A whole category accepted in one click, and one undo

Five article fixes were five clicks, five confetti bursts and five undo
steps. "Accept all Tidy-up (5)" makes them one of each.

The single undo decided the implementation: every replacement goes into one
Tiptap chain, which applies as one transaction and so undoes as one history
event. That only works if the spans can't move under each other, so the
plan resolves every span against the document as it stands and applies them
last-first.

Three outcomes rather than one, because a batch that quietly dropped a card
would be reporting edits it never made: a span she already fixed herself is
settled without an edit (what a single Accept does too), and a card quoting
the same words as one already taken is left on screen, since findRange
would resolve both to the same place.

The control sits on the first card of its kind — the rail can't carry a
category header, its cards are anchored to their own sentences — and only
when the category has company. It is outlined rather than filled: it acts
on cards she can't see from where she's standing.

UX_REVIEW_2026-07-27.md

@@ -861,6 +861,105 @@ one from many.
 **Still open in item 8:** Accept All per category, and the keyboard triage
 flow. Both untouched.
 
+### 8 — Accept All per category DONE (tenth session). The undo step decided the shape.
+
+The item's one design question — the batch must be a single undo — turned out to
+decide the implementation rather than follow it. Five accepts is five
+transactions, and prosemirror-history would group them only by *timing*, which is
+not a guarantee. So every replacement goes into one Tiptap `chain()`: a chain is
+applied as a single transaction, and the history plugin undoes a transaction as
+one event. That is the whole mechanism, and the rest of the work is making a set
+of spans safe to apply in one go.
+
+**Implemented:**
+
+- `acceptBatch.ts` — `planBatch(list, resolve)`, pure, takes the span resolver as
+  an argument so the arithmetic can be tested without a ProseMirror document.
+  Every span is resolved against the doc as it stands *before* any edit, and the
+  steps are returned **last-span-first**: applying from the end backwards means no
+  earlier position can be shifted by a later replacement, which is what lets them
+  share one transaction with no position mapping.
+- Three outcomes, not one. A card whose span is **gone** (she fixed it herself) is
+  settled without an edit — exactly what a single Accept already does with an
+  unresolvable span. A card that **overlaps** one already taken is left on screen:
+  `findRange` resolves to the *first* occurrence, so two cards quoting the same
+  words would have the second overwrite the first. A batch that silently dropped
+  either would be reporting edits it never made.
+- `EditorCore.handleAcceptAll(type)` — builds the plan, applies the chain, fires
+  **one** confetti burst over the topmost changed span (read before the edit
+  removes the highlight), and hands the settled cards to the parent.
+  `burstAt`/`showConfetti` were lifted out of `handleAccept`, which now shares
+  them.
+- `App.handleAcceptMany` — the bookkeeping only. Each row is filed individually
+  (there is no batch endpoint, and each accept plants its own word in the garden)
+  but the kitten cheers **once**: five cheers for one click would read as five
+  separate congratulations for a decision she made once.
+- `batchLeaders` — which card carries the control. The item asks for a category
+  *header* in the rail, and the rail cannot have one: cards are anchored to their
+  own sentence, so a type's cards are scattered down the column. The closest
+  honest stand-in is the first card of its kind. The first version put the button
+  on every card of the type, which in the browser was five identical buttons in
+  one column saying one thing.
+- The control appears only at **two or more** — a lone Grammar card doesn't grow a
+  second button saying the same thing as the first — and on both surfaces, since
+  item 7 made the anchored popover primary in both layouts. It is outlined in the
+  category's colour rather than filled like Accept: it acts on cards she can't see
+  from where she's standing, and an equally loud button would invite the click she
+  meant to give the one suggestion in front of her.
+- The label stays English (`Accept all Tidy-up (4)`) even on a translation card,
+  whose pill is bilingual. The pill names the kind of advice she is reading; this
+  names an action over the rest of the queue, and every other word in that row —
+  Accept, Dismiss, Ask Petal — is English. A control that changed language between
+  cards would read as a different control.
+
+**Verified in a real browser at the review's own 1517×810**, against a stub model
+server (no VPN, no GPU) that adds two `grammar` findings so the document carries
+two categories at once:
+
+- *Single undo, measured.* `view.dispatch` hooked to count doc-changing
+  transactions: accepting five Tidy-up cards produced **one** transaction and took
+  the history's `done` count from 1 to 2. One Ctrl+Z restored all five originals;
+  one Ctrl+Shift+Z put all five corrections back.
+- *Categories don't touch each other.* With three Tidy-up and one Grammar card,
+  "Accept all Tidy-up (3)" fixed exactly its three; the Grammar span stayed
+  underlined and its card stayed put.
+- *One control per category*, on the first card of its kind, on the rail — and the
+  Grammar card, alone in its type, offered none on either surface.
+- *Both surfaces click.* The rail's button and the anchored popover's own button
+  were each clicked and each applied the whole category, closing the card and
+  leaving one cheer behind.
+
+**A measurement trap that cost an hour, and it is the sixth and ninth sessions'
+trap in a third disguise.** After the accept the rail appeared to keep its
+accepted cards on screen indefinitely — server said nothing pending, the status
+bar agreed, the DOM still had five cards. It is not a bug: **`recomputeRail` does
+its work inside `requestAnimationFrame`, and Chrome does not fire rAF in a hidden
+tab.** Driving the page through the JS tool leaves the window backgrounded, so
+the rail is simply frozen; every screenshot (which forces a paint) showed it
+correctly empty. Anything measured through rAF is invalid unless the tab is
+foreground — check `document.visibilityState` before believing a rail reading.
+
+And the bundle-hash check caught its third variant: **the Go binary embeds
+`web/dist`**, so rebuilding the frontend alone changes nothing. `npm run build`
+without `go build` served the previous bundle from a server that had just been
+restarted and looked entirely healthy.
+
+Coverage: `acceptBatch.test.ts` — the ordering property asserted by actually
+applying the plan to a string and checking the result; document order rather than
+card order; the missing-span case; two cards quoting the same words; adjacent
+spans not counting as overlapping; an awareness-only card with nothing to insert;
+and `batchLeaders` (one leader per category, follows the stack order it is given,
+empty stack). **The wiring itself has no unit test**, for the reason items 6 and 7
+recorded: jsdom has no layout, and a rail test there would pass whatever the code
+did. It is browser-verified only, and is written down as such.
+
+**Deliberately not done:** no "accept everything" across all categories. The
+categories are the unit she can reason about — five article fixes are one
+decision, but her whole queue is not — and a single button that rewrites the
+document in one press is the opposite of a tool that teaches.
+
+**Still open in item 8:** the keyboard triage flow.
+
 ---
 
 ## Explicit non-goals (from this review)
@@ -960,6 +1059,18 @@ unbound and the old bundle keeps serving — the eighth session's lesson with a
 different cause, and the same bundle-hash check catches it. Untouched: item
 8's Accept All and keyboard flow, item 3's incremental half.)*
 
+*(Tenth session: item 8's Accept All per category done — see the subsection under
+item 8. **Still not pushed and not deployed**, and the undeployed stack is now
+four commits: the ninth session's settled-spans work plus this. Neither needs a
+migration and neither touches the schema, so deploying is still `git push` then
+`git pull && docker compose up -d --build`. Two things to carry forward. First,
+**a hidden Chrome tab fires no `requestAnimationFrame`**, and `recomputeRail`
+lives inside one — so the rail freezes under JS-tool automation and every reading
+of it is stale until a screenshot forces a paint. Second, **the binary embeds
+`web/dist`**: rebuild the frontend and you must rebuild the binary, or the
+bundle-hash check will (rightly) fail. Untouched: item 8's keyboard flow, item 3's
+incremental half.)*
+
 **Migration 0015 on the live database.** It rebuilds the suggestions table, so
 unlike 0014 it could have dropped her rows. Backed up first — and the backup
 had to be the whole WAL set (`petal.db`, `-wal`, `-shm` in
@@ -974,10 +1085,18 @@ CHECK, and every existing row still carrying the label she has already read (the
 seven `clarity` rows include the mislabelled Chinese one — by design, only new
 findings get the new type; if you want that card relabelled, edit the sentence).
 
-**Suggested next (ninth session onward):** three things remain in the whole
+**Suggested next (tenth session onward):** two things remain in the whole review.
-review. **Accept All per category** is the one with real value left — five
+**Keyboard triage** is the one to take: it is the last of item 8, and Accept All
-tense fixes are still five clicks, and item 2's stable ids make a batch safe to
+just built half of what it needs — a category is now a thing the UI can act on in
-reason about; the one design question it has to answer is that its undo must be
+one step, so "triage without the mouse" is mostly about driving the anchored
+popover between spans. **Item 3's incremental surfacing** is the last item of any
+size, and still needs a streaming `/check`. It remains the only one left that
+changes how the app *feels* rather than what it can do.
+
+*(Superseded, kept for the reading list: the ninth session's advice.)* Three
+things remained. **Accept All per category** was the one with real value left —
+five tense fixes are still five clicks, and item 2's stable ids make a batch safe
+to reason about; the one design question it has to answer is that its undo must be
 a single step. **Keyboard triage** is next, and is bigger than it looks: item 7
 made the anchored popover the primary surface in both layouts, so "cycle the
 underlines" now means driving that popover, not the rail. **Item 3's

web/src/App.tsx

@@ -367,6 +367,30 @@ export default function App() {
     [removeSuggestion, resolveServerId],
   )
 
+  // Accept-all: EditorCore has already applied the whole category in one editor
+  // transaction, so this is only the bookkeeping. Each row is filed individually
+  // (there's no batch endpoint, and each accept plants its own word in the
+  // garden), but the kitten cheers once — five cheers for one click would read as
+  // five separate congratulations for a decision she made once.
+  const handleAcceptMany = useCallback(
+    async (list: Suggestion[]) => {
+      if (list.length === 0) return
+      for (const s of list) removeSuggestion(s.id)
+      setAcceptTick((n) => n + 1)
+      await Promise.all(
+        list.map(async (s) => {
+          try {
+            const id = await resolveServerId(s)
+            if (id) await api.acceptSuggestion(id)
+          } catch (err) {
+            console.error('accept failed', err)
+          }
+        }),
+      )
+    },
+    [removeSuggestion, resolveServerId],
+  )
+
   // After restoring a version, swap the restored doc into the editor. Bumping
   // editorEpoch remounts EditorCore so it picks up the restored content.
   const handleRestored = useCallback(
@@ -569,6 +593,7 @@ export default function App() {
                     onChange={handleEditorChange}
                     suggestions={suggestions}
                     onAccept={handleAccept}
+                    onAcceptMany={handleAcceptMany}
                     onDismiss={handleDismiss}
                     onVoiceCheck={runVoice}
                     voicing={voicing}

web/src/components/Editor/EditorCore.tsx

@@ -15,7 +15,7 @@ import TableHeader from '@tiptap/extension-table-header'
 import TableCell from '@tiptap/extension-table-cell'
 import { FontSize } from './FontSize'
 import type { EditorView } from '@tiptap/pm/view'
-import { useCallback, useEffect, useLayoutEffect, useRef, useState } from 'react'
+import { useCallback, useEffect, useLayoutEffect, useMemo, useRef, useState } from 'react'
 import { Toolbar } from '../Toolbar/Toolbar'
 import { SuggestionCard } from './SuggestionCard'
 import { SuggestionRail, type RailItem } from './SuggestionRail'
@@ -30,7 +30,8 @@ import { SearchHighlight } from './SearchHighlight'
 import { FindReplace } from './FindReplace'
 import { Typography } from './Typography'
 import { RewritePreview, type RewriteStatus } from './RewritePreview'
-import { api, type Suggestion, type WordInfo } from '../../api/client'
+import { planBatch } from './acceptBatch'
+import { api, type Suggestion, type SuggestionType, type WordInfo } from '../../api/client'
 import { speak, speechSupported } from '../../audio/speech'
 import type { SpellChecker } from '../../hooks/useSpellChecker'
 import { usePack } from '../../i18n'
@@ -54,6 +55,9 @@ interface Props {
   // call + state removal (accept's text replacement happens here in the editor).
   suggestions: Suggestion[]
   onAccept: (s: Suggestion) => void
+  // A whole category accepted at once. The text replacement happens here, in one
+  // transaction; the parent files each row and cheers once for the batch.
+  onAcceptMany: (list: Suggestion[]) => void
   onDismiss: (s: Suggestion) => void
   // Triggers the whole-document voice-consistency pass; `voicing` is true while
   // it runs (drives the toolbar button's loading state).
@@ -221,6 +225,7 @@ export function EditorCore({
   onChange,
   suggestions,
   onAccept,
+  onAcceptMany,
   onDismiss,
   onVoiceCheck,
   voicing,
@@ -638,18 +643,28 @@ export function EditorCore({
   // anchored to the highlight itself (captured before the replacement removes it),
   // so it fires in the right spot whether the accept came from the hover card or
   // the margin rail.
-  const handleAccept = useCallback(
+  // Where a suggestion's highlight sits, in wrapper coordinates — captured before
-    (s: Suggestion) => {
+  // the replacement removes it, so the confetti fires over the words that changed.
+  const burstAt = useCallback((id: string): { top: number; left: number } | null => {
     const wrapper = wrapperRef.current
     const el = wrapper?.querySelector(
-        `.petal-suggestion[data-suggestion-id="${CSS.escape(s.id)}"]`,
+      `.petal-suggestion[data-suggestion-id="${CSS.escape(id)}"]`,
     ) as HTMLElement | null
-      let burst: { top: number; left: number } | null = null
+    if (!wrapper || !el) return null
-      if (wrapper && el) {
     const wrapRect = wrapper.getBoundingClientRect()
     const elRect = el.getBoundingClientRect()
-        burst = { top: elRect.top - wrapRect.top, left: elRect.right - wrapRect.left }
+    return { top: elRect.top - wrapRect.top, left: elRect.right - wrapRect.left }
-      }
+  }, [])
+
+  const showConfetti = useCallback((burst: { top: number; left: number }) => {
+    setConfetti(burst)
+    clearTimeout(confettiTimer.current)
+    confettiTimer.current = setTimeout(() => setConfetti(null), 720)
+  }, [])
+
+  const handleAccept = useCallback(
+    (s: Suggestion) => {
+      let burst = burstAt(s.id)
       if (editor && s.replacement.trim() !== '') {
         const range = findRange(editor.state.doc, s.original)
         if (range) {
@@ -658,16 +673,57 @@ export function EditorCore({
       }
       // Fall back to the hover card's position if the highlight wasn't found.
       if (!burst && hover) burst = { top: hover.top, left: hover.left + 16 }
-      if (burst) {
+      if (burst) showConfetti(burst)
-        setConfetti(burst)
-        clearTimeout(confettiTimer.current)
-        confettiTimer.current = setTimeout(() => setConfetti(null), 720)
-      }
       closeCard()
       setRailExpandedId(null)
       onAccept(s)
     },
-    [editor, onAccept, closeCard, hover],
+    [editor, onAccept, closeCard, hover, burstAt, showConfetti],
+  )
+
+  // How many pending cards of each type could be accepted in one go. A card
+  // offers "Accept all" only when it has company, so a lone Grammar card doesn't
+  // grow a second button saying the same thing as the first.
+  const batchCounts = useMemo(() => {
+    const counts: Partial<Record<SuggestionType, number>> = {}
+    for (const s of suggestions) {
+      if (s.replacement.trim() === '') continue // awareness-only: nothing to accept
+      counts[s.type] = (counts[s.type] ?? 0) + 1
+    }
+    return counts
+  }, [suggestions])
+
+  // Accept a whole category at once. Five tense fixes were five clicks, five
+  // confetti bursts and five separate undo steps; this is one of each. The single
+  // undo step is the reason every replacement goes into ONE chain: Tiptap applies
+  // a chain as a single transaction, and prosemirror-history undoes it as a
+  // single event, so Ctrl+Z takes back the batch rather than unpicking it.
+  const handleAcceptAll = useCallback(
+    (type: SuggestionType) => {
+      if (!editor) return
+      const family = suggestions.filter((s) => s.type === type)
+      const plan = planBatch(family, (original) => findRange(editor.state.doc, original))
+      const settled = [...plan.steps.map((step) => step.suggestion), ...plan.missing]
+      if (settled.length === 0) return
+
+      // The topmost span that's about to change — the last step, since steps run
+      // bottom-up. Read before the edit, while the highlights still exist.
+      const first = plan.steps[plan.steps.length - 1]
+      const burst = first ? burstAt(first.suggestion.id) : null
+
+      if (plan.steps.length > 0) {
+        let chain = editor.chain().focus()
+        for (const step of plan.steps) {
+          chain = chain.insertContentAt({ from: step.from, to: step.to }, step.suggestion.replacement)
+        }
+        chain.run()
+      }
+      if (burst) showConfetti(burst)
+      closeCard()
+      setRailExpandedId(null)
+      onAcceptMany(settled)
+    },
+    [editor, suggestions, onAcceptMany, closeCard, burstAt, showConfetti],
   )
 
   const handleDismiss = useCallback(
@@ -1248,7 +1304,9 @@ export function EditorCore({
           <SuggestionCard
             suggestion={hover.suggestion}
             style={{ top: hover.top, left: hover.left }}
+            batchCount={batchCounts[hover.suggestion.type] ?? 0}
             onAccept={handleAccept}
+            onAcceptAll={handleAcceptAll}
             onDismiss={handleDismiss}
             onPointerEnter={keepOpen}
             onPointerLeave={scheduleClose}
@@ -1261,7 +1319,9 @@ export function EditorCore({
             items={railItems}
             activeId={activeId}
             expandedId={railExpandedId}
+            batchCounts={batchCounts}
             onAccept={handleAccept}
+            onAcceptAll={handleAcceptAll}
             onDismiss={handleDismiss}
             onHover={setActiveId}
             onActivate={activateRailCard}

web/src/components/Editor/SuggestionCard.tsx

@@ -1,13 +1,17 @@
 import { useEffect, useRef, useState } from 'react'
-import type { Suggestion } from '../../api/client'
+import type { Suggestion, SuggestionType } from '../../api/client'
 import { usePack } from '../../i18n'
 import { AskPetal } from './AskPetal'
-import { TYPE_META, typeLabel } from './suggestionMeta'
+import { TYPE_META, batchLabel, typeLabel } from './suggestionMeta'
 
 interface Props {
   suggestion: Suggestion
   style: React.CSSProperties
+  // How many pending suggestions share this card's type (including this one).
+  // Two or more offers to take the whole category in one step.
+  batchCount: number
   onAccept: (s: Suggestion) => void
+  onAcceptAll: (type: SuggestionType) => void
   onDismiss: (s: Suggestion) => void
   onPointerEnter: () => void
   onPointerLeave: () => void
@@ -30,7 +34,9 @@ interface Props {
 export function SuggestionCard({
   suggestion,
   style,
+  batchCount,
   onAccept,
+  onAcceptAll,
   onDismiss,
   onPointerEnter,
   onPointerLeave,
@@ -144,6 +150,17 @@ export function SuggestionCard({
           Dismiss
         </button>
       </div>
+
+      {hasReplacement && batchCount > 1 && (
+        <button
+          type="button"
+          onClick={() => onAcceptAll(suggestion.type)}
+          className="petal-accept-all mt-2 w-full rounded-full py-1.5 text-xs font-bold"
+          style={{ color: 'var(--color-plum)', borderColor: meta.color }}
+        >
+          {batchLabel(suggestion.type, batchCount)}
+        </button>
+      )}
     </div>
   )
 }

web/src/components/Editor/SuggestionRail.tsx

@@ -1,8 +1,9 @@
 import { forwardRef, useLayoutEffect, useRef, useState } from 'react'
-import type { Suggestion } from '../../api/client'
+import type { Suggestion, SuggestionType } from '../../api/client'
 import { usePack } from '../../i18n'
 import { AskPetal } from './AskPetal'
-import { TYPE_META, typeLabel } from './suggestionMeta'
+import { batchLeaders } from './acceptBatch'
+import { TYPE_META, batchLabel, typeLabel } from './suggestionMeta'
 
 // Vertical breathing room kept between stacked cards when their natural anchors
 // would otherwise collide.
@@ -22,7 +23,11 @@ interface Props {
   activeId: string | null
   // The card expanded to show the full explanation + Ask Petal, or null.
   expandedId: string | null
+  // Pending suggestions per type. A card whose type has company offers to take
+  // the whole category in one step (and one undo step).
+  batchCounts: Partial<Record<SuggestionType, number>>
   onAccept: (s: Suggestion) => void
+  onAcceptAll: (type: SuggestionType) => void
   onDismiss: (s: Suggestion) => void
   // Pointer entering/leaving a card, so the matching highlight can light up.
   onHover: (id: string | null) => void
@@ -46,7 +51,9 @@ export function SuggestionRail({
   items,
   activeId,
   expandedId,
+  batchCounts,
   onAccept,
+  onAcceptAll,
   onDismiss,
   onHover,
   onActivate,
@@ -93,6 +100,9 @@ export function SuggestionRail({
     // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [layoutKey, expandedId, measureTick])
 
+  // One card per category carries the batch control — see batchLeaders.
+  const batchLead = batchLeaders(ordered.map(({ suggestion }) => suggestion))
+
   return (
     <div className="petal-rail petal-no-print" aria-label="Suggestions">
       {ordered.map(({ suggestion }) => (
@@ -112,7 +122,9 @@ export function SuggestionRail({
           top={tops[suggestion.id] ?? 0}
           active={activeId === suggestion.id}
           expanded={expandedId === suggestion.id}
+          batchCount={batchLead.has(suggestion.id) ? (batchCounts[suggestion.type] ?? 0) : 0}
           onAccept={onAccept}
+          onAcceptAll={onAcceptAll}
           onDismiss={onDismiss}
           onHover={onHover}
           onActivate={onActivate}
@@ -128,7 +140,9 @@ interface CardProps {
   top: number
   active: boolean
   expanded: boolean
+  batchCount: number
   onAccept: (s: Suggestion) => void
+  onAcceptAll: (type: SuggestionType) => void
   onDismiss: (s: Suggestion) => void
   onHover: (id: string | null) => void
   onActivate: (id: string) => void
@@ -136,7 +150,7 @@ interface CardProps {
 }
 
 const RailCard = forwardRef<HTMLDivElement, CardProps>(function RailCard(
-  { suggestion, top, active, expanded, onAccept, onDismiss, onHover, onActivate, onToggleExpand },
+  { suggestion, top, active, expanded, batchCount, onAccept, onAcceptAll, onDismiss, onHover, onActivate, onToggleExpand },
   ref,
 ) {
   const pack = usePack()
@@ -230,6 +244,17 @@ const RailCard = forwardRef<HTMLDivElement, CardProps>(function RailCard(
           {expanded ? 'Hide Petal' : 'Ask Petal ✨'}
         </button>
       </div>
+
+      {hasReplacement && batchCount > 1 && (
+        <button
+          type="button"
+          onClick={() => onAcceptAll(suggestion.type)}
+          className="petal-accept-all mt-2 w-full rounded-full py-1 text-[0.7rem] font-bold"
+          style={{ color: 'var(--color-plum)', borderColor: meta.color }}
+        >
+          {batchLabel(suggestion.type, batchCount)}
+        </button>
+      )}
     </div>
   )
 })

web/src/components/Editor/acceptBatch.test.ts

@@ -0,0 +1,128 @@
+import { describe, expect, it } from 'vitest'
+import type { Suggestion, SuggestionType } from '../../api/client'
+import { batchLeaders, planBatch } from './acceptBatch'
+
+function sug(id: string, original: string, replacement: string, type: SuggestionType = 'grammar'): Suggestion {
+  return {
+    id,
+    doc_id: 'd1',
+    from_pos: 0,
+    to_pos: 0,
+    original,
+    replacement,
+    explanation: '',
+    type,
+    status: 'pending',
+    created_at: '2026-07-28T00:00:00Z',
+  }
+}
+
+// A stand-in for findRange: first occurrence in a flat string, or null.
+function finder(text: string) {
+  return (needle: string) => {
+    const i = text.indexOf(needle)
+    return i < 0 ? null : { from: i, to: i + needle.length }
+  }
+}
+
+describe('planBatch', () => {
+  it('applies from the end backwards, so earlier spans keep their positions', () => {
+    const text = 'a apple and a orange and a egg'
+    const plan = planBatch(
+      [sug('1', 'a apple', 'an apple'), sug('2', 'a orange', 'an orange'), sug('3', 'a egg', 'an egg')],
+      finder(text),
+    )
+    expect(plan.steps.map((s) => s.suggestion.id)).toEqual(['3', '2', '1'])
+    expect(plan.missing).toEqual([])
+    expect(plan.skipped).toEqual([])
+
+    // Applying the plan in order against a mutable string must land every
+    // replacement where it belongs — this is the property the ordering exists
+    // for, and the one a single shared transaction depends on.
+    let out = text
+    for (const step of plan.steps) {
+      out = out.slice(0, step.from) + step.suggestion.replacement + out.slice(step.to)
+    }
+    expect(out).toBe('an apple and an orange and an egg')
+  })
+
+  it('orders by position in the document, not by the order the cards arrived', () => {
+    const plan = planBatch(
+      [sug('late', 'a egg', 'an egg'), sug('early', 'a apple', 'an apple')],
+      finder('a apple and a egg'),
+    )
+    expect(plan.steps.map((s) => s.suggestion.id)).toEqual(['late', 'early'])
+  })
+
+  it('settles a card whose span is gone — she already fixed it herself', () => {
+    const plan = planBatch(
+      [sug('1', 'a apple', 'an apple'), sug('2', 'a orange', 'an orange')],
+      finder('an apple and a orange'),
+    )
+    expect(plan.steps.map((s) => s.suggestion.id)).toEqual(['2'])
+    expect(plan.missing.map((s) => s.id)).toEqual(['1'])
+  })
+
+  it('leaves the second of two cards quoting the same words on screen', () => {
+    // findRange resolves both to the first occurrence, so applying both would
+    // overwrite the first edit with the second. The batch takes one and says so.
+    const plan = planBatch(
+      [sug('1', 'a apple', 'an apple'), sug('2', 'a apple', 'the apple')],
+      finder('a apple, a apple'),
+    )
+    expect(plan.steps.map((s) => s.suggestion.id)).toEqual(['1'])
+    expect(plan.skipped.map((s) => s.id)).toEqual(['2'])
+    expect(plan.missing).toEqual([])
+  })
+
+  it('keeps both of two spans that merely touch', () => {
+    // Adjacent is not overlapping: "to" ends exactly where "the" begins.
+    const plan = planBatch(
+      [sug('1', 'aa', 'AA'), sug('2', 'bb', 'BB')],
+      finder('aabb'),
+    )
+    expect(plan.steps.map((s) => s.suggestion.id)).toEqual(['2', '1'])
+    expect(plan.skipped).toEqual([])
+  })
+
+  it('skips an awareness-only card with nothing to insert', () => {
+    const plan = planBatch([sug('v', 'a apple', '  ', 'voice')], finder('a apple'))
+    expect(plan.steps).toEqual([])
+    expect(plan.skipped.map((s) => s.id)).toEqual(['v'])
+  })
+
+  it('returns an empty plan for an empty family', () => {
+    const plan = planBatch([], finder('anything'))
+    expect(plan).toEqual({ steps: [], missing: [], skipped: [] })
+  })
+})
+
+describe('batchLeaders', () => {
+  const card = (id: string, type: SuggestionType) => ({ id, type })
+
+  it('gives each category exactly one leader, the first of its kind', () => {
+    const leaders = batchLeaders([
+      card('g1', 'grammar'),
+      card('m1', 'mechanics'),
+      card('g2', 'grammar'),
+      card('m2', 'mechanics'),
+      card('g3', 'grammar'),
+    ])
+    expect([...leaders]).toEqual(['g1', 'm1'])
+  })
+
+  it('follows the stacking order it is given, not the order types appear elsewhere', () => {
+    // The rail sorts by anchor, so "first" means highest on the page — the card
+    // she reads first, which is where the batch control belongs.
+    const leaders = batchLeaders([card('lower', 'grammar'), card('upper', 'grammar')])
+    expect([...leaders]).toEqual(['lower'])
+  })
+
+  it('leads a lone card of its type too — the count is what hides the button', () => {
+    expect([...batchLeaders([card('only', 'voice')])]).toEqual(['only'])
+  })
+
+  it('has no leaders for an empty stack', () => {
+    expect(batchLeaders([]).size).toBe(0)
+  })
+})

web/src/components/Editor/acceptBatch.ts

@@ -0,0 +1,84 @@
+import type { Suggestion, SuggestionType } from '../../api/client'
+
+// batchLeaders picks the one card per category that carries the "Accept all
+// Grammar (5)" control, given the cards in the order they are stacked. The rail
+// can't carry a category header — cards are anchored to their own sentence, so a
+// type's cards are scattered down the column — and the closest honest stand-in
+// is the first card of its kind. Offering the same batch on all five would be
+// five buttons saying one thing.
+export function batchLeaders(ordered: { type: SuggestionType; id: string }[]): Set<string> {
+  const leaders = new Set<string>()
+  const claimed = new Set<SuggestionType>()
+  for (const { type, id } of ordered) {
+    if (claimed.has(type)) continue
+    claimed.add(type)
+    leaders.add(id)
+  }
+  return leaders
+}
+
+export interface BatchStep {
+  suggestion: Suggestion
+  from: number
+  to: number
+}
+
+export interface BatchPlan {
+  // The edits to apply, ordered LAST span first. Every position is resolved
+  // against the document as it stands *before* any of them are applied, and
+  // applying from the end backwards means an earlier span's position can't be
+  // shifted by a later one — so the whole batch can go into a single
+  // transaction, which is the point: one Ctrl+Z puts it all back.
+  steps: BatchStep[]
+  // Cards whose span is no longer in the document — she fixed it herself, or
+  // edited around it since the check ran. There is nothing to replace, but the
+  // card is stale and settling it is what a single Accept already does.
+  missing: Suggestion[]
+  // Cards deliberately left on screen: two suggestions quoting the same words
+  // (findRange resolves both to the first occurrence, so applying the second
+  // would overwrite the first), and anything with nothing to insert. A batch
+  // that silently dropped these would report edits it never made.
+  skipped: Suggestion[]
+}
+
+// planBatch decides what one "Accept all <category>" click actually does. It is
+// pure and takes the span resolver as an argument so the ordering and overlap
+// rules can be tested without a ProseMirror document — the part that goes wrong
+// is the arithmetic, not the lookup.
+export function planBatch(
+  list: Suggestion[],
+  resolve: (original: string) => { from: number; to: number } | null,
+): BatchPlan {
+  const located: BatchStep[] = []
+  const missing: Suggestion[] = []
+  const skipped: Suggestion[] = []
+
+  for (const suggestion of list) {
+    if (suggestion.replacement.trim() === '') {
+      skipped.push(suggestion) // awareness-only (voice): nothing to accept
+      continue
+    }
+    const range = resolve(suggestion.original)
+    if (!range) {
+      missing.push(suggestion)
+      continue
+    }
+    located.push({ suggestion, from: range.from, to: range.to })
+  }
+
+  // Document order, then greedily keep the non-overlapping ones.
+  located.sort((a, b) => a.from - b.from || a.to - b.to)
+  const steps: BatchStep[] = []
+  let cursor = -Infinity
+  for (const step of located) {
+    if (step.from < cursor) {
+      skipped.push(step.suggestion)
+      continue
+    }
+    steps.push(step)
+    cursor = step.to
+  }
+  steps.reverse()
+
+  return { steps, missing, skipped }
+}

web/src/components/Editor/suggestionMeta.ts

@@ -28,3 +28,13 @@ export const TYPE_META: Record<SuggestionType, { color: string; label: string }>
 export function typeLabel(type: SuggestionType, pack: Pack): string {
   return type === 'translate' ? pack.editor.translateLabel : TYPE_META[type].label
 }
+
+// The label on the accept-a-whole-category button. It stays English even on a
+// translation card, whose pill is bilingual: the pill names the kind of advice
+// she is reading, while this names an action taken over the rest of the queue,
+// and every other word in the action row — Accept, Dismiss, Ask Petal — is
+// English too. A control that switched languages between cards would read as a
+// different control.
+export function batchLabel(type: SuggestionType, count: number): string {
+  return `Accept all ${TYPE_META[type].label} (${count})`
+}

web/src/index.css

@@ -314,6 +314,19 @@ button, a, input {
   color: var(--color-plum);
 }
 
+/* Accept-all: the quieter sibling of Accept, on both card surfaces. It's outlined
+   in its category's own colour rather than filled like Accept, because it acts on
+   cards she can't see from here — an equally loud button for a larger action would
+   invite the click she meant to give the one suggestion in front of her. */
+.petal-accept-all {
+  border: 1px solid;
+  background: transparent;
+  transition: background-color 120ms ease;
+}
+.petal-accept-all:hover {
+  background: var(--color-surface-alt);
+}
+
 /* --- Find & Replace ---------------------------------------------------------
    In-document search (Ctrl/Cmd+F). Every match gets a soft honey wash; the
    current match is brighter with a rose ring so it stands out as you step