diff --git a/README.md b/README.md
index 600a308..3d28fde 100644
--- a/README.md
+++ b/README.md
@@ -71,6 +71,9 @@ no database, no data leaving your machine except the requests you send to Ollama
 - Arrow-key / PageUp / PageDown / Home / End navigation, plus swipe on touch devices
 - Progress bar, activity counter, XP counter and answer streak in the chrome
 - Instant feedback with a short explanation for every activity, confetti when you finish
+- Results screen: **Retry missed ones** (fresh, answerable items), **Study the whole deck again**,
+  and **Regenerate deck** — rebuilds the activities from the stored notes so you can keep drilling
+  until you're ready
 
 **Activity types** (each one is its own React component, rendered from JSON by a `type` field)
 
@@ -98,7 +101,8 @@ no database, no data leaving your machine except the requests you send to Ollama
 - **Folders**: create named groups ("History", "Biology", …) from the sidebar or the create screen.
   New decks freeze straight into the selected folder; anything unfiled lives in **Unorganized**.
   Rename, delete (decks survive — they return to Unorganized) and move decks between folders
-- Rename, delete (with confirm), regenerate, and **add more activities** to an existing deck
+- Rename, delete (with confirm), regenerate (from the library or the results screen), and
+  **add more activities** to an existing deck
 - Export a deck as JSON and import it back — the folder name travels with the file and is recreated
   on import
 
diff --git a/components/ResultsScreen.tsx b/components/ResultsScreen.tsx
index 0042190..55f73a8 100644
--- a/components/ResultsScreen.tsx
+++ b/components/ResultsScreen.tsx
@@ -6,7 +6,7 @@ import { motion } from "framer-motion";
 import type { Activity, Deck } from "@/lib/schemas";
 import { downloadDeck } from "@/lib/storage";
 import { cn } from "@/lib/utils";
-import { BoltIcon, DownloadIcon, FlameIcon, HomeIcon, RefreshIcon, SnowflakeIcon } from "@/components/icons";
+import { BoltIcon, DownloadIcon, FlameIcon, HomeIcon, RefreshIcon, SnowflakeIcon, SparklesIcon } from "@/components/icons";
 import { Button, Chip } from "@/components/ui";
 
 import type { AttemptResult } from "./StudyFeed";
@@ -60,6 +60,9 @@ export function ResultsScreen({
   retryMode,
   onRetryMissed,
   onRestart,
+  onRegenerate,
+  regenerating,
+  regenerateError,
 }: {
   deck: Deck;
   xp: number;
@@ -71,6 +74,9 @@ export function ResultsScreen({
   retryMode: boolean;
   onRetryMissed: () => void;
   onRestart: () => void;
+  onRegenerate: () => void;
+  regenerating: boolean;
+  regenerateError: string | null;
 }) {
   const percent = total > 0 ? Math.round((correct / total) * 100) : 0;
   const perfect = missed.length === 0 && total > 0;
@@ -151,6 +157,21 @@ export function ResultsScreen({
             <RefreshIcon className="size-5" /> Study the whole deck again
           </Button>
         ) : null}
+        <Button
+          size="lg"
+          variant="secondary"
+          onClick={onRegenerate}
+          loading={regenerating}
+          disabled={regenerating}
+        >
+          <SparklesIcon className="size-5" />
+          {regenerating ? "Regenerating…" : "Regenerate deck"}
+        </Button>
+        {regenerateError ? (
+          <p className="text-center text-xs text-rose-600 dark:text-rose-300" role="alert">
+            {regenerateError}
+          </p>
+        ) : null}
         <div className="grid grid-cols-2 gap-2">
           <Button variant="ghost" onClick={() => downloadDeck(deck)}>
             <DownloadIcon className="size-4" /> Export JSON
diff --git a/components/StudyFeed.tsx b/components/StudyFeed.tsx
index 55c38b6..84708a6 100644
--- a/components/StudyFeed.tsx
+++ b/components/StudyFeed.tsx
@@ -6,8 +6,10 @@ import { useCallback, useEffect, useMemo, useRef, useState } from "react";
 
 import { ActivityRenderer } from "@/components/activities";
 import { celebrate } from "@/lib/confetti";
+import { ApiError, generateDeck } from "@/lib/client-api";
+import { materialiseActivities } from "@/lib/deck-ops";
 import type { Activity, Deck } from "@/lib/schemas";
-import { recordAttempt } from "@/lib/storage";
+import { recordAttempt, upsertDeck } from "@/lib/storage";
 import { cn } from "@/lib/utils";
 import { BoltIcon, FlameIcon, XIcon } from "@/components/icons";
 import { Button, ProgressBar } from "@/components/ui";
@@ -17,6 +19,7 @@ import { ResultsScreen } from "./ResultsScreen";
 export type AttemptResult = { correct: boolean; userAnswer?: string };
 
 export function StudyFeed({ deck }: { deck: Deck }) {
+  const [currentDeck, setCurrentDeck] = useState(deck);
   const [queue, setQueue] = useState<Activity[]>(deck.activities);
   const [index, setIndex] = useState(0);
   const [results, setResults] = useState<Record<string, AttemptResult>>({});
@@ -24,6 +27,8 @@ export function StudyFeed({ deck }: { deck: Deck }) {
   const [streak, setStreak] = useState(0);
   const [bestStreak, setBestStreak] = useState(0);
   const [retryMode, setRetryMode] = useState(false);
+  const [regenerating, setRegenerating] = useState(false);
+  const [regenerateError, setRegenerateError] = useState<string | null>(null);
   /** Bumped on every restart so the activity components remount fresh. */
   const [runId, setRunId] = useState(0);
 
@@ -95,10 +100,15 @@ export function StudyFeed({ deck }: { deck: Deck }) {
       const answered = Object.values(results);
       const correct = answered.filter((entry) => entry.correct).length;
       const score = total > 0 ? (correct / total) * 100 : 0;
-      recordAttempt(deck.id, { scorePercent: score, correct, total });
-      void celebrate("big");
+      // Only a run with at least one answer counts. Without this, the scroll
+      // anchoring shift when a regenerated deck swaps in could fire a bogus
+      // 0% attempt before the feed had snapped back to the first slide.
+      if (answered.length > 0) {
+        recordAttempt(currentDeck.id, { scorePercent: score, correct, total });
+        void celebrate("big");
+      }
     }
-  }, [finished, results, total, deck.id, retryMode]);
+  }, [finished, results, total, currentDeck.id, retryMode]);
 
   const handleAnswered = useCallback(
     (activity: Activity, correct: boolean, meta?: { userAnswer?: string }) => {
@@ -136,7 +146,11 @@ export function StudyFeed({ deck }: { deck: Deck }) {
       setStreak(0);
       setRetryMode(retry);
       recordedRef.current = retry; // don't overwrite the best score with a retry run
-      // A new run: remount every activity (fresh state) and jump to the top.
+      // A new run: remount every activity (fresh state) and jump to the top
+      // *before* the new slides render, so scroll anchoring can't hold the
+      // old results position.
+      const container = scrollRef.current;
+      if (container) container.scrollTop = 0;
       setRunId((value) => value + 1);
       setIndex(0);
     },
@@ -149,6 +163,45 @@ export function StudyFeed({ deck }: { deck: Deck }) {
     if (container) container.scrollTop = 0;
   }, [runId]);
 
+  /** Rebuild the deck from its stored source notes and start it over. */
+  const regenerate = useCallback(async () => {
+    if (regenerating) return;
+    const source = currentDeck.sourceText?.trim();
+    if (!source) {
+      setRegenerateError("This deck has no stored source notes, so it can't be regenerated.");
+      return;
+    }
+    setRegenerating(true);
+    setRegenerateError(null);
+    try {
+      const { deck: generated } = await generateDeck({
+        text: source,
+        difficulty: currentDeck.difficulty,
+        language: currentDeck.language,
+        count: Math.max(8, currentDeck.activities.length),
+      });
+      const activities = materialiseActivities(generated.activities, currentDeck.difficulty);
+      const updated: Deck = {
+        ...currentDeck,
+        title: generated.title || currentDeck.title,
+        subject: generated.subject ?? currentDeck.subject,
+        summary: generated.summary ?? currentDeck.summary,
+        language: generated.language ?? currentDeck.language,
+        activities,
+        attempts: 0,
+        bestScore: undefined,
+        updatedAt: Date.now(),
+      };
+      upsertDeck(updated);
+      setCurrentDeck(updated);
+      restart(activities, false);
+    } catch (error) {
+      setRegenerateError(error instanceof ApiError ? error.message : "Could not regenerate the deck.");
+    } finally {
+      setRegenerating(false);
+    }
+  }, [currentDeck, regenerating, restart]);
+
   const progressValue = Math.min(index + 1, slideCount);
   const progressMax = slideCount;
 
@@ -197,7 +250,7 @@ export function StudyFeed({ deck }: { deck: Deck }) {
             </div>
           </div>
           <div className="flex items-center justify-between gap-2 text-[11px] font-semibold text-[var(--text-muted)]">
-            <span className="min-w-0 truncate">{deck.title}</span>
+            <span className="min-w-0 truncate">{currentDeck.title}</span>
             <span className="shrink-0">
               {retryMode ? "Retry · " : ""}
               {Math.min(index + 1, total)}/{total}
@@ -210,7 +263,7 @@ export function StudyFeed({ deck }: { deck: Deck }) {
       <div
         ref={scrollRef}
         data-testid="study-feed"
-        className="no-scrollbar h-full snap-y snap-mandatory overflow-y-scroll overscroll-y-contain"
+        className="no-scrollbar h-full snap-y snap-mandatory overflow-y-scroll overscroll-y-contain [overflow-anchor:none]"
       >
         {queue.map((activity, activityIndex) => (
           <section
@@ -222,7 +275,7 @@ export function StudyFeed({ deck }: { deck: Deck }) {
               <ActivityRenderer
                 activity={activity}
                 active={index === activityIndex}
-                language={deck.language}
+                language={currentDeck.language}
                 isLast={activityIndex === total - 1}
                 onAnswered={(correct, meta) => handleAnswered(activity, correct, meta)}
                 onNext={() => goTo(activityIndex + 1)}
@@ -237,7 +290,7 @@ export function StudyFeed({ deck }: { deck: Deck }) {
         >
           <div className="mx-auto flex min-h-full w-full max-w-xl flex-col gap-4 px-4 pb-24 pt-24 [justify-content:safe_center]">
             <ResultsScreen
-              deck={deck}
+              deck={currentDeck}
               xp={xp}
               bestStreak={bestStreak}
               total={total}
@@ -246,7 +299,10 @@ export function StudyFeed({ deck }: { deck: Deck }) {
               results={results}
               retryMode={retryMode}
               onRetryMissed={() => restart(missed, true)}
-              onRestart={() => restart(deck.activities, false)}
+              onRestart={() => restart(currentDeck.activities, false)}
+              onRegenerate={regenerate}
+              regenerating={regenerating}
+              regenerateError={regenerateError}
             />
           </div>
         </section>
diff --git a/components/activities/TrueFalse.tsx b/components/activities/TrueFalse.tsx
index 8947f6b..7a27ce1 100644
--- a/components/activities/TrueFalse.tsx
+++ b/components/activities/TrueFalse.tsx
@@ -19,8 +19,13 @@ type TrueFalseActivity = Extract<Activity, { type: "true_false" }>;
 export function TrueFalse({ activity, onAnswered, onNext, isLast }: ActivityComponentProps<TrueFalseActivity>) {
   const [result, setResult] = useState<boolean | null>(null);
   const x = useMotionValue(0);
-  const trueOpacity = useTransform(x, [30, 110], [0, 1]);
-  const falseOpacity = useTransform(x, [-30, -110], [1, 0]);
+  // The drag is elastic, so the card moves less than the pointer — these
+  // ranges make each label reach full strength on a normal swipe.
+  const trueOpacity = useTransform(x, [20, 70], [0, 1]);
+  // Left drags go negative: at rest (0) this must be invisible, only fading
+  // in as the card is dragged left. The old range was inverted, so the
+  // "false ✕" label sat on the card permanently.
+  const falseOpacity = useTransform(x, [-70, -20], [1, 0]);
 
   const answer = (value: boolean) => {
     if (result !== null) return;
diff --git a/lib/ai.ts b/lib/ai.ts
index 136c6e6..db049c5 100644
--- a/lib/ai.ts
+++ b/lib/ai.ts
@@ -250,6 +250,13 @@ const VALID_TYPES = [
   "short_answer",
 ] as const;
 
+/**
+ * Giveaways that make an ordering card solvable without knowing the material:
+ * years, clock times, "first/next/finally", "step 2", "1.".
+ */
+const ORDERING_GIVEAWAY =
+  /\b(1[0-9]{3}|20[0-9]{2})\b|\b\d{1,2}[:.]\d{2}\b|\b(first|second|third|fourth|fifth|sixth|seventh|eighth|1st|2nd|3rd|4th|5th|6th|7th|8th)\b|\b(step|stage|phase)\s*\d+\b|^\s*\d+[.)]\s/i;
+
 /**
  * Be forgiving: validate each activity on its own and keep the good ones.
  * Only if too few survive does the step get retried.
@@ -280,7 +287,16 @@ function validateDeck(parsed: unknown, minimum: number, requested: number) {
   rawActivities.slice(0, 24).forEach((item, index) => {
     const result = generatedActivitySchema.safeParse(item);
     if (result.success) {
-      activities.push(sanitiseActivity(result.data));
+      const activity = sanitiseActivity(result.data);
+      // An ordering card with dates/ordinals on it gives the answer away —
+      // drop it rather than ship a card the learner can solve without knowing.
+      if (activity.type === "ordering" && activity.items.some((entry) => ORDERING_GIVEAWAY.test(entry))) {
+        rejected.push(
+          `#${index + 1} (ordering): items must not contain dates, years, times or ordinals — the order has to come from the content.`,
+        );
+        return;
+      }
+      activities.push(activity);
     } else {
       const type = (item as { type?: unknown } | null)?.type;
       rejected.push(`#${index + 1} (${typeof type === "string" ? type : "no type"}): ${summariseZodError(result.error)}`);
diff --git a/lib/prompts.ts b/lib/prompts.ts
index 000c214..ca5a444 100644
--- a/lib/prompts.ts
+++ b/lib/prompts.ts
@@ -85,7 +85,9 @@ export const ACTIVITY_JSON_SPEC = `Each activity is one JSON object with a "type
    { "type": "matching", "topic": "short label", "instruction": "Match each term to its definition.", "pairs": [{ "term": "Osmosis", "definition": "diffusion of water across a membrane" }], "explanation": "..." }
 
 6. ordering - 3 to 7 items given in the CORRECT order (the app shuffles them)
-   { "type": "ordering", "topic": "short label", "instruction": "Put the stages of mitosis in order.", "items": ["Prophase", "Metaphase", "Anaphase", "Telophase"], "explanation": "..." }
+   Items must NEVER contain dates, years, clock times, step numbers or ordinals ("first", "step 2") —
+   the learner has to work out the order from what each item says, not read it off.
+   { "type": "ordering", "topic": "short label", "instruction": "Put the events in the order they happened.", "items": ["The Allies land in Normandy", "Paris is liberated", "Germany surrenders"], "explanation": "..." }
 
 7. short_answer - typed answer, graded by AI
    { "type": "short_answer", "topic": "short label", "question": "...", "referenceAnswer": "a model answer in one or two sentences", "keyPoints": ["idea that must appear", "second idea"], "explanation": "..." }
@@ -102,6 +104,8 @@ Hard rules:
 - Vary the activity types. Use a healthy mix of all seven types below, and never put more than two activities of the same type in a row.
 - Cover the whole notes, not just the first paragraph. Spread activities across the different topics, definitions, formulas and facts.
 - Make the wrong options plausible (common mix-ups from the same notes), never silly filler.
+- Ordering items must never reveal their position: no dates, years, times, step numbers or ordinals
+  ("first", "next", "finally", "step 2"). Describe the event or stage itself.
 - Keep questions short enough to read on a phone screen. No walls of text.
 - Never reference "the notes", "the text above" or "the student". Speak directly to the learner.
 - Match the requested difficulty: easy = recall and simple recognition; medium = understanding and application; hard = analysis, edge cases and multi-step reasoning.