fn);
+	if (allUserThinkingOpen !== null) fn(allUserThinkingOpen); // late mounts honor the last set
+	return () => userThinkingListeners.delete(fn);
+}
 
 export function ThinkingBlock({
 	label = "YOUR THINKING",
 	text,
 	testId,
 	dataCommit,
+	mirror = false,
+	defaultOpen = false,
 }: {
 	label?: string;
 	text: string;
 	testId?: string;
 	/** The practice marker's register flag (commit vs nudge). */
 	dataCommit?: boolean;
+	/** Right-side placement (transcript rows): mirrored rail, right edge. */
+	mirror?: boolean;
+	/** Transcript rows arrive expanded (owner, 2026-10-07). */
+	defaultOpen?: boolean;
 }) {
-	const [open, setOpen] = useState(false); // arrives collapsed, always
+	const [open, setOpen] = useState(() => resolveUserThinkingOpen(defaultOpen));
+	// ⌘T: an explicit global set overrides every row (once set, rows re-sync).
+	const syncedRef = useRef(false);
+	useEffect(
+		() => subscribeUserThinking((o) => { syncedRef.current = true; setOpen(o); }),
+		[],
+	);
 	return (
 		/* The rail GROUPS and the rail IS THE MISSING CHARACTER (owner
 		   rounds, 2026-10-05): the heading sits INSIDE the rail, and the
@@ -44,7 +95,16 @@ export function ThinkingBlock({
 		<div
 			data-testid={testId}
 			data-commit={dataCommit === undefined ? undefined : dataCommit ? "true" : "false"}
-			className="lh-card-enter my-1 w-full border-l-2 border-muted-foreground/30 pl-[calc(3ch-2px)] font-ui text-xs"
+			data-side={mirror ? "right" : "left"}
+			className={cn(
+				// my-3 (owner, 2026-10-07): the row's own rhythm — 4px hugged the
+				// options on non-graded answered cards; 12px matches the card's
+				// mt-2/mt-3 spacing grammar.
+				"lh-card-enter my-3 font-ui text-xs",
+				mirror
+					? "ml-auto w-[85%] border-r-2 border-muted-foreground/30 pr-[calc(3ch-2px)]"
+					: "w-full border-l-2 border-muted-foreground/30 pl-[calc(3ch-2px)]",
+			)}
 		>
 			<button
 				type="button"