diff --git a/mobile/src/session/mobile-dictation-refused-start-routing.test.tsx b/mobile/src/session/mobile-dictation-refused-start-routing.test.tsx
index b073eda047d..41dc8b5d11f 100644
--- a/mobile/src/session/mobile-dictation-refused-start-routing.test.tsx
+++ b/mobile/src/session/mobile-dictation-refused-start-routing.test.tsx
@@ -92,7 +92,7 @@ function mount(client: FakeRpcClient): Mounted {
       worktreeId: 'w1',
       client,
       connState: 'connected',
-      agentSessionPromptCancelSupported: true,
+      agentSessionHostSupport: { promptCancel: true, questionAnswers: false },
       setInput: () => {},
       liveInputTerminalHandles: new Set<string>(),
       activeHandle: 't1',
diff --git a/mobile/src/session/mobile-session-route-parity.test.ts b/mobile/src/session/mobile-session-route-parity.test.ts
index e2f8b478cc0..7523976132b 100644
--- a/mobile/src/session/mobile-session-route-parity.test.ts
+++ b/mobile/src/session/mobile-session-route-parity.test.ts
@@ -93,7 +93,8 @@ const HOST_COMPONENT_NAMES = new Set([
 // Moved, count unchanged, when the Markdown actions' Back `useEffect` became `useBackClaim`, the
 // seam that also claims the key on the page while a draft is dirty.
 const HEAD_MAIN_HOOK_SHA256 = 'f161e14a9c53d80c3dc75f51dd8ecb339b59b7239c9c3e067791b8612f51ede2'
-const HEAD_HOOK_BINDING_SHA256 = '7a9e256e2058635850253a74b9faa05bcafbbcaa1e5b0b3ef7f677dde85753a2'
+// Moved when the prompt-cancel flag became one structured-session host support object.
+const HEAD_HOOK_BINDING_SHA256 = 'db9f32cc60fc68adbcbb2acf9f9384ad0d78bbc6feef5d581449fadb647405fc'
 const HEAD_CALLBACK_IDENTITY_SHA256 =
   '373dca17a060e63d8cb4e32416ca2889b8404cee78f7b47e632940a9980baf23'
 // Pins that no callback body in the route changed unnoticed. Body text, not behaviour: the sends
@@ -125,7 +126,8 @@ const HEAD_CALLBACK_BODY_SHA256 = '2ccbfb5ee57e7dfeb07dafaee6fa592b95862b3bc898a
 // the diff-comments effect, which now catches the loader's rejection. Count unchanged.
 // Moved again by the keyboard seam above, which is the +1 effect.
 // -1 effect for the Markdown actions' Back registration, which is `useBackClaim`'s own now.
-const HEAD_EFFECT_SHA256 = '69096e20a44fa03a2c364e00a617eeabc437a79bb850b9135408d35fbd4d4a71'
+// Moved by the capability probe setting that host support object.
+const HEAD_EFFECT_SHA256 = '9b045a547ed269acf95db16cc87e33a9035a20c6888fd30e0363e58bb6b7d883'
 const HEAD_CONTENT_HOOK_SHA256 = '9c3b612fef3f370d66873aefdbe1d701f20cb64ded31fef5cc45fde6f8189581'
 // Same pin for the 12 bodies that sit in nested functions rather than callbacks, moved by the same
 // rewrite of those send and read expressions. Count unchanged. Refreshed again in step 6 for
@@ -181,7 +183,9 @@ const HEAD_STYLE_REFERENCE_SHA256 =
 const HEAD_IDENTITY_FIELD_SHA256 =
   '91146853930a34dd1f3d80e5c97fbacd7cf19fb93dd26fe8fc6f29169622f9d6'
 const HEAD_NAVIGATION_SHA256 = '9d96f5dad7de555d6553eac39c0fab00efad507470fd562cb9beaa32db16f512'
-const HEAD_CAPABILITY_SHA256 = '67c3154b71b542bb63a4365d3ea75aef19ef133c02f509318619618221786fab'
+// Moved when structured-session features became one helper call; the capability strings it reads
+// are pinned by that helper's own test.
+const HEAD_CAPABILITY_SHA256 = 'ec1159d6e726383bf7121e9c642657fee5b4ebec049cb44ba132a303bb9a61e2'
 
 type Definition = { declaration: ts.FunctionDeclaration; sourceFile: ts.SourceFile }
 type HookFacts = {
@@ -550,7 +554,11 @@ function readCompatibilityFacts(definitions: ReadonlyMap<string, Definition>): {
         : ''
     const callText = canonical(node, sourceFile)
     if (
-      ['startRuntimeCapabilityProbe', 'supportsMobileQuickCommands'].includes(callName) ||
+      [
+        'startRuntimeCapabilityProbe',
+        'supportsMobileQuickCommands',
+        'structuredAgentSessionHostSupport'
+      ].includes(callName) ||
       (callName === 'includes' && callText.includes('capabilities.includes'))
     ) {
       capabilities.push(callText)
diff --git a/mobile/src/session/mobile-structured-agent-prompts.ts b/mobile/src/session/mobile-structured-agent-prompts.ts
index d11a2fc0317..649fc06b774 100644
--- a/mobile/src/session/mobile-structured-agent-prompts.ts
+++ b/mobile/src/session/mobile-structured-agent-prompts.ts
@@ -1,4 +1,8 @@
 import type { AgentJournalRenderItem } from '../../../src/shared/agent-session-journal-types'
+import {
+  agentSessionPromptQuestions,
+  type AgentSessionQuestionAnswer
+} from '../../../src/shared/agent-session-question-answer'
 import type { MobileChatPermission } from './mobile-native-chat-permission'
 import type { MobileChatQuestion } from './mobile-native-chat-question'
 import {
@@ -21,6 +25,12 @@ export type StructuredPromptResponseTarget = {
   optionId: string
 }
 
+export type StructuredQuestionResponseTarget = {
+  itemId: string
+  expectedRevision: number
+  answer: AgentSessionQuestionAnswer
+}
+
 type PromptTokenPayload =
   | {
       kind: 'approval'
@@ -32,6 +42,7 @@ type PromptTokenPayload =
       kind: 'question-option'
       itemId: string
       revision: number
+      questionId: string
       optionId: string
     }
   | {
@@ -55,10 +66,6 @@ export function pendingStructuredQuestion(
   return item.body.kind === 'question' && item.body.resolution.state === 'pending'
 }
 
-function encodeQuestionAnswer(questionId: string, answer: string): string {
-  return `${encodeURIComponent(questionId)}:${encodeURIComponent(answer)}`
-}
-
 function encodePromptToken(payload: PromptTokenPayload): string {
   return `${STRUCTURED_PROMPT_TOKEN_PREFIX}${encodeURIComponent(JSON.stringify(payload))}`
 }
@@ -86,11 +93,16 @@ function decodePromptToken(value: string): PromptTokenPayload | null {
         optionId: decoded.optionId
       }
     }
-    if (decoded.kind === 'question-option' && typeof decoded.optionId === 'string') {
+    if (
+      decoded.kind === 'question-option' &&
+      typeof decoded.questionId === 'string' &&
+      typeof decoded.optionId === 'string'
+    ) {
       return {
         kind: decoded.kind,
         itemId: decoded.itemId,
         revision: decoded.revision,
+        questionId: decoded.questionId,
         optionId: decoded.optionId
       }
     }
@@ -169,6 +181,10 @@ export function projectStructuredQuestion(
       { itemId: prompt.itemId, expectedRevision: prompt.revision }
     )
   }
+  const [question] = agentSessionPromptQuestions(prompt.body)
+  if (!question) {
+    return null
+  }
   const optionDescriptions = prompt.body.options.map((option) => option.description)
   return {
     question: prompt.body.question,
@@ -182,6 +198,7 @@ export function projectStructuredQuestion(
         kind: 'question-option',
         itemId: prompt.itemId,
         revision: prompt.revision,
+        questionId: question.id,
         optionId: option.id
       })
     ),
@@ -228,13 +245,13 @@ export function structuredApprovalResponseTarget(
 export function structuredQuestionResponseTarget(
   response: string,
   currentPrompt: StructuredQuestionItem | null
-): StructuredPromptResponseTarget | null {
+): StructuredQuestionResponseTarget | null {
   const token = decodePromptToken(response)
   if (token?.kind === 'question-option') {
     return {
       itemId: token.itemId,
       expectedRevision: token.revision,
-      optionId: token.optionId
+      answer: { questionId: token.questionId, optionIds: [token.optionId] }
     }
   }
   if (token) {
@@ -247,29 +264,30 @@ export function structuredQuestionResponseTarget(
       ? {
           itemId: freeText.payload.itemId,
           expectedRevision: freeText.payload.revision,
-          optionId: encodeQuestionAnswer(freeText.payload.questionId, answer)
+          answer: { questionId: freeText.payload.questionId, optionIds: [], other: answer }
         }
       : null
   }
   if (!currentPrompt) {
     return null
   }
+  const [question] = agentSessionPromptQuestions(currentPrompt.body)
   const trimmed = response.trim()
   const option = currentPrompt.body.options.find(
     (candidate) => candidate.id === response || candidate.label === trimmed
   )
-  if (option) {
+  if (question && option) {
     return {
       itemId: currentPrompt.itemId,
       expectedRevision: currentPrompt.revision,
-      optionId: option.id
+      answer: { questionId: question.id, optionIds: [option.id] }
     }
   }
-  return currentPrompt.body.freeTextQuestionId && trimmed
+  return question && currentPrompt.body.freeTextQuestionId && trimmed
     ? {
         itemId: currentPrompt.itemId,
         expectedRevision: currentPrompt.revision,
-        optionId: encodeQuestionAnswer(currentPrompt.body.freeTextQuestionId, trimmed)
+        answer: { questionId: question.id, optionIds: [], other: trimmed }
       }
     : null
 }
diff --git a/mobile/src/session/mobile-structured-agent-session-host-support.test.ts b/mobile/src/session/mobile-structured-agent-session-host-support.test.ts
new file mode 100644
index 00000000000..6857855c9b4
--- /dev/null
+++ b/mobile/src/session/mobile-structured-agent-session-host-support.test.ts
@@ -0,0 +1,21 @@
+import { describe, expect, it } from 'vitest'
+import {
+  AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY,
+  AGENT_SESSION_QUESTION_ANSWERS_RUNTIME_CAPABILITY
+} from '../../../src/shared/protocol-version'
+import { structuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
+
+describe('structuredAgentSessionHostSupport', () => {
+  it('reads each structured-session feature from the host capability list', () => {
+    expect(structuredAgentSessionHostSupport([])).toEqual({
+      promptCancel: false,
+      questionAnswers: false
+    })
+    expect(
+      structuredAgentSessionHostSupport([AGENT_SESSION_QUESTION_ANSWERS_RUNTIME_CAPABILITY])
+    ).toEqual({ promptCancel: false, questionAnswers: true })
+    expect(
+      structuredAgentSessionHostSupport([AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY])
+    ).toEqual({ promptCancel: true, questionAnswers: false })
+  })
+})
diff --git a/mobile/src/session/mobile-structured-agent-session-host-support.ts b/mobile/src/session/mobile-structured-agent-session-host-support.ts
new file mode 100644
index 00000000000..939f7d559f4
--- /dev/null
+++ b/mobile/src/session/mobile-structured-agent-session-host-support.ts
@@ -0,0 +1,19 @@
+import {
+  AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY,
+  AGENT_SESSION_QUESTION_ANSWERS_RUNTIME_CAPABILITY
+} from '../../../src/shared/protocol-version'
+
+/** Structured-session features the connected host advertised; null until the status probe answers. */
+export type StructuredAgentSessionHostSupport = {
+  promptCancel: boolean
+  questionAnswers: boolean
+}
+
+export function structuredAgentSessionHostSupport(
+  capabilities: readonly string[]
+): StructuredAgentSessionHostSupport {
+  return {
+    promptCancel: capabilities.includes(AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY),
+    questionAnswers: capabilities.includes(AGENT_SESSION_QUESTION_ANSWERS_RUNTIME_CAPABILITY)
+  }
+}
diff --git a/mobile/src/session/mobile-structured-grouped-question.test.ts b/mobile/src/session/mobile-structured-grouped-question.test.ts
index 45a365864d9..37a4dc125c2 100644
--- a/mobile/src/session/mobile-structured-grouped-question.test.ts
+++ b/mobile/src/session/mobile-structured-grouped-question.test.ts
@@ -1,6 +1,5 @@
 import { describe, expect, it } from 'vitest'
 import type { AgentJournalQuestion } from '../../../src/shared/agent-session-journal-types'
-import { decodeAgentSessionQuestionAnswers } from '../../../src/shared/agent-session-question-answer'
 import {
   formatQuestionAnswer,
   formatQuestionFreeTextAnswer,
@@ -85,7 +84,7 @@ describe('mobile structured grouped questions', () => {
     expect(second).toMatchObject({ question: 'Which regions? (2 of 2)', multiSelect: true })
   })
 
-  it('submits the whole group as one encoded answer on the last step', () => {
+  it('submits the whole group as one set of answers on the last step', () => {
     const questions = [question(), SECOND]
     const draft: GroupedQuestionDraft = {
       promptKey: PROMPT_KEY,
@@ -102,9 +101,7 @@ describe('mobile structured grouped questions', () => {
     })
 
     expect(result?.kind).toBe('submit')
-    expect(
-      decodeAgentSessionQuestionAnswers(result?.kind === 'submit' ? result.optionId : '')
-    ).toEqual([
+    expect(result?.kind === 'submit' ? result.answers : null).toEqual([
       { questionId: 'q1', optionIds: ['q1:choice-1'] },
       { questionId: 'q2', optionIds: ['q2:choice-1', 'q2:choice-2'] }
     ])
@@ -121,9 +118,9 @@ describe('mobile structured grouped questions', () => {
       promptKey: PROMPT_KEY
     })
 
-    expect(
-      decodeAgentSessionQuestionAnswers(result?.kind === 'submit' ? result.optionId : '')
-    ).toEqual([{ questionId: 'q1', optionIds: [], other: 'DuckDB' }])
+    expect(result?.kind === 'submit' ? result.answers : null).toEqual([
+      { questionId: 'q1', optionIds: [], other: 'DuckDB' }
+    ])
   })
 
   it('keeps selected options and other text for grouped multi-select answers', () => {
@@ -137,9 +134,9 @@ describe('mobile structured grouped questions', () => {
       promptKey: PROMPT_KEY
     })
 
-    expect(
-      decodeAgentSessionQuestionAnswers(result?.kind === 'submit' ? result.optionId : '')
-    ).toEqual([{ questionId: 'q2', optionIds: ['q2:choice-1'], other: 'ap-south' }])
+    expect(result?.kind === 'submit' ? result.answers : null).toEqual([
+      { questionId: 'q2', optionIds: ['q2:choice-1'], other: 'ap-south' }
+    ])
   })
 
   it('gives each step a distinct card key so a selection cannot carry into the next question', () => {
diff --git a/mobile/src/session/mobile-structured-grouped-question.ts b/mobile/src/session/mobile-structured-grouped-question.ts
index 031cdf3f10f..d111b4de95a 100644
--- a/mobile/src/session/mobile-structured-grouped-question.ts
+++ b/mobile/src/session/mobile-structured-grouped-question.ts
@@ -1,6 +1,5 @@
 import type { AgentJournalQuestion } from '../../../src/shared/agent-session-journal-types'
 import {
-  encodeAgentSessionQuestionAnswers,
   isValidAgentSessionQuestionAnswers,
   type AgentSessionQuestionAnswer
 } from '../../../src/shared/agent-session-question-answer'
@@ -11,7 +10,7 @@ import type { MobileChatQuestion } from './mobile-native-chat-question'
  * prompt. The host then leaves the flat `question.options` EMPTY and puts the real content in
  * `questions`, so a client that reads only the flat shape renders an unanswerable card and the turn
  * stalls. The phone has room for one question at a time, so the group is answered as steps and
- * submitted once — the host accepts the whole group as one encoded option id.
+ * submitted once as one set of answers.
  */
 export type GroupedQuestionDraft = {
   /** Identifies the exact prompt revision these answers belong to; a revised prompt discards them. */
@@ -21,7 +20,7 @@ export type GroupedQuestionDraft = {
 
 export type GroupedQuestionAdvance =
   | { kind: 'advance'; draft: GroupedQuestionDraft }
-  | { kind: 'submit'; optionId: string }
+  | { kind: 'submit'; answers: AgentSessionQuestionAnswer[] }
 
 const GROUPED_TOKEN_PREFIX = 'structured-grouped-question:'
 
@@ -195,7 +194,7 @@ function answerFromResponse(
 
 /**
  * Fold one answer into the draft. Returns `advance` while questions remain and `submit` with the
- * encoded group once the last one lands; null when the response does not answer this prompt step.
+ * whole group once the last one lands; null when the response does not answer this prompt step.
  */
 export function advanceGroupedQuestion(args: {
   response: string
@@ -218,6 +217,6 @@ export function advanceGroupedQuestion(args: {
   }
   // Never send a group the host would refuse — the user would see a silent failure with no way back.
   return isValidAgentSessionQuestionAnswers(args.questions, answers)
-    ? { kind: 'submit', optionId: encodeAgentSessionQuestionAnswers(answers) }
+    ? { kind: 'submit', answers }
     : null
 }
diff --git a/mobile/src/session/use-mobile-native-chat-controller.ts b/mobile/src/session/use-mobile-native-chat-controller.ts
index a3a49e53552..dfefcf47627 100644
--- a/mobile/src/session/use-mobile-native-chat-controller.ts
+++ b/mobile/src/session/use-mobile-native-chat-controller.ts
@@ -2,6 +2,7 @@ import { useLayoutEffect, useRef, type MutableRefObject } from 'react'
 import type { RpcClient } from '../transport/rpc-client'
 import type { ConnectionState } from '../transport/types'
 import type { MobileNativeChatTab } from './mobile-native-chat-eligibility'
+import type { StructuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
 import { useMobileNativeChatAskDismiss } from './use-mobile-native-chat-ask-dismiss'
 import { useMobileNativeChatDrafts } from './use-mobile-native-chat-drafts'
 import { useMobileNativeChatFileSearch } from './use-mobile-native-chat-file-search'
@@ -36,7 +37,7 @@ export function useMobileNativeChatController(args: {
   /** Live socket state; the lease collapses on disconnect but one render later. */
   connState: ConnectionState
   /** Host capability fact from the shared runtime status probe. */
-  agentSessionPromptCancelSupported?: boolean | null
+  agentSessionHostSupport?: StructuredAgentSessionHostSupport | null
   onSendError: (message: string) => void
   /** Retires a held failure banner. Any accepted chat write clears it — a delivered
    *  answer or permission reply must not sit under a stale "not sent". */
@@ -53,7 +54,7 @@ export function useMobileNativeChatController(args: {
     nativeChatTranscriptIsLocalReadable,
     nativeChatInputLeaseReady,
     connState,
-    agentSessionPromptCancelSupported = null,
+    agentSessionHostSupport = null,
     onSendError,
     onSendResolved
   } = args
@@ -93,7 +94,7 @@ export function useMobileNativeChatController(args: {
       callerIdentity: deviceTokenRef.current ?? '',
       enabled: showNativeChat,
       connState,
-      promptCancelSupported: agentSessionPromptCancelSupported,
+      hostSupport: agentSessionHostSupport,
       onSendError
     })
   const {
diff --git a/mobile/src/session/use-mobile-native-chat-session-lane.ts b/mobile/src/session/use-mobile-native-chat-session-lane.ts
index bfeb1b06945..147cf017ac0 100644
--- a/mobile/src/session/use-mobile-native-chat-session-lane.ts
+++ b/mobile/src/session/use-mobile-native-chat-session-lane.ts
@@ -1,5 +1,6 @@
 import type { RpcClient } from '../transport/rpc-client'
 import type { ConnectionState } from '../transport/types'
+import type { StructuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
 import { useMobileNativeChatSession } from './use-mobile-native-chat-session'
 import { useMobileStructuredAgentSession } from './use-mobile-structured-agent-session'
 
@@ -15,7 +16,7 @@ export function useMobileNativeChatSessionLane({
   sessionId,
   sourceIdentity,
   callerIdentity,
-  promptCancelSupported,
+  hostSupport,
   enabled,
   connState,
   onSendError
@@ -30,7 +31,7 @@ export function useMobileNativeChatSessionLane({
   sessionId: string | null
   sourceIdentity: Parameters<typeof useMobileNativeChatSession>[0]['sourceIdentity']
   callerIdentity: string
-  promptCancelSupported?: boolean | null
+  hostSupport: StructuredAgentSessionHostSupport | null
   enabled: boolean
   connState: ConnectionState
   onSendError: (message: string) => void
@@ -50,7 +51,7 @@ export function useMobileNativeChatSessionLane({
     sessionId: structured ? sessionId : null,
     sourceIdentity,
     callerIdentity,
-    promptCancelSupported,
+    hostSupport,
     enabled,
     // Holds are connection-scoped; dropping this on transport loss lets the hook
     // reacquire the provider without clearing the cached transcript.
diff --git a/mobile/src/session/use-mobile-session-feedback-capabilities.ts b/mobile/src/session/use-mobile-session-feedback-capabilities.ts
index f0231189309..ea8d6438bf4 100644
--- a/mobile/src/session/use-mobile-session-feedback-capabilities.ts
+++ b/mobile/src/session/use-mobile-session-feedback-capabilities.ts
@@ -2,6 +2,7 @@ import { useState, useRef, useCallback } from 'react'
 import { Animated } from 'react-native'
 import { reconcileMobileSessionCreateWarningState } from './mobile-session-create-warning-state'
 import type { MobileSessionTerminalRuntimeModel } from './use-mobile-session-terminal-runtime'
+import type { StructuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
 
 export function useMobileSessionFeedbackCapabilities(scope: MobileSessionTerminalRuntimeModel) {
   const {
@@ -32,11 +33,10 @@ export function useMobileSessionFeedbackCapabilities(scope: MobileSessionTermina
     null
   )
   const [quickCommandsSupported, setQuickCommandsSupported] = useState<boolean | null>(null)
-  // Prompt cancellation is negotiated with the same host capability probe as
+  // Structured-session features are negotiated with the same host capability probe as
   // the other session surfaces; consumers never maintain a second status cache.
-  const [agentSessionPromptCancelSupported, setAgentSessionPromptCancelSupported] = useState<
-    boolean | null
-  >(null)
+  const [agentSessionHostSupport, setAgentSessionHostSupport] =
+    useState<StructuredAgentSessionHostSupport | null>(null)
   // Why: stable callbacks (handleFileTap) read the live value via this ref, since
   // the capability probe resolves after the callbacks are created.
   const browserScreencastSupportedRef = useRef(browserScreencastSupported)
@@ -120,8 +120,8 @@ export function useMobileSessionFeedbackCapabilities(scope: MobileSessionTermina
     setAgentSessionHistorySupported,
     quickCommandsSupported,
     setQuickCommandsSupported,
-    agentSessionPromptCancelSupported,
-    setAgentSessionPromptCancelSupported,
+    agentSessionHostSupport,
+    setAgentSessionHostSupport,
     browserScreencastSupportedRef,
     reconciledCreateWarningState,
     createWarning,
diff --git a/mobile/src/session/use-mobile-session-native-chat-dictation.ts b/mobile/src/session/use-mobile-session-native-chat-dictation.ts
index 8ea9de2b9f0..4dc1743e221 100644
--- a/mobile/src/session/use-mobile-session-native-chat-dictation.ts
+++ b/mobile/src/session/use-mobile-session-native-chat-dictation.ts
@@ -27,7 +27,7 @@ export function useMobileSessionNativeChatDictation(
     worktreeId,
     client,
     connState,
-    agentSessionPromptCancelSupported,
+    agentSessionHostSupport,
     setInput,
     liveInputTerminalHandles,
     activeHandle,
@@ -73,7 +73,7 @@ export function useMobileSessionNativeChatDictation(
     nativeChatTranscriptIsLocalReadable,
     nativeChatInputLeaseReady,
     connState,
-    agentSessionPromptCancelSupported,
+    agentSessionHostSupport,
     onSendError: nativeChatSendError.show,
     onSendResolved: nativeChatSendError.clear
   })
diff --git a/mobile/src/session/use-mobile-session-tab-reconciliation.ts b/mobile/src/session/use-mobile-session-tab-reconciliation.ts
index da7a48e035c..9f9af59833d 100644
--- a/mobile/src/session/use-mobile-session-tab-reconciliation.ts
+++ b/mobile/src/session/use-mobile-session-tab-reconciliation.ts
@@ -2,10 +2,8 @@ import { useEffect, useRef, useCallback, useMemo, useState } from 'react'
 import { startRuntimeCapabilityProbe } from '../transport/runtime-capability-probe'
 import { supportsMobileQuickCommands } from '../terminal/quick-commands'
 import { MOBILE_AI_VAULT_CAPABILITY } from '../agent-history/agent-history-capability'
-import {
-  AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY,
-  TERMINAL_QUERY_REPLY_INPUT_RUNTIME_CAPABILITY
-} from '../../../src/shared/protocol-version'
+import { TERMINAL_QUERY_REPLY_INPUT_RUNTIME_CAPABILITY } from '../../../src/shared/protocol-version'
+import { structuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
 import { runAcceptedMobileSessionTabsEffects } from './mobile-session-tabs-accepted-effects'
 import type { SessionTabsStreamSource } from './mobile-session-tabs-stream-health'
 import { useMobileSessionTabsFetchReporting } from './use-mobile-session-tabs-fetch-reporting'
@@ -34,7 +32,7 @@ export function useMobileSessionTabReconciliation(scope: MobileSessionMarkdownAc
     switchSessionTabRef,
     setBrowserScreencastSupported,
     setAgentSessionHistorySupported,
-    setAgentSessionPromptCancelSupported,
+    setAgentSessionHostSupport,
     setQuickCommandsSupported,
     nativeChatStream,
     fetchTerminals,
@@ -152,7 +150,7 @@ export function useMobileSessionTabReconciliation(scope: MobileSessionMarkdownAc
     if (!client || connState !== 'connected') {
       setBrowserScreencastSupported(null)
       setAgentSessionHistorySupported(null)
-      setAgentSessionPromptCancelSupported(null)
+      setAgentSessionHostSupport(null)
       setQuickCommandsSupported(null)
       setShowQuickCommands(false)
       hostQueryReplyInputSupportedRef.current = false
@@ -162,7 +160,7 @@ export function useMobileSessionTabReconciliation(scope: MobileSessionMarkdownAc
     // host; clear the prior capability before exposing host-specific actions.
     setBrowserScreencastSupported(null)
     setAgentSessionHistorySupported(null)
-    setAgentSessionPromptCancelSupported(null)
+    setAgentSessionHostSupport(null)
     setQuickCommandsSupported(null)
     setShowQuickCommands(false)
     hostQueryReplyInputSupportedRef.current = false
@@ -171,9 +169,7 @@ export function useMobileSessionTabReconciliation(scope: MobileSessionMarkdownAc
     return startRuntimeCapabilityProbe(client, (capabilities) => {
       setBrowserScreencastSupported(capabilities.includes('browser.screencast.v1'))
       setAgentSessionHistorySupported(capabilities.includes(MOBILE_AI_VAULT_CAPABILITY))
-      setAgentSessionPromptCancelSupported(
-        capabilities.includes(AGENT_SESSION_PROMPT_CANCEL_RUNTIME_CAPABILITY)
-      )
+      setAgentSessionHostSupport(structuredAgentSessionHostSupport(capabilities))
       setQuickCommandsSupported(supportsMobileQuickCommands(capabilities))
       // Why: hosts without this capability strip inputKind from terminal.send,
       // so a forwarded xterm reply would become floor-stealing shell input.
diff --git a/mobile/src/session/use-mobile-structured-agent-session-prompt-cancel.test.tsx b/mobile/src/session/use-mobile-structured-agent-session-prompt-cancel.test.tsx
index 2ce48ff908d..2e18bce750c 100644
--- a/mobile/src/session/use-mobile-structured-agent-session-prompt-cancel.test.tsx
+++ b/mobile/src/session/use-mobile-structured-agent-session-prompt-cancel.test.tsx
@@ -5,7 +5,14 @@ import type { AgentJournalRenderItem } from '../../../src/shared/agent-session-j
 import type { StructuredAgentSessionState } from '../../../src/shared/structured-agent-session-reducer'
 import type { RpcClient } from '../transport/rpc-client'
 
-const mocks = vi.hoisted(() => ({ sendRequest: vi.fn() }))
+const mocks = vi.hoisted(() => ({
+  sendRequest: vi.fn(),
+  promptResponses: vi.fn(() => ({
+    groupedDraft: null,
+    respondPermission: vi.fn(),
+    respondQuestion: vi.fn()
+  }))
+}))
 vi.mock('./use-mobile-structured-agent-state', () => ({
   useMobileStructuredAgentState: () => ({
     state,
@@ -25,11 +32,7 @@ vi.mock('./use-mobile-structured-agent-options', () => ({
   })
 }))
 vi.mock('./use-mobile-structured-prompt-responses', () => ({
-  useMobileStructuredPromptResponses: () => ({
-    groupedDraft: null,
-    respondPermission: vi.fn(),
-    respondQuestion: vi.fn()
-  })
+  useMobileStructuredPromptResponses: mocks.promptResponses
 }))
 vi.mock('./use-mobile-structured-send-operation-reconciliation', () => ({
   useMobileStructuredSendOperationReconciliation: vi.fn()
@@ -104,7 +107,13 @@ const client: RpcClient = {
   close: () => {}
 }
 
-function Harness({ promptCancelSupported }: { promptCancelSupported: boolean }): null {
+function Harness({
+  promptCancelSupported,
+  questionAnswersSupported = false
+}: {
+  promptCancelSupported: boolean
+  questionAnswersSupported?: boolean
+}): null {
   hook = useMobileStructuredAgentSession({
     client,
     sessionId: 'session-1',
@@ -112,7 +121,7 @@ function Harness({ promptCancelSupported }: { promptCancelSupported: boolean }):
     enabled: true,
     connected: true,
     agent: 'codex',
-    promptCancelSupported,
+    hostSupport: { promptCancel: promptCancelSupported, questionAnswers: questionAnswersSupported },
     onSendError: vi.fn()
   })
   return null
@@ -151,6 +160,17 @@ describe('mobile structured prompt cancellation', () => {
     renderer = null
   })
 
+  it('hands question answering the host answers capability', () => {
+    act(() => {
+      renderer = create(
+        createElement(Harness, { promptCancelSupported: false, questionAnswersSupported: true })
+      )
+    })
+    expect(mocks.promptResponses).toHaveBeenLastCalledWith(
+      expect.objectContaining({ questionAnswersSupported: true })
+    )
+  })
+
   it('sends the clicked prompt identity on capable hosts', async () => {
     act(() => {
       renderer = create(createElement(Harness, { promptCancelSupported: true }))
diff --git a/mobile/src/session/use-mobile-structured-agent-session.ts b/mobile/src/session/use-mobile-structured-agent-session.ts
index 2a24d38a38e..f04d78cc9a7 100644
--- a/mobile/src/session/use-mobile-structured-agent-session.ts
+++ b/mobile/src/session/use-mobile-structured-agent-session.ts
@@ -32,6 +32,7 @@ import type { MobileNativeChatSession } from './use-mobile-native-chat-session'
 import type { NativeChatLiveTurnIndicator } from '../../../src/shared/native-chat-turn-status'
 import { useMobileStructuredAgentState } from './use-mobile-structured-agent-state'
 import { useMobileStructuredPromptResponses } from './use-mobile-structured-prompt-responses'
+import type { StructuredAgentSessionHostSupport } from './mobile-structured-agent-session-host-support'
 import { useMobileStructuredAgentOptions } from './use-mobile-structured-agent-options'
 import { useMobileStructuredAgentTurnTiming } from './use-mobile-structured-agent-turn-timing'
 import { sendMobileStructuredAgentSessionMessage } from './mobile-structured-agent-session-send'
@@ -77,8 +78,8 @@ export function useMobileStructuredAgentSession(args: {
   enabled: boolean
   /** Live transport only; gates the connection-scoped hold, nothing else. */
   connected: boolean
-  /** Capability fact from the shared runtime status probe; null follows legacy cancellation. */
-  promptCancelSupported?: boolean | null
+  /** Capability facts from the shared runtime status probe; null follows the legacy wire. */
+  hostSupport: StructuredAgentSessionHostSupport | null
   agent: string | null
   onSendError: (message: string) => void
 }): StructuredMobileSession {
@@ -91,8 +92,9 @@ export function useMobileStructuredAgentSession(args: {
     sourceIdentity = '',
     enabled,
     onSendError,
-    promptCancelSupported = null
+    hostSupport
   } = args
+  const promptCancelSupported = hostSupport?.promptCancel ?? null
   const sessionKey = encodeNativeChatTranscriptIdentity([sourceIdentity, agent, sessionId])
   const operationIdsRef = useRef(new Map<string, string>())
   const commandPendingRef = useRef(false)
@@ -240,6 +242,7 @@ export function useMobileStructuredAgentSession(args: {
     stateRef,
     sessionKey,
     mutate,
+    questionAnswersSupported: hostSupport?.questionAnswers ?? null,
     onSendError
   })
 
diff --git a/mobile/src/session/use-mobile-structured-prompt-responses.test.tsx b/mobile/src/session/use-mobile-structured-prompt-responses.test.tsx
index 05a2b7fc380..67e15635aa5 100644
--- a/mobile/src/session/use-mobile-structured-prompt-responses.test.tsx
+++ b/mobile/src/session/use-mobile-structured-prompt-responses.test.tsx
@@ -7,7 +7,12 @@ import {
   EMPTY_STRUCTURED_AGENT_SESSION,
   type StructuredAgentSessionState
 } from '../../../src/shared/structured-agent-session-reducer'
-import { projectStructuredQuestion } from './mobile-structured-agent-prompts'
+import { encodeAgentSessionQuestionAnswers } from '../../../src/shared/agent-session-question-answer'
+import { formatQuestionFreeTextAnswer } from './mobile-native-chat-question'
+import {
+  projectStructuredQuestion,
+  type StructuredQuestionItem
+} from './mobile-structured-agent-prompts'
 import type {
   StructuredAgentSessionMutate,
   StructuredAgentSessionMutationResult
@@ -72,6 +77,8 @@ function Probe(props: {
   sessionKey: string
   state: StructuredAgentSessionState
   mutate: StructuredAgentSessionMutate
+  questionAnswersSupported?: boolean | null
+  onSendError?: (message: string) => void
 }) {
   const stateRef = useRef(props.state)
   stateRef.current = props.state
@@ -79,7 +86,8 @@ function Probe(props: {
     stateRef,
     sessionKey: props.sessionKey,
     mutate: props.mutate,
-    onSendError: vi.fn()
+    questionAnswersSupported: props.questionAnswersSupported ?? null,
+    onSendError: props.onSendError ?? vi.fn()
   })
   return null
 }
@@ -172,4 +180,298 @@ describe('useMobileStructuredPromptResponses', () => {
       expect(hook().groupedDraft?.answers).toHaveLength(1)
     }
   )
+
+  describe('answer wire', () => {
+    const LONG_ANSWER = 'Proceed with the replacement, but wait for the capture. '.repeat(30).trim()
+
+    function singlePrompt(): StructuredQuestionItem {
+      return {
+        itemId: 'item-s',
+        revision: 3,
+        sequence: 1,
+        observedAt: 1,
+        body: {
+          kind: 'question',
+          question: 'Anything else?',
+          options: [
+            { id: 'yes', label: 'Yes' },
+            { id: 'no', label: 'No' }
+          ],
+          freeTextQuestionId: 'notes',
+          resolution: {
+            state: 'pending',
+            selectedOptionId: null,
+            resolvedBy: null,
+            resolvedAt: null
+          }
+        }
+      }
+    }
+
+    // Records what was sent; the verdict is irrelevant to the wire shape under test.
+    function recordingMutate() {
+      const sent: { method: string; fields: Record<string, unknown> }[] = []
+      const mutate: StructuredAgentSessionMutate = async (method, _fingerprint, fields) => {
+        sent.push({ method, fields })
+        return { status: 'rejected' }
+      }
+      return { mutate, sent }
+    }
+
+    function mount(
+      prompt: AgentJournalRenderItem,
+      mutate: StructuredAgentSessionMutate,
+      questionAnswersSupported: boolean | null,
+      onSendError?: (message: string) => void
+    ): void {
+      act(() => {
+        renderer = create(
+          createElement(Probe, {
+            sessionKey: 'session-a',
+            state: sessionState(prompt),
+            mutate,
+            questionAnswersSupported,
+            onSendError
+          })
+        )
+      })
+    }
+
+    function sentFields(sent: ReturnType<typeof recordingMutate>['sent']): Record<string, unknown> {
+      expect(sent).toHaveLength(1)
+      expect(sent[0]!.method).toBe('agentSession.respondToQuestion')
+      return sent[0]!.fields
+    }
+
+    it('sends a long typed answer as structured answers to a capable host', async () => {
+      const prompt = singlePrompt()
+      const { mutate, sent } = recordingMutate()
+      mount(prompt, mutate, true)
+      const card = projectStructuredQuestion(prompt)!
+
+      await act(async () => {
+        await hook().respondQuestion(formatQuestionFreeTextAnswer(card, LONG_ANSWER))
+      })
+
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-s',
+        expectedRevision: 3,
+        answers: [{ questionId: 'notes', optionIds: [], other: LONG_ANSWER }]
+      })
+    })
+
+    it('packs a typed answer into the option id for a host that predates answers', async () => {
+      const prompt = singlePrompt()
+      const { mutate, sent } = recordingMutate()
+      mount(prompt, mutate, null)
+      const card = projectStructuredQuestion(prompt)!
+
+      await act(async () => {
+        await hook().respondQuestion(formatQuestionFreeTextAnswer(card, '  DuckDB  '))
+      })
+
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-s',
+        expectedRevision: 3,
+        optionId: 'notes:DuckDB'
+      })
+    })
+
+    it('tries structured answers for an answer too long to pack while support is unknown', async () => {
+      const prompt = singlePrompt()
+      const { mutate, sent } = recordingMutate()
+      mount(prompt, mutate, null)
+
+      await act(async () => {
+        await hook().respondQuestion(
+          formatQuestionFreeTextAnswer(projectStructuredQuestion(prompt)!, LONG_ANSWER)
+        )
+      })
+
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-s',
+        expectedRevision: 3,
+        answers: [{ questionId: 'notes', optionIds: [], other: LONG_ANSWER }]
+      })
+    })
+
+    it('tells the user to update an older host instead of sending an answer it must refuse', async () => {
+      const prompt = singlePrompt()
+      const { mutate, sent } = recordingMutate()
+      const onSendError = vi.fn()
+      mount(prompt, mutate, false, onSendError)
+
+      let accepted = true
+      await act(async () => {
+        accepted = await hook().respondQuestion(
+          formatQuestionFreeTextAnswer(projectStructuredQuestion(prompt)!, LONG_ANSWER)
+        )
+      })
+
+      expect(accepted).toBe(false)
+      expect(sent).toEqual([])
+      expect(onSendError).toHaveBeenCalledWith(
+        'Update Orca on your computer to send answers this long'
+      )
+    })
+
+    it.each([
+      [true, { answers: [{ questionId: 'notes', optionIds: ['no'] }] }],
+      [false, { optionId: 'no' }]
+    ])(
+      'answers an option tap for the question it was shown on (answers: %s)',
+      async (supported, wire) => {
+        const prompt = singlePrompt()
+        const { mutate, sent } = recordingMutate()
+        mount(prompt, mutate, supported)
+
+        await act(async () => {
+          await hook().respondQuestion(projectStructuredQuestion(prompt)!.optionTokens[1]!)
+        })
+
+        expect(sentFields(sent)).toEqual({ itemId: 'item-s', expectedRevision: 3, ...wire })
+      }
+    )
+
+    it.each([true, false])('submits a grouped question once (answers: %s)', async (supported) => {
+      const prompt = groupedPrompt('item-g', 1)
+      const { mutate, sent } = recordingMutate()
+      mount(prompt, mutate, supported)
+
+      await act(async () => {
+        await hook().respondQuestion(projectedResponse(prompt, null))
+      })
+      await act(async () => {
+        await hook().respondQuestion(projectedResponse(prompt, hook().groupedDraft))
+      })
+
+      const answers = [
+        { questionId: 'q1', optionIds: ['q1:choice-1'] },
+        { questionId: 'q2', optionIds: ['q2:choice-1'] }
+      ]
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-g',
+        expectedRevision: 1,
+        ...(supported ? { answers } : { optionId: encodeAgentSessionQuestionAnswers(answers) })
+      })
+    })
+
+    // Claude always sends a question list, so this is the path a long typed Claude answer takes.
+    it.each([
+      [true, true],
+      [null, true],
+      [false, false]
+    ])('submits a grouped long typed answer (answers: %s)', async (supported, expectSent) => {
+      const base = groupedPrompt('item-g', 1)
+      if (base.body.kind !== 'question' || !base.body.questions) {
+        throw new Error('expected a grouped question')
+      }
+      const [first, second] = base.body.questions
+      const prompt: AgentJournalRenderItem = {
+        ...base,
+        body: { ...base.body, questions: [first!, { ...second!, freeTextQuestionId: 'q2' }] }
+      }
+      const { mutate, sent } = recordingMutate()
+      const onSendError = vi.fn()
+      mount(prompt, mutate, supported, onSendError)
+
+      await act(async () => {
+        await hook().respondQuestion(projectedResponse(prompt, null))
+      })
+      const step = projectStructuredQuestion(prompt, hook().groupedDraft)!
+      await act(async () => {
+        await hook().respondQuestion(formatQuestionFreeTextAnswer(step, LONG_ANSWER))
+      })
+
+      if (!expectSent) {
+        expect(sent).toEqual([])
+        expect(onSendError).toHaveBeenCalledWith(
+          'Update Orca on your computer to send answers this long'
+        )
+        return
+      }
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-g',
+        expectedRevision: 1,
+        answers: [
+          { questionId: 'q1', optionIds: ['q1:choice-1'] },
+          { questionId: 'q2', optionIds: [], other: LONG_ANSWER }
+        ]
+      })
+    })
+
+    it('keeps a grouped draft an older host cannot take so it sends after an update', async () => {
+      const base = groupedPrompt('item-g', 1)
+      if (base.body.kind !== 'question' || !base.body.questions) {
+        throw new Error('expected a grouped question')
+      }
+      const [first, second] = base.body.questions
+      const prompt: AgentJournalRenderItem = {
+        ...base,
+        body: { ...base.body, questions: [{ ...first!, freeTextQuestionId: 'q1' }, second!] }
+      }
+      const { mutate, sent } = recordingMutate()
+      const onSendError = vi.fn()
+      mount(prompt, mutate, false, onSendError)
+
+      let advanced = false
+      await act(async () => {
+        advanced = await hook().respondQuestion(
+          formatQuestionFreeTextAnswer(projectStructuredQuestion(prompt, null)!, LONG_ANSWER)
+        )
+      })
+      expect(advanced).toBe(true)
+      await act(async () => {
+        await hook().respondQuestion(projectedResponse(prompt, hook().groupedDraft))
+      })
+
+      expect(sent).toEqual([])
+      expect(onSendError).toHaveBeenCalledWith(
+        'Update Orca on your computer to send answers this long'
+      )
+      expect(hook().groupedDraft).not.toBeNull()
+
+      act(() => {
+        renderer?.update(
+          createElement(Probe, {
+            sessionKey: 'session-a',
+            state: sessionState(prompt),
+            mutate,
+            questionAnswersSupported: true,
+            onSendError
+          })
+        )
+      })
+      await act(async () => {
+        await hook().respondQuestion(projectedResponse(prompt, hook().groupedDraft))
+      })
+
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-g',
+        expectedRevision: 1,
+        answers: [
+          { questionId: 'q1', optionIds: [], other: LONG_ANSWER },
+          { questionId: 'q2', optionIds: ['q2:choice-1'] }
+        ]
+      })
+    })
+
+    it('names the single question an option tap answers when the prompt has no typed field', async () => {
+      const base = singlePrompt()
+      const { freeTextQuestionId: _omitted, ...body } = base.body
+      const prompt: StructuredQuestionItem = { ...base, body }
+      const { mutate, sent } = recordingMutate()
+      mount(prompt, mutate, true)
+
+      await act(async () => {
+        await hook().respondQuestion(projectStructuredQuestion(prompt)!.optionTokens[0]!)
+      })
+
+      expect(sentFields(sent)).toEqual({
+        itemId: 'item-s',
+        expectedRevision: 3,
+        answers: [{ questionId: 'q1', optionIds: ['yes'] }]
+      })
+    })
+  })
 })
diff --git a/mobile/src/session/use-mobile-structured-prompt-responses.ts b/mobile/src/session/use-mobile-structured-prompt-responses.ts
index 8340b7edee8..093029b2be5 100644
--- a/mobile/src/session/use-mobile-structured-prompt-responses.ts
+++ b/mobile/src/session/use-mobile-structured-prompt-responses.ts
@@ -1,5 +1,11 @@
 import { useCallback, useState } from 'react'
 import type { AgentSessionPromptResult } from '../../../src/shared/agent-session-wire'
+import type { AgentJournalQuestionItem } from '../../../src/shared/agent-session-journal-types'
+import {
+  AGENT_SESSION_RESPONSE_OPTION_ID_MAX_LENGTH,
+  legacyAgentSessionSelectedOptionId,
+  type AgentSessionQuestionAnswer
+} from '../../../src/shared/agent-session-question-answer'
 import type { StructuredAgentSessionState } from '../../../src/shared/structured-agent-session-reducer'
 import {
   pendingStructuredApproval,
@@ -7,7 +13,10 @@ import {
   structuredApprovalResponseTarget,
   structuredQuestionResponseTarget
 } from './mobile-structured-agent-prompts'
-import type { StructuredAgentSessionMutate } from './mobile-structured-agent-session-rpc'
+import type {
+  StructuredAgentSessionMutate,
+  StructuredAgentSessionMutationResult
+} from './mobile-structured-agent-session-rpc'
 import {
   advanceGroupedQuestion,
   groupedQuestionPromptKey,
@@ -23,13 +32,15 @@ export function useMobileStructuredPromptResponses(args: {
   stateRef: { readonly current: StructuredAgentSessionState }
   sessionKey: string
   mutate: StructuredAgentSessionMutate
+  /** Host takes structured `answers`; otherwise the answer is packed into `optionId`. */
+  questionAnswersSupported: boolean | null
   onSendError: (message: string) => void
 }): {
   groupedDraft: GroupedQuestionDraft | null
   respondPermission: (optionId: string) => Promise<boolean>
   respondQuestion: (answer: string) => Promise<boolean>
 } {
-  const { mutate, onSendError, sessionKey, stateRef } = args
+  const { mutate, onSendError, questionAnswersSupported, sessionKey, stateRef } = args
   // Partially answered grouped question, held only until its last step is submitted. The session it
   // was collected in is stored with it and checked on read, so switching sessions drops the draft
   // without an effect that would render the stale one for a frame first.
@@ -62,6 +73,38 @@ export function useMobileStructuredPromptResponses(args: {
     [mutate, onSendError, stateRef]
   )
 
+  const sendAnswers = useCallback(
+    (
+      target: { itemId: string; expectedRevision: number },
+      body: Pick<AgentJournalQuestionItem, 'questions'>,
+      answers: AgentSessionQuestionAnswer[]
+    ): Promise<StructuredAgentSessionMutationResult<AgentSessionPromptResult> | null> => {
+      const send = (fields: Record<string, unknown>) =>
+        mutate<AgentSessionPromptResult>(
+          'agentSession.respondToQuestion',
+          'agentSession.respondTo:question',
+          { ...target, ...fields }
+        )
+      if (questionAnswersSupported === true) {
+        return send({ answers })
+      }
+      const optionId = legacyAgentSessionSelectedOptionId(body, answers)
+      if (optionId === null) {
+        return Promise.resolve(null)
+      }
+      if (optionId.length > AGENT_SESSION_RESPONSE_OPTION_ID_MAX_LENGTH) {
+        // Too long to pack for any host: while support is unknown, `answers` is the only form that can land.
+        if (questionAnswersSupported === null) {
+          return send({ answers })
+        }
+        onSendError('Update Orca on your computer to send answers this long')
+        return Promise.resolve(null)
+      }
+      return send({ optionId })
+    },
+    [mutate, onSendError, questionAnswersSupported]
+  )
+
   const respondQuestion = useCallback(
     async (answer: string): Promise<boolean> => {
       const prompt = stateRef.current.items.find(pendingStructuredQuestion) ?? null
@@ -80,11 +123,14 @@ export function useMobileStructuredPromptResponses(args: {
           setCollected({ sessionKey, draft: grouped.draft })
           return true
         }
-        const result = await mutate<AgentSessionPromptResult>(
-          'agentSession.respondToQuestion',
-          'agentSession.respondTo:question',
-          { itemId: prompt.itemId, expectedRevision: prompt.revision, optionId: grouped.optionId }
+        const result = await sendAnswers(
+          { itemId: prompt.itemId, expectedRevision: prompt.revision },
+          prompt.body,
+          grouped.answers
         )
+        if (!result) {
+          return false
+        }
         if (result.status !== 'rejected') {
           // The group left the phone; a retry must start from the first question, not a stale tail.
           setCollected((current) =>
@@ -103,18 +149,22 @@ export function useMobileStructuredPromptResponses(args: {
       if (!target) {
         return false
       }
-      const result = await mutate<AgentSessionPromptResult>(
-        'agentSession.respondToQuestion',
-        'agentSession.respondTo:question',
-        target
+      const result = await sendAnswers(
+        { itemId: target.itemId, expectedRevision: target.expectedRevision },
+        // Grouped questions returned above, so this is a single question.
+        {},
+        [target.answer]
       )
+      if (!result) {
+        return false
+      }
       if (result.status === 'unknown') {
         onSendError('Answer unconfirmed — check chat before retrying')
         return false
       }
       return result.status === 'accepted'
     },
-    [groupedDraft, mutate, onSendError, sessionKey, stateRef]
+    [groupedDraft, onSendError, sendAnswers, sessionKey, stateRef]
   )
 
   return { groupedDraft, respondPermission, respondQuestion }
diff --git a/src/shared/agent-session-question-answer.ts b/src/shared/agent-session-question-answer.ts
index cc815acf542..08374adc3d5 100644
--- a/src/shared/agent-session-question-answer.ts
+++ b/src/shared/agent-session-question-answer.ts
@@ -5,6 +5,9 @@ const GROUP_ANSWER_PREFIX = 'question-group:'
 /** A single-question item has no question list; the client and host must agree on its id. */
 const SINGLE_QUESTION_ID = 'q1'
 
+/** Longest option id a host takes, which also bounds an answer packed into one. */
+export const AGENT_SESSION_RESPONSE_OPTION_ID_MAX_LENGTH = 1024
+
 /** Largest typed answer to one question, in UTF-8 bytes. */
 export const AGENT_SESSION_QUESTION_ANSWER_MAX_BYTES = 64 * 1024
 
diff --git a/src/shared/rpc-contract/structured-agent-session-params.ts b/src/shared/rpc-contract/structured-agent-session-params.ts
index bd9ce3db82a..9d240ac578c 100644
--- a/src/shared/rpc-contract/structured-agent-session-params.ts
+++ b/src/shared/rpc-contract/structured-agent-session-params.ts
@@ -2,7 +2,10 @@ import { z } from 'zod'
 import { isAgentSessionSurfaceTabId } from '../agent-session-surface-tab-id'
 import { isAgentSessionId } from '../agent-session-record'
 import { normalizeExecutionHostId } from '../execution-host'
-import { AGENT_SESSION_QUESTION_ANSWER_MAX_BYTES } from '../agent-session-question-answer'
+import {
+  AGENT_SESSION_QUESTION_ANSWER_MAX_BYTES,
+  AGENT_SESSION_RESPONSE_OPTION_ID_MAX_LENGTH
+} from '../agent-session-question-answer'
 import {
   AGENT_SESSION_ID_MAX_LENGTH,
   AGENT_SESSION_HISTORY_DIRECTIONS,
@@ -13,7 +16,7 @@ import {
 export const MAX_ID_LENGTH = AGENT_SESSION_ID_MAX_LENGTH
 
 // Four Claude questions with all four generated choices occupy 610 chars when fully percent-encoded.
-export const MAX_RESPONSE_OPTION_ID_LENGTH = 1024
+export const MAX_RESPONSE_OPTION_ID_LENGTH = AGENT_SESSION_RESPONSE_OPTION_ID_MAX_LENGTH
 
 export const MAX_PROMPT_BYTES = 256 * 1024