fix(native-chat): separate literal and image echo identity

mobile/src/session/mobile-native-chat-draft-reconcile.test.ts

@@ -1,6 +1,7 @@
 import { describe, expect, it } from 'vitest'
 import type { NativeChatMessage } from '../../../src/shared/native-chat-types'
 import {
+  countUserTextOccurrences,
   findLandedImagePreviewEchoes,
   findLandedUnconfirmedSends,
   migrateImagePreviewMessageIds,
@@ -89,6 +90,35 @@ describe('mobile native chat image preview reconciliation', () => {
     ).toEqual([unconfirmed])
   })
 
+  it('does not treat literal marker text as a markerless unknown-ack echo', () => {
+    const unconfirmed: UnconfirmedSend = {
+      draftKey: 'draft',
+      pendingKey: 'pending-key',
+      text: 'keep literal',
+      normalizedText: 'keep literal',
+      imageCount: 0,
+      baselineTailMessageId: null,
+      deadline: null
+    }
+
+    expect(
+      findLandedUnconfirmedSends([userText('wrong', 'keep [Image #1] literal')], [unconfirmed])
+    ).toEqual([])
+    expect(findLandedUnconfirmedSends([userText('right', 'keep literal')], [unconfirmed])).toEqual([
+      unconfirmed
+    ])
+  })
+
+  it('keeps no-image occurrence baselines separate from attached-image turns', () => {
+    const messages = [
+      userText('source', '[Image: source: /tmp/a.png]'),
+      userText('prompt', 'keep [Image #1] literal')
+    ]
+
+    expect(countUserTextOccurrences(messages, 'keep literal')).toBe(0)
+    expect(countUserTextOccurrences(messages, 'keep literal', 1)).toBe(1)
+  })
+
   it('reconciles multiple transcript text blocks with desktop separators', () => {
     const prompt: NativeChatMessage = {
       ...userText('prompt', 'unused'),
@@ -114,6 +144,15 @@ describe('mobile native chat image preview reconciliation', () => {
     ).toEqual([{ pendingId: 'pending', messageId: 'prompt', images: ['file:///a.jpg'] }])
   })
 
+  it('hands a captioned preview to a marker-only compatibility echo', () => {
+    expect(
+      findLandedImagePreviewEchoes(
+        [userText('prompt', 'look [Image #1] here')],
+        [{ ...pending('pending', ['file:///a.jpg']), text: 'look here' }]
+      )
+    ).toEqual([{ pendingId: 'pending', messageId: 'prompt', images: ['file:///a.jpg'] }])
+  })
+
   it('keeps separate adjacent image-only sends independently reconcilable', () => {
     const landed = findLandedImagePreviewEchoes(
       [

mobile/src/session/mobile-native-chat-draft-reconcile.ts

@@ -34,8 +34,14 @@ export function countUserTextOccurrences(
 ): number {
   let count = 0
   for (const message of normalizeImageTranscriptMessages(messages)) {
+    const hasImageRefs = message.blocks.some(isImageRefBlock)
+    const matchText =
+      imageCount > 0
+        ? normalizedNativeChatUserMessageText(message)
+        : nativeChatUserMessageMatchText(message)
     if (
-      nativeChatUserMessageMatchText(message) === text &&
+      matchText === text &&
+      (imageCount > 0 || !hasImageRefs) &&
       nativeChatUserMessageImageEvidenceCount(message) >= imageCount
     ) {
       count++
@@ -176,7 +182,7 @@ export function findLandedImagePreviewEchoes(
       const imageCount = nativeChatUserMessageImageEvidenceCount(message)
       if (targetText) {
         return (
-          nativeChatUserMessageMatchText(message) === targetText &&
+          normalizedNativeChatUserMessageText(message) === targetText &&
           imageCount >= entry.images!.length
         )
       }
@@ -222,7 +228,11 @@ export function findLandedUnconfirmedSends(
   // (`[Image: source: …]` or no text) keys under '' so an empty-text send can
   // claim it.
   const messageIndexById = new Map(messages.map((message, index) => [message.id, index]))
-  const userMessagesByText = new Map<string, { id: string; index: number; imageCount: number }[]>()
+  const literalMessagesByText = new Map<
+    string,
+    { id: string; index: number; imageCount: number }[]
+  >()
+  const imageMessagesByText = new Map<string, { id: string; index: number; imageCount: number }[]>()
   for (const message of normalizeImageTranscriptMessages(messages)) {
     const index = messageIndexById.get(message.id)
     if (index === undefined) {
@@ -232,15 +242,19 @@ export function findLandedUnconfirmedSends(
       continue
     }
     const isImageSource = isImageSourceUserTurn(message)
-    const key = isImageSource ? '' : (normalizedUserText(message) ?? '')
-    // A marker-only turn keys both ways: '' claims it for a caption-less image
-    // send, and its literal text claims it for a send of that same text.
-    const literalKey = isImageSource ? key : (nativeChatUserMessageMatchText(message) ?? key)
     const imageCount = nativeChatUserMessageImageEvidenceCount(message)
-    for (const candidateKey of literalKey === key ? [key] : [key, literalKey]) {
-      const current = userMessagesByText.get(candidateKey) ?? []
-      current.push({ id: message.id, index, imageCount })
-      userMessagesByText.set(candidateKey, current)
+    const candidate = { id: message.id, index, imageCount }
+    if (!message.blocks.some(isImageRefBlock)) {
+      const literalKey = isImageSource ? '' : (nativeChatUserMessageMatchText(message) ?? '')
+      const current = literalMessagesByText.get(literalKey) ?? []
+      current.push(candidate)
+      literalMessagesByText.set(literalKey, current)
+    }
+    if (imageCount > 0) {
+      const imageKey = isImageSource ? '' : (normalizedUserText(message) ?? '')
+      const current = imageMessagesByText.get(imageKey) ?? []
+      current.push(candidate)
+      imageMessagesByText.set(imageKey, current)
     }
   }
 
@@ -253,7 +267,8 @@ export function findLandedUnconfirmedSends(
     if (tailIndex === undefined) {
       continue
     }
-    const echo = userMessagesByText
+    const messagesByText = entry.imageCount > 0 ? imageMessagesByText : literalMessagesByText
+    const echo = messagesByText
       .get(entry.normalizedText)
       ?.find(
         (message) =>

mobile/src/session/mobile-native-chat-literal-marker-echo.test.ts

@@ -130,6 +130,18 @@ describe('mobile literal image-marker turns', () => {
     expect(rows(exact).map(rowText)).toEqual(['keep [Image #1] literal'])
   })
 
+  it('does not let an attached-image send shift a markerless send ordinal', async () => {
+    await mount()
+    send('keep [Image #1] literal', ['file:///a.jpg'])
+    send('keep literal')
+
+    const messages = [userTextMessage('plain', 'keep literal')]
+    await transcript(messages)
+
+    expect(rows(messages).map(rowText)).toEqual(['keep literal', 'keep [Image #1] literal'])
+    expect(rows(messages).map(rowImages)).toEqual([[], ['file:///a.jpg']])
+  })
+
   it('holds every preview on the optimistic echo until the marker echo covers them all', async () => {
     await mount()
     send('', ['file:///a.jpg', 'file:///b.jpg'])

mobile/src/session/mobile-native-chat-pending-echo.ts

@@ -55,7 +55,9 @@ export function appendMobileNativeChatPending(
   const earlierOutstanding = current.filter(
     (pending) =>
       pendingMatchText(pending) === origin.normalizedText &&
-      (pending.images?.length ?? 0) >= origin.imageCount &&
+      (origin.imageCount > 0
+        ? (pending.images?.length ?? 0) >= origin.imageCount
+        : !pending.images?.length) &&
       pending.expectedOccurrence > origin.baselineOccurrences
   ).length
   // An image send whose caption is only markers still reconciles by image
@@ -63,6 +65,7 @@ export function appendMobileNativeChatPending(
   const expectedImageEchoOrdinal =
     current.filter(
       (pending) =>
+        origin.imageCount > 0 &&
         (pending.images?.length ?? 0) >= origin.imageCount &&
         normalizeNativeChatUserText(pending.text) === ''
     ).length + 1

mobile/src/session/mobile-native-chat-pending-retirement.test.ts

@@ -282,4 +282,28 @@ describe('retireLandedMobileNativeChatPending', () => {
     ).toEqual(['p1'])
     expect(retireLandedMobileNativeChatPending(messages, pending, new Set(['p1']))).toEqual([])
   })
+
+  it('keeps literal and attached-image turns in separate retirement lanes', () => {
+    const markerPending = [pendingSend('marker', 'keep [Image #1] literal', null)]
+    const plainPending = [pendingSend('plain', 'keep literal', null)]
+    const literalMarker = [userTurn('literal', 'keep [Image #1] literal', 1000)]
+    const attachedImage = [
+      userTurn('source', '[Image: source: /tmp/a.png]', 1000),
+      userTurn('prompt', 'keep [Image #1] literal', 1001)
+    ]
+
+    expect(
+      retireLandedMobileNativeChatPending(literalMarker, plainPending, NO_IMAGE_ECHOES).map(
+        (item) => item.id
+      )
+    ).toEqual(['plain'])
+    expect(
+      retireLandedMobileNativeChatPending(literalMarker, markerPending, NO_IMAGE_ECHOES)
+    ).toEqual([])
+    expect(
+      retireLandedMobileNativeChatPending(attachedImage, plainPending, NO_IMAGE_ECHOES).map(
+        (item) => item.id
+      )
+    ).toEqual(['plain'])
+  })
 })

mobile/src/session/mobile-native-chat-pending-retirement.ts

@@ -1,4 +1,4 @@
-import type { NativeChatMessage } from '../../../src/shared/native-chat-types'
+import { isImageRefBlock, type NativeChatMessage } from '../../../src/shared/native-chat-types'
 import { countImageSourceTurnsAfter } from './mobile-native-chat-draft-reconcile'
 import {
   nativeChatUserMessageMatchText,
@@ -33,7 +33,9 @@ export function selectGluedPendingIds(
     if (index === undefined) {
       continue
     }
-    const text = nativeChatUserMessageMatchText(message)
+    const text = message.blocks.some(isImageRefBlock)
+      ? null
+      : nativeChatUserMessageMatchText(message)
     if (text) {
       turns.push({ index, text })
     }
@@ -120,6 +122,9 @@ export function retireLandedMobileNativeChatPending(
 ): MobileNativeChatPendingMessage[] {
   const landedCounts = new Map<string, number>()
   for (const message of normalizeImageTranscriptMessages(messages)) {
+    if (message.blocks.some(isImageRefBlock)) {
+      continue
+    }
     const text = nativeChatUserMessageMatchText(message)
     if (text) {
       landedCounts.set(text, (landedCounts.get(text) ?? 0) + 1)

mobile/src/session/use-mobile-native-chat-drafts-launch-draft.test.ts

@@ -129,7 +129,7 @@ describe('useMobileNativeChatDrafts launch draft', () => {
       renderer?.update(
         createElement(Harness, {
           tabId: 'a',
-          messages: [userTextMessage('m1', '[Image #1]')],
+          messages: [userTextMessage('m1', 'keep [Image #1] literal')],
           launchDraft: 'issue link'
         })
       )
@@ -149,7 +149,7 @@ describe('useMobileNativeChatDrafts launch draft', () => {
       renderer?.update(
         createElement(Harness, {
           tabId: 'a',
-          messages: [userTextMessage('m1', '[Image #1]')],
+          messages: [userTextMessage('m1', 'keep [Image #1] literal')],
           launchDraft: 'issue link'
         })
       )

mobile/src/session/use-mobile-native-chat-drafts.test.ts

@@ -219,7 +219,7 @@ describe('useMobileNativeChatDrafts', () => {
         createElement(Harness, { tabId: 'a', messages: [assistantTextMessage('a1', 'hi')] })
       )
     )
-    const origin = state?.captureSendOrigin('')
+    const origin = state?.captureSendOrigin('', ['file:///a.jpg'])
     act(() => {
       if (origin) {
         state?.acceptSend(origin, '', ['file:///a.jpg'])
@@ -265,7 +265,7 @@ describe('useMobileNativeChatDrafts', () => {
       )
     )
     const textOrigin = state?.captureSendOrigin('ping')
-    const imageOrigin = state?.captureSendOrigin('')
+    const imageOrigin = state?.captureSendOrigin('', ['file:///a.jpg'])
     act(() => {
       if (textOrigin && imageOrigin) {
         state?.acceptSend(textOrigin, 'ping')
@@ -308,7 +308,7 @@ describe('useMobileNativeChatDrafts', () => {
         createElement(Harness, { tabId: 'a', messages: [assistantTextMessage('a1', 'hi')] })
       )
     )
-    const origin = state?.captureSendOrigin('look at this')
+    const origin = state?.captureSendOrigin('look at this', ['file:///a.jpg'])
     act(() => {
       if (origin) {
         state?.acceptSend(origin, 'look at this', ['file:///a.jpg'])
@@ -341,7 +341,7 @@ describe('useMobileNativeChatDrafts', () => {
         createElement(Harness, { tabId: 'a', messages: [assistantTextMessage('a1', 'hi')] })
       )
     )
-    const origin = state?.captureSendOrigin('look at this')
+    const origin = state?.captureSendOrigin('look at this', ['file:///a.jpg'])
     act(() => {
       if (origin) {
         state?.acceptSend(origin, 'look at this', ['file:///a.jpg'])
@@ -367,7 +367,7 @@ describe('useMobileNativeChatDrafts', () => {
 
   it('hands a marker-only image preview to the authoritative user bubble', async () => {
     await mount('a')
-    const origin = state?.captureSendOrigin('')
+    const origin = state?.captureSendOrigin('', ['file:///a.jpg'])
     act(() => {
       if (origin) {
         state?.acceptSend(origin, '', ['file:///a.jpg'])
@@ -475,7 +475,7 @@ describe('useMobileNativeChatDrafts', () => {
         )
       )
       // Image-only send: empty text, so it can only reconcile against a new user turn.
-      const origin = state?.captureSendOrigin('')
+      const origin = state?.captureSendOrigin('', ['file:///a.jpg'])
       const onUnconfirmed = vi.fn()
       act(() => {
         if (origin) {
@@ -492,7 +492,7 @@ describe('useMobileNativeChatDrafts', () => {
           })
         )
       )
-      // ...but the user's own turn landing does, so the deadline never warns.
+      // ...but the user's own image turn landing does, so the deadline never warns.
       await act(async () =>
         renderer?.update(
           createElement(Harness, {
@@ -500,7 +500,7 @@ describe('useMobileNativeChatDrafts', () => {
             messages: [
               assistantTextMessage('a1', 'hi'),
               assistantTextMessage('a2', 'ok'),
-              userTextMessage('u1', '')
+              userTextMessage('u1', '[Image #1]')
             ]
           })
         )
@@ -519,7 +519,7 @@ describe('useMobileNativeChatDrafts', () => {
         createElement(Harness, { tabId: 'a', messages: [assistantTextMessage('a1', 'hi')] })
       )
     )
-    const origin = state?.captureSendOrigin('')
+    const origin = state?.captureSendOrigin('', ['file:///a.jpg'])
     act(() => {
       if (origin) {
         state?.acceptSend(origin, '', ['file:///a.jpg'])
@@ -768,7 +768,7 @@ describe('useMobileNativeChatDrafts', () => {
     const images = ['file:///a.jpg', 'file:///b.jpg', 'file:///c.jpg']
     act(() => state?.setComposerText('look'))
 
-    const origin = state?.captureSendOrigin('look')
+    const origin = state?.captureSendOrigin('look', images)
     expect(origin).toMatchObject({ pendingKey: null })
     act(() => {
       if (origin) {

src/renderer/src/components/native-chat/native-chat-literal-marker-echo.test.ts

@@ -88,6 +88,27 @@ describe('literal image-marker turns in native chat', () => {
     ])
   })
 
+  it('does not let literal or attached-image marker rows hide markerless text', () => {
+    const pending: NativeChatPendingSend[] = [{ id: 'p1', text: 'keep literal', sentAt: 100 }]
+    const literalTranscript = [userMessage('literal', 'keep [Image #1] literal')]
+    const imageTranscript = [
+      userMessage('source', '[Image: source: /tmp/a.png]'),
+      userMessage('prompt', 'keep[Image #1] literal')
+    ]
+
+    expect(renderedUserRows(literalTranscript, pending).map(rowText)).toEqual([
+      'keep [Image #1] literal',
+      'keep literal'
+    ])
+    expect(renderedUserRows(imageTranscript, pending).map(rowText)).toEqual([
+      'keep literal',
+      'keep literal'
+    ])
+    expect(
+      prunePendingSends(pending, assembled([...imageTranscript, assistantMessage('a1', 'done')]))
+    ).toEqual(pending)
+  })
+
   it('prunes the marker-only echo once the assistant answers it', () => {
     const pending: NativeChatPendingSend[] = [{ id: 'p1', text: '[Image #1]', sentAt: 100 }]
     const transcript = [userMessage('m1', '[Image #1]'), assistantMessage('m2', 'preserved')]

src/renderer/src/components/native-chat/native-chat-pending-occurrence.ts

@@ -51,7 +51,8 @@ function nativeChatUserMessageContentKeys(message: NativeChatMessage): readonly
   }
   const keys = new Set<string>()
   const matchText = nativeChatUserMessageMatchText(message) ?? ''
-  if (matchText) {
+  const hasImageRefs = message.blocks.some(isImageRefBlock)
+  if (matchText && !hasImageRefs) {
     keys.add(nativeChatTextContentKey(matchText, 0))
   }
   const imageCount = nativeChatUserMessageImageEvidenceCount(message)
@@ -110,7 +111,9 @@ export function advancedNativeChatUserTexts(
   const waiting: string[] = []
   for (const message of messages) {
     if (message.role === 'user') {
-      const text = nativeChatUserMessageMatchText(message)
+      const text = message.blocks.some(isImageRefBlock)
+        ? null
+        : nativeChatUserMessageMatchText(message)
       if (text) {
         waiting.push(text)
       }
@@ -128,7 +131,9 @@ export function matchingNativeChatUserTexts(
 ): readonly string[] {
   const texts: string[] = []
   for (const message of messages) {
-    const text = nativeChatUserMessageMatchText(message)
+    const text = message.blocks.some(isImageRefBlock)
+      ? null
+      : nativeChatUserMessageMatchText(message)
     if (text) {
       texts.push(text)
     }

src/renderer/src/components/native-chat/native-chat-pending.test.ts

@@ -116,7 +116,7 @@ describe('prunePendingSends', () => {
       ]
     }
     const next = prunePendingSends(
-      [pendingOf('p1', 'what do you see')],
+      [{ ...pendingOf('p1', 'what do you see'), imagePaths: ['/tmp/a.png'] }],
       [prompt, assistantMessage('m2', 'an image')]
     )
 
@@ -468,7 +468,7 @@ describe('launchPromptAsMessage', () => {
     ).toBeNull()
   })
 
-  it('uses pending-send normalization for large multiline generated prompts', () => {
+  it('does not strip literal markers from launch-prompt identity', () => {
     const prompt = [
       '[Image #1] Resolve the failing checks:',
       '',
@@ -488,17 +488,17 @@ describe('launchPromptAsMessage', () => {
       { ...assistantMessage('a1', 'I will fix it'), timestamp: 44 }
     ]
 
-    expect(
-      shouldPruneLaunchPrompt(
-        {
-          tabId: 'tab-1',
-          agent: 'codex',
-          text: prompt,
-          createdAt: 42
-        },
-        transcript
-      )
-    ).toBe(true)
+    const entry = { tabId: 'tab-1', agent: 'codex' as const, text: prompt, createdAt: 42 }
+
+    expect(launchPromptAsMessage(entry, transcript)).not.toBeNull()
+    expect(shouldPruneLaunchPrompt(entry, transcript)).toBe(false)
+
+    const exact = [
+      { ...userMessage('u1', prompt), timestamp: 43 },
+      { ...assistantMessage('a1', 'I will fix it'), timestamp: 44 }
+    ]
+    expect(launchPromptAsMessage(entry, exact)).toBeNull()
+    expect(shouldPruneLaunchPrompt(entry, exact)).toBe(true)
   })
 
   it('keeps the launch prompt until the transcript advances past the user turn', () => {

src/shared/native-chat-image-transcript-markers.test.ts

@@ -2,6 +2,7 @@ import { describe, expect, it } from 'vitest'
 import type { NativeChatMessage } from './native-chat-types'
 import {
   isImageSourceUserTurn,
+  nativeChatUserMessageMatchText,
   nativeChatUserTextMatchText,
   normalizeImageTranscriptMessages,
   normalizeNativeChatUserText,
@@ -110,6 +111,14 @@ describe('normalizeImageTranscriptMessages', () => {
       'keep [Image #1] literal'
     )
     expect(nativeChatUserTextMatchText(' keep [Image #1] literal ', true)).toBe('keep literal')
+
+    const literal = userText('literal', 'keep [Image #1] literal')
+    const attached = {
+      ...literal,
+      blocks: [{ type: 'image-ref' as const, path: '/tmp/a.png' }, ...literal.blocks]
+    }
+    expect(nativeChatUserMessageMatchText(literal)).toBe('keep [Image #1] literal')
+    expect(nativeChatUserMessageMatchText(attached)).toBe('keep literal')
   })
 
   it('recognizes only sole-text image-source user turns', () => {