From a0c269b0fbc4ee1226e57308303bd8c47f814646 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Wed, 22 Jul 2026 13:02:44 +0800
Subject: [PATCH 01/36] feat(gui): fold todo/write into
 ConversationSnapshot.todos

Session consumes the todo/write session event as a per-event side effect
(last write wins), rebuilds it on window replay/paging/resync, and exposes
snapshot.todos. TodoItem re-exported through the runtime surface.
---
 packages/client/runtime/src/client/index.ts     |  2 +-
 .../runtime/src/client/sessions/conversation.ts |  5 +++++
 .../runtime/src/client/sessions/session.ts      | 10 +++++++++-
 packages/client/runtime/tests/event-script.ts   |  2 ++
 packages/client/runtime/tests/session.spec.ts   | 17 +++++++++++++++++
 .../tests/gate-branch-tails.spec.tsx            |  2 +-
 .../tests/skeleton-branches.spec.tsx            |  2 +-
 7 files changed, 36 insertions(+), 4 deletions(-)

diff --git a/packages/client/runtime/src/client/index.ts b/packages/client/runtime/src/client/index.ts
index 33097d4573..645575a893 100644
--- a/packages/client/runtime/src/client/index.ts
+++ b/packages/client/runtime/src/client/index.ts
@@ -33,7 +33,7 @@ export type {
 export type {
   AssistantBlock, AssistantMessageNode, ContextMessageNode, ConversationNode, ConversationSnapshot,
   RunningToolCall, SteeringMessageNode,
-  ToolResultNode, UnknownSurfaceNode, UserMessageNode,
+  TodoItem, ToolResultNode, UnknownSurfaceNode, UserMessageNode,
 } from './sessions/conversation.ts'
 // PendingWait is a value export: tests construct fixture waits directly.
 export { PendingWait } from './sessions/pending.ts'
diff --git a/packages/client/runtime/src/client/sessions/conversation.ts b/packages/client/runtime/src/client/sessions/conversation.ts
index 08b80f2a26..b77a6fdcac 100644
--- a/packages/client/runtime/src/client/sessions/conversation.ts
+++ b/packages/client/runtime/src/client/sessions/conversation.ts
@@ -4,9 +4,12 @@
 // string here (narrow to real brands when convenient).
 
 import type { ContentBlock } from '@deepseek-ai/dsh-llm/types'
+import type { TodoItem } from '@deepseek-ai/dsh-session/types'
 import type { RpcError, SessionId, ToolCallView, ToolResultView } from '@deepseek-ai/dsh-client-connection/client'
 import type { PendingInteraction } from './pending.ts'
 
+export type { TodoItem }
+
 /** Assistant content blocks sorted by what the UI cares about
  *  (text body / collapsible reasoning / tool-call card head / other fallback). */
 export type AssistantBlock =
@@ -174,4 +177,6 @@ export interface ConversationSnapshot {
   loadingOlder: boolean
   promptError: PromptError | null
   lastAgentError: string | null
+  /** Latest `todo/write` whole-list snapshot in the window (last write wins); empty = no plan. */
+  todos: readonly TodoItem[]
 }
diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts
index 6b773e0903..304b2e6e02 100644
--- a/packages/client/runtime/src/client/sessions/session.ts
+++ b/packages/client/runtime/src/client/sessions/session.ts
@@ -4,7 +4,7 @@
 // subscribe/getSnapshot.
 
 import type { ContentBlock } from '@deepseek-ai/dsh-llm/types'
-import type { SessionEvent } from '@deepseek-ai/dsh-session/types'
+import type { SessionEvent, TodoItem } from '@deepseek-ai/dsh-session/types'
 import type {
   HistoryEntry, IApiClient, MuxFrame, RpcError, RpcId, RpcResult,
   SessionId, ToolEventView,
@@ -65,6 +65,8 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
   private pendingCache: { rev: number; value: PendingInteraction[] } | null = null
   private frozenRev = 0
   private nodesCache: { folded: readonly ConversationNode[]; frozenRev: number; value: readonly ConversationNode[] } | null = null
+  /** Latest todo/write whole-list snapshot in the window (last write wins on replay). */
+  private todos: readonly TodoItem[] = []
   private running = false
   private removed = false
   private promptError: PromptError | null = null
@@ -444,6 +446,10 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
         if (this.openCalls.delete(String(event.data.callId))) this.callsRev++
         return
       }
+      case 'todo/write': {
+        this.todos = event.data.todos
+        return
+      }
       case 'turn/end': {
         // Aborted turns never finalize. The accumulated partial is VALUE, not residue: freeze it
         // into an interrupted terminal node (pulse stops, text survives) instead of deleting it.
@@ -495,6 +501,7 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
     this.callsRev++
     this.frozenNodes = []
     this.frozenRev++
+    this.todos = []
     for (let i = 0; i < this.events.length; i++) {
       const event = this.events[i]
       /* v8 ignore next -- dense-array guard: i stays within events.length, so the undefined arm needs a sparse array no caller builds. */
@@ -542,6 +549,7 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
       loadingOlder: this.loadingOlder,
       promptError: this.promptError,
       lastAgentError: this.lastAgentError,
+      todos: this.todos,
     }
   }
 }
diff --git a/packages/client/runtime/tests/event-script.ts b/packages/client/runtime/tests/event-script.ts
index b567800c9b..8b3b6f59ee 100644
--- a/packages/client/runtime/tests/event-script.ts
+++ b/packages/client/runtime/tests/event-script.ts
@@ -30,6 +30,8 @@ export const ev = {
     at(seq, { type: 'step/end', data: { turn, step } }),
   turnEnd: (seq: number, turn: number, reason: 'completed' | 'cancelled' = 'completed'): SessionEvent =>
     at(seq, { type: 'turn/end', data: { turn, reason: { kind: reason } } }),
+  todoWrite: (seq: number, todos: { content: string; status: 'pending' | 'in_progress' | 'completed' }[]): SessionEvent =>
+    at(seq, { type: 'todo/write', data: { todos } }),
 }
 
 /** One complete plain turn (turn/start → user → step → assistant → turn/end), 6 events from startSeq. */
diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts
index b980a674fe..786bb9793b 100644
--- a/packages/client/runtime/tests/session.spec.ts
+++ b/packages/client/runtime/tests/session.spec.ts
@@ -153,6 +153,23 @@ describe('live event path', () => {
     })
   })
 
+  it('folds todo/write into snapshot.todos last-write-wins, live and on window replay', async () => {
+    const listA = [{ content: '搭骨架', status: 'completed' as const }, { content: '写组件', status: 'in_progress' as const }]
+    const listB = [{ content: '搭骨架', status: 'completed' as const }, { content: '写组件', status: 'completed' as const }]
+    const { session } = await opened()
+    expect(session.getSnapshot().todos).toEqual([])
+    const feed = (event: SessionEvent) => { session.handleMuxEnvelope('r' as never, { type: 'session/event', sessionId: SID, event }) }
+    feed(ev.todoWrite(6, listA))
+    expect(session.getSnapshot().todos).toEqual(listA)
+    feed(ev.todoWrite(7, listB))
+    expect(session.getSnapshot().todos).toEqual(listB)
+    // Window replay converges on the same last snapshot (history contains both writes).
+    const replayed = makeSession()
+    replayed.api.onHistory = () => histResponse([...plainTurn(0, 0, 'a', 'b'), ev.todoWrite(6, listA), ev.todoWrite(7, listB)])
+    await replayed.session.open()
+    expect(replayed.session.getSnapshot().todos).toEqual(listB)
+  })
+
   it('repairs a seq gap by repulling the tail page instead of appending a hole', async () => {
     const { api, session } = await opened(plainTurn(0, 0, 'a', 'b')) // tail seq = 5
     const repaired = [...plainTurn(0, 0, 'a', 'b'), ...plainTurn(6, 1, 'c', 'd')]
diff --git a/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx b/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx
index c59b1c7527..62c4213a6d 100644
--- a/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx
+++ b/packages/client/ui-conversation/tests/gate-branch-tails.spec.tsx
@@ -25,7 +25,7 @@ function snapshotBase(): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [],
     pending: [], running: false, removed: false, openState: 'open', openError: null,
-    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null,
+    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null, todos: [],
   } as ConversationSnapshot
 }
 
diff --git a/packages/client/ui-conversation/tests/skeleton-branches.spec.tsx b/packages/client/ui-conversation/tests/skeleton-branches.spec.tsx
index 4eb70ea39f..c3461cc0c3 100644
--- a/packages/client/ui-conversation/tests/skeleton-branches.spec.tsx
+++ b/packages/client/ui-conversation/tests/skeleton-branches.spec.tsx
@@ -29,7 +29,7 @@ function snapshotBase(): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [],
     pending: [], running: false, removed: false, openState: 'open', openError: null,
-    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null,
+    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null, todos: [],
   } as ConversationSnapshot
 }
 

From 63109dab66ad93ec4e274e15624fa3ebf9b603f7 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Wed, 22 Jul 2026 13:02:53 +0800
Subject: [PATCH 02/36] =?UTF-8?q?feat(gui):=20todo=20display=20=E2=80=94?=
 =?UTF-8?q?=20TodoPanel=20plan=20strip=20+=20todo=5Fwrite=20toolview=20row?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

TodoPanel pins above the composer (776px card axis), hidden while empty,
collapsible with the active item as the collapsed hint; status glyphs
mirror the TUI plan panel. todo_write rows render a plan-flavored summary
(counts + active item) via the toolview registry, generic fallback on
malformed args. Existing fake snapshots gain the required todos field.
---
 .../ui-conversation/src/client/apply.ts       |   4 +
 .../src/client/skeleton/ConversationRoot.tsx  |   3 +
 .../src/client/skeleton/TodoPanel.module.css  | 111 +++++++++++++++
 .../src/client/skeleton/TodoPanel.tsx         |  59 ++++++++
 .../src/client/toolviews/todo-row.module.css  |  42 ++++++
 .../src/client/toolviews/todo-row.tsx         |  71 ++++++++++
 .../ui-conversation/tests/chat-apply.spec.tsx |   6 +-
 .../tests/chat-stats-bash-sample.spec.tsx     |   2 +-
 .../tests/chat-toolview-slot.spec.tsx         |   2 +-
 .../ui-conversation/tests/chat-view.spec.tsx  |   2 +-
 .../ui-conversation/tests/skeleton.spec.tsx   |   3 +-
 .../ui-conversation/tests/todo-panel.spec.tsx | 128 ++++++++++++++++++
 .../client/ui-trajectory/tests/views.spec.tsx |   1 +
 13 files changed, 427 insertions(+), 7 deletions(-)
 create mode 100644 packages/client/ui-conversation/src/client/skeleton/TodoPanel.module.css
 create mode 100644 packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
 create mode 100644 packages/client/ui-conversation/src/client/toolviews/todo-row.module.css
 create mode 100644 packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
 create mode 100644 packages/client/ui-conversation/tests/todo-panel.spec.tsx

diff --git a/packages/client/ui-conversation/src/client/apply.ts b/packages/client/ui-conversation/src/client/apply.ts
index 372eb36c80..70ec78722c 100644
--- a/packages/client/ui-conversation/src/client/apply.ts
+++ b/packages/client/ui-conversation/src/client/apply.ts
@@ -21,6 +21,7 @@ import { createChatStore } from './stores.ts'
 import { ConversationService } from './service.ts'
 import { ChatView } from './chat/ChatView.tsx'
 import { bashToolviewSample } from './toolviews/bash-sample.tsx'
+import { todoToolview } from './toolviews/todo-row.tsx'
 import { ConversationRoot } from './skeleton/ConversationRoot.tsx'
 import { DetailsPanel } from './skeleton/DetailsPanel.tsx'
 import { EmptyState } from './skeleton/EmptyState.tsx'
@@ -140,6 +141,9 @@ export function apply(ctx: Context): void {
   // The bash sample rides that exact seam, in third-party posture.
   ctx.plugin(bashToolviewSample)
 
+  // The todo_write row rides the same seam (a product registration, not a sample).
+  ctx.plugin(todoToolview)
+
   slots.register({
     name: 'details',
     store: chatStore,
diff --git a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
index 59346a557b..ca7fc526e4 100644
--- a/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/ConversationRoot.tsx
@@ -15,6 +15,7 @@ import type { SessionId, SessionListState, SessionSummary } from '@deepseek-ai/d
 import type { ConversationSlotProps } from '../contract/slots.ts'
 import { InputBar } from './InputBar.tsx'
 import type { InputBarError } from './InputBar.tsx'
+import { TodoPanel } from './TodoPanel.tsx'
 import css from './ConversationRoot.module.css'
 
 /** Full props = the automatic shares & injected share — composed by reference
@@ -123,6 +124,8 @@ export function ConversationRoot({
         {active !== undefined && renderSlot('conversation.view', {}, { only: active.id })}
       </div>
 
+      <TodoPanel useSession={useSession} />
+
       {renderSlotChain('conversation.composer', { interactions: pending }, { fallback: composerBar })}
     </div>
   )
diff --git a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.module.css b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.module.css
new file mode 100644
index 0000000000..17c9c890a7
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.module.css
@@ -0,0 +1,111 @@
+/* Plan strip pinned above the composer: bordered card on the composer card's
+   axis (776px column inside 32px side padding). Colors resolve through
+   --dsw-alias-* tokens only; the active row rides the business blue, done
+   rows fade to tertiary. */
+
+.root {
+  flex: none;
+  overflow: hidden;
+  margin: 8px auto 0;
+  width: calc(100% - 64px);
+  max-width: 776px;
+  border: 1px solid var(--dsw-alias-border-l2);
+  border-radius: 12px;
+  background: var(--dsw-alias-bg-base);
+}
+
+.header {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  width: 100%;
+  padding: 8px 12px;
+  border: none;
+  background: transparent;
+  text-align: left;
+  cursor: pointer;
+}
+
+.header:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.title {
+  font-size: 13px;
+  line-height: 16px;
+  font-weight: 510;
+  color: var(--dsw-alias-label-primary);
+}
+
+.progress {
+  font-size: 12px;
+  line-height: 16px;
+  color: var(--dsw-alias-label-tertiary);
+}
+
+.activeHint {
+  flex: 1;
+  min-width: 0;
+  overflow: hidden;
+  font-size: 12px;
+  line-height: 16px;
+  color: var(--dsw-alias-label-secondary);
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.chevron {
+  display: grid;
+  flex: none;
+  place-items: center;
+  margin-left: auto;
+  color: var(--dsw-alias-label-secondary);
+}
+
+.list {
+  margin: 0;
+  padding: 0 12px 8px;
+  list-style: none;
+  max-height: 180px;
+  overflow-y: auto;
+}
+
+.item {
+  display: flex;
+  align-items: baseline;
+  gap: 8px;
+  padding: 2px 0;
+  font-size: 13px;
+  line-height: 20px;
+  color: var(--dsw-alias-label-secondary);
+}
+
+.glyph {
+  flex: none;
+  width: 14px;
+  text-align: center;
+  color: var(--dsw-alias-label-tertiary);
+}
+
+.item[data-status='completed'] .content {
+  color: var(--dsw-alias-label-tertiary);
+  text-decoration: line-through;
+}
+
+.item[data-status='completed'] .glyph {
+  color: var(--dsw-alias-state-success-primary);
+}
+
+.item[data-status='in_progress'] .content {
+  font-weight: 510;
+  color: var(--dsw-alias-label-primary);
+}
+
+.item[data-status='in_progress'] .glyph {
+  color: var(--dsw-alias-state-business-primary);
+}
+
+.content {
+  min-width: 0;
+  overflow-wrap: anywhere;
+}
diff --git a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
new file mode 100644
index 0000000000..efaa2599b8
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
@@ -0,0 +1,59 @@
+// TodoPanel: persistent plan strip pinned above the composer (the web
+// counterpart of the TUI plan panel; ACP maps the same event to its native
+// plan). Renders the latest todo/write whole-list snapshot off the session
+// snapshot — no data of its own, hidden while the list is empty. Zero
+// framework imports: useSession arrives via props from ConversationRoot.
+
+import { useState } from 'react'
+import type { TodoItem } from '@deepseek-ai/dsh-client-runtime/client'
+import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
+import { IconChevronDownOutline14, IconChevronUpOutline14 } from '@deepseek-ai/dsh-client-ui-primitives'
+import css from './TodoPanel.module.css'
+
+export interface TodoPanelProps {
+  useSession: UseSession
+}
+
+/** Status glyphs mirror the TUI plan panel (✓ done / ● active / ○ pending). */
+const STATUS_GLYPHS: Record<TodoItem['status'], string> = {
+  completed: '✓', in_progress: '●', pending: '○',
+}
+
+export function TodoPanel({ useSession }: TodoPanelProps) {
+  const todos = useSession(s => (s as { todos: readonly TodoItem[] }).todos)
+  const [collapsed, setCollapsed] = useState(false)
+  if (todos.length === 0) return null
+
+  const done = todos.filter(t => t.status === 'completed').length
+  const active = todos.find(t => t.status === 'in_progress')
+
+  return (
+    <section className={css.root} data-testid="todo-panel" aria-label="任务清单">
+      <button
+        type="button"
+        className={css.header}
+        aria-expanded={!collapsed}
+        onClick={() => { setCollapsed(v => !v) }}
+      >
+        <span className={css.title}>Plan</span>
+        <span className={css.progress}>{done}/{todos.length}</span>
+        {collapsed && active !== undefined && (
+          <span className={css.activeHint}>{active.content}</span>
+        )}
+        <span className={css.chevron} aria-hidden>
+          {collapsed ? <IconChevronUpOutline14 /> : <IconChevronDownOutline14 />}
+        </span>
+      </button>
+      {!collapsed && (
+        <ul className={css.list}>
+          {todos.map(item => (
+            <li key={item.content} className={css.item} data-status={item.status}>
+              <span className={css.glyph} aria-hidden>{STATUS_GLYPHS[item.status]}</span>
+              <span className={css.content}>{item.content}</span>
+            </li>
+          ))}
+        </ul>
+      )}
+    </section>
+  )
+}
diff --git a/packages/client/ui-conversation/src/client/toolviews/todo-row.module.css b/packages/client/ui-conversation/src/client/toolviews/todo-row.module.css
new file mode 100644
index 0000000000..ff4068d49c
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/toolviews/todo-row.module.css
@@ -0,0 +1,42 @@
+/* todo_write plan-update row: title + progress summary on one line. */
+
+.row {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  height: 24px;
+  min-width: 0;
+  cursor: pointer;
+  border-radius: 6px;
+  font-size: 13px;
+}
+
+.row:hover {
+  background: var(--dsw-alias-interactive-bg-hover);
+}
+
+.badge {
+  flex: none;
+  color: var(--dsw-alias-state-business-primary);
+}
+
+.title {
+  flex: none;
+  font-weight: 510;
+  color: var(--dsw-alias-label-primary);
+}
+
+.summary {
+  flex: 1 1 auto;
+  min-width: 0;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+  color: var(--dsw-alias-label-secondary);
+}
+
+.err {
+  flex: none;
+  color: var(--dsw-alias-state-error-primary);
+  font-size: 11px;
+}
diff --git a/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
new file mode 100644
index 0000000000..390361d20b
--- /dev/null
+++ b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
@@ -0,0 +1,71 @@
+// todo_write toolview: plan-flavored summary row replacing the generic
+// "Tool call" card, registered into the keyed 'conversation.chat.toolview'
+// hole like the bash sample (a product registration, not a sample). The row
+// summarizes the written list (counts + active item) from the call args; the
+// durable list itself renders in the TodoPanel above the composer, so the
+// row stays one line.
+
+import type { Context } from 'cordis'
+import type { ToolRowProps } from '../contract/slots.ts'
+import { toolRowModel } from '../contract/tool-call-model.ts'
+import css from './todo-row.module.css'
+
+/** One parsed args item, shape-checked (model JSON: any field may be missing or mistyped). */
+interface TodoWriteItem { content?: unknown; status?: unknown }
+
+function isItem(value: unknown): value is TodoWriteItem {
+  return typeof value === 'object' && value !== null
+}
+
+function summarize(argsRaw: string): string | null {
+  let parsed: unknown
+  try {
+    parsed = JSON.parse(argsRaw)
+  } catch {
+    // Mid-stream truncation or malformed model JSON: fall back to the generic summary.
+    return null
+  }
+  // Valid JSON with an invalid shape (null root, non-array todos, null items —
+  // a rejected tool/call retains such args verbatim): same generic fallback.
+  if (typeof parsed !== 'object' || parsed === null) return null
+  const todos = (parsed as { todos?: unknown }).todos
+  if (!Array.isArray(todos) || !todos.every(isItem)) return null
+  const done = todos.filter(t => t.status === 'completed').length
+  const active = todos.find(t => t.status === 'in_progress')
+  const head = `${done}/${todos.length} 已完成`
+  return typeof active?.content === 'string' && active.content !== ''
+    ? `${head} · ${active.content}`
+    : head
+}
+
+/** One-line plan update row (click opens the raw args in details). */
+export function TodoRow({ toolName, block, openDetails }: ToolRowProps) {
+  const model = toolRowModel(toolName, block)
+  const argsRaw = ('kind' in block ? block.call?.argsRaw : block.argsRaw) ?? ''
+  const summary = summarize(argsRaw) ?? model.summary
+  return (
+    <div className={css.row} data-sample="todo-row" onClick={openDetails}>
+      <span className={css.badge} aria-hidden>☰</span>
+      <span className={css.title}>更新任务清单</span>
+      <span className={css.summary}>{summary}</span>
+      {model.state === 'error' && <span className={css.err}>failed</span>}
+    </div>
+  )
+}
+
+/**
+ * The todo row as a plain registrant plugin, riding the same load-order seam
+ * as the bash sample: `inject: ['conversation']` guarantees the chat entry
+ * (and with it the 'conversation.chat.toolview' declaration) is on the ledger.
+ */
+export const todoToolview = {
+  name: 'todo-toolview',
+  inject: ['slots', 'conversation'],
+  /**
+   * Register the todo row into the chat view's keyed toolview hole.
+   * @param ctx - registrant context (disposal rides ctx.effect inside slots.register).
+   */
+  apply(ctx: Context): void {
+    ctx.slots.register({ name: 'conversation.chat.toolview', key: 'todo_write' }, TodoRow)
+  },
+}
diff --git a/packages/client/ui-conversation/tests/chat-apply.spec.tsx b/packages/client/ui-conversation/tests/chat-apply.spec.tsx
index adb9271c71..bcd5cec2ed 100644
--- a/packages/client/ui-conversation/tests/chat-apply.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-apply.spec.tsx
@@ -103,13 +103,13 @@ describe('apply wiring', () => {
     expect(empty?.store).toBeUndefined()
   })
 
-  it('mounts the bash sample as a keyed entry through the load-order seam', async () => {
+  it('mounts the bash sample and the todo row as keyed entries through the load-order seam', async () => {
     const b = await bench()
     await b.fiber.await()
-    // The sample plugin's inject: ['slots', 'conversation'] resolved — the
+    // Both registrant plugins' inject: ['slots', 'conversation'] resolved — the
     // service being present implies the chat entry declared the hole first.
     const entries = b.slots.entries('conversation.chat.toolview')
-    expect(entries.map((e) => e.options.key)).toEqual(['bash'])
+    expect(entries.map((e) => e.options.key)).toEqual(['bash', 'todo_write'])
   })
 
   it('plugin fiber disposal collects every registration (unload cascade, ring and hole included)', async () => {
diff --git a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
index 4d3383b2d1..6efb63fa00 100644
--- a/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-stats-bash-sample.spec.tsx
@@ -28,7 +28,7 @@ function snapshotBase(): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [],
     pending: [], running: false, removed: false, openState: 'open', openError: null,
-    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null,
+    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null, todos: [],
   }
 }
 
diff --git a/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx b/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
index 5d2b3408a2..73e9478431 100644
--- a/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-toolview-slot.spec.tsx
@@ -41,7 +41,7 @@ function snapshotWith(nodes: ToolResultNode[]): ConversationSnapshot {
   return {
     sessionId: SID, nodes, foldDegraded: false, partial: null, runningCalls: [],
     pending: [], running: false, removed: false, openState: 'open', openError: null,
-    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null,
+    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null, todos: [],
   } as ConversationSnapshot
 }
 
diff --git a/packages/client/ui-conversation/tests/chat-view.spec.tsx b/packages/client/ui-conversation/tests/chat-view.spec.tsx
index 3f1db55199..8930f3171d 100644
--- a/packages/client/ui-conversation/tests/chat-view.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-view.spec.tsx
@@ -30,7 +30,7 @@ function snapshotBase(): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [],
     pending: [], running: false, removed: false, openState: 'open', openError: null,
-    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null,
+    hasMore: false, loadingOlder: false, promptError: null, lastAgentError: null, todos: [],
   }
 }
 
diff --git a/packages/client/ui-conversation/tests/skeleton.spec.tsx b/packages/client/ui-conversation/tests/skeleton.spec.tsx
index a598803a25..a3ffd3cb11 100644
--- a/packages/client/ui-conversation/tests/skeleton.spec.tsx
+++ b/packages/client/ui-conversation/tests/skeleton.spec.tsx
@@ -59,11 +59,12 @@ interface FakeSnapshot {
   removed: boolean
   promptError: { op: 'send' | 'stop'; error: { message: string; code: string } } | null
   pending: readonly PendingInteraction[]
+  todos: readonly { content: string; status: 'pending' | 'in_progress' | 'completed' }[]
 }
 
 function fakeSession(init: Partial<FakeSnapshot> = {}) {
   const store = createSnapshotStore<FakeSnapshot>({
-    nodes: [], runningCalls: [], running: false, removed: false, promptError: null, pending: [], ...init,
+    nodes: [], runningCalls: [], running: false, removed: false, promptError: null, pending: [], todos: [], ...init,
   })
   return { store, useSession: bindSnapshotSelector(store) as unknown as UseSession<ConversationSnapshot> }
 }
diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
new file mode 100644
index 0000000000..761cb4dfd3
--- /dev/null
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -0,0 +1,128 @@
+// @vitest-environment jsdom
+/**
+ * Todo display acceptance: the TodoPanel plan strip (empty-hidden, status
+ * rows, collapse with active hint) and the todo_write toolview row (progress
+ * summary from args, generic fallback on malformed JSON, error badge).
+ */
+import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
+import { afterEach, describe, expect, it, vi } from 'vitest'
+import { hookOf } from './hook.ts'
+import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
+import type { TodoItem, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
+import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
+import type { ToolRowProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
+// Export discipline: packages/client/AGENTS.md.
+import { TodoRow, todoToolview } from '../src/client/toolviews/todo-row.tsx'
+import { TodoPanel } from '../src/client/skeleton/TodoPanel.tsx'
+
+afterEach(cleanup)
+
+function sessionWith(todos: readonly TodoItem[]) {
+  const store = createSnapshotStore<{ todos: readonly TodoItem[] }>({ todos })
+  return { store, useSession: hookOf(store) as unknown as UseSession }
+}
+
+const LIST: TodoItem[] = [
+  { content: '搭骨架', status: 'completed' },
+  { content: '写组件', status: 'in_progress' },
+  { content: '补测试', status: 'pending' },
+]
+
+describe('TodoPanel', () => {
+  it('renders nothing while the list is empty, appears when todos land', () => {
+    const { store, useSession } = sessionWith([])
+    render(<TodoPanel useSession={useSession} />)
+    expect(screen.queryByTestId('todo-panel')).toBeNull()
+    act(() => { store.set({ todos: LIST }) })
+    expect(screen.getByTestId('todo-panel')).toBeTruthy()
+  })
+
+  it('shows progress, one row per item with its status, and strikes done items', () => {
+    const { useSession } = sessionWith(LIST)
+    render(<TodoPanel useSession={useSession} />)
+    expect(screen.getByText('1/3')).toBeTruthy()
+    const items = screen.getAllByRole('listitem')
+    expect(items.map(li => li.getAttribute('data-status'))).toEqual(['completed', 'in_progress', 'pending'])
+    expect(screen.getByText('搭骨架')).toBeTruthy()
+    expect(screen.getByText('写组件')).toBeTruthy()
+  })
+
+  it('collapse hides the list and surfaces the active item in the header; expand restores', () => {
+    const { useSession } = sessionWith(LIST)
+    render(<TodoPanel useSession={useSession} />)
+    const header = screen.getByRole('button', { expanded: true })
+    fireEvent.click(header)
+    expect(screen.queryByRole('list')).toBeNull()
+    // Collapsed header carries the in-progress content as the one-line hint.
+    expect(screen.getByText('写组件')).toBeTruthy()
+    fireEvent.click(screen.getByRole('button', { expanded: false }))
+    expect(screen.getAllByRole('listitem')).toHaveLength(3)
+  })
+
+  it('collapsed header omits the hint when nothing is in progress', () => {
+    const { useSession } = sessionWith([{ content: '都完了', status: 'completed' }])
+    render(<TodoPanel useSession={useSession} />)
+    fireEvent.click(screen.getByRole('button', { expanded: true }))
+    expect(screen.queryByText('都完了')).toBeNull()
+    expect(screen.getByText('1/1')).toBeTruthy()
+  })
+})
+
+const resultNode = (argsRaw: string, over?: Partial<ToolResultNode>): ToolResultNode => ({
+  kind: 'tool-result', seq: 10, callId: 'c1',
+  call: { name: 'todo_write', argsRaw },
+  content: [], isError: false, callView: null, resultView: null, ...over,
+})
+
+function rowProps(block: unknown, openDetails = vi.fn()): ToolRowProps {
+  return {
+    callId: 'c1', toolName: 'todo_write', block,
+    openDetails,
+    sessionId: 's1',
+    useSessions: () => undefined,
+  } as unknown as ToolRowProps
+}
+
+describe('TodoRow', () => {
+  const ARGS = JSON.stringify({ todos: LIST })
+
+  it('summarizes counts and the active item from the call args', () => {
+    render(<TodoRow {...rowProps(resultNode(ARGS))} />)
+    expect(screen.getByText('更新任务清单')).toBeTruthy()
+    expect(screen.getByText('1/3 已完成 · 写组件')).toBeTruthy()
+  })
+
+  it('omits the active clause when no item is in progress and reads running-call args', () => {
+    const args = JSON.stringify({ todos: [{ content: 'x', status: 'completed' }] })
+    render(<TodoRow {...rowProps({ callId: 'c1', name: 'todo_write', argsRaw: args, turn: 1, step: 1, callView: null })} />)
+    expect(screen.getByText('1/1 已完成')).toBeTruthy()
+  })
+
+  it('falls back to the generic summary on malformed args and flags errors', () => {
+    render(<TodoRow {...rowProps(resultNode('not json', { isError: true }))} />)
+    expect(screen.getByText('failed')).toBeTruthy()
+    // Generic others summary: "<tool> · <raw>".
+    expect(screen.getByText('todo_write · not json')).toBeTruthy()
+  })
+
+  it('falls back when parsed args carry no todos array, and click opens details', () => {
+    const openDetails = vi.fn()
+    render(<TodoRow {...rowProps(resultNode('{"other":1}'), openDetails)} />)
+    expect(screen.getByText('todo_write · {"other":1}')).toBeTruthy()
+    fireEvent.click(screen.getByText('更新任务清单'))
+    expect(openDetails).toHaveBeenCalledTimes(1)
+  })
+
+  it('window-truncated result (call head lost) falls back to the callId summary', () => {
+    render(<TodoRow {...rowProps(resultNode('', { call: null }))} />)
+    expect(screen.getByText('todo_write · c1')).toBeTruthy()
+  })
+
+  it('todoToolview is a plain registrant riding the conversation load-order seam', () => {
+    expect(todoToolview.name).toBe('todo-toolview')
+    expect(todoToolview.inject).toEqual(['slots', 'conversation'])
+    const register = vi.fn()
+    todoToolview.apply({ slots: { register } } as never)
+    expect(register).toHaveBeenCalledWith({ name: 'conversation.chat.toolview', key: 'todo_write' }, TodoRow)
+  })
+})
diff --git a/packages/client/ui-trajectory/tests/views.spec.tsx b/packages/client/ui-trajectory/tests/views.spec.tsx
index 4818e67db5..ec37507443 100644
--- a/packages/client/ui-trajectory/tests/views.spec.tsx
+++ b/packages/client/ui-trajectory/tests/views.spec.tsx
@@ -109,6 +109,7 @@ function mount(slots: SlotsService, nodes: ConversationSnapshot['nodes'] = NODES
   const sessionSnapshot = createSnapshotStore({
     running: false, removed: false, promptError: null, nodes,
     partial: null, runningCalls: [] as ConversationSnapshot['runningCalls'],
+    todos: [] as ConversationSnapshot['todos'],
   })
   const useSession = bindSnapshotSelector(sessionSnapshot) as unknown as UseSession<ConversationSnapshot>
   const chat = createChatStore().create()

From de4f818c80b11dc56128087344d18b97f2e3d7ca Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Wed, 22 Jul 2026 13:15:51 +0800
Subject: [PATCH 03/36] test(gui): todo display fixture sample + browser
 acceptance script

fx-alpha gains turn 63: a todo_write call/result pair plus the todo/write
snapshot event, feeding both the TodoRow toolview and the TodoPanel strip
in ?fixture mode. verify-todo-display.mjs drives chromium through panel
visibility, content, row summary, details linkage, collapse and dark.
---
 .../client/connection/src/client/fixture.ts   |  12 ++
 scripts/verify-todo-display.mjs               | 107 ++++++++++++++++++
 2 files changed, 119 insertions(+)
 create mode 100644 scripts/verify-todo-display.mjs

diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts
index e1e21dfd78..82b3925aac 100644
--- a/packages/client/connection/src/client/fixture.ts
+++ b/packages/client/connection/src/client/fixture.ts
@@ -124,6 +124,18 @@ function buildAlphaLog(): SessionEvent[] {
   toolTurn(61, 'fx-write', '{"path":"notes/demo.txt","content":"hello fixture\\n"}', 'wrote notes/demo.txt')
   toolTurn(62, 'edit', '{"file_path":"notes/demo.txt","old_string":"hello","new_string":"hello fixture"}', '已编辑')
   toolTurn(63, 'write', '{"file_path":"notes/new-demo.txt","content":"hello fixture\\n"}', '已写入')
+  // Turn 64: todo_write sample — the TodoRow toolview in the flow plus the
+  // todo/write snapshot event feeding the TodoPanel plan strip.
+  const fixtureTodos = [
+    { content: '梳理需求', status: 'completed' },
+    { content: '实现 fixture 样本', status: 'in_progress' },
+    { content: '浏览器验收', status: 'pending' },
+  ]
+  const todoArgs = JSON.stringify({ todos: fixtureTodos })
+  toolTurn(64, 'todo_write', todoArgs, 'Updated todo list: 1 pending, 1 in progress, 1 completed.')
+  // The tool appends the snapshot event inside its own turn; splice it before the trailing turn/end.
+  events.splice(events.length - 1, 0, { type: 'todo/write', time: time += 800, data: { todos: fixtureTodos } })
+  events.forEach((e, i) => { e.seq = i })
   return events as unknown as SessionEvent[]
 }
 
diff --git a/scripts/verify-todo-display.mjs b/scripts/verify-todo-display.mjs
new file mode 100644
index 0000000000..3e8f42e5fe
--- /dev/null
+++ b/scripts/verify-todo-display.mjs
@@ -0,0 +1,107 @@
+// Manual acceptance probe: boot the real shell + 8 bundles in ?fixture mode,
+// open fx-alpha, assert the TodoPanel strip and the todo_write row render.
+import { existsSync } from 'node:fs'
+import { fileURLToPath } from 'node:url'
+import { createRequire } from 'node:module'
+import { startWebServer } from '@deepseek-ai/dsh-host-webserver'
+
+// playwright is a dependency of apps/web (the browser test owner), not the root.
+const { chromium } = createRequire(new URL('../apps/web/package.json', import.meta.url)).call(undefined, 'playwright')
+
+const root = fileURLToPath(new URL('..', import.meta.url))
+const bundle = (dir) => `${root}packages/client/${dir}/lib/client.js`
+const PLUGINS = [
+  { id: '@deepseek-ai/dsh-client-connection', dir: 'connection', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-runtime', dir: 'runtime', inject: ['@deepseek-ai/dsh-client-connection'], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-theme', dir: 'ui-theme', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-i18n', dir: 'i18n', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-layout', dir: 'ui-layout', inject: ['@deepseek-ai/dsh-client-runtime'] },
+  { id: '@deepseek-ai/dsh-client-ui-sidebar', dir: 'ui-sidebar', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-conversation', dir: 'ui-conversation', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-trajectory', dir: 'ui-trajectory', inject: ['@deepseek-ai/dsh-client-ui-conversation'] },
+]
+for (const p of PLUGINS) if (!existsSync(bundle(p.dir))) throw new Error(`bundle missing: ${p.dir}`)
+
+const rows = PLUGINS.map(p => ({
+  id: p.id, url: `/plugins/${p.id}/client.js?rev=verify`, rev: 'verify',
+  ...(p.inject.length > 0 ? { inject: p.inject } : {}),
+  ...(p.immediately ? { immediately: true } : {}),
+}))
+const graph = { rev: 'verify', entries: rows }
+const byId = new Map(PLUGINS.map(p => [p.id, bundle(p.dir)]))
+const port = 34567
+const errors = []
+const server = await startWebServer({
+  host: '127.0.0.1',
+  port,
+  distIndex: `${root}apps/web/dist/index.html`,
+  apiHandler: { fetch: () => Promise.resolve(new Response('fixture mode must not call /api', { status: 500 })) },
+  webPlugins: { graph: () => graph, clientPath: (id) => byId.get(id), onRebuilt: () => () => undefined },
+}, (err) => errors.push(`server: ${String(err)}`))
+
+const browser = await chromium.launch()
+const page = await browser.newPage({ viewport: { width: 1680, height: 1000 } })
+page.on('pageerror', e => errors.push(String(e)))
+await page.goto(`http://127.0.0.1:${port}/?fixture`, { waitUntil: 'load' })
+try {
+  await page.waitForSelector('[class*="frame"]', { timeout: 15000 })
+} catch (e) {
+  console.error('BODY:', (await page.evaluate(() => document.body.innerText)).slice(0, 400))
+  console.error('STATUS:', await page.evaluate(() => JSON.stringify(globalThis
+
+.__DSH_LOADER_STATUS__ ?? 'n/a')))
+  console.error('BOOT:', await page.evaluate(() => JSON.stringify(window.__DSH_BOOT__)))
+  const reqs = await page.evaluate(() => performance.getEntriesByType('resource').map(r => `${r.name.split('/').slice(-2).join('/')}=${r.responseStatus ?? '?'}`))
+  console.error('RES:', reqs.join(' '))
+  console.error('ERRORS:', errors.join(' ;; '))
+  throw e
+}
+
+// Open fx-alpha: expand the workspace group, then click the newest session row.
+await page.locator('[role="treeitem"]').first().click()
+// fx-alpha is the newest (running) session — the first option row.
+const sessionRow = page.locator('[role="treeitem"][aria-selected]').first()
+await sessionRow.waitFor({ timeout: 5000 })
+await sessionRow.click()
+await page.waitForSelector('[data-testid="todo-panel"]', { timeout: 15000 })
+console.log('✓ TodoPanel visible')
+
+const panelText = await page.locator('[data-testid="todo-panel"]').innerText()
+for (const expected of ['Plan', '1/3', '梳理需求', '实现 fixture 样本', '浏览器验收']) {
+  if (!panelText.includes(expected)) throw new Error(`TodoPanel missing "${expected}"; got: ${panelText}`)
+}
+console.log('✓ TodoPanel content: counts + all three items')
+
+await page.screenshot({ path: `${root}.artifacts/todo-01-panel.png` })
+
+// The todo_write row in the flow (turn 63 sample, already at the bottom).
+const row = page.locator('[data-sample="todo-row"]')
+await row.waitFor({ timeout: 10000 })
+const rowText = await row.innerText()
+if (!rowText.includes('更新任务清单') || !rowText.includes('1/3 已完成')) throw new Error(`TodoRow wrong: ${rowText}`)
+console.log('✓ TodoRow renders plan summary:', rowText.replace(/\n/g, ' '))
+await page.screenshot({ path: `${root}.artifacts/todo-02-row.png` })
+
+// Row click opens details with the raw args.
+await row.click()
+await page.waitForSelector('text=Input', { timeout: 5000 })
+console.log('✓ TodoRow click opens details')
+await page.screenshot({ path: `${root}.artifacts/todo-03-details.png` })
+
+// Collapse: list hides, active item hint appears in the header.
+await page.locator('[data-testid="todo-panel"] button').first().click()
+const collapsed = await page.locator('[data-testid="todo-panel"]').innerText()
+if (collapsed.includes('梳理需求')) throw new Error('collapse failed: list still visible')
+if (!collapsed.includes('实现 fixture 样本')) throw new Error('collapsed hint missing the active item')
+console.log('✓ Collapse hides list, shows active hint')
+await page.screenshot({ path: `${root}.artifacts/todo-04-collapsed.png` })
+
+// Dark theme spot check.
+await page.evaluate(() => document.body.setAttribute('data-ds-dark-theme', ''))
+await page.screenshot({ path: `${root}.artifacts/todo-05-dark.png` })
+console.log('✓ Dark screenshot taken')
+
+if (errors.length > 0) throw new Error(`page errors: ${errors.join('; ')}`)
+console.log('✓ No page errors — todo display acceptance PASSED')
+await browser.close()
+await server.close()

From 0a3c7f2b39af0865cf647108e0ea449f806a9922 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 13:54:33 +0800
Subject: [PATCH 04/36] =?UTF-8?q?docs(agents):=20web=20todo=20display=20Ag?=
 =?UTF-8?q?ent=20Note=20=E2=80=94=20side-effect=20channel=20+=20two=20surf?=
 =?UTF-8?q?aces?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 .../2026-07-23-web-todo-display.i18n.yaml     |  6 ++++
 .../feature/2026-07-23-web-todo-display.md    | 35 +++++++++++++++++++
 .../feature/2026-07-23-web-todo-display.zh.md | 35 +++++++++++++++++++
 3 files changed, 76 insertions(+)
 create mode 100644 .agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
 create mode 100644 .agents/notes/implemented/feature/2026-07-23-web-todo-display.md
 create mode 100644 .agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
new file mode 100644
index 0000000000..d8a6a517bf
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write
+2026-07-23-web-todo-display.md: 6b6af215016b2f73baa088653fcf133dbbfd3449
+2026-07-23-web-todo-display.zh.md: a06083190ba7b9b030a48aa7dc2c3177853f3cee
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
new file mode 100644
index 0000000000..6b6af21501
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -0,0 +1,35 @@
+# Agent Note: Web todo display — snapshot side-effect channel + two render surfaces
+
+Status: implemented
+
+English | [中文](2026-07-23-web-todo-display.zh.md)
+
+## Problem
+
+`todo_write` appends `todo/write` whole-list snapshots to the session log; the TUI renders a persistent plan panel and the ACP bridge maps the event to native `plan` updates. The web client dropped the event entirely: the host mux stream already forwards every session event, but `todo/write` is not a surface type (it never folds into `ConversationSnapshot.nodes`), and no side-effect branch accumulated it — the browser had no consumption point and no display surface.
+
+## Decision
+
+Consume `todo/write` as a Session side effect, not a surface node, and render it on two surfaces matching the split the TUI and ACP already draw.
+
+### Side-effect channel, converging with window replay
+
+`applyEventSideEffects` gains a `todo/write` case (whole list, last write wins) and `rebuildDerivedFromWindow` resets it, so the live path and every window rebuild (paging, reconnect stitch, resync) converge on the same latest snapshot — the same shape partial/openCalls already use. `ConversationSnapshot.todos` is the read surface. This follows the event's own contract ("log-only UI state; never derived history"): surfacing each write as a conversation node would render superseded lists as if they were still standing.
+
+### TodoPanel: the durable list as a persistent strip
+
+The skeleton pins the panel between the view area and the composer (the composer-card axis), hidden while empty, collapsible with the in-progress item as the collapsed one-line hint; ✓/●/○ glyphs mirror the TUI plan panel. It reads `snapshot.todos` via the framework `useSession` hook — no store, no service, no ctx. It lives inside `ConversationRoot` rather than the details column or its own slot: the details slot is single-occupant and selection-driven (a different lifetime than an always-on strip), and the slot table reserves no plan seat. The component is props-complete and framework-free, so a later relocation to a dedicated slot touches nothing inside it.
+
+### TodoRow: the per-call row through the toolview registry
+
+The dedicated `todo_write` chat row registers through the named `ctx.toolviews` registry from `apply` (the cross-domain assembly point, the same posture as the bash samples but a product registration). The summary derives from call args (`N/M done · active item`); unparseable args fall back to the generic row summary; clicking opens the details column with the raw args. No `ToolEventView` is added for todo — presentation is client-owned, and the durable list renders from the session event, not the tool card.
+
+## Alternatives considered
+
+- **Fold todo writes into `nodes` as surface entries** — replayed windows would render every superseded list; the event is deliberately not a surface type.
+- **Details column or a dedicated slot for the panel** — the details slot is single-occupant and selection-driven; a new slot key needs a slot-table seat that design has not assigned. The panel is framework-free, so the relocation stays cheap if one lands.
+- **Host-computed view (a todo `ToolEventView`)** — presentation belongs to the client; the wire already carries the whole snapshot in the event payload.
+
+## Consequences
+
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
new file mode 100644
index 0000000000..a06083190b
--- /dev/null
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -0,0 +1,35 @@
+# Agent Note: Web todo 展示——快照副作用通道 + 两个渲染面
+
+Status: implemented
+
+[English](2026-07-23-web-todo-display.md) | 中文
+
+## Problem
+
+`todo_write` 把 `todo/write` 的整份列表快照追加进会话日志；TUI 渲染一块常驻的 plan 面板，ACP 桥接把该事件映射为原生 `plan` 更新。Web 客户端把这个事件整个丢弃了：host mux 流本已转发每一个会话事件，但 `todo/write` 不是 surface 类型（它从不 fold 进 `ConversationSnapshot.nodes`），也没有任何副作用分支累积它——浏览器既无消费点，也无展示面。
+
+## Decision
+
+把 `todo/write` 当作 Session 副作用消费，而非 surface 节点，并在两个面上渲染它，这两个面正对应 TUI 与 ACP 已经绘制的那套划分。
+
+### 副作用通道，与窗口回放收敛
+
+`applyEventSideEffects` 新增一个 `todo/write` 分支（整份列表，后写覆盖先写），`rebuildDerivedFromWindow` 将其重置，于是实时路径与每一次窗口重建（分页、重连缝合、resync）都收敛到同一份最新快照——partial/openCalls 已在用的正是这个形态。`ConversationSnapshot.todos` 是读取面。这遵循事件自身的契约（「仅日志 UI 状态，绝非派生历史」）：把每次写入作为对话节点呈现，会让已被取代的列表看起来仍然有效。
+
+### TodoPanel：长驻列表作为一条常驻横条
+
+骨架把面板钉在视图区与 composer 之间（composer-card 轴），空列表时隐藏，可折叠——折叠态以进行中项作为单行提示；✓/●/○ 字形与 TUI plan 面板一致。它经框架 `useSession` hook 读取 `snapshot.todos`——无 store、无 service、无 ctx。它住在 `ConversationRoot` 之内，而非 details 列或自建 slot：details slot 单占用且由选中驱动（生命周期不同于一条常开横条），且 slot 表没有为 plan 预留席位。组件 props 完备且框架无关，因此日后迁往专用 slot 不触及其内部任何东西。
+
+### TodoRow：经 toolview 注册表的逐调用行
+
+专用的 `todo_write` 对话行经具名的 `ctx.toolviews` 注册表在 `apply` 中注册（跨域装配点，与 bash 样例同一姿态，但属产品级注册）。摘要由调用 args 推导（`N/M done · active item`）；无法解析的 args 回退到通用行摘要；点击会以原始 args 打开 details 列。todo 不新增任何 `ToolEventView`——呈现归客户端所有，常驻列表从会话事件渲染，而非工具卡。
+
+## Alternatives considered
+
+- **把 todo 写入作为 surface 条目折叠进 `nodes`**——回放的窗口会渲染每一份已被取代的列表；该事件被刻意设计成非 surface 类型。
+- **面板放进 details 列或专用 slot**——details slot 单占用且由选中驱动；新增一个 slot 键需要一个 slot 表席位，而设计尚未分配。面板框架无关，所以真要迁移，代价依然很低。
+- **host 计算的视图（一个 todo `ToolEventView`）**——呈现属于客户端；协议已在事件载荷里携带整份快照。
+
+## Consequences
+
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。

From 2d95a60ac9a012a9e01a86aca6b52adcedb36c21 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 14:10:00 +0800
Subject: [PATCH 05/36] fix(tool-todo): reject unknown item keys instead of
 silently dropping them

An item carrying keys beyond content/status (ids, children, priority) was
flattened to {content, status} on append, so the logged snapshot diverged
from what the model believed it wrote (model-visible must equal logged).
Reject loudly; the isError result lets the model self-correct.
---
 packages/todo/tool-todo/README.md               |  2 +-
 packages/todo/tool-todo/src/index.ts            | 11 +++++++++--
 packages/todo/tool-todo/tests/tool-todo.spec.ts |  1 +
 3 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/packages/todo/tool-todo/README.md b/packages/todo/tool-todo/README.md
index febb2c9ce0..2e06c7dde5 100644
--- a/packages/todo/tool-todo/README.md
+++ b/packages/todo/tool-todo/README.md
@@ -14,7 +14,7 @@ The list belongs to the ONE agent session that called the tool. There is no suba
 
 ## Validation
 
-Beyond the schema's type/required/enum checks, `execute` rejects an empty or duplicate `content` and more than one `in_progress` task (a coherent plan has at most one task active). Ordering and the discipline of keeping the list current are left to the model via the tool description.
+Beyond the schema's type/required/enum checks, `execute` rejects an empty or duplicate `content`, more than one `in_progress` task (a coherent plan has at most one task active), and any item key beyond `content`/`status` — an extended item shape (ids, nesting) fails loud instead of silently flattening, keeping the logged snapshot equal to what the model believes it wrote. Ordering and the discipline of keeping the list current are left to the model via the tool description.
 
 ## Rendering
 
diff --git a/packages/todo/tool-todo/src/index.ts b/packages/todo/tool-todo/src/index.ts
index 1da9914ac9..5297103e77 100644
--- a/packages/todo/tool-todo/src/index.ts
+++ b/packages/todo/tool-todo/src/index.ts
@@ -28,14 +28,21 @@ const DESCRIPTION =
 
 /**
  * Validate the value constraints the ParameterSchemaSpec can't express and build the canonical {@link
- * TodoItem}[]: trimmed non-empty unique content and at most one in-progress item. The registry
- * has already enforced the status enum; the cast below records that guarantee.
+ * TodoItem}[]: known keys only, trimmed non-empty unique content, and at most one in-progress
+ * item. The registry has already enforced the status enum; the cast below records that
+ * guarantee. Unknown keys are rejected rather than dropped — the logged snapshot must equal
+ * what the model believes it wrote (model-visible ⟺ logged), so a nested/extended item shape
+ * fails loud instead of silently flattening.
  */
 function toTodoList(raw: { content: string; status: string }[]): TodoItem[] {
   const todos: TodoItem[] = []
   const seen = new Set<string>()
   let inProgress = 0
   for (const item of raw) {
+    const unknown = Object.keys(item).filter(key => key !== 'content' && key !== 'status')
+    if (unknown.length > 0) {
+      throw new Error(`invalid todo: unknown key(s) ${unknown.map(k => JSON.stringify(k)).join(', ')} — each item is exactly { content, status }`)
+    }
     const content = item.content.trim()
     if (content.length === 0) {
       throw new Error('invalid todo: `content` must be a non-empty string')
diff --git a/packages/todo/tool-todo/tests/tool-todo.spec.ts b/packages/todo/tool-todo/tests/tool-todo.spec.ts
index 79758cb3ea..f83b202d32 100644
--- a/packages/todo/tool-todo/tests/tool-todo.spec.ts
+++ b/packages/todo/tool-todo/tests/tool-todo.spec.ts
@@ -126,6 +126,7 @@ describe('dsh-tool-todo', () => {
     { label: 'empty content', todos: [{ content: '   ', status: 'pending' }], fragment: 'non-empty' },
     { label: 'duplicate content', todos: [{ content: 'dup', status: 'pending' }, { content: 'dup', status: 'completed' }], fragment: 'duplicate' },
     { label: 'two in_progress', todos: [{ content: 'a', status: 'in_progress' }, { content: 'b', status: 'in_progress' }], fragment: 'in_progress' },
+    { label: 'unknown item keys', todos: [{ content: 'a', status: 'pending', children: [] }], fragment: 'unknown key' },
   ])('rejects $label as an isError result', async ({ todos, fragment }) => {
     const ctx = await setup()
     const result = await callTodo(ctx, { todos })

From 02abe3c821b811582f9cf4b572234d997fe297c3 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 15:56:28 +0800
Subject: [PATCH 06/36] fix(gui): todo-row guards valid-JSON invalid-shape args
 before dereferencing

null roots, non-object roots, and null array items (retained verbatim on a
rejected tool/call) now take the documented generic-summary fallback instead
of throwing into the row error boundary.
---
 .../client/ui-conversation/tests/todo-panel.spec.tsx   | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
index 761cb4dfd3..d3c5b93ecf 100644
--- a/packages/client/ui-conversation/tests/todo-panel.spec.tsx
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -113,6 +113,16 @@ describe('TodoRow', () => {
     expect(openDetails).toHaveBeenCalledTimes(1)
   })
 
+  it.each([
+    { label: 'null root', argsRaw: 'null' },
+    { label: 'non-object root', argsRaw: '42' },
+    { label: 'null items', argsRaw: '{"todos":[null]}' },
+  ])('falls back to the generic summary on valid JSON with an invalid shape ($label)', ({ argsRaw }) => {
+    render(<TodoRow {...rowProps(resultNode(argsRaw))} />)
+    // No throw, and the generic others summary carries the raw args verbatim.
+    expect(screen.getByText(`todo_write · ${argsRaw}`)).toBeTruthy()
+  })
+
   it('window-truncated result (call head lost) falls back to the callId summary', () => {
     render(<TodoRow {...rowProps(resultNode('', { call: null }))} />)
     expect(screen.getByText('todo_write · c1')).toBeTruthy()

From f2a9c09429d4826d05d751c9d69add16a26bf7d4 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 15:56:28 +0800
Subject: [PATCH 07/36] fix(gui): fixture emits todo/write at the real tool
 boundary
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The tool appends the snapshot mid-execution, between tool/call and
tool/result; the fixture spliced it after step/end with a post-turn
timestamp, so acceptance never exercised the production ordering. A spec
pins call → snapshot → result with monotonic times.
---
 packages/client/connection/src/client/fixture.ts |  8 ++++++--
 packages/client/connection/tests/fixture.spec.ts | 15 +++++++++++++++
 2 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts
index 82b3925aac..062a39a2cd 100644
--- a/packages/client/connection/src/client/fixture.ts
+++ b/packages/client/connection/src/client/fixture.ts
@@ -133,8 +133,12 @@ function buildAlphaLog(): SessionEvent[] {
   ]
   const todoArgs = JSON.stringify({ todos: fixtureTodos })
   toolTurn(64, 'todo_write', todoArgs, 'Updated todo list: 1 pending, 1 in progress, 1 completed.')
-  // The tool appends the snapshot event inside its own turn; splice it before the trailing turn/end.
-  events.splice(events.length - 1, 0, { type: 'todo/write', time: time += 800, data: { todos: fixtureTodos } })
+  // The real tool appends the snapshot mid-execution — between tool/call and
+  // tool/result — so the fixture reproduces that exact ordering (the last
+  // toolTurn events run ... tool/call, tool/result, step/end, turn/end).
+  const callIndex = events.length - 4
+  const callTime = events[callIndex]?.time as number
+  events.splice(callIndex + 1, 0, { type: 'todo/write', time: callTime + 400, data: { todos: fixtureTodos } })
   events.forEach((e, i) => { e.seq = i })
   return events as unknown as SessionEvent[]
 }
diff --git a/packages/client/connection/tests/fixture.spec.ts b/packages/client/connection/tests/fixture.spec.ts
index ac32955c36..ac363334bd 100644
--- a/packages/client/connection/tests/fixture.spec.ts
+++ b/packages/client/connection/tests/fixture.spec.ts
@@ -71,6 +71,21 @@ describe('createFixtureApi', () => {
     expect(empty.result.value).toEqual({ events: [], hasMore: false })
   })
 
+  it('emits the todo/write snapshot at the real tool boundary: between tool/call and tool/result, timestamps monotonic', async () => {
+    const api = createFixtureApi()
+    const tail = await api.sessions.history(req({ sessionId: sid('fx-alpha'), maxMessages: 10 }))
+    if (!tail.result.ok) throw new Error('history failed')
+    const events = tail.result.value.events.map(e => e.event)
+    const todoAt = events.findIndex(e => e.type === 'todo/write')
+    expect(todoAt).toBeGreaterThan(0)
+    // Production ordering (the tool appends mid-execution): call → snapshot → result.
+    expect(events[todoAt - 1]?.type).toBe('tool/call')
+    expect(events[todoAt + 1]?.type).toBe('tool/result')
+    const times = events.slice(todoAt - 1, todoAt + 2).map(e => e.time)
+    expect(times[0]).toBeLessThanOrEqual(times[1] ?? 0)
+    expect(times[1]).toBeLessThanOrEqual(times[2] ?? 0)
+  })
+
   it('create adds a session and pushes host/session-added to open host streams', async () => {
     const api = createFixtureApi()
     const abort = new AbortController()

From 3b1ad3ee6c37332f5cb7ddad21f4169220b754d3 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 15:56:28 +0800
Subject: [PATCH 08/36] test(gui): keyless assembled todo-display pass in the
 fixture smoke
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Eight real bundles through the DI chain in ?fixture mode: plan strip
content, dedicated row summary + details linkage, collapse hint, zero page
errors — the CI-gated assembled-surface coverage for the todo display.
---
 apps/web/tests/smoke-fixture.e2e.ts | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/apps/web/tests/smoke-fixture.e2e.ts b/apps/web/tests/smoke-fixture.e2e.ts
index 0726d14c8b..3b997442f5 100644
--- a/apps/web/tests/smoke-fixture.e2e.ts
+++ b/apps/web/tests/smoke-fixture.e2e.ts
@@ -285,6 +285,28 @@ describe('web boot chain success pass (keyless, nine real bundles, ?fixture)', (
     expect(await restoredInput.getAttribute('placeholder')).toBe('回复生成中，可停止后再输入')
   })
 
+  it('renders the todo plan strip and the dedicated todo_write row off the session events', async () => {
+    onTestFailed(() => saveFailureShot(page, 'smoke-todo-display'))
+    // The question pass above already opened the fixture session; the strip
+    // reads the todo/write snapshot (tail-page projection + window replay).
+    const panel = page.locator('[data-testid="todo-panel"]')
+    await panel.waitFor({ timeout: 15_000 })
+    const text = await panel.innerText()
+    for (const expected of ['Plan', '1/3', '梳理需求', '实现 fixture 样本', '浏览器验收']) {
+      expect(text).toContain(expected)
+    }
+    // The dedicated row renders through the keyed toolview hole with the plan summary.
+    const row = page.locator('[data-sample="todo-row"]')
+    await row.scrollIntoViewIfNeeded()
+    expect(await row.innerText()).toContain('1/3 已完成')
+    // Collapse hides the list and surfaces the active item as the header hint.
+    await panel.locator('button').first().click()
+    const collapsed = await panel.innerText()
+    expect(collapsed).not.toContain('梳理需求')
+    expect(collapsed).toContain('实现 fixture 样本')
+    await panel.locator('button').first().click() // restore for later passes
+  })
+
   it('stayed clean: no page errors across the whole load chain', () => {
     expect(pageErrors).toEqual([])
   })

From 356be9710ad03fef76b2952108f0bb7df0d9f32f Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 15:56:28 +0800
Subject: [PATCH 09/36] docs(gui): record the window-scoped todos gap and the
 web consumer

runtime README documents ConversationSnapshot.todos and its window-scoped
limitation; the todo tool README and Agent Note name the web client among
the event consumers; the web display note records the cold-load gap and
fix directions (bilingual pair re-recorded).
---
 .../notes/implemented/feature/2026-06-29-todo-write-tool.md   | 4 ++--
 .../implemented/feature/2026-07-23-web-todo-display.i18n.yaml | 4 ++--
 .../notes/implemented/feature/2026-07-23-web-todo-display.md  | 2 +-
 .../implemented/feature/2026-07-23-web-todo-display.zh.md     | 2 +-
 packages/client/runtime/README.md                             | 3 ++-
 packages/todo/tool-todo/README.md                             | 2 +-
 6 files changed, 9 insertions(+), 8 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
index bf5fceaafd..aa281a22a2 100644
--- a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
+++ b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
@@ -10,7 +10,7 @@ The harness gives the model bash and subagent tools but no way to record a struc
 
 ## Decision
 
-Add a model-facing `todo_write(todos: [{ content, status }])` tool whose whole-list state lives on the event-sourced session log as a new `todo/write` `SessionEventMap` variant. Both the stdio UI and the ACP bridge render off the existing `session/event` — the ACP bridge maps the list to a `plan` sessionUpdate.
+Add a model-facing `todo_write(todos: [{ content, status }])` tool whose whole-list state lives on the event-sourced session log as a new `todo/write` `SessionEventMap` variant. Every UI renders off the existing `session/event`: the stdio/TUI front doors show a persistent plan, the ACP bridge maps the list to a `plan` sessionUpdate, and the web client projects it into `ConversationSnapshot.todos` ([web todo display](2026-07-23-web-todo-display.md)).
 
 ### Whole-list replace, three-state status
 
@@ -18,7 +18,7 @@ The model sends the ENTIRE list every call; the new list replaces the old (last-
 
 ### State on the session log, not a service
 
-The list is appended as a `todo/write` event carrying the full `{ todos }` snapshot. The harness is event-sourced — the LLM history, tool calls, and turn structure all live on the log — so the todo list lives there too. This buys durability, replay, and `session/load` reconstruction for free: a reopened session re-derives the current list (the last `todo/write`) and the ACP bridge re-emits the `plan` on load, with no separate persistence backend, no in-memory service to rehydrate, and no extra wiring. An in-memory `ctx.todos` service would have had to reinvent all of that.
+The list is appended as a `todo/write` event carrying the full `{ todos }` snapshot. The harness is event-sourced — the LLM history, tool calls, and turn structure all live on the log — so the todo list lives there too. This buys durability, replay, and `session/load` reconstruction for free: a reopened session re-derives the current list (the last `todo/write`) and the ACP bridge re-emits the `plan` on load, with no separate persistence backend, no in-memory service to rehydrate, and no extra wiring. An in-memory `ctx.todos` service would have had to reinvent all of that. (Full-log consumers get this reconstruction outright; the web client's paged window currently sees only the tail page — the gap and its fix directions are recorded in the [web todo display note](2026-07-23-web-todo-display.md).)
 
 ### NOT a surface event
 
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index d8a6a517bf..1bfe2b59f3 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-web-todo-display.md: 6b6af215016b2f73baa088653fcf133dbbfd3449
-2026-07-23-web-todo-display.zh.md: a06083190ba7b9b030a48aa7dc2c3177853f3cee
+2026-07-23-web-todo-display.md: 003ab29546e8098331496641a28d095a1dd95fed
+2026-07-23-web-todo-display.zh.md: 200fd1d44dca7c5bcaa4f47ac31e26cb1f0f8d3d
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index 6b6af21501..003ab29546 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -32,4 +32,4 @@ The dedicated `todo_write` chat row registers through the named `ctx.toolviews`
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary.
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Known gap: the projection is window-scoped — reopening a session whose last `todo/write` precedes the tail history page shows an empty plan until the user pages back to it; restoring the tool note's cold-load reconstruction promise needs the current projection independent of the display window (host-attached on the history response, or a dedicated read).
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index a06083190b..200fd1d44d 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -32,4 +32,4 @@ Status: implemented
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。已知缺口：该投影以窗口为界——重新打开一个最后一次 `todo/write` 落在尾页之前的会话时，计划面板为空，直到用户翻页翻到它；要兑现工具 Note 里冷加载重建的承诺，需要一份独立于显示窗口的当前投影（history 响应由 host 附带，或提供专门的读取口）。
diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md
index 4fd5d15905..a9d03279f1 100644
--- a/packages/client/runtime/README.md
+++ b/packages/client/runtime/README.md
@@ -1,6 +1,6 @@
 # @deepseek-ai/dsh-client-runtime
 
-Client cordis boot + core services: SlotsService (Service wrapper over SlotCore + 'slots/changed' bridge), SessionsService (list store projection, scope tree, bindings, ancestry), the Session object layer (exported as a type; instances are owned and handed out by SessionsService — the manager/paging internals stay package-internal, tests reach them via src), ClientLoader (`./loader` subpath, statically held by the shell). Contract: api-contracts v3 §4.
+Client cordis boot + core services: SlotsService (Service wrapper over SlotCore + 'slots/changed' bridge), SessionsService (list store projection, scope tree, bindings, ancestry), the Session object layer (exported as a type; instances are owned and handed out by SessionsService — the manager/paging internals stay package-internal, tests reach them via src), ClientLoader (`./loader` subpath, statically held by the shell). Contract: api-contracts v3 §4. `ConversationSnapshot` carries `todos` — the latest `todo/write` whole-list snapshot in the window, consumed as a per-event side effect (last write wins) and reset on every window rebuild.
 
 ## Session title projection
 
@@ -19,3 +19,4 @@ None; this package neither assembles nor sends a provider request.
 - **`loader.unload` is a stub (throws not-implemented)** — the full chain (fiber dispose → registration cascade → style removal) lands with the HMR project.
 - **Scope teardown is stage-driven, single-occupant today** — the staged session follows `list.current` exactly (staging is the open signal: the event window opens ⟺ the session is on stage); a removed-while-staged session's scope survives frozen until the stage moves on, not until true observer count reaches zero. Resolution (`cell()`/`binding()`/`scope()`) is pure addressing, render-safe. The staged state can widen to a multi-pane list when concurrent panes land.
 - **Value imports of this package from plugin bundles must use the `/client` subpath** — the bare package name is not in the loader externals table and inlines a second module instance, whose private scope-tag Symbol never matches (the empty-state P0 postmortem).
+- **`todos` is window-scoped** — the projection scans only the paged display window, so reopening a session whose last `todo/write` precedes the tail page shows an empty plan until the user pages back to it. Restoring the tool's cold-load reconstruction promise needs the current projection independent of the window (host-attached on history, or a dedicated read).
diff --git a/packages/todo/tool-todo/README.md b/packages/todo/tool-todo/README.md
index 2e06c7dde5..d7479a3bb2 100644
--- a/packages/todo/tool-todo/README.md
+++ b/packages/todo/tool-todo/README.md
@@ -18,7 +18,7 @@ Beyond the schema's type/required/enum checks, `execute` rejects an empty or dup
 
 ## Rendering
 
-The canonical result is `{ todos, counts: { pending, inProgress, completed } }`; its Native renderer returns the compact update acknowledgement. The tool also writes the full `todo/write` session event. UIs subscribe to `session/event` and render that durable list themselves: the [TUI app](../../examples/tui-demo) shows a persistent plan, and the [ACP bridge](../../ui/acp) maps the list to a `plan` sessionUpdate (synthesizing the `priority` ACP requires).
+The canonical result is `{ todos, counts: { pending, inProgress, completed } }`; its Native renderer returns the compact update acknowledgement. The tool also writes the full `todo/write` session event. UIs subscribe to `session/event` and render that durable list themselves: the [TUI app](../../examples/tui-demo) shows a persistent plan, the [ACP bridge](../../ui/acp) maps the list to a `plan` sessionUpdate (synthesizing the `priority` ACP requires), and the [web client](../../client/ui-conversation) renders a plan strip plus a dedicated tool row off `ConversationSnapshot.todos` ([Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)).
 
 ## Export shape
 

From 1687c2c15c8f09d6ac74a04e824bc5718e3be388 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Thu, 23 Jul 2026 16:40:48 +0800
Subject: [PATCH 10/36] fix(gui): tail history page carries the full-log todo
 projection
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The client's todos projection derived only from the paged display window,
so reopening a session whose last todo/write preceded the tail page showed
an empty plan until the user paged back — session-level state cannot be
reconstructed from an arbitrary window. The host owns the full log, so the
tail history response now attaches todos (latest todo/write backscan, the
same posture as the view pairing); installWindow seeds it, window rebuilds
preserve it, and any in-window or live write keeps overwriting it. The
fixture mirrors the host; docs and both Agent Notes record the mechanism.
---
 .../feature/2026-06-29-todo-write-tool.md     |  2 +-
 .../2026-07-23-web-todo-display.i18n.yaml     |  4 +--
 .../feature/2026-07-23-web-todo-display.md    |  2 +-
 .../feature/2026-07-23-web-todo-display.zh.md |  2 +-
 .../client/connection/src/client/fixture.ts   | 15 +++++++--
 packages/client/runtime/README.md             |  4 +--
 .../runtime/src/client/sessions/session.ts    | 16 ++++++---
 packages/client/runtime/tests/fake-api.ts     |  2 +-
 packages/client/runtime/tests/session.spec.ts | 23 +++++++++++--
 packages/host/apiproxy/src/api/sessions.ts    |  8 +++--
 packages/host/runtime/src/api-proxy.ts        | 17 ++++++++--
 .../host/runtime/tests/api-proxy-view.spec.ts | 33 +++++++++++++++++++
 12 files changed, 107 insertions(+), 21 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
index aa281a22a2..8f9cf7fd9c 100644
--- a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
+++ b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.md
@@ -18,7 +18,7 @@ The model sends the ENTIRE list every call; the new list replaces the old (last-
 
 ### State on the session log, not a service
 
-The list is appended as a `todo/write` event carrying the full `{ todos }` snapshot. The harness is event-sourced — the LLM history, tool calls, and turn structure all live on the log — so the todo list lives there too. This buys durability, replay, and `session/load` reconstruction for free: a reopened session re-derives the current list (the last `todo/write`) and the ACP bridge re-emits the `plan` on load, with no separate persistence backend, no in-memory service to rehydrate, and no extra wiring. An in-memory `ctx.todos` service would have had to reinvent all of that. (Full-log consumers get this reconstruction outright; the web client's paged window currently sees only the tail page — the gap and its fix directions are recorded in the [web todo display note](2026-07-23-web-todo-display.md).)
+The list is appended as a `todo/write` event carrying the full `{ todos }` snapshot. The harness is event-sourced — the LLM history, tool calls, and turn structure all live on the log — so the todo list lives there too. This buys durability, replay, and `session/load` reconstruction for free: a reopened session re-derives the current list (the last `todo/write`) and the ACP bridge re-emits the `plan` on load, with no separate persistence backend, no in-memory service to rehydrate, and no extra wiring. An in-memory `ctx.todos` service would have had to reinvent all of that. (Full-log consumers get this reconstruction outright; the web client's paged window gets it from the tail history page's host-computed projection — see the [web todo display note](2026-07-23-web-todo-display.md).)
 
 ### NOT a surface event
 
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index 1bfe2b59f3..e3c6868165 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-web-todo-display.md: 003ab29546e8098331496641a28d095a1dd95fed
-2026-07-23-web-todo-display.zh.md: 200fd1d44dca7c5bcaa4f47ac31e26cb1f0f8d3d
+2026-07-23-web-todo-display.md: a5b45f288518cd53594aea724fb8eba532a1142d
+2026-07-23-web-todo-display.zh.md: d960a5ab402d9f1c53ddecc7838b19da0743ef3b
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index 003ab29546..a5b45f2885 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -32,4 +32,4 @@ The dedicated `todo_write` chat row registers through the named `ctx.toolviews`
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Known gap: the projection is window-scoped — reopening a session whose last `todo/write` precedes the tail history page shows an empty plan until the user pages back to it; restoring the tool note's cold-load reconstruction promise needs the current projection independent of the display window (host-attached on the history response, or a dedicated read).
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index 200fd1d44d..d960a5ab40 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -32,4 +32,4 @@ Status: implemented
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。已知缺口：该投影以窗口为界——重新打开一个最后一次 `todo/write` 落在尾页之前的会话时，计划面板为空，直到用户翻页翻到它；要兑现工具 Note 里冷加载重建的承诺，需要一份独立于显示窗口的当前投影（history 响应由 host 附带，或提供专门的读取口）。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
diff --git a/packages/client/connection/src/client/fixture.ts b/packages/client/connection/src/client/fixture.ts
index 062a39a2cd..3a84961677 100644
--- a/packages/client/connection/src/client/fixture.ts
+++ b/packages/client/connection/src/client/fixture.ts
@@ -6,7 +6,7 @@
 // approval/question requests exercise replay and composer takeover with stable rpcIds.
 
 import type { ContentBlock } from '@deepseek-ai/dsh-llm/types'
-import type { SessionEvent, SessionId } from '@deepseek-ai/dsh-session/types'
+import type { SessionEvent, SessionId, TodoItem } from '@deepseek-ai/dsh-session/types'
 import type {
   ApiProxy, ClientRequest, ClientResponse, HistoryEntry, HostFrame, MuxFrame, RpcReceipt,
   RpcRequest, RpcResponse, RpcResult, ServerRequest, ServerResponse, SessionSummary,
@@ -254,6 +254,15 @@ function pageOf(
   return { events, hasMore: start > 0 }
 }
 
+/** Current todo projection over the full log (host parallel: latest todo/write, last write wins). */
+function backscanTodos(log: readonly SessionEvent[]): TodoItem[] | undefined {
+  for (let i = log.length - 1; i >= 0; i--) {
+    const event = log[i]
+    if (event !== undefined && event.type === 'todo/write') return event.data.todos
+  }
+  return undefined
+}
+
 interface StreamConn<F> {
   push(envelope: RpcRequest<F>): void
 }
@@ -492,12 +501,14 @@ export function createFixtureApi(): ApiProxy {
         const log = logs.get(request.payload.sessionId) ?? []
         // Snapshot at request time, deliver after the transit delay (mirrors a real host under latency).
         const page = pageOf(log, request.payload.beforeSeq, request.payload.maxMessages ?? 50)
+        // Tail page carries the session-level todo projection (host parallel: full-log backscan).
+        const todos = request.payload.beforeSeq === undefined ? backscanTodos(log) : undefined
         const doomed = failNextHistory
         failNextHistory = false
         const delay = historyDelayMs
         if (delay > 0) await new Promise(resolve => setTimeout(resolve, delay))
         if (doomed) throw new Error('fixture: simulated history transport failure')
-        return ok(request, page)
+        return ok(request, { ...page, ...todos === undefined ? {} : { todos } })
       },
       prompt: (request) => {
         const { sessionId: id, mode, content } = request.payload
diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md
index a9d03279f1..72c4f1f059 100644
--- a/packages/client/runtime/README.md
+++ b/packages/client/runtime/README.md
@@ -1,6 +1,6 @@
 # @deepseek-ai/dsh-client-runtime
 
-Client cordis boot + core services: SlotsService (Service wrapper over SlotCore + 'slots/changed' bridge), SessionsService (list store projection, scope tree, bindings, ancestry), the Session object layer (exported as a type; instances are owned and handed out by SessionsService — the manager/paging internals stay package-internal, tests reach them via src), ClientLoader (`./loader` subpath, statically held by the shell). Contract: api-contracts v3 §4. `ConversationSnapshot` carries `todos` — the latest `todo/write` whole-list snapshot in the window, consumed as a per-event side effect (last write wins) and reset on every window rebuild.
+Client cordis boot + core services: SlotsService (Service wrapper over SlotCore + 'slots/changed' bridge), SessionsService (list store projection, scope tree, bindings, ancestry), the Session object layer (exported as a type; instances are owned and handed out by SessionsService — the manager/paging internals stay package-internal, tests reach them via src), ClientLoader (`./loader` subpath, statically held by the shell). Contract: api-contracts v3 §4. `ConversationSnapshot` carries `todos` — the session-level todo projection: seeded from the tail history page's full-log projection (`history` response `todos`), overwritten by every in-window or live `todo/write` (last write wins), and preserved across window rebuilds.
 
 ## Session title projection
 
@@ -19,4 +19,4 @@ None; this package neither assembles nor sends a provider request.
 - **`loader.unload` is a stub (throws not-implemented)** — the full chain (fiber dispose → registration cascade → style removal) lands with the HMR project.
 - **Scope teardown is stage-driven, single-occupant today** — the staged session follows `list.current` exactly (staging is the open signal: the event window opens ⟺ the session is on stage); a removed-while-staged session's scope survives frozen until the stage moves on, not until true observer count reaches zero. Resolution (`cell()`/`binding()`/`scope()`) is pure addressing, render-safe. The staged state can widen to a multi-pane list when concurrent panes land.
 - **Value imports of this package from plugin bundles must use the `/client` subpath** — the bare package name is not in the loader externals table and inlines a second module instance, whose private scope-tag Symbol never matches (the empty-state P0 postmortem).
-- **`todos` is window-scoped** — the projection scans only the paged display window, so reopening a session whose last `todo/write` precedes the tail page shows an empty plan until the user pages back to it. Restoring the tool's cold-load reconstruction promise needs the current projection independent of the window (host-attached on history, or a dedicated read).
+- **`SessionSummary.title` is a display projection** — the wire summary carries no title yet; the cwd basename stands in, then the raw id.
diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts
index 304b2e6e02..aff1bc1b0f 100644
--- a/packages/client/runtime/src/client/sessions/session.ts
+++ b/packages/client/runtime/src/client/sessions/session.ts
@@ -327,13 +327,13 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
         this.openError = result.error
         return
       }
-      this.installWindow(result.value.events, result.value.hasMore)
+      this.installWindow(result.value.events, result.value.hasMore, result.value.todos)
       // Gap detection (§D.3-4): baseline past the window tail and liveBuffer did not cover it -> pull the tail page once more.
       const tailSeq = this.windowTailSeq()
       if (this.subscribedLastSeq !== null && tailSeq !== null && this.subscribedLastSeq > tailSeq) {
         result = (await this.api.sessions.history({ sessionId: this.sessionId, maxMessages: PAGE_MESSAGES })).result
         if (generation !== this.openGeneration) return
-        if (result.ok) this.installWindow(result.value.events, result.value.hasMore)
+        if (result.ok) this.installWindow(result.value.events, result.value.hasMore, result.value.todos)
       }
       this.openState = 'open'
     } catch (error) {
@@ -351,11 +351,15 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
    *  Stitching MUST NOT route through acceptLiveEvent: openState is still 'loading' here
    *  (doOpen flips it after install), so recursing would push every buffered event straight
    *  back into liveBuffer where nothing ever drains it — a silent drop loop (audit S1). */
-  private installWindow(entries: HistoryEntry[], hasMore: boolean): void {
+  private installWindow(entries: HistoryEntry[], hasMore: boolean, todos?: readonly TodoItem[]): void {
     this.events = entries.map(e => e.event)
     this.views = entries.map(e => e.view)
     this.baseSeq = this.events[0]?.seq ?? 0
     this.hasMore = hasMore
+    // Session-level projection from the tail page (full-log latest todo/write,
+    // independent of the window); an in-window write below re-derives the same
+    // value, and later live events keep overwriting it.
+    if (todos !== undefined) this.todos = todos
     this.foldAdapter.reset(this.events, this.baseSeq, this.views)
     this.rebuildDerivedFromWindow()
     const buffered = this.liveBuffer
@@ -494,14 +498,16 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
 
   /** Re-derive state (partial/openCalls/frozenNodes) from raw window events after a rebuild — keeps
    *  paging/stitching consistent, and makes the live freeze and the history replay converge on the
-   *  same interrupted nodes (chunks are logged, so the replayed sweep re-freezes identical text). */
+   *  same interrupted nodes (chunks are logged, so the replayed sweep re-freezes identical text).
+   *  todos is deliberately NOT reset: it is session-level (seeded by the tail page's full-log
+   *  projection, not derivable from an arbitrary window). The window always extends to the log
+   *  tail, so an in-window todo/write can only overwrite it with the same latest value. */
   private rebuildDerivedFromWindow(): void {
     this.partial = null
     this.openCalls.clear()
     this.callsRev++
     this.frozenNodes = []
     this.frozenRev++
-    this.todos = []
     for (let i = 0; i < this.events.length; i++) {
       const event = this.events[i]
       /* v8 ignore next -- dense-array guard: i stays within events.length, so the undefined arm needs a sparse array no caller builds. */
diff --git a/packages/client/runtime/tests/fake-api.ts b/packages/client/runtime/tests/fake-api.ts
index 25f12c2b70..67907443f0 100644
--- a/packages/client/runtime/tests/fake-api.ts
+++ b/packages/client/runtime/tests/fake-api.ts
@@ -47,7 +47,7 @@ export class FakeApiClient implements IApiClient {
   onList: (payload: unknown) => Promise<RpcResponse<{ items: never[] }>> = () => Promise.resolve(ok({ items: [] }))
   onCreate: (payload: unknown) => Promise<RpcResponse<{ sessionId: SessionId }>> = () => Promise.resolve(ok({ sessionId: 'fk-new' as SessionId }))
   onHistory: (payload: { sessionId: SessionId; beforeSeq?: number; maxMessages?: number })
-  => Promise<RpcResponse<{ events: never[]; hasMore: boolean }>> =
+  => Promise<RpcResponse<{ events: never[]; hasMore: boolean; todos?: { content: string; status: 'pending' | 'in_progress' | 'completed' }[] }>> =
     () => Promise.resolve(ok({ events: [], hasMore: false }))
 
   onPrompt: (payload: unknown) => Promise<RpcResponse<{ accepted: true }>> = () => Promise.resolve(ok({ accepted: true as const }))
diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts
index 786bb9793b..990e14d1b4 100644
--- a/packages/client/runtime/tests/session.spec.ts
+++ b/packages/client/runtime/tests/session.spec.ts
@@ -22,9 +22,9 @@ function makeSession(api = new FakeApiClient()): { api: FakeApiClient; session:
   return { api, session: new Session(SID, api) }
 }
 
-function histResponse(events: SessionEvent[], hasMore = false) {
+function histResponse(events: SessionEvent[], hasMore = false, todos?: { content: string; status: 'pending' | 'in_progress' | 'completed' }[]) {
   // history now returns HistoryEntry[] ({event, view?}); these tests are view-less.
-  return Promise.resolve(ok({ events: entries(events) as never[], hasMore }))
+  return Promise.resolve(ok({ events: entries(events) as never[], hasMore, ...todos === undefined ? {} : { todos } }))
 }
 
 describe('open', () => {
@@ -170,6 +170,25 @@ describe('live event path', () => {
     expect(replayed.session.getSnapshot().todos).toEqual(listB)
   })
 
+  it('seeds todos from the tail page projection when the last write precedes the window', async () => {
+    const list = [{ content: '窗口外的计划', status: 'in_progress' as const }]
+    // Cold open: the page window carries NO todo/write; the projection rides the response.
+    const { api, session } = makeSession()
+    api.onHistory = () => histResponse(plainTurn(100, 9, '问', '答'), true, list)
+    await session.open()
+    expect(session.getSnapshot().todos).toEqual(list)
+    // Paging an older window in must not clear the session-level projection.
+    api.onHistory = () => histResponse(plainTurn(94, 8, '旧问', '旧答'), false)
+    await session.loadOlder()
+    expect(session.getSnapshot().todos).toEqual(list)
+    // A later live write still overrides the seeded projection.
+    session.handleMuxEnvelope('r' as never, {
+      type: 'session/event', sessionId: SID,
+      event: ev.todoWrite(106, [{ content: '新计划', status: 'pending' as const }]),
+    })
+    expect(session.getSnapshot().todos).toEqual([{ content: '新计划', status: 'pending' }])
+  })
+
   it('repairs a seq gap by repulling the tail page instead of appending a hole', async () => {
     const { api, session } = await opened(plainTurn(0, 0, 'a', 'b')) // tail seq = 5
     const repaired = [...plainTurn(0, 0, 'a', 'b'), ...plainTurn(6, 1, 'c', 'd')]
diff --git a/packages/host/apiproxy/src/api/sessions.ts b/packages/host/apiproxy/src/api/sessions.ts
index 393303f817..ee97405063 100644
--- a/packages/host/apiproxy/src/api/sessions.ts
+++ b/packages/host/apiproxy/src/api/sessions.ts
@@ -5,7 +5,7 @@
  */
 
 import type { ContentBlock } from '@deepseek-ai/dsh-llm/types'
-import type { SessionEvent, SessionId } from '@deepseek-ai/dsh-session/types'
+import type { SessionEvent, SessionId, TodoItem } from '@deepseek-ai/dsh-session/types'
 import type { RpcId, RpcRequest, RpcResponse } from './rpc.ts'
 import type { ToolEventView } from './events.ts'
 
@@ -60,9 +60,13 @@ export interface SessionsApi {
    * Each entry pairs the raw SessionEvent with the host-computed view (tool events whose
    * presenter produced one, evaluated against the registry at pagination time); the client
    * rebuilds the surface from the events with the shared fold.
+   * The tail page (beforeSeq absent) also carries `todos` — the session's current todo
+   * projection (latest `todo/write` over the FULL log, independent of the page window) —
+   * so a paged client restores the plan without walking history; absent when the session
+   * never wrote one. Older pages omit it (the projection is session-level, not per-page).
    */
   history(request: RpcRequest<{ sessionId: SessionId; beforeSeq?: number; maxMessages?: number }>):
-  Promise<RpcResponse<{ events: HistoryEntry[]; hasMore: boolean }>>
+  Promise<RpcResponse<{ events: HistoryEntry[]; hasMore: boolean; todos?: TodoItem[] }>>
 
   /** Sends a message. content is core's ContentBlock[] verbatim; mode maps 1:1 — queue→send, steer→steer. */
   prompt(request: RpcRequest<{ sessionId: SessionId; mode: 'queue' | 'steer'; content: ContentBlock[] }>):
diff --git a/packages/host/runtime/src/api-proxy.ts b/packages/host/runtime/src/api-proxy.ts
index acf77751af..1b794c0c4b 100644
--- a/packages/host/runtime/src/api-proxy.ts
+++ b/packages/host/runtime/src/api-proxy.ts
@@ -8,7 +8,7 @@ import { mkdir, stat } from 'node:fs/promises'
 import type { Context } from 'cordis'
 import type { Agent, AgentStatus } from '@deepseek-ai/dsh-agent'
 import type { ContentBlock, MessageSource } from '@deepseek-ai/dsh-llm'
-import type { JsonValue, Session, SessionEvent, SessionHeader, SessionId } from '@deepseek-ai/dsh-session'
+import type { JsonValue, Session, SessionEvent, SessionHeader, SessionId, TodoItem } from '@deepseek-ai/dsh-session'
 import type { SessionPersistence } from '@deepseek-ai/dsh-session-persistence'
 import { foldSessionTitle } from '@deepseek-ai/dsh-session-title'
 import type {
@@ -265,6 +265,15 @@ function backscanArgs(events: readonly SessionEvent[], callId: string): { name:
   return undefined
 }
 
+/** Current todo projection: the latest `todo/write` over the full log (whole-list replace ⇒ last write wins); undefined when none. */
+function backscanTodos(events: readonly SessionEvent[]): TodoItem[] | undefined {
+  for (let i = events.length - 1; i >= 0; i--) {
+    const event = events[i]
+    if (event !== undefined && event.type === 'todo/write') return event.data.todos
+  }
+  return undefined
+}
+
 /**
  * Thrown by the cold-resume path when the id names no servable session
  * (absent from the store, or a pre-project legacy log without a cwd).
@@ -435,7 +444,11 @@ export function createApiProxy(ctx: Context, defaults: ApiProxyDefaults): ApiPro
           const view = viewFor(ctx, event, callId => backscanArgs(page.events, callId))
           return { event, ...view === undefined ? {} : { view } }
         })
-        return ok(request, { events: entries, hasMore: page.hasMore })
+        // Tail page carries the session-level todo projection over the FULL
+        // log (the page window may not contain the last todo/write; a paged
+        // client cannot reconstruct session-level state from it).
+        const todos = beforeSeq === undefined ? backscanTodos(found.agent.session.events) : undefined
+        return ok(request, { events: entries, hasMore: page.hasMore, ...todos === undefined ? {} : { todos } })
       },
 
       async prompt(request) {
diff --git a/packages/host/runtime/tests/api-proxy-view.spec.ts b/packages/host/runtime/tests/api-proxy-view.spec.ts
index a7dcdc73c5..41f064a06f 100644
--- a/packages/host/runtime/tests/api-proxy-view.spec.ts
+++ b/packages/host/runtime/tests/api-proxy-view.spec.ts
@@ -154,6 +154,39 @@ describe('mux live view computation', () => {
     expect('view' in (byKey.get('tool/result:h-plain') ?? {})).toBe(false)
   })
 
+  it('tail page carries the full-log todo projection; older pages and todo-less sessions omit it', async () => {
+    const { ctx } = await harness()
+    const api = createApiProxy(ctx, { provider: 'p', model: 'm', cwd: '/tmp' })
+    const session = ctx.sessions.create()
+    ctx.agents.register({ id: session.id, session, status: 'idle', ctx } as Agent)
+    // Superseded write early in the log, latest write later; enough messages to page.
+    session.append('todo/write', { todos: [{ content: 'old', status: 'pending' }] })
+    for (let turn = 0; turn < 6; turn++) {
+      session.append('turn/start', { turn, trigger: { kind: 'message', source: { kind: 'user' } } })
+      session.append('user/message', { content: [{ type: 'text', text: `q${turn}` }], source: { kind: 'user' } }, { surfaceOp: 'append' })
+      session.append('assistant/message', { turn, step: 0, content: [{ type: 'text', text: `a${turn}` }], provenance: { provider: 'p', model: 'm' } }, { surfaceOp: 'append' })
+      session.append('turn/end', { turn, reason: { kind: 'completed' } })
+    }
+    session.append('todo/write', { todos: [{ content: 'current', status: 'in_progress' }] })
+
+    // Tail page limited to 2 messages: the latest todo/write may or may not sit
+    // in the window — the projection must come from the FULL log either way.
+    const tail = await api.sessions.history({ rpcId: RpcId('t-todos'), payload: { sessionId: session.id, maxMessages: 2 } })
+    if (!tail.result.ok) throw new Error('history failed')
+    expect(tail.result.value.todos).toEqual([{ content: 'current', status: 'in_progress' }])
+    // An older page omits the projection (session-level, tail-page-only).
+    const boundary = tail.result.value.events[0]?.event.seq ?? 0
+    const older = await api.sessions.history({ rpcId: RpcId('t-todos-2'), payload: { sessionId: session.id, beforeSeq: boundary, maxMessages: 2 } })
+    if (!older.result.ok) throw new Error('older failed')
+    expect('todos' in older.result.value).toBe(false)
+    // A session with no todo/write anywhere omits the field.
+    const bare = ctx.sessions.create()
+    ctx.agents.register({ id: bare.id, session: bare, status: 'idle', ctx } as Agent)
+    const bareTail = await api.sessions.history({ rpcId: RpcId('t-todos-3'), payload: { sessionId: bare.id } })
+    if (!bareTail.result.ok) throw new Error('bare failed')
+    expect('todos' in bareTail.result.value).toBe(false)
+  })
+
   it('drops a disposed session from the live open-call table (result after dispose gets no view)', async () => {
     const { ctx } = await harness()
     const api = createApiProxy(ctx, { provider: 'p', model: 'm', cwd: '/tmp' })

From beb191d87fe16b5ad69e6848f4b88d51c2e7089d Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 11:47:29 +0800
Subject: [PATCH 11/36] fix(gui): admit todos in the history wire schema
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

sessionHistoryValueSchema declared only events/hasMore, so the fetch
carrier's Zod parse stripped the tail page's todos projection — the
in-process and fixture paths carried it while a real WebApiClient lost it.
The fetch-carrier spec pins the field through the wire round trip.
---
 packages/host/apiproxy/src/api/sessions.schema.ts  |  7 +++++++
 packages/host/apiproxy/tests/fetch-carrier.spec.ts | 12 ++++++++++++
 2 files changed, 19 insertions(+)

diff --git a/packages/host/apiproxy/src/api/sessions.schema.ts b/packages/host/apiproxy/src/api/sessions.schema.ts
index 3edf0e6014..bb77383388 100644
--- a/packages/host/apiproxy/src/api/sessions.schema.ts
+++ b/packages/host/apiproxy/src/api/sessions.schema.ts
@@ -78,10 +78,17 @@ export const historyEntrySchema = z.object({
   view: toolEventViewSchema.optional(),
 }) satisfies z.ZodType<Wire<HistoryEntry>>
 
+/** One todo item of the tail page's session-level projection (the todo/write payload shape). */
+export const todoItemSchema = z.object({
+  content: z.string(),
+  status: z.union([z.literal('pending'), z.literal('in_progress'), z.literal('completed')]),
+})
+
 /** session.history response value. */
 export const sessionHistoryValueSchema = z.object({
   events: z.array(historyEntrySchema),
   hasMore: z.boolean(),
+  todos: z.array(todoItemSchema).optional(),
 }) satisfies z.ZodType<Wire<ResponseValue<'session.history'>>>
 
 /** ContentBlock passthrough: core is merge-extensible — the type discriminant envelope is strict, the rest stays wide. */
diff --git a/packages/host/apiproxy/tests/fetch-carrier.spec.ts b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
index d097daecef..3c2260a2ae 100644
--- a/packages/host/apiproxy/tests/fetch-carrier.spec.ts
+++ b/packages/host/apiproxy/tests/fetch-carrier.spec.ts
@@ -25,6 +25,12 @@ function fakeApi(overrides: Partial<{ muxFrames: MuxFrame[]; hostFrames: HostFra
         return { rpcId: request.rpcId, result: { ok: true, value: { sessionId: 's-new' as never } } }
       },
       async history(request) {
+        if (request.payload.sessionId === ('with-todos' as never)) {
+          return {
+            rpcId: request.rpcId,
+            result: { ok: true, value: { events: [], hasMore: false, todos: [{ content: 'current', status: 'in_progress' as const }] } },
+          }
+        }
         return {
           rpcId: request.rpcId,
           result: { ok: false, error: { code: 'session-not-found', message: 'nope', details: { sessionId: request.payload.sessionId } } },
@@ -69,6 +75,12 @@ describe('unary round trip (handler ⇄ client, no network)', () => {
     expect(response.rpcId).toMatch(/[0-9a-f-]{36}/)
   })
 
+  it('carries the tail-page todos projection through the wire schema (Zod must not strip it)', async () => {
+    const response = await client().sessions.history({ sessionId: 'with-todos' as never })
+    expect(response.result.ok).toBe(true)
+    if (response.result.ok) expect(response.result.value.todos).toEqual([{ content: 'current', status: 'in_progress' }])
+  })
+
   it('carries a business error as 200 + error result', async () => {
     const response = await client().sessions.history({ sessionId: 'missing' as never })
     expect(response.result.ok).toBe(false)

From c1ae98940cd4536d9c0ac00ddebac76de3f7f3db Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 11:47:29 +0800
Subject: [PATCH 12/36] fix(gui): gap repair adopts the repull response's todos
 projection

repairGap installed the repulled window without the response projection;
a todo/write missed during the gap and already outside the new tail page
kept the stale list. The spec pins adoption through the repair path.
---
 .../client/runtime/src/client/sessions/session.ts |  2 +-
 packages/client/runtime/tests/session.spec.ts     | 15 +++++++++++++++
 2 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts
index aff1bc1b0f..abe2a4b30c 100644
--- a/packages/client/runtime/src/client/sessions/session.ts
+++ b/packages/client/runtime/src/client/sessions/session.ts
@@ -410,7 +410,7 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
       const { result } = await this.api.sessions.history({ sessionId: this.sessionId, maxMessages: PAGE_MESSAGES })
       // Failure or superseded by a full resync: drop — the resync path rebuilds and clears the buffer itself.
       if (result.ok && generation === this.openGeneration && this.openState === 'open') {
-        this.installWindow(result.value.events, result.value.hasMore)
+        this.installWindow(result.value.events, result.value.hasMore, result.value.todos)
       }
     } catch (error) {
       console.error('[web-runtime] gap repair failed:', error)
diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts
index 990e14d1b4..8940e7fe78 100644
--- a/packages/client/runtime/tests/session.spec.ts
+++ b/packages/client/runtime/tests/session.spec.ts
@@ -202,6 +202,21 @@ describe('live event path', () => {
     const seqs = session.getSnapshot().nodes.map(n => n.seq)
     expect(seqs).toEqual([1, 3, 7, 9]) // both turns' user/assistant, no hole, no duplicate 9
   })
+
+  it('gap repair adopts the repull response projection (a missed todo/write outside the new tail page)', async () => {
+    const { api, session } = await opened(plainTurn(0, 0, 'a', 'b')) // tail seq = 5
+    expect(session.getSnapshot().todos).toEqual([])
+    // The missed range contained a todo/write that the repulled page no longer
+    // covers; the response's session-level projection is the only carrier.
+    const current = [{ content: '断线期间写的', status: 'in_progress' as const }]
+    api.onHistory = () => histResponse([...plainTurn(0, 0, 'a', 'b'), ...plainTurn(8, 1, 'c', 'd')], false, current)
+    session.handleMuxEnvelope('r' as never, { type: 'session/event', sessionId: SID, event: ev.assistant(11, 1, 'd') })
+    await vi.waitFor(() => {
+      expect(api.callsOf('session.history').length).toBe(2)
+    })
+    await Promise.resolve()
+    expect(session.getSnapshot().todos).toEqual(current)
+  })
 })
 
 describe('paging', () => {

From ba75229638ba659117e3b395c4399f07da5914b9 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 22:47:01 +0800
Subject: [PATCH 13/36] fix(tool-todo): declare the unknown-key rejection in
 the item schema
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

additionalProperties stays true in the published schema while execute
rejected extra keys, so generated typings and validation disagreed with
runtime behavior. The item schema now declares additionalProperties:
false — the registry's arg validation rejects extra keys with a
path-qualified violation before execute runs — and the redundant manual
check is dropped (tool catalog regenerated).
---
 docs/tool-catalog.md                            |  2 +-
 packages/todo/tool-todo/src/index.ts            | 16 ++++++----------
 packages/todo/tool-todo/tests/tool-todo.spec.ts |  2 +-
 3 files changed, 8 insertions(+), 12 deletions(-)

diff --git a/docs/tool-catalog.md b/docs/tool-catalog.md
index c10b0c90a7..864b5378d8 100644
--- a/docs/tool-catalog.md
+++ b/docs/tool-catalog.md
@@ -898,7 +898,7 @@ Record and update a structured task list for the current work. Send the ENTIRE l
       "description": "The COMPLETE task list, replacing any previous list.",
       "items": {
         "type": "object",
-        "additionalProperties": true,
+        "additionalProperties": false,
         "properties": {
           "content": {
             "type": "string",
diff --git a/packages/todo/tool-todo/src/index.ts b/packages/todo/tool-todo/src/index.ts
index 5297103e77..66b0a8ab12 100644
--- a/packages/todo/tool-todo/src/index.ts
+++ b/packages/todo/tool-todo/src/index.ts
@@ -28,21 +28,17 @@ const DESCRIPTION =
 
 /**
  * Validate the value constraints the ParameterSchemaSpec can't express and build the canonical {@link
- * TodoItem}[]: known keys only, trimmed non-empty unique content, and at most one in-progress
- * item. The registry has already enforced the status enum; the cast below records that
- * guarantee. Unknown keys are rejected rather than dropped — the logged snapshot must equal
- * what the model believes it wrote (model-visible ⟺ logged), so a nested/extended item shape
- * fails loud instead of silently flattening.
+ * TodoItem}[]: trimmed non-empty unique content and at most one in-progress item. The registry
+ * has already enforced the status enum and rejected unknown item keys (`additionalProperties:
+ * false` — the logged snapshot must equal what the model believes it wrote, so a nested/extended
+ * item shape fails loud at the schema boundary instead of silently flattening); the cast below
+ * records that guarantee.
  */
 function toTodoList(raw: { content: string; status: string }[]): TodoItem[] {
   const todos: TodoItem[] = []
   const seen = new Set<string>()
   let inProgress = 0
   for (const item of raw) {
-    const unknown = Object.keys(item).filter(key => key !== 'content' && key !== 'status')
-    if (unknown.length > 0) {
-      throw new Error(`invalid todo: unknown key(s) ${unknown.map(k => JSON.stringify(k)).join(', ')} — each item is exactly { content, status }`)
-    }
     const content = item.content.trim()
     if (content.length === 0) {
       throw new Error('invalid todo: `content` must be a non-empty string')
@@ -73,7 +69,7 @@ export function apply(ctx: Context): void {
         description: 'The COMPLETE task list, replacing any previous list.',
         items: {
           type: 'object',
-          additionalProperties: true,
+          additionalProperties: false,
           properties: {
             content: { type: 'string', required: true, description: 'What the task is — a short imperative line.' },
             status: {
diff --git a/packages/todo/tool-todo/tests/tool-todo.spec.ts b/packages/todo/tool-todo/tests/tool-todo.spec.ts
index f83b202d32..2883cfdc02 100644
--- a/packages/todo/tool-todo/tests/tool-todo.spec.ts
+++ b/packages/todo/tool-todo/tests/tool-todo.spec.ts
@@ -126,7 +126,7 @@ describe('dsh-tool-todo', () => {
     { label: 'empty content', todos: [{ content: '   ', status: 'pending' }], fragment: 'non-empty' },
     { label: 'duplicate content', todos: [{ content: 'dup', status: 'pending' }, { content: 'dup', status: 'completed' }], fragment: 'duplicate' },
     { label: 'two in_progress', todos: [{ content: 'a', status: 'in_progress' }, { content: 'b', status: 'in_progress' }], fragment: 'in_progress' },
-    { label: 'unknown item keys', todos: [{ content: 'a', status: 'pending', children: [] }], fragment: 'unknown key' },
+    { label: 'unknown item keys', todos: [{ content: 'a', status: 'pending', children: [] }], fragment: 'not a declared property' },
   ])('rejects $label as an isError result', async ({ todos, fragment }) => {
     const ctx = await setup()
     const result = await callTodo(ctx, { todos })

From 729dd3e1b75db4fab4e5f71641f7d9b7633adc13 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 22:47:01 +0800
Subject: [PATCH 14/36] docs(agents): correct the web todo note to the shipped
 mechanisms
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The projection sentence prescribed resetting todos on window rebuild —
the implementation deliberately preserves the tail-page seed and lets
only in-window/live writes overwrite. The toolview section named a
nonexistent ctx.toolviews registry — the shipped seam is the keyed
conversation.chat.toolview slot via ctx.slots.register. Both sides of
the bilingual pair re-recorded.
---
 .../feature/2026-07-23-web-todo-display.i18n.yaml           | 4 ++--
 .../implemented/feature/2026-07-23-web-todo-display.md      | 6 +++---
 .../implemented/feature/2026-07-23-web-todo-display.zh.md   | 6 +++---
 3 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index e3c6868165..86de69efbf 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-web-todo-display.md: a5b45f288518cd53594aea724fb8eba532a1142d
-2026-07-23-web-todo-display.zh.md: d960a5ab402d9f1c53ddecc7838b19da0743ef3b
+2026-07-23-web-todo-display.md: 15368ee44c3a3aee1f1854964b9f3575d1b2fa92
+2026-07-23-web-todo-display.zh.md: c9ac6cda23721296a9cae5431adac4182feab72c
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index a5b45f2885..15368ee44c 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -14,15 +14,15 @@ Consume `todo/write` as a Session side effect, not a surface node, and render it
 
 ### Side-effect channel, converging with window replay
 
-`applyEventSideEffects` gains a `todo/write` case (whole list, last write wins) and `rebuildDerivedFromWindow` resets it, so the live path and every window rebuild (paging, reconnect stitch, resync) converge on the same latest snapshot — the same shape partial/openCalls already use. `ConversationSnapshot.todos` is the read surface. This follows the event's own contract ("log-only UI state; never derived history"): surfacing each write as a conversation node would render superseded lists as if they were still standing.
+`applyEventSideEffects` gains a `todo/write` case (whole list, last write wins). Unlike partial/openCalls, `rebuildDerivedFromWindow` deliberately does NOT reset it: the value is session-level — seeded by the tail history page's full-log projection — and an arbitrary window may not contain the latest write, so rebuilds (paging, reconnect stitch, resync) preserve it and only an in-window or live write overwrites it. `ConversationSnapshot.todos` is the read surface. This follows the event's own contract ("log-only UI state; never derived history"): surfacing each write as a conversation node would render superseded lists as if they were still standing.
 
 ### TodoPanel: the durable list as a persistent strip
 
 The skeleton pins the panel between the view area and the composer (the composer-card axis), hidden while empty, collapsible with the in-progress item as the collapsed one-line hint; ✓/●/○ glyphs mirror the TUI plan panel. It reads `snapshot.todos` via the framework `useSession` hook — no store, no service, no ctx. It lives inside `ConversationRoot` rather than the details column or its own slot: the details slot is single-occupant and selection-driven (a different lifetime than an always-on strip), and the slot table reserves no plan seat. The component is props-complete and framework-free, so a later relocation to a dedicated slot touches nothing inside it.
 
-### TodoRow: the per-call row through the toolview registry
+### TodoRow: the per-call row through the keyed toolview slot
 
-The dedicated `todo_write` chat row registers through the named `ctx.toolviews` registry from `apply` (the cross-domain assembly point, the same posture as the bash samples but a product registration). The summary derives from call args (`N/M done · active item`); unparseable args fall back to the generic row summary; clicking opens the details column with the raw args. No `ToolEventView` is added for todo — presentation is client-owned, and the durable list renders from the session event, not the tool card.
+The dedicated `todo_write` chat row is a plain registrant plugin (`todoToolview`, mounted from `apply`) that registers into the keyed `conversation.chat.toolview` slot via `ctx.slots.register` — the same seam and load-order posture as the bash sample (`inject: ['slots', 'conversation']`), but a product registration. The summary derives from call args (`N/M done · active item`); unparseable args fall back to the generic row summary; clicking opens the details column with the raw args. No `ToolEventView` is added for todo — presentation is client-owned, and the durable list renders from the session event, not the tool card.
 
 ## Alternatives considered
 
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index d960a5ab40..c9ac6cda23 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -14,15 +14,15 @@ Status: implemented
 
 ### 副作用通道，与窗口回放收敛
 
-`applyEventSideEffects` 新增一个 `todo/write` 分支（整份列表，后写覆盖先写），`rebuildDerivedFromWindow` 将其重置，于是实时路径与每一次窗口重建（分页、重连缝合、resync）都收敛到同一份最新快照——partial/openCalls 已在用的正是这个形态。`ConversationSnapshot.todos` 是读取面。这遵循事件自身的契约（「仅日志 UI 状态，绝非派生历史」）：把每次写入作为对话节点呈现，会让已被取代的列表看起来仍然有效。
+`applyEventSideEffects` 新增一个 `todo/write` 分支（整份列表，后写覆盖先写）。与 partial/openCalls 不同，`rebuildDerivedFromWindow` 刻意不重置它：该值是会话级的——由尾页 history 携带的全量 log 投影播种——而任意窗口未必包含最近一次写入，因此窗口重建（分页、重连缝合、resync）保留它，只有窗口内或实时的写入才会覆盖。`ConversationSnapshot.todos` 是读取面。这遵循事件自身的契约（「仅日志 UI 状态，绝非派生历史」）：把每次写入作为对话节点呈现，会让已被取代的列表看起来仍然有效。
 
 ### TodoPanel：长驻列表作为一条常驻横条
 
 骨架把面板钉在视图区与 composer 之间（composer-card 轴），空列表时隐藏，可折叠——折叠态以进行中项作为单行提示；✓/●/○ 字形与 TUI plan 面板一致。它经框架 `useSession` hook 读取 `snapshot.todos`——无 store、无 service、无 ctx。它住在 `ConversationRoot` 之内，而非 details 列或自建 slot：details slot 单占用且由选中驱动（生命周期不同于一条常开横条），且 slot 表没有为 plan 预留席位。组件 props 完备且框架无关，因此日后迁往专用 slot 不触及其内部任何东西。
 
-### TodoRow：经 toolview 注册表的逐调用行
+### TodoRow：经 keyed toolview slot 的逐调用行
 
-专用的 `todo_write` 对话行经具名的 `ctx.toolviews` 注册表在 `apply` 中注册（跨域装配点，与 bash 样例同一姿态，但属产品级注册）。摘要由调用 args 推导（`N/M done · active item`）；无法解析的 args 回退到通用行摘要；点击会以原始 args 打开 details 列。todo 不新增任何 `ToolEventView`——呈现归客户端所有，常驻列表从会话事件渲染，而非工具卡。
+专用的 `todo_write` 对话行是一个普通注册者插件（`todoToolview`，由 `apply` 挂载），经 `ctx.slots.register` 注册进 keyed 的 `conversation.chat.toolview` slot——与 bash 样例同一接缝、同一载序姿态（`inject: ['slots', 'conversation']`），但属产品级注册。摘要由调用 args 推导（`N/M done · active item`）；无法解析的 args 回退到通用行摘要；点击会以原始 args 打开 details 列。todo 不新增任何 `ToolEventView`——呈现归客户端所有，常驻列表从会话事件渲染，而非工具卡。
 
 ## Alternatives considered
 

From 62bfc6b4fb49b41c403a6556a18050702cea2fc9 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Fri, 24 Jul 2026 23:10:19 +0800
Subject: [PATCH 15/36] docs(gui): bring the todo tool note's Chinese side
 along

master gave 2026-06-29-todo-write-tool a Chinese counterpart; the English
side's web-consumer sentences now translate across (in-body links keep
their .md targets per the pairing contract) and the pair is re-recorded.
The todo-panel fake gains the time/callTime fields ToolResultNode now
requires.
---
 .../implemented/feature/2026-06-29-todo-write-tool.i18n.yaml  | 4 ++--
 .../implemented/feature/2026-06-29-todo-write-tool.zh.md      | 4 ++--
 packages/client/ui-conversation/tests/todo-panel.spec.tsx     | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.i18n.yaml b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.i18n.yaml
index e6fc4aba97..afec670de8 100644
--- a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-06-29-todo-write-tool.md: bf5fceaafd475224914212beb460fdbc42e3dd68
-2026-06-29-todo-write-tool.zh.md: 3afc03a393284e2a9c2d6e234bc95f0faebbfb48
+2026-06-29-todo-write-tool.md: 8f9cf7fd9c102871566bded4db6fddf47519f130
+2026-06-29-todo-write-tool.zh.md: 7abfe1b0f3433d047c72e1e8a17527cb4b10fdd9
diff --git a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.zh.md b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.zh.md
index 3afc03a393..7abfe1b0f3 100644
--- a/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.zh.md
+++ b/.agents/notes/implemented/feature/2026-06-29-todo-write-tool.zh.md
@@ -10,7 +10,7 @@ harness 为模型提供了 bash 和 subagent 工具，却没有办法记录结
 
 ## 决策
 
-新增一个面向模型的 `todo_write(todos: [{ content, status }])` 工具，其整列表状态作为新的 `todo/write` `SessionEventMap` 变体存储在事件溯源的会话日志上。stdio UI 和 ACP bridge 均从现有的 `session/event` 渲染；ACP bridge 将列表映射为 `plan` sessionUpdate。
+新增一个面向模型的 `todo_write(todos: [{ content, status }])` 工具，其整列表状态作为新的 `todo/write` `SessionEventMap` 变体存储在事件溯源的会话日志上。每个 UI 都从现有的 `session/event` 渲染：stdio/TUI 前门展示常驻计划，ACP bridge 将列表映射为 `plan` sessionUpdate，web 客户端将其投影进 `ConversationSnapshot.todos`（[web todo 展示](2026-07-23-web-todo-display.md)）。
 
 ### 整列表替换，三态 status
 
@@ -18,7 +18,7 @@ harness 为模型提供了 bash 和 subagent 工具，却没有办法记录结
 
 ### 状态在会话日志上，而非服务
 
-列表作为 `todo/write` 事件追加到日志，携带完整的 `{ todos }` 快照。harness 是事件溯源的——LLM（大语言模型）历史、工具调用和轮次结构都在日志上——所以 todo 列表也在那里。这免费获得了持久性、回放和 `session/load` 重建：重新打开的会话从最后一条 `todo/write` 重新推导当前列表，ACP bridge 在加载时重新发出 `plan`，无需独立的持久化后端、无需重新注水的内存服务、无需额外接线。一个内存中的 `ctx.todos` 服务需要重新发明以上所有。
+列表作为 `todo/write` 事件追加到日志，携带完整的 `{ todos }` 快照。harness 是事件溯源的——LLM（大语言模型）历史、工具调用和轮次结构都在日志上——所以 todo 列表也在那里。这免费获得了持久性、回放和 `session/load` 重建：重新打开的会话从最后一条 `todo/write` 重新推导当前列表，ACP bridge 在加载时重新发出 `plan`，无需独立的持久化后端、无需重新注水的内存服务、无需额外接线。一个内存中的 `ctx.todos` 服务需要重新发明以上所有。（全量 log 消费者直接获得这份重建；web 客户端的分页窗口则从尾页 history 携带的 host 计算投影获得——见 [web todo 展示 Note](2026-07-23-web-todo-display.md)。）
 
 ### 不是 surface 事件
 
diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
index d3c5b93ecf..196077e6a4 100644
--- a/packages/client/ui-conversation/tests/todo-panel.spec.tsx
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -69,7 +69,7 @@ describe('TodoPanel', () => {
 })
 
 const resultNode = (argsRaw: string, over?: Partial<ToolResultNode>): ToolResultNode => ({
-  kind: 'tool-result', seq: 10, callId: 'c1',
+  kind: 'tool-result', seq: 10, time: 2_000, callTime: 1_000, callId: 'c1',
   call: { name: 'todo_write', argsRaw },
   content: [], isError: false, callView: null, resultView: null, ...over,
 })

From 34eef10f045cb8874db6426e3847465c6409afd0 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sat, 25 Jul 2026 00:00:50 +0800
Subject: [PATCH 16/36] test(snapshot): re-record tool schemas for the todo
 item key tightening

additionalProperties: false on the todo_write item schema is model-visible
(tool schemas ride the request header and the code-mode prompt types), so
the pinned ACP/headless expected outputs re-record. Keyless refresh; the
two locally-failing scenarios are this machine's known environment issues
(HOME-symlink cwd normalization, SQLite ExperimentalWarning), not the diff.
---
 .../snapshots/advanced-toolchain/system-prompt.expected.md    | 2 +-
 .../snapshots/advanced-toolchain/tool-schemas.expected.json   | 2 +-
 .../tests/snapshots/both-mode-turn/system-prompt.expected.md  | 2 +-
 .../tests/snapshots/both-mode-turn/tool-schemas.expected.json | 2 +-
 .../tests/snapshots/code-mode-turn/system-prompt.expected.md  | 2 +-
 .../code-mode-workspace-context/system-prompt.expected.md     | 2 +-
 .../tests/snapshots/lsp-definition/tool-schemas.expected.json | 2 +-
 .../snapshots/model-switching/tool-schemas.expected.json      | 4 ++--
 .../snapshots/permission-switching/tool-schemas.expected.json | 4 ++--
 .../tests/snapshots/plan-mode/tool-schemas.expected.json      | 4 ++--
 .../tests/snapshots/pty-tools/tool-schemas.expected.json      | 2 +-
 .../tests/snapshots/skill-load/tool-schemas.expected.json     | 2 +-
 .../tests/snapshots/text-turn/tool-schemas.expected.json      | 2 +-
 .../snapshots/workspace-context/tool-schemas.expected.json    | 2 +-
 .../tests/snapshots/advanced-toolchain/session.1.jsonl        | 2 +-
 .../tests/snapshots/advanced-toolchain/session.2.jsonl        | 2 +-
 16 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/examples/acp-agent/tests/snapshots/advanced-toolchain/system-prompt.expected.md b/examples/acp-agent/tests/snapshots/advanced-toolchain/system-prompt.expected.md
index 73c31413bf..672948ecbe 100644
--- a/examples/acp-agent/tests/snapshots/advanced-toolchain/system-prompt.expected.md
+++ b/examples/acp-agent/tests/snapshots/advanced-toolchain/system-prompt.expected.md
@@ -188,7 +188,7 @@ interface ToolArgsMap {
       content: string;
       /** pending (not started) | in_progress (now) | completed (done). */
       status: "pending" | "in_progress" | "completed";
-    } & Record<string, JsonValue>)[];
+    })[];
   } & Record<string, JsonValue>;
   /** Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason. */
   update_goal: {
diff --git a/examples/acp-agent/tests/snapshots/advanced-toolchain/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/advanced-toolchain/tool-schemas.expected.json
index 6b50a5d220..e5173eb3f9 100644
--- a/examples/acp-agent/tests/snapshots/advanced-toolchain/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/advanced-toolchain/tool-schemas.expected.json
@@ -440,7 +440,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/both-mode-turn/system-prompt.expected.md b/examples/acp-agent/tests/snapshots/both-mode-turn/system-prompt.expected.md
index 8744029272..9556f9eec5 100644
--- a/examples/acp-agent/tests/snapshots/both-mode-turn/system-prompt.expected.md
+++ b/examples/acp-agent/tests/snapshots/both-mode-turn/system-prompt.expected.md
@@ -171,7 +171,7 @@ interface ToolArgsMap {
       content: string;
       /** pending (not started) | in_progress (now) | completed (done). */
       status: "pending" | "in_progress" | "completed";
-    } & Record<string, JsonValue>)[];
+    })[];
   } & Record<string, JsonValue>;
   /** Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason. */
   update_goal: {
diff --git a/examples/acp-agent/tests/snapshots/both-mode-turn/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/both-mode-turn/tool-schemas.expected.json
index 7ceeec4042..a0bd131348 100644
--- a/examples/acp-agent/tests/snapshots/both-mode-turn/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/both-mode-turn/tool-schemas.expected.json
@@ -383,7 +383,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/code-mode-turn/system-prompt.expected.md b/examples/acp-agent/tests/snapshots/code-mode-turn/system-prompt.expected.md
index 8744029272..9556f9eec5 100644
--- a/examples/acp-agent/tests/snapshots/code-mode-turn/system-prompt.expected.md
+++ b/examples/acp-agent/tests/snapshots/code-mode-turn/system-prompt.expected.md
@@ -171,7 +171,7 @@ interface ToolArgsMap {
       content: string;
       /** pending (not started) | in_progress (now) | completed (done). */
       status: "pending" | "in_progress" | "completed";
-    } & Record<string, JsonValue>)[];
+    })[];
   } & Record<string, JsonValue>;
   /** Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason. */
   update_goal: {
diff --git a/examples/acp-agent/tests/snapshots/code-mode-workspace-context/system-prompt.expected.md b/examples/acp-agent/tests/snapshots/code-mode-workspace-context/system-prompt.expected.md
index 8744029272..9556f9eec5 100644
--- a/examples/acp-agent/tests/snapshots/code-mode-workspace-context/system-prompt.expected.md
+++ b/examples/acp-agent/tests/snapshots/code-mode-workspace-context/system-prompt.expected.md
@@ -171,7 +171,7 @@ interface ToolArgsMap {
       content: string;
       /** pending (not started) | in_progress (now) | completed (done). */
       status: "pending" | "in_progress" | "completed";
-    } & Record<string, JsonValue>)[];
+    })[];
   } & Record<string, JsonValue>;
   /** Update the exact current goal revision. edit, pause, and resume require a direct top-level human request. During an automatic continuation of the current goal, complete and blocked are also allowed. blocked is rejected before the configured minimum round count; the model remains responsible for judging that the same condition persisted across those rounds and must explain it in blocked_reason. */
   update_goal: {
diff --git a/examples/acp-agent/tests/snapshots/lsp-definition/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/lsp-definition/tool-schemas.expected.json
index b42a434388..c78fe8c63f 100644
--- a/examples/acp-agent/tests/snapshots/lsp-definition/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/lsp-definition/tool-schemas.expected.json
@@ -404,7 +404,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/model-switching/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/model-switching/tool-schemas.expected.json
index ef40784fa5..d4abb8f48b 100644
--- a/examples/acp-agent/tests/snapshots/model-switching/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/model-switching/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
@@ -917,7 +917,7 @@
               "description": "The COMPLETE task list, replacing any previous list.",
               "items": {
                 "type": "object",
-                "additionalProperties": true,
+                "additionalProperties": false,
                 "properties": {
                   "content": {
                     "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/permission-switching/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/permission-switching/tool-schemas.expected.json
index ef40784fa5..d4abb8f48b 100644
--- a/examples/acp-agent/tests/snapshots/permission-switching/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/permission-switching/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
@@ -917,7 +917,7 @@
               "description": "The COMPLETE task list, replacing any previous list.",
               "items": {
                 "type": "object",
-                "additionalProperties": true,
+                "additionalProperties": false,
                 "properties": {
                   "content": {
                     "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/plan-mode/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/plan-mode/tool-schemas.expected.json
index ef40784fa5..d4abb8f48b 100644
--- a/examples/acp-agent/tests/snapshots/plan-mode/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/plan-mode/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
@@ -917,7 +917,7 @@
               "description": "The COMPLETE task list, replacing any previous list.",
               "items": {
                 "type": "object",
-                "additionalProperties": true,
+                "additionalProperties": false,
                 "properties": {
                   "content": {
                     "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/pty-tools/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/pty-tools/tool-schemas.expected.json
index 529b1419da..b98e501452 100644
--- a/examples/acp-agent/tests/snapshots/pty-tools/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/pty-tools/tool-schemas.expected.json
@@ -496,7 +496,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/skill-load/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/skill-load/tool-schemas.expected.json
index b01e7683d1..9ce90ffd83 100644
--- a/examples/acp-agent/tests/snapshots/skill-load/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/skill-load/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/text-turn/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/text-turn/tool-schemas.expected.json
index b01e7683d1..9ce90ffd83 100644
--- a/examples/acp-agent/tests/snapshots/text-turn/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/text-turn/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/workspace-context/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/workspace-context/tool-schemas.expected.json
index b01e7683d1..9ce90ffd83 100644
--- a/examples/acp-agent/tests/snapshots/workspace-context/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/workspace-context/tool-schemas.expected.json
@@ -367,7 +367,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl
index 045b9bb736..6cae860e36 100644
--- a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl
+++ b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.1.jsonl
@@ -3,7 +3,7 @@
 {"type":"user/message","seq":1,"time":1783957884563,"data":{"content":[{"type":"text","text":"Reply with exactly DIRECT_CHILD_OK and nothing else."}],"source":{"kind":"user"}},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1783957884563,"data":{"title":"Reply with exactly DIRECT_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1783957884564,"data":{"turn":1,"step":1}}
-{"type":"request/header","seq":4,"time":1783957884564,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Calls execute sequentially, even under `Promise.all`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n  /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n  bash: {\n    /** The bash command to execute. */\n    command: string;\n    /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n    description: string;\n    /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n    timeoutMs?: number;\n    /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n    workdir?: string;\n    /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n  cordis_inspect: {\n    /** Limit the report to one section. Omit for all sections. */\n    what?: \"services\" | \"plugins\" | \"tools\" | \"dynamic\" | \"api\" | \"events\";\n    /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n    name?: string;\n  } & Record<string, JsonValue>;\n  /** Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n  cordis_mount: {\n    /** Body of an async JS function; must `return` the plugin to mount. */\n    code: string;\n  } & Record<string, JsonValue>;\n  /** Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop). */\n  cordis_unmount: {\n    /** The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\"). */\n    id: string;\n  } & Record<string, JsonValue>;\n  /** Edit an existing UTF-8 text file by replacing literal text. */\n  edit: {\n    /** Path to edit, resolved by the filesystem backend. */\n    file_path: string;\n    /** Literal text to replace. Must match exactly. */\n    old_string: string;\n    /** Literal replacement text. Use an empty string to delete the match. */\n    new_string: string;\n    /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n    replace_all?: boolean;\n  } & Record<string, JsonValue>;\n  /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n  ralph: {\n    /** The immutable completion objective for every fresh Ralph round. */\n    objective: string;\n    /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n    maxRounds?: number;\n  } & Record<string, JsonValue>;\n  /** Read a UTF-8 text file and return line-numbered content. */\n  read: {\n    /** Path to read, resolved by the filesystem backend. */\n    file_path: string;\n    /** 1-based first line to return. Defaults to 1. */\n    offset?: number;\n    /** Maximum number of lines to return. Defaults to 2000. */\n    limit?: number;\n  } & Record<string, JsonValue>;\n  /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n  skill: {\n    /** The exact skill name from the available skills list. */\n    name: string;\n  } & Record<string, JsonValue>;\n  /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent_fork: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n  task_kill: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Optional short reason, recorded in the log and forwarded to the task. */\n    reason?: string;\n  } & Record<string, JsonValue>;\n  /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n  task_list: Record<string, JsonValue>;\n  /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n  task_output: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n    wait?: boolean;\n    /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n    timeout_ms?: number;\n  } & Record<string, JsonValue>;\n  /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n  todo_write: {\n    /** The COMPLETE task list, replacing any previous list. */\n    todos: ({\n      /** What the task is — a short imperative line. */\n      content: string;\n      /** pending (not started) | in_progress (now) | completed (done). */\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    } & Record<string, JsonValue>)[];\n  } & Record<string, JsonValue>;\n  /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n  workflow: {\n    /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`). */\n    script: string;\n    /** The workflow identity block (plain JSON — never code). */\n    meta: {\n      /** Short kebab-case workflow name. */\n      name: string;\n      /** One-line description of what the workflow does. */\n      description: string;\n      /** Optional guidance on when this workflow applies. */\n      whenToUse?: string;\n      /** Optional phase declarations matched by phase() calls. */\n      phases?: ({\n        /** The phase title phase() calls match by exact string. */\n        title: string;\n        /** Optional one-line description of the phase. */\n        detail?: string;\n        /** Optional provider override this phase is expected to use. */\n        provider?: string;\n        /** Optional model override this phase is expected to use. */\n        model?: string;\n      } & Record<string, JsonValue>)[];\n    } & Record<string, JsonValue>;\n    /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n    args?: Record<string, JsonValue>;\n  } & Record<string, JsonValue>;\n  /** Create or fully replace a UTF-8 text file. */\n  write: {\n    /** Path to write, resolved by the filesystem backend. */\n    file_path: string;\n    /** Full UTF-8 text content to write. */\n    content: string;\n  } & Record<string, JsonValue>;\n}\n\ninterface ToolOutputMap {\n  bash: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    exitCode: number | null;\n    signal: string | null;\n    timedOut: boolean;\n    aborted: boolean;\n    timeoutMs: number;\n    stdout: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    stderr: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    sandbox?: {\n      mode: string;\n      denied: boolean;\n      enforcement?: string;\n      runnerFailed?: boolean;\n    };\n  };\n  cordis_inspect: string;\n  cordis_mount: {\n    id: string;\n    pluginName: string;\n    state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n    provides: string[];\n    waitingFor: string[];\n  };\n  cordis_unmount: {\n    id: string;\n    pluginName: string;\n  };\n  edit: {\n    path: string;\n    before: string;\n    after: string;\n  };\n  ralph: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  read: {\n    path: string;\n    offset: number;\n    lines: {\n      number: number;\n      text: string;\n    }[];\n    totalLines: number;\n  };\n  skill: {\n    name: string;\n    provider: string;\n    resourceBase?: {\n      kind: \"directory\";\n      path: string;\n    } | {\n      kind: \"url\";\n      url: string;\n    } | {\n      kind: \"opaque\";\n      description: string;\n    };\n    content: string;\n  };\n  subagent: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  subagent_fork: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  task_kill: {\n    outcome: \"cancellation-requested\" | \"already-finished\";\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  task_list: ({\n    id: string;\n    kind: string;\n    label: string;\n    status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n    detail?: string;\n    startedAt: number;\n    finishedAt?: number;\n  })[];\n  task_output: {\n    text: string;\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  todo_write: {\n    todos: ({\n      content: string;\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n    counts: {\n      pending: number;\n      inProgress: number;\n      completed: number;\n    };\n  };\n  workflow: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  write: {\n    path: string;\n    operation: \"create\" | \"update\";\n    before: string | null;\n    after: string;\n  };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n  readonly name: \"ToolCallError\";\n  readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n  [K in ToolName]: (args: ToolArgsMap[K]) => Promise<ToolOutputMap[K]>;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","dynamic","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"Body of an async JS function; must `return` the plugin to mount."}},"required":["code"]}},{"name":"cordis_unmount","description":"Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop).","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\")."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."}},"required":["code"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":true,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}}
+{"type":"request/header","seq":4,"time":1783957884564,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Calls execute sequentially, even under `Promise.all`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n  /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n  bash: {\n    /** The bash command to execute. */\n    command: string;\n    /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n    description: string;\n    /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n    timeoutMs?: number;\n    /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n    workdir?: string;\n    /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n  cordis_inspect: {\n    /** Limit the report to one section. Omit for all sections. */\n    what?: \"services\" | \"plugins\" | \"tools\" | \"dynamic\" | \"api\" | \"events\";\n    /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n    name?: string;\n  } & Record<string, JsonValue>;\n  /** Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n  cordis_mount: {\n    /** Body of an async JS function; must `return` the plugin to mount. */\n    code: string;\n  } & Record<string, JsonValue>;\n  /** Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop). */\n  cordis_unmount: {\n    /** The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\"). */\n    id: string;\n  } & Record<string, JsonValue>;\n  /** Edit an existing UTF-8 text file by replacing literal text. */\n  edit: {\n    /** Path to edit, resolved by the filesystem backend. */\n    file_path: string;\n    /** Literal text to replace. Must match exactly. */\n    old_string: string;\n    /** Literal replacement text. Use an empty string to delete the match. */\n    new_string: string;\n    /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n    replace_all?: boolean;\n  } & Record<string, JsonValue>;\n  /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n  ralph: {\n    /** The immutable completion objective for every fresh Ralph round. */\n    objective: string;\n    /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n    maxRounds?: number;\n  } & Record<string, JsonValue>;\n  /** Read a UTF-8 text file and return line-numbered content. */\n  read: {\n    /** Path to read, resolved by the filesystem backend. */\n    file_path: string;\n    /** 1-based first line to return. Defaults to 1. */\n    offset?: number;\n    /** Maximum number of lines to return. Defaults to 2000. */\n    limit?: number;\n  } & Record<string, JsonValue>;\n  /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n  skill: {\n    /** The exact skill name from the available skills list. */\n    name: string;\n  } & Record<string, JsonValue>;\n  /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent_fork: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n  task_kill: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Optional short reason, recorded in the log and forwarded to the task. */\n    reason?: string;\n  } & Record<string, JsonValue>;\n  /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n  task_list: Record<string, JsonValue>;\n  /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n  task_output: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n    wait?: boolean;\n    /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n    timeout_ms?: number;\n  } & Record<string, JsonValue>;\n  /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n  todo_write: {\n    /** The COMPLETE task list, replacing any previous list. */\n    todos: ({\n      /** What the task is — a short imperative line. */\n      content: string;\n      /** pending (not started) | in_progress (now) | completed (done). */\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n  } & Record<string, JsonValue>;\n  /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n  workflow: {\n    /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`). */\n    script: string;\n    /** The workflow identity block (plain JSON — never code). */\n    meta: {\n      /** Short kebab-case workflow name. */\n      name: string;\n      /** One-line description of what the workflow does. */\n      description: string;\n      /** Optional guidance on when this workflow applies. */\n      whenToUse?: string;\n      /** Optional phase declarations matched by phase() calls. */\n      phases?: ({\n        /** The phase title phase() calls match by exact string. */\n        title: string;\n        /** Optional one-line description of the phase. */\n        detail?: string;\n        /** Optional provider override this phase is expected to use. */\n        provider?: string;\n        /** Optional model override this phase is expected to use. */\n        model?: string;\n      } & Record<string, JsonValue>)[];\n    } & Record<string, JsonValue>;\n    /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n    args?: Record<string, JsonValue>;\n  } & Record<string, JsonValue>;\n  /** Create or fully replace a UTF-8 text file. */\n  write: {\n    /** Path to write, resolved by the filesystem backend. */\n    file_path: string;\n    /** Full UTF-8 text content to write. */\n    content: string;\n  } & Record<string, JsonValue>;\n}\n\ninterface ToolOutputMap {\n  bash: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    exitCode: number | null;\n    signal: string | null;\n    timedOut: boolean;\n    aborted: boolean;\n    timeoutMs: number;\n    stdout: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    stderr: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    sandbox?: {\n      mode: string;\n      denied: boolean;\n      enforcement?: string;\n      runnerFailed?: boolean;\n    };\n  };\n  cordis_inspect: string;\n  cordis_mount: {\n    id: string;\n    pluginName: string;\n    state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n    provides: string[];\n    waitingFor: string[];\n  };\n  cordis_unmount: {\n    id: string;\n    pluginName: string;\n  };\n  edit: {\n    path: string;\n    before: string;\n    after: string;\n  };\n  ralph: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  read: {\n    path: string;\n    offset: number;\n    lines: {\n      number: number;\n      text: string;\n    }[];\n    totalLines: number;\n  };\n  skill: {\n    name: string;\n    provider: string;\n    resourceBase?: {\n      kind: \"directory\";\n      path: string;\n    } | {\n      kind: \"url\";\n      url: string;\n    } | {\n      kind: \"opaque\";\n      description: string;\n    };\n    content: string;\n  };\n  subagent: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  subagent_fork: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  task_kill: {\n    outcome: \"cancellation-requested\" | \"already-finished\";\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  task_list: ({\n    id: string;\n    kind: string;\n    label: string;\n    status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n    detail?: string;\n    startedAt: number;\n    finishedAt?: number;\n  })[];\n  task_output: {\n    text: string;\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  todo_write: {\n    todos: ({\n      content: string;\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n    counts: {\n      pending: number;\n      inProgress: number;\n      completed: number;\n    };\n  };\n  workflow: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  write: {\n    path: string;\n    operation: \"create\" | \"update\";\n    before: string | null;\n    after: string;\n  };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n  readonly name: \"ToolCallError\";\n  readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n  [K in ToolName]: (args: ToolArgsMap[K]) => Promise<ToolOutputMap[K]>;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","dynamic","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"Body of an async JS function; must `return` the plugin to mount."}},"required":["code"]}},{"name":"cordis_unmount","description":"Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop).","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\")."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."}},"required":["code"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}}
 {"type":"assistant/chunk","seq":5,"time":1783950001005,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"text"}}}
 {"type":"assistant/chunk","seq":6,"time":1783957884564,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":0,"text":"DIRECT_CHILD_OK"}}}
 {"type":"assistant/chunk","seq":7,"time":1783957884564,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"DIRECT_CHILD_OK"}}}}
diff --git a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl
index 8193973bee..c00a4119c7 100644
--- a/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl
+++ b/examples/headless-agent/tests/snapshots/advanced-toolchain/session.2.jsonl
@@ -3,7 +3,7 @@
 {"type":"user/message","seq":1,"time":1783957884700,"data":{"content":[{"type":"text","text":"Reply with exactly WORKFLOW_CHILD_OK and nothing else."}],"source":{"kind":"user"}},"surfaceOp":"append"}
 {"type":"session/title","seq":2,"time":1783957884700,"data":{"title":"Reply with exactly WORKFLOW_CHILD_OK and","messageSeqs":[1],"source":{"kind":"fallback"}}}
 {"type":"step/start","seq":3,"time":1783957884700,"data":{"turn":1,"step":1}}
-{"type":"request/header","seq":4,"time":1783957884701,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Calls execute sequentially, even under `Promise.all`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n  /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n  bash: {\n    /** The bash command to execute. */\n    command: string;\n    /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n    description: string;\n    /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n    timeoutMs?: number;\n    /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n    workdir?: string;\n    /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n  cordis_inspect: {\n    /** Limit the report to one section. Omit for all sections. */\n    what?: \"services\" | \"plugins\" | \"tools\" | \"dynamic\" | \"api\" | \"events\";\n    /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n    name?: string;\n  } & Record<string, JsonValue>;\n  /** Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n  cordis_mount: {\n    /** Body of an async JS function; must `return` the plugin to mount. */\n    code: string;\n  } & Record<string, JsonValue>;\n  /** Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop). */\n  cordis_unmount: {\n    /** The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\"). */\n    id: string;\n  } & Record<string, JsonValue>;\n  /** Edit an existing UTF-8 text file by replacing literal text. */\n  edit: {\n    /** Path to edit, resolved by the filesystem backend. */\n    file_path: string;\n    /** Literal text to replace. Must match exactly. */\n    old_string: string;\n    /** Literal replacement text. Use an empty string to delete the match. */\n    new_string: string;\n    /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n    replace_all?: boolean;\n  } & Record<string, JsonValue>;\n  /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n  ralph: {\n    /** The immutable completion objective for every fresh Ralph round. */\n    objective: string;\n    /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n    maxRounds?: number;\n  } & Record<string, JsonValue>;\n  /** Read a UTF-8 text file and return line-numbered content. */\n  read: {\n    /** Path to read, resolved by the filesystem backend. */\n    file_path: string;\n    /** 1-based first line to return. Defaults to 1. */\n    offset?: number;\n    /** Maximum number of lines to return. Defaults to 2000. */\n    limit?: number;\n  } & Record<string, JsonValue>;\n  /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n  skill: {\n    /** The exact skill name from the available skills list. */\n    name: string;\n  } & Record<string, JsonValue>;\n  /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent_fork: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n  task_kill: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Optional short reason, recorded in the log and forwarded to the task. */\n    reason?: string;\n  } & Record<string, JsonValue>;\n  /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n  task_list: Record<string, JsonValue>;\n  /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n  task_output: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n    wait?: boolean;\n    /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n    timeout_ms?: number;\n  } & Record<string, JsonValue>;\n  /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n  todo_write: {\n    /** The COMPLETE task list, replacing any previous list. */\n    todos: ({\n      /** What the task is — a short imperative line. */\n      content: string;\n      /** pending (not started) | in_progress (now) | completed (done). */\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    } & Record<string, JsonValue>)[];\n  } & Record<string, JsonValue>;\n  /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n  workflow: {\n    /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`). */\n    script: string;\n    /** The workflow identity block (plain JSON — never code). */\n    meta: {\n      /** Short kebab-case workflow name. */\n      name: string;\n      /** One-line description of what the workflow does. */\n      description: string;\n      /** Optional guidance on when this workflow applies. */\n      whenToUse?: string;\n      /** Optional phase declarations matched by phase() calls. */\n      phases?: ({\n        /** The phase title phase() calls match by exact string. */\n        title: string;\n        /** Optional one-line description of the phase. */\n        detail?: string;\n        /** Optional provider override this phase is expected to use. */\n        provider?: string;\n        /** Optional model override this phase is expected to use. */\n        model?: string;\n      } & Record<string, JsonValue>)[];\n    } & Record<string, JsonValue>;\n    /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n    args?: Record<string, JsonValue>;\n  } & Record<string, JsonValue>;\n  /** Create or fully replace a UTF-8 text file. */\n  write: {\n    /** Path to write, resolved by the filesystem backend. */\n    file_path: string;\n    /** Full UTF-8 text content to write. */\n    content: string;\n  } & Record<string, JsonValue>;\n}\n\ninterface ToolOutputMap {\n  bash: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    exitCode: number | null;\n    signal: string | null;\n    timedOut: boolean;\n    aborted: boolean;\n    timeoutMs: number;\n    stdout: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    stderr: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    sandbox?: {\n      mode: string;\n      denied: boolean;\n      enforcement?: string;\n      runnerFailed?: boolean;\n    };\n  };\n  cordis_inspect: string;\n  cordis_mount: {\n    id: string;\n    pluginName: string;\n    state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n    provides: string[];\n    waitingFor: string[];\n  };\n  cordis_unmount: {\n    id: string;\n    pluginName: string;\n  };\n  edit: {\n    path: string;\n    before: string;\n    after: string;\n  };\n  ralph: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  read: {\n    path: string;\n    offset: number;\n    lines: {\n      number: number;\n      text: string;\n    }[];\n    totalLines: number;\n  };\n  skill: {\n    name: string;\n    provider: string;\n    resourceBase?: {\n      kind: \"directory\";\n      path: string;\n    } | {\n      kind: \"url\";\n      url: string;\n    } | {\n      kind: \"opaque\";\n      description: string;\n    };\n    content: string;\n  };\n  subagent: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  subagent_fork: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  task_kill: {\n    outcome: \"cancellation-requested\" | \"already-finished\";\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  task_list: ({\n    id: string;\n    kind: string;\n    label: string;\n    status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n    detail?: string;\n    startedAt: number;\n    finishedAt?: number;\n  })[];\n  task_output: {\n    text: string;\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  todo_write: {\n    todos: ({\n      content: string;\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n    counts: {\n      pending: number;\n      inProgress: number;\n      completed: number;\n    };\n  };\n  workflow: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  write: {\n    path: string;\n    operation: \"create\" | \"update\";\n    before: string | null;\n    after: string;\n  };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n  readonly name: \"ToolCallError\";\n  readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n  [K in ToolName]: (args: ToolArgsMap[K]) => Promise<ToolOutputMap[K]>;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","dynamic","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"Body of an async JS function; must `return` the plugin to mount."}},"required":["code"]}},{"name":"cordis_unmount","description":"Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop).","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\")."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."}},"required":["code"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":true,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}}
+{"type":"request/header","seq":4,"time":1783957884701,"data":{"header":{"config":{"provider":"deepseek","model":"deepseek-v4-flash"},"system":"You are an AI agent powered by the DeepSeek Harness SDK.\n\nYou are headless-agent, a coding assistant powered by the deepseek-v4-flash model. Your working directory is /tmp/advanced-headless.\n\nVerify your work by running the code or tests. Keep answers brief and factual.\n\n\nUse the read tool — not shell commands like cat — to inspect text files. Results include line numbers. Use offset and limit to continue reading large files.\n\nUse the write tool to create files or completely replace file contents. Existing files are overwritten, so read an existing file first (the default fs-policy requires it) and prefer edit for targeted changes.\n\nUse the edit tool for targeted changes to existing UTF-8 text files. It replaces literal old_string with new_string; by default old_string must appear exactly once. If old_string appears multiple times, provide a more specific old_string or set replace_all to true. Read the file first (the default fs-policy requires it), unless you just created or edited it in this session.\n\nCheck the [exit code: N] marker on every bash result; investigate failures before moving on.\n\nTrack every background task id you start. You are notified in-session when a task finishes — do not busy-poll or sleep on one; keep working on independent steps and do not duplicate a running task's work. Before giving a final answer, collect every still-relevant task with task_output (set wait: true only when you are genuinely blocked on it), and task_kill tasks that stopped mattering.\n\nUse the workflow tool ONLY when the user explicitly asks for a workflow or for large multi-agent orchestration: you write a JavaScript script (the tool description documents the exact format) that fans work out across many subagents with phases and structured results. For one or two delegations, prefer plain subagent calls.\n\nUse the ralph tool ONLY when the direct human explicitly asks for a Ralph loop or fresh-agent iterative execution. Each Ralph round starts a fresh child with no conversation seed and uses the shared workspace as durable memory. Completion and blockers are worker reports, not independent evaluation. Use same-session goal tools for ordinary long-running objectives, and plain subagents or workflows for bounded delegation and fan-out.\n\n## Writing code for run_code\n\nPass `run_code` the body of an async TypeScript function (erasable syntax only — no `enum` or namespaces; type annotations are advisory, the code runs type-stripped). Inside the program:\n\n- Call tools as `await tools.name(args)` — quoted access for exotic names: `tools[\"my-tool\"](args)`. Every call resolves to the tool's typed canonical JSON value. Tool arguments must be lossless JSON.\n- A FAILED tool call rejects with `ToolCallError`, whose `toolName` identifies the failed tool and whose `message` is human-readable — `try/catch` it to handle and continue.\n- Calls execute sequentially, even under `Promise.all`.\n- Emit results with `return` and/or `console.log(...)`. ONLY what you print or return comes back to you — intermediate tool results never enter the conversation, so extract just what you need.\n\nThe available tools:\n\n```ts\ntype JsonValue = null | boolean | number | string | JsonValue[] | { [key: string]: JsonValue }\n\ninterface ToolArgsMap {\n  /** Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`. */\n  bash: {\n    /** The bash command to execute. */\n    command: string;\n    /** Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\". */\n    description: string;\n    /** Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry. */\n    timeoutMs?: number;\n    /** Working directory for this command. Defaults to the session workspace; a relative path is resolved against it. */\n    workdir?: string;\n    /** Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc. */\n  cordis_inspect: {\n    /** Limit the report to one section. Omit for all sections. */\n    what?: \"services\" | \"plugins\" | \"tools\" | \"dynamic\" | \"api\" | \"events\";\n    /** Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\". */\n    name?: string;\n  } & Record<string, JsonValue>;\n  /** Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime. */\n  cordis_mount: {\n    /** Body of an async JS function; must `return` the plugin to mount. */\n    code: string;\n  } & Record<string, JsonValue>;\n  /** Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop). */\n  cordis_unmount: {\n    /** The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\"). */\n    id: string;\n  } & Record<string, JsonValue>;\n  /** Edit an existing UTF-8 text file by replacing literal text. */\n  edit: {\n    /** Path to edit, resolved by the filesystem backend. */\n    file_path: string;\n    /** Literal text to replace. Must match exactly. */\n    old_string: string;\n    /** Literal replacement text. Use an empty string to delete the match. */\n    new_string: string;\n    /** Replace all matches. Defaults to false; when false, old_string must appear exactly once. */\n    replace_all?: boolean;\n  } & Record<string, JsonValue>;\n  /** Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools. */\n  ralph: {\n    /** The immutable completion objective for every fresh Ralph round. */\n    objective: string;\n    /** Optional positive safe-integer round cap, bounded by the deployment ceiling. */\n    maxRounds?: number;\n  } & Record<string, JsonValue>;\n  /** Read a UTF-8 text file and return line-numbered content. */\n  read: {\n    /** Path to read, resolved by the filesystem backend. */\n    file_path: string;\n    /** 1-based first line to return. Defaults to 1. */\n    offset?: number;\n    /** Maximum number of lines to return. Defaults to 2000. */\n    limit?: number;\n  } & Record<string, JsonValue>;\n  /** Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill. */\n  skill: {\n    /** The exact skill name from the available skills list. */\n    name: string;\n  } & Record<string, JsonValue>;\n  /** Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`. */\n  subagent_fork: {\n    /** A short (3-5 word) description of the delegated task, for display. */\n    description: string;\n    /** The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new. */\n    prompt: string;\n    /** Run as a background task and return its id; collect with task_output or stop with task_kill. */\n    run_in_background?: boolean;\n  } & Record<string, JsonValue>;\n  /** Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops. */\n  task_kill: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Optional short reason, recorded in the log and forwarded to the task. */\n    reason?: string;\n  } & Record<string, JsonValue>;\n  /** List your background tasks (running and finished) with their ids, kinds, and statuses. */\n  task_list: Record<string, JsonValue>;\n  /** Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap. */\n  task_output: {\n    /** Task id returned by the tool that started the background work. */\n    task_id: string;\n    /** Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive. */\n    wait?: boolean;\n    /** Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum. */\n    timeout_ms?: number;\n  } & Record<string, JsonValue>;\n  /** Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished). */\n  todo_write: {\n    /** The COMPLETE task list, replacing any previous list. */\n    todos: ({\n      /** What the task is — a short imperative line. */\n      content: string;\n      /** pending (not started) | in_progress (now) | completed (done). */\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n  } & Record<string, JsonValue>;\n  /** Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn. The workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result. Script-body hooks: - `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly. - `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages. - `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`. - `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim. Misused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`. Constraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes. */\n  workflow: {\n    /** The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`). */\n    script: string;\n    /** The workflow identity block (plain JSON — never code). */\n    meta: {\n      /** Short kebab-case workflow name. */\n      name: string;\n      /** One-line description of what the workflow does. */\n      description: string;\n      /** Optional guidance on when this workflow applies. */\n      whenToUse?: string;\n      /** Optional phase declarations matched by phase() calls. */\n      phases?: ({\n        /** The phase title phase() calls match by exact string. */\n        title: string;\n        /** Optional one-line description of the phase. */\n        detail?: string;\n        /** Optional provider override this phase is expected to use. */\n        provider?: string;\n        /** Optional model override this phase is expected to use. */\n        model?: string;\n      } & Record<string, JsonValue>)[];\n    } & Record<string, JsonValue>;\n    /** Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}). */\n    args?: Record<string, JsonValue>;\n  } & Record<string, JsonValue>;\n  /** Create or fully replace a UTF-8 text file. */\n  write: {\n    /** Path to write, resolved by the filesystem backend. */\n    file_path: string;\n    /** Full UTF-8 text content to write. */\n    content: string;\n  } & Record<string, JsonValue>;\n}\n\ninterface ToolOutputMap {\n  bash: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    exitCode: number | null;\n    signal: string | null;\n    timedOut: boolean;\n    aborted: boolean;\n    timeoutMs: number;\n    stdout: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    stderr: {\n      text: string;\n      truncated: boolean;\n      spillPath?: string;\n    };\n    sandbox?: {\n      mode: string;\n      denied: boolean;\n      enforcement?: string;\n      runnerFailed?: boolean;\n    };\n  };\n  cordis_inspect: string;\n  cordis_mount: {\n    id: string;\n    pluginName: string;\n    state: \"pending\" | \"loading\" | \"active\" | \"failed\" | \"disposed\" | \"unloading\";\n    provides: string[];\n    waitingFor: string[];\n  };\n  cordis_unmount: {\n    id: string;\n    pluginName: string;\n  };\n  edit: {\n    path: string;\n    before: string;\n    after: string;\n  };\n  ralph: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  read: {\n    path: string;\n    offset: number;\n    lines: {\n      number: number;\n      text: string;\n    }[];\n    totalLines: number;\n  };\n  skill: {\n    name: string;\n    provider: string;\n    resourceBase?: {\n      kind: \"directory\";\n      path: string;\n    } | {\n      kind: \"url\";\n      url: string;\n    } | {\n      kind: \"opaque\";\n      description: string;\n    };\n    content: string;\n  };\n  subagent: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  subagent_fork: {\n    kind: \"background\";\n    taskId: string;\n  } | {\n    kind: \"foreground\";\n    runId: string;\n    output: JsonValue[];\n  };\n  task_kill: {\n    outcome: \"cancellation-requested\" | \"already-finished\";\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  task_list: ({\n    id: string;\n    kind: string;\n    label: string;\n    status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n    detail?: string;\n    startedAt: number;\n    finishedAt?: number;\n  })[];\n  task_output: {\n    text: string;\n    task: {\n      id: string;\n      kind: string;\n      label: string;\n      status: \"running\" | \"stopping\" | \"completed\" | \"killed\" | \"failed\";\n      detail?: string;\n      startedAt: number;\n      finishedAt?: number;\n    };\n  };\n  todo_write: {\n    todos: ({\n      content: string;\n      status: \"pending\" | \"in_progress\" | \"completed\";\n    })[];\n    counts: {\n      pending: number;\n      inProgress: number;\n      completed: number;\n    };\n  };\n  workflow: {\n    runId: string;\n    agentsStarted: number;\n    result: JsonValue;\n  };\n  write: {\n    path: string;\n    operation: \"create\" | \"update\";\n    before: string | null;\n    after: string;\n  };\n}\n\ntype ToolName = keyof ToolOutputMap\n\ndeclare class ToolCallError extends Error {\n  readonly name: \"ToolCallError\";\n  readonly toolName: ToolName;\n}\n\ndeclare const tools: {\n  [K in ToolName]: (args: ToolArgsMap[K]) => Promise<ToolOutputMap[K]>;\n}\n```","tools":[{"name":"bash","description":"Execute a bash command (`bash -c`) and return its stdout/stderr. Each call runs in a fresh shell: no state (cwd, variables, functions) persists between calls — pass `workdir` instead of using `cd`. Non-zero exits are reported as `[exit code: N]`. Current harness environment facts are exposed through managed `$DSH_*` variables; inspect them when needed. Commands may run under a file sandbox; a blocked file operation is reported as `[sandbox: file access denied under <mode> mode]` — a policy denial, not a bug in the command; do not retry another way. Long output is truncated to its tail; the full output is saved to a file whose path is reported when available. Set `run_in_background: true` for long-running commands: the call returns a task id immediately; read its output with `task_output` and stop it with `task_kill`.","parameters":{"type":"object","properties":{"command":{"type":"string","description":"The bash command to execute."},"description":{"type":"string","description":"Clear, concise description of what this command does in active voice, 5-10 words (shown in the UI). Examples: \"ls\" → \"List files in current directory\"; \"git status\" → \"Show working tree status\"; \"npm install\" → \"Install package dependencies\"."},"timeoutMs":{"type":"number","description":"Timeout in milliseconds. The executor applies its configured default and cap, and kills the command on expiry."},"workdir":{"type":"string","description":"Working directory for this command. Defaults to the session workspace; a relative path is resolved against it."},"run_in_background":{"type":"boolean","description":"Run in the background and return a task id immediately (collect with task_output, stop with task_kill). No timeout applies."}},"required":["command","description"]}},{"name":"cordis_inspect","description":"Inspect the live cordis runtime that is running THIS agent. Read-only. Sections: `services` (every provided ctx service and the plugin fiber that owns it), `plugins` (a flat list of the loaded plugins with their lifecycle states), `tools` (the model-facing tools currently registered, i.e. what you can call), `dynamic` (plugins you mounted via cordis_mount: id, name, state, provided services, awaited services), `api` (method signatures AND argument/return type shapes for every LIVE service — read this before writing plugin code that calls a service), `events` (every harness event with its dispatch mode and exact signature — pick listener targets here). Omit `what` to get all six sections. With `what:\"api\"` or `what:\"events\"`, pass an exact `name` to narrow to one service/event and include its original source JSDoc.","parameters":{"type":"object","properties":{"what":{"type":"string","description":"Limit the report to one section. Omit for all sections.","enum":["services","plugins","tools","dynamic","api","events"]},"name":{"type":"string","description":"Exact service key or event name whose original JSDoc to include; valid only with what:\"api\" or what:\"events\"."}}}},{"name":"cordis_mount","description":"Mount a NEW cordis plugin into the live runtime that is running THIS agent (self-modification). `code` runs as the body of an async JavaScript function in an isolated sandbox and MUST `return` a plugin. Two forms: FUNCTION form `return (ctx) => { … }` — declares no inject, so it can register tools, listen to events, and provide services, but reaching ANY service (e.g. ctx.bash) throws; use it only when you need no services. OBJECT form `return { name?, inject: ['bash', 'llm', …], apply(ctx) { … } }` — declares dependencies, and cordis activates the plugin only after the services exist; PREFER this form. You may reach ONLY the services you list in inject: an undeclared service throws even if it exists, because an undeclared dependency would not be cleaned up if its provider is unmounted. BEFORE calling a service from your code, read cordis_inspect what:\"api\" — it lists method signatures AND the type shapes of their arguments/returns (do not guess a field's type; e.g. a bash run's stdout is an object, not a string). Inside `apply`, use the standard cordis API: `ctx.on(event, listener)` to observe events (see cordis_inspect what:\"events\"), or call `harness.registerTool(ctx, harness.defineTool({ name, description, parameters: { text: { type: 'string', required: true } }, output: { schema: { type: 'string' }, render(_args, value) { return [{ type: 'text', text: value }] } }, async execute(args) { return args.text } }))` to give yourself a new tool — it becomes callable on your NEXT step. Tool parameters: each key IS a property — { type: 'string'|'number'|'integer'|'boolean'|'null'|'object'|'array'|'json', required?: true, description?, enum?, const?, items?, properties? }; every direct DSL object declares additionalProperties: true|false, and oneOf: [schema, schema, ...] replaces type for an exact-one union. A raw JSON-Schema { type: 'object', properties, required?: […] } wrapper is also accepted with open-by-default objects. A tool's `execute` MUST return the lossless JSON value declared by `output.schema`; `output.render(args, value)` separately returns Native/model content blocks. Mounts can COMPOSE: one plugin may `ctx.provide('name', value)` a service and another may declare `inject: ['name']` to consume it — the consumer stays pending until the provider exists and returns to pending when the provider is unmounted. Everything registered inside `apply` is cleaned up automatically on unmount. Sandbox globals: `console` (tagged `[cordis:<id>]`, writes through to the harness terminal), `harness.defineTool`, `harness.registerTool`, `btoa`, `atob`, `TextEncoder`, `TextDecoder`. Node APIs are DISABLED — do filesystem/network/timer work through the cordis services, never Node built-ins: `require`, `setTimeout`/`setInterval`, and `fetch` throw redirect errors; `process` and `Buffer` are undefined. Instead use inject: ['fs'] + ctx.fs for files, inject: ['web'] + ctx.web for HTTP, inject: ['bash'] + ctx.bash for processes, and inject: ['timer'] + ctx.setTimeout/ctx.setInterval for timing (fiber effects, auto-cleaned on unmount) — cordis_inspect what:\"api\" shows what THIS runtime provides. Write PLAIN JavaScript, not TypeScript (no `as`, no type annotations). Cautions: (1) waterfall events (e.g. tools/pre-execute) hand the listener a trailing `next` callback which MUST be called — returning without `next()` VETOES the call; prefer plain notification events unless you intend to intercept. (2) Never await something that only resolves after the current turn (your code runs INSIDE a tool call of that turn — it would deadlock). (3) Your `ctx` is a restricted façade: you can register tools, observe events, provide/consume services, and use timers, but framework internals (ctx.root, ctx.fiber, ctx.extend, ctx.plugin, …) are withheld. It is not a security boundary though — the services you inject (e.g. ctx.bash) reach the real runtime.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"Body of an async JS function; must `return` the plugin to mount."}},"required":["code"]}},{"name":"cordis_unmount","description":"Dispose a plugin previously mounted with cordis_mount, by id. All its registrations (event listeners, tools, services) are cleaned up through the cordis effect lifecycle. Returns only after disposal has fully completed (quiescence, not just a request to stop).","parameters":{"type":"object","properties":{"id":{"type":"string","description":"The dynamic mount id returned by cordis_mount (e.g. \"dyn-1\")."}},"required":["id"]}},{"name":"edit","description":"Edit an existing UTF-8 text file by replacing literal text.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to edit, resolved by the filesystem backend."},"old_string":{"type":"string","description":"Literal text to replace. Must match exactly."},"new_string":{"type":"string","description":"Literal replacement text. Use an empty string to delete the match."},"replace_all":{"type":"boolean","description":"Replace all matches. Defaults to false; when false, old_string must appear exactly once."}},"required":["file_path","old_string","new_string"]}},{"name":"ralph","description":"Run a foreground fresh-agent Ralph loop toward one immutable objective. Use only when the direct human explicitly asks for Ralph or fresh-agent iteration. Each round opens a new child with no parent conversation or prior child session; the shared workspace is long-term memory, and only a bounded structured report crosses rounds. The call returns when a worker reports completion or a concrete blocker, or at the round limit. Ordinary long-running same-session work belongs to goal tools.","parameters":{"type":"object","properties":{"objective":{"type":"string","description":"The immutable completion objective for every fresh Ralph round."},"maxRounds":{"type":"number","description":"Optional positive safe-integer round cap, bounded by the deployment ceiling."}},"required":["objective"]}},{"name":"read","description":"Read a UTF-8 text file and return line-numbered content.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to read, resolved by the filesystem backend."},"offset":{"type":"number","description":"1-based first line to return. Defaults to 1."},"limit":{"type":"number","description":"Maximum number of lines to return. Defaults to 2000."}},"required":["file_path"]}},{"name":"run_code","description":"Execute a TypeScript program against the available tools. Write the BODY of an async function (erasable syntax only; top-level `await` and `return` work) and call tools as `await tools.name(args)` per the declarations in the system prompt. Only what you print or return comes back — curate it.","parameters":{"type":"object","properties":{"code":{"type":"string","description":"The program: the body of an async TypeScript function."}},"required":["code"]}},{"name":"skill","description":"Load the full instructions for an available skill. Call this with the exact skill name from the session skill catalog before acting on a task that names or clearly matches that skill.","parameters":{"type":"object","properties":{"name":{"type":"string","description":"The exact skill name from the available skills list."}},"required":["name"]}},{"name":"subagent","description":"Delegate a self-contained task to a subagent (a separate agent that works in its own context) and return its final result. Use this to offload focused, independent work — research, a scoped implementation, an analysis — so it does not consume this conversation's context. The subagent runs to completion and you receive only its final answer, not its intermediate steps. Give it a complete, standalone prompt: it does not see this conversation. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The complete, self-contained task for the subagent. It does not share this conversation's context, so include everything it needs."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"subagent_fork","description":"Delegate a task to a subagent that inherits this conversation: a child agent seeded with all completed turns so far (it does not see the current in-flight turn), returning only its final result. Use this when the subtask builds on this conversation's context — a follow-up analysis, a review, a continuation — without consuming this conversation's context for the work itself. You receive only its final answer, not its intermediate steps. Set `run_in_background: true` to return a task id; collect with `task_output` and stop with `task_kill`.","parameters":{"type":"object","properties":{"description":{"type":"string","description":"A short (3-5 word) description of the delegated task, for display."},"prompt":{"type":"string","description":"The task for the subagent. It already sees this conversation's completed turns, so build on them freely and state only what is new."},"run_in_background":{"type":"boolean","description":"Run as a background task and return its id; collect with task_output or stop with task_kill."}},"required":["description","prompt"]}},{"name":"task_kill","description":"Request cancellation of a running background task by task id. Returns immediately; the task settles as killed once its work actually stops.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"reason":{"type":"string","description":"Optional short reason, recorded in the log and forwarded to the task."}},"required":["task_id"]}},{"name":"task_list","description":"List your background tasks (running and finished) with their ids, kinds, and statuses.","parameters":{"type":"object","properties":{}}},{"name":"task_output","description":"Read a background task. Stream tasks return only output since the previous read; final-output tasks return their result after settlement. Every response ends with `[status: ...]`. Reads are non-blocking unless `wait: true`, which waits up to the configured cap.","parameters":{"type":"object","properties":{"task_id":{"type":"string","description":"Task id returned by the tool that started the background work."},"wait":{"type":"boolean","description":"Block until the task reaches a terminal status or the timeout expires. A timed-out wait returns [status: running] and leaves the task alive."},"timeout_ms":{"type":"number","description":"Max wait in milliseconds (only meaningful with wait: true). Defaults to the configured wait timeout; capped by the configured maximum."}},"required":["task_id"]}},{"name":"todo_write","description":"Record and update a structured task list for the current work. Send the ENTIRE list every call — it REPLACES the previous list (there are no partial updates, no per-item edits). Use it to plan multi-step work and show progress: add one todo per concrete step before you start. Keep AT MOST ONE todo `in_progress` at a time; while work remains, exactly one active task should be `in_progress`. Mark a todo `completed` the moment it is done (do not batch completions), and allow no `in_progress` item only once all work is complete. Skip the list for trivial single-step tasks. Statuses: `pending` (not started), `in_progress` (being worked on now), `completed` (finished).","parameters":{"type":"object","properties":{"todos":{"type":"array","description":"The COMPLETE task list, replacing any previous list.","items":{"type":"object","additionalProperties":false,"properties":{"content":{"type":"string","description":"What the task is — a short imperative line."},"status":{"type":"string","description":"pending (not started) | in_progress (now) | completed (done).","enum":["pending","in_progress","completed"]}},"required":["content","status"]}}},"required":["todos"]}},{"name":"workflow","description":"Run a JavaScript workflow script that orchestrates subagents at scale. Use this for work that fans out across many independent pieces — an audit over many files, a migration, multi-angle research, adversarial verification of findings — where you write the orchestration as a script instead of delegating turn by turn.\n\nThe workflow's identity rides the `meta` parameter as JSON: required `name` (short kebab-case) and `description` strings, optional `whenToUse` string and `phases` array (`{title, detail?, provider?, model?}`). The `script` parameter is the plain JavaScript body ONLY (NOT TypeScript, and NO `export const meta` statement — meta is a parameter, not code), running with top-level await; end with `return <value>` — the value must be JSON-serializable and is this tool's result.\n\nScript-body hooks:\n- `agent(prompt, opts?): Promise<any>` — run one subagent to completion. Without `opts.schema` it resolves to the child's final text; with `opts.schema` (an object-rooted JSON Schema using ONLY type/properties/required/additionalProperties/items/enum/const/oneOf — no pattern/format/numeric bounds) it resolves to the validated object. Resolves `null` when the child fails (filter with `.filter(Boolean)`). Other opts: `label` (display), `phase` (progress group), and independent `provider`/`model` LLM target overrides (either may be provided alone). Anything else (`effort`/`isolation`/`agentType`) is rejected loudly.\n- `pipeline(items, ...stages): Promise<any[]>` — run each item through the stages independently with NO barrier between stages (prefer this for multi-stage work). Each stage receives `(prev, item, index)`. An ordinary stage throw drops that ITEM to `null` and skips its remaining stages.\n- `parallel(thunks): Promise<any[]>` — run zero-argument functions concurrently and await ALL of them (a barrier; use only when a stage genuinely needs every prior result together). A throwing thunk resolves to `null`.\n- `phase(title)` — start a progress phase; `log(message)` — narrate progress; `args` — the tool call's `args` input, verbatim.\n\nMisused hooks (bad arguments, unknown options, unsupported schemas, tripped caps) throw errors that ALWAYS kill the script — they never dissolve into a per-item `null`.\n\nConstraints: concurrency and total-agent caps apply; no filesystem, network, timers, or Node.js APIs are provided — the agents do the work, the script only coordinates them. The run executes in the foreground: this call returns when the whole script finishes.","parameters":{"type":"object","properties":{"script":{"type":"string","description":"The plain-JS workflow script body (top-level await allowed; NO `export const meta` statement; end with `return <json-value>`)."},"meta":{"type":"object","description":"The workflow identity block (plain JSON — never code).","additionalProperties":true,"properties":{"name":{"type":"string","description":"Short kebab-case workflow name."},"description":{"type":"string","description":"One-line description of what the workflow does."},"whenToUse":{"type":"string","description":"Optional guidance on when this workflow applies."},"phases":{"type":"array","description":"Optional phase declarations matched by phase() calls.","items":{"type":"object","additionalProperties":true,"properties":{"title":{"type":"string","description":"The phase title phase() calls match by exact string."},"detail":{"type":"string","description":"Optional one-line description of the phase."},"provider":{"type":"string","description":"Optional provider override this phase is expected to use."},"model":{"type":"string","description":"Optional model override this phase is expected to use."}},"required":["title"]}}},"required":["name","description"]},"args":{"type":"object","description":"Optional JSON input exposed to the script as the `args` global (wrap a bare list as a field, e.g. {\"files\": [...]}).","additionalProperties":true}},"required":["script","meta"]}},{"name":"write","description":"Create or fully replace a UTF-8 text file.","parameters":{"type":"object","properties":{"file_path":{"type":"string","description":"Path to write, resolved by the filesystem backend."},"content":{"type":"string","description":"Full UTF-8 text content to write."}},"required":["file_path","content"]}}]},"reason":"initial"}}
 {"type":"assistant/chunk","seq":5,"time":1783950002005,"data":{"turn":1,"step":1,"chunk":{"type":"block-start","index":0,"blockType":"text"}}}
 {"type":"assistant/chunk","seq":6,"time":1783957884701,"data":{"turn":1,"step":1,"chunk":{"type":"text-delta","index":0,"text":"WORKFLOW_CHILD_OK"}}}
 {"type":"assistant/chunk","seq":7,"time":1783957884701,"data":{"turn":1,"step":1,"chunk":{"type":"block-end","index":0,"block":{"type":"text","text":"WORKFLOW_CHILD_OK"}}}}

From 1f094ae0762a36e103448aa82a35812ced721256 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sat, 25 Jul 2026 00:42:40 +0800
Subject: [PATCH 17/36] fix(gui): todo row keeps running/stopped execution
 states visible
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The row rendered a status only for error, so a call cancelled before
tool/result read as a completed plan update even though no todo/write
occurred. Non-ok states now ride the generic row's StateDot semantics
(ongoing dot while running, warning dot + 已中断 marker when interrupted);
the ok badge stays for settled successful updates.
---
 .../src/client/toolviews/todo-row.tsx             | 12 +++++++++---
 .../ui-conversation/tests/todo-panel.spec.tsx     | 15 ++++++++++++++-
 2 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
index 390361d20b..665fc4dfab 100644
--- a/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
+++ b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
@@ -6,6 +6,7 @@
 // row stays one line.
 
 import type { Context } from 'cordis'
+import { StateDot } from '@deepseek-ai/dsh-client-ui-primitives'
 import type { ToolRowProps } from '../contract/slots.ts'
 import { toolRowModel } from '../contract/tool-call-model.ts'
 import css from './todo-row.module.css'
@@ -38,17 +39,22 @@ function summarize(argsRaw: string): string | null {
     : head
 }
 
-/** One-line plan update row (click opens the raw args in details). */
+/** One-line plan update row (click opens the raw args in details). Non-ok
+ *  execution states keep the generic row's dot semantics — a cancelled call
+ *  wrote no todo/write, so it must not read as a completed update. */
 export function TodoRow({ toolName, block, openDetails }: ToolRowProps) {
   const model = toolRowModel(toolName, block)
   const argsRaw = ('kind' in block ? block.call?.argsRaw : block.argsRaw) ?? ''
   const summary = summarize(argsRaw) ?? model.summary
   return (
-    <div className={css.row} data-sample="todo-row" onClick={openDetails}>
-      <span className={css.badge} aria-hidden>☰</span>
+    <div className={css.row} data-sample="todo-row" data-state={model.state} onClick={openDetails}>
+      {model.state === 'ok'
+        ? <span className={css.badge} aria-hidden>☰</span>
+        : <StateDot state={model.state === 'running' ? 'ongoing' : model.state === 'stopped' ? 'warning' : 'error'} />}
       <span className={css.title}>更新任务清单</span>
       <span className={css.summary}>{summary}</span>
       {model.state === 'error' && <span className={css.err}>failed</span>}
+      {model.state === 'stopped' && <span className={css.err}>已中断</span>}
     </div>
   )
 }
diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
index 196077e6a4..8971bba2cd 100644
--- a/packages/client/ui-conversation/tests/todo-panel.spec.tsx
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -94,10 +94,23 @@ describe('TodoRow', () => {
 
   it('omits the active clause when no item is in progress and reads running-call args', () => {
     const args = JSON.stringify({ todos: [{ content: 'x', status: 'completed' }] })
-    render(<TodoRow {...rowProps({ callId: 'c1', name: 'todo_write', argsRaw: args, turn: 1, step: 1, callView: null })} />)
+    render(<TodoRow {...rowProps({ callId: 'c1', name: 'todo_write', argsRaw: args, turn: 1, step: 1, time: 1_000, callView: null })} />)
     expect(screen.getByText('1/1 已完成')).toBeTruthy()
   })
 
+  it('keeps the non-ok execution states visible: running dot, interrupted marker', () => {
+    // A running call (no result yet) shows the ongoing dot, never the ok badge.
+    const args = JSON.stringify({ todos: LIST })
+    const running = render(<TodoRow {...rowProps({ callId: 'c1', name: 'todo_write', argsRaw: args, turn: 1, step: 1, time: 1_000, callView: null })} />)
+    expect(running.container.querySelector('[data-state="running"]')).not.toBeNull()
+    expect(running.container.querySelector('[data-state="running"] svg')).not.toBeNull()
+    running.unmount()
+    // A cancelled call wrote no todo/write: the row must not read as a completed update.
+    const stopped = render(<TodoRow {...rowProps(resultNode(args, { isError: true, error: { name: 'Interrupted', code: 'interrupted' } }))} />)
+    expect(stopped.container.querySelector('[data-state="stopped"]')).not.toBeNull()
+    expect(stopped.getByText('已中断')).toBeTruthy()
+  })
+
   it('falls back to the generic summary on malformed args and flags errors', () => {
     render(<TodoRow {...rowProps(resultNode('not json', { isError: true }))} />)
     expect(screen.getByText('failed')).toBeTruthy()

From 8372340f9c58cd12d73559ec8426846c39e14471 Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sat, 25 Jul 2026 07:47:51 +0800
Subject: [PATCH 18/36] feat(llm): add model-specific reasoning effort controls

---
 ...ed-reasoning-effort-capabilities.i18n.yaml |   6 +
 ...ter-owned-reasoning-effort-capabilities.md |  33 +++++
 ...-owned-reasoning-effort-capabilities.zh.md |  33 +++++
 docs/architecture.i18n.yaml                   |   4 +-
 docs/architecture.md                          |   2 +-
 docs/architecture.zh.md                       |   2 +-
 docs/config-catalog.md                        |   9 +-
 docs/cookbook/adding-an-llm-adapter.i18n.yaml |   4 +-
 docs/cookbook/adding-an-llm-adapter.md        |   2 +-
 docs/cookbook/adding-an-llm-adapter.zh.md     |   2 +-
 docs/cordis-catalog/events.md                 |   2 +-
 docs/cordis-catalog/services.md               |  25 +++-
 docs/core-data-structures/core.md             |  46 ++++++-
 docs/core-data-structures/llm-streaming.md    |  13 +-
 docs/core-data-structures/session.md          |   2 +-
 docs/event-producer-consumer.md               |   2 +-
 .../develop/practice/llm-adapter.i18n.yaml    |   4 +-
 docs/user/develop/practice/llm-adapter.md     |   4 +-
 docs/user/develop/practice/llm-adapter.zh.md  |   4 +-
 .../tests/fixtures/cli-mock-llm.ts            |  26 +++-
 .../headless-agent/tests/headless.snapshot.ts |  41 ++++++
 .../cordis/tool-cordis/src/api-catalog.ts     |  24 +++-
 packages/core/agent-loop/README.md            |   2 +
 packages/core/agent-loop/src/loop.ts          |  39 +++++-
 .../core/agent-loop/tests/mock-adapter.ts     |  14 +-
 .../tests/request-reconstruction.spec.ts      |  78 ++++++++++-
 packages/core/session/src/index.ts            |   5 +
 packages/core/session/src/types.ts            |   2 +-
 .../core/session/tests/request-header.spec.ts |   5 +
 packages/core/session/tests/session.spec.ts   |  36 ++++-
 packages/llm/llm-deepseek/README.md           |   8 +-
 packages/llm/llm-deepseek/src/adapter.ts      |  25 +++-
 packages/llm/llm-deepseek/src/index.ts        |  10 +-
 packages/llm/llm-deepseek/src/serialize.ts    |  18 ++-
 .../llm/llm-deepseek/tests/adapter.e2e.ts     |   6 +-
 .../llm/llm-deepseek/tests/adapter.spec.ts    |  76 ++++++++++-
 .../llm/llm-deepseek/tests/serialize.spec.ts  |  22 ++-
 packages/llm/llm-pi-ai/README.md              |   3 +
 packages/llm/llm-pi-ai/src/adapter.ts         |  81 ++++++++++-
 packages/llm/llm-pi-ai/tests/adapter.e2e.ts   |  13 +-
 packages/llm/llm-pi-ai/tests/adapter.spec.ts  |  70 +++++++++-
 packages/llm/llm/README.md                    |  10 +-
 packages/llm/llm/src/brand.ts                 |  12 ++
 packages/llm/llm/src/call-config.ts           |  23 +++-
 packages/llm/llm/src/index.ts                 | 112 ++++++++++++++-
 packages/llm/llm/src/types.ts                 |  25 +++-
 packages/llm/llm/tests/call-config.spec.ts    |   6 +
 packages/llm/llm/tests/service.spec.ts        | 127 +++++++++++++++++-
 scripts/gen-cordis-catalog.ts                 |   1 +
 scripts/type-equiv.manifest.json              |  15 +++
 50 files changed, 1046 insertions(+), 88 deletions(-)
 create mode 100644 .agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
 create mode 100644 .agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
 create mode 100644 .agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md

diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
new file mode 100644
index 0000000000..489f9845f6
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
@@ -0,0 +1,6 @@
+# Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
+# side as of the last confirmed-consistent state. Both languages carry equal authority;
+# after editing either side, bring the other along and re-record with:
+#   pnpm run verify-translation-pairing --write
+2026-07-24-adapter-owned-reasoning-effort-capabilities.md: 806e808ab18d617003f757200f0cdee5853476f5
+2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 9be8b6f4f7bcbe063ff90492ac44b9590860f67f
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
new file mode 100644
index 0000000000..806e808ab1
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
@@ -0,0 +1,33 @@
+# Agent Note: Adapter-owned reasoning effort capabilities
+
+Status: implemented
+
+English | [中文](2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md)
+
+## Problem
+
+Reasoning strength was adapter configuration only, so a conversation could not discover or change the selected model's supported levels between requests. Promoting one adapter's level union into `dsh-llm` would make every provider and model adopt names it may not support, while a provider-specific options bag would make the loop unable to validate or durably reconstruct the effective request.
+
+## Decision
+
+`dsh-llm` represents a reasoning effort as the opaque branded `ReasoningEffortId`. An adapter's `resolveModelReasoning(provider, model)` returns a non-empty ordered list of ids with display metadata and may name one configured default. The core validates metadata, requires an explicit or configured effort to appear exactly in that list, and never clamps or aliases a value.
+
+`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop resolves the post-`agent/request` config before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id. The terminal `LlmService` adapter boundary repeats resolution for direct calls that do not pass through the loop.
+
+The native DeepSeek adapter advertises `high` and `max`, defaults to configured effort or `high`, and exposes no effort capability while thinking is disabled. The pi-ai adapter derives each exact model's list from `getSupportedThinkingLevels()`, excludes `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai.
+
+## Alternatives considered
+
+**Define the pi-ai `ThinkingLevel` union in core.** Rejected because current pi-ai canonical names are an adapter implementation detail; a future provider can expose a different identifier without requiring a core release.
+
+**Carry an untyped provider options object.** Rejected because the loop could neither validate a selected value nor put a stable provider-neutral fact in the request header.
+
+**Clamp unsupported levels.** Rejected because a silent substitution makes the user's selected control differ from the logged request intent and hides stale deployment configuration.
+
+**Include `off` as an effort.** Rejected because disabling reasoning is a mode capability with different request and output semantics, not a reasoning-strength level.
+
+## Consequences
+
+Clients can query one exact route and render the adapter's order and names without knowing a global enum. Adapter configuration remains the deployment-default owner, while `agent/request` can replace the effective effort on each step. Invalid metadata fails with `INVALID_MODEL_REASONING`, and unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+
+The capability query is asynchronous and exact-model resolution may fail for adapters backed by authoritative catalogs. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, and resume behavior; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
new file mode 100644
index 0000000000..9be8b6f4f7
--- /dev/null
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
@@ -0,0 +1,33 @@
+# Agent Note：适配器持有的推理强度能力
+
+Status: implemented
+
+[English](2026-07-24-adapter-owned-reasoning-effort-capabilities.md) | 中文
+
+## 问题
+
+推理强度过去只能在适配器中配置，因此对话无法在多次请求之间发现或更改所选模型支持的等级。若将某个适配器的等级联合类型提升到 `dsh-llm`，所有提供方和模型都必须采用一套自身可能并不支持的名称；若改用提供方特有的 options 对象，主循环又无法校验最终生效的请求，也无法通过持久化记录准确重建该请求。
+
+## 决策
+
+`dsh-llm` 使用不透明的品牌类型 `ReasoningEffortId` 表示推理强度。适配器的 `resolveModelReasoning(provider, model)` 返回非空的有序 ID 列表及其展示元数据，并可指定一个由配置确定的默认值。核心会校验元数据，要求显式指定或配置指定的推理强度与列表中的某个 ID 完全一致，且绝不自动调整或为值提供别名。
+
+`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在 `agent/request` 处理完成后、写入 `request/header` 前解析配置，因此默认值和动态变更只有成为持久化事实后才对模型可见。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。最终的 `LlmService` 适配器边界会再次执行解析，以覆盖未经过主循环的直接调用。
+
+原生 DeepSeek 适配器声明 `high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`；禁用思考时不暴露推理强度能力。pi-ai 适配器通过 `getSupportedThinkingLevels()` 按具体模型推导等级列表，排除 `off`，在 profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。
+
+## 备选方案
+
+**在核心中定义 pi-ai 的 `ThinkingLevel` 联合类型。** 不予采纳：pi-ai 当前的规范名称属于适配器实现细节；未来的提供方可以暴露不同的标识符，而无需为此发布新的核心版本。
+
+**携带无类型约束的提供方 options 对象。** 不予采纳：主循环既无法校验选定值，也无法在请求头中写入稳定且与提供方无关的事实。
+
+**自动调整不支持的等级。** 不予采纳：静默替换会导致用户选定的控制项与日志记录的请求意图不一致，还会掩盖陈旧的部署配置。
+
+**将 `off` 列为推理强度。** 不予采纳：禁用推理属于具有不同请求和输出语义的模式能力，而不是推理强度等级。
+
+## 影响
+
+客户端可以查询一条确切路由，并按适配器给出的顺序和名称渲染等级，而无需了解全局枚举。适配器配置仍负责提供部署默认值，`agent/request` 则可以在每个步骤替换实际生效的推理强度。元数据无效时抛出 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
+
+能力查询采用异步方式；对于由权威目录支持的适配器，确切模型解析可能失败。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录和恢复行为提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
diff --git a/docs/architecture.i18n.yaml b/docs/architecture.i18n.yaml
index 20553e1b90..e7c0dd02b3 100644
--- a/docs/architecture.i18n.yaml
+++ b/docs/architecture.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-architecture.md: d1051eecf51d8d1c7f8c235b0c5dd80478b43316
-architecture.zh.md: 502c0248a9d2c62af165ce07eb19489b76c5f6ee
+architecture.md: 0bc015afcf583224b7162540ae53f2bb19952cd6
+architecture.zh.md: 20ed3229e0a8d5997cc982ae5c82fb90f900c948
diff --git a/docs/architecture.md b/docs/architecture.md
index d1051eecf5..0bc015afcf 100644
--- a/docs/architecture.md
+++ b/docs/architecture.md
@@ -91,7 +91,7 @@ forever:
       agent/pre-step
       snapshot the derived messages (the reconstruction boundary)
       'step/start'
-      agent/request (config only) -> log request/header -> checkpoint -> llm/stream (frozen)
+      agent/request -> resolve reasoning/default -> log request/header -> checkpoint -> llm/stream (frozen)
       on final adapter-path or terminal in-band failure:
         'step/end'
         agent/request-error(original error, failure facts, immutable prior failures, signal)
diff --git a/docs/architecture.zh.md b/docs/architecture.zh.md
index 502c0248a9..20ed3229e0 100644
--- a/docs/architecture.zh.md
+++ b/docs/architecture.zh.md
@@ -91,7 +91,7 @@ forever:
       agent/pre-step
       snapshot the derived messages (the reconstruction boundary)
       'step/start'
-      agent/request (config only) -> log request/header -> checkpoint -> llm/stream (frozen)
+      agent/request -> resolve reasoning/default -> log request/header -> checkpoint -> llm/stream (frozen)
       on final adapter-path or terminal in-band failure:
         'step/end'
         agent/request-error(original error, failure facts, immutable prior failures, signal)
diff --git a/docs/config-catalog.md b/docs/config-catalog.md
index 985e493184..b1f3f969c4 100644
--- a/docs/config-catalog.md
+++ b/docs/config-catalog.md
@@ -520,8 +520,9 @@ Requires: `llm`
 /**
  * Plugin config, validated by the same-named schemastery schema. Every field
  * is optional in yml: credentials/endpoint fall back to the environment (a
- * missing API key fails plugin load, not the first call), and omitted
- * thinking fields send nothing on the wire, so the provider default applies.
+ * missing API key fails plugin load, not the first call), omitted thinking
+ * mode uses the provider default, and omitted reasoning effort resolves to
+ * `high`.
  */
 export interface Config {
   /** API key; falls back to $DEEPSEEK_API_KEY. Required one way or the other. */
@@ -530,7 +531,7 @@ export interface Config {
   baseURL?: string
   /** Thinking-mode default for every request (provider default: enabled). */
   thinking?: 'enabled' | 'disabled'
-  /** Thinking effort (only meaningful with thinking enabled). */
+  /** Default thinking effort when thinking is enabled (default `high`). */
   reasoningEffort?: 'high' | 'max'
   /** Positive context capacity used when the selected model has no exact value. */
   defaultContextWindow?: number
@@ -553,7 +554,7 @@ export interface DeepSeekCatalogModel {
 }
 ```
 
-Source: [`packages/llm/llm-deepseek/src/index.ts:34`](../packages/llm/llm-deepseek/src/index.ts)
+Source: [`packages/llm/llm-deepseek/src/index.ts:35`](../packages/llm/llm-deepseek/src/index.ts)
 
 ## `@deepseek-ai/dsh-llm-pi-ai`
 
diff --git a/docs/cookbook/adding-an-llm-adapter.i18n.yaml b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
index 497ae08c32..15c62ae966 100644
--- a/docs/cookbook/adding-an-llm-adapter.i18n.yaml
+++ b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-adding-an-llm-adapter.md: f20442b8c2ce823452a3ea13409f202185d12d04
-adding-an-llm-adapter.zh.md: 2864dd1e18742c7449e24f22504a5a38976ab450
+adding-an-llm-adapter.md: 3adae89f360cd81e264912d74bd86f5a06cd42f8
+adding-an-llm-adapter.zh.md: 80329da5e7c89bfb5ee2e2aa5b9931a3ef0e83f6
diff --git a/docs/cookbook/adding-an-llm-adapter.md b/docs/cookbook/adding-an-llm-adapter.md
index f20442b8c2..3adae89f36 100644
--- a/docs/cookbook/adding-an-llm-adapter.md
+++ b/docs/cookbook/adding-an-llm-adapter.md
@@ -32,7 +32,7 @@ Registration is effect-based (HMR-safe); one adapter per provider route — dupl
 - A `GenerateOptions` field your provider cannot honor (e.g. a `stop` list on a provider without stop sequences): throw `LlmError(..., 'UNSUPPORTED')` rather than silently dropping it.
 - If the provider requires response ids, signatures, or other native metadata on follow-up calls, emit the minimal lossless-JSON projection as `finish.replayState`. Validate it when rebuilding history. `LlmService` passes it only when the historical provider route and target provider route are currently owned by the exact same adapter instance; your adapter decides whether same-model, cross-model, or cross-provider restoration is legal. Never infer native replay from provider/model names alone when state is absent.
 
-Provider-specific request knobs (thinking modes, effort levels) belong in the ADAPTER's Config, not in `GenerateOptions` — the core vocabulary stays provider-neutral.
+Provider-specific thinking-mode toggles remain in the adapter's Config. Selectable reasoning strength uses the provider-neutral capability seam: return ordered opaque ids from `resolveModelReasoning()`, declare a configured `defaultEffort` only when one exists, and map `GenerateOptions.reasoningEffort` to the provider wire value. Do not expose provider wire spellings, clamp unsupported values, or include an `off` mode as an effort.
 
 ## Structure that worked
 
diff --git a/docs/cookbook/adding-an-llm-adapter.zh.md b/docs/cookbook/adding-an-llm-adapter.zh.md
index 2864dd1e18..80329da5e7 100644
--- a/docs/cookbook/adding-an-llm-adapter.zh.md
+++ b/docs/cookbook/adding-an-llm-adapter.zh.md
@@ -32,7 +32,7 @@ export function apply(ctx: Context, config: Config) {
 - 如果 `GenerateOptions` 中某个字段你的提供方无法支持（例如提供方不支持 stop sequences 时收到 `stop` 列表）：抛出 `LlmError(..., 'UNSUPPORTED')`，而非静默丢弃。
 - 如果提供方在后续调用中需要响应 ID、签名或其他原生元数据，请将其最小无损 JSON 投影作为 `finish.replayState` 发出。重建历史时验证该状态。只有历史提供方路由和目标提供方路由当前由完全相同的适配器实例拥有时，`LlmService` 才会传递该状态；由适配器决定同模型、跨模型或跨提供方恢复是否合法。状态缺失时，切勿仅根据提供方/模型名称推断原生回放。
 
-提供方特有的请求旋钮（thinking 模式、effort 级别）放在**适配器**的 Config 中，而非 `GenerateOptions` 中——核心词汇保持提供方无关。
+提供方特有的 thinking 模式开关仍放在适配器的 Config 中。可选的推理强度使用提供方无关的能力 seam：`resolveModelReasoning()` 返回有序的不透明 ID；仅当存在配置指定的默认值时才声明 `defaultEffort`；并将 `GenerateOptions.reasoningEffort` 映射为提供方协议值。不得暴露提供方协议值的具体拼写、自动调整不支持的值，也不得把 `off` 模式列为推理强度。
 
 ## 经验证有效的结构
 
diff --git a/docs/cordis-catalog/events.md b/docs/cordis-catalog/events.md
index 6becfd9434..30cc12af6b 100644
--- a/docs/cordis-catalog/events.md
+++ b/docs/cordis-catalog/events.md
@@ -547,7 +547,7 @@ Waterfall around every streaming model call (retry, replay, routing). Bound to t
 
 Types: [GenerateOptions](../core-data-structures/core.md) · [LlmService](../core-data-structures/llm-streaming.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
 
-Source: [`packages/llm/llm/src/index.ts:52`](../../packages/llm/llm/src/index.ts)
+Source: [`packages/llm/llm/src/index.ts:54`](../../packages/llm/llm/src/index.ts)
 
 ## `session/*`
 
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index c5d6d88d5c..718a45189b 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -671,6 +671,25 @@ async listModels(provider: string): Promise<LlmModelInfo[]>
  */
 async resolveModelContext( provider: string, model: string, ): Promise<LlmModelContext | undefined>
 
+/**
+ * Resolve selectable reasoning efforts from the adapter that owns one exact
+ * route. Metadata is validated and detached; an absent result means an
+ * effort selector is unsupported for that model.
+ * @param provider - registered provider route to inspect.
+ * @param model - exact model id passed to the adapter.
+ * @returns detached reasoning metadata, or `undefined` when unsupported.
+ */
+async resolveModelReasoning( provider: string, model: string, ): Promise<LlmModelReasoningInfo | undefined>
+
+/**
+ * Validate a conversation call config against its exact model capability and
+ * materialize an adapter-configured default. Unsupported explicit efforts
+ * reject before provider I/O; no clamping or aliasing is performed.
+ * @param config - provider/model route and optional request controls.
+ * @returns a detached config only when a default must be materialized.
+ */
+async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>
+
 /**
  * Stream one model call as raw chunks (token-level deltas). Throws
  * `LlmError` with code `NO_ADAPTER` if no adapter is registered for
@@ -686,9 +705,9 @@ async resolveModelContext( provider: string, model: string, ): Promise<LlmModelC
 stream(options: GenerateOptions): AsyncIterable<StreamChunk>
 ```
 
-Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmModelContext](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
+Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmCallConfig](../core-data-structures/core.md) · [LlmModelContext](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmModelReasoningInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
 
-Source: [`packages/llm/llm/src/index.ts:159`](../../packages/llm/llm/src/index.ts)
+Source: [`packages/llm/llm/src/index.ts:175`](../../packages/llm/llm/src/index.ts)
 
 ## `ctx.permission` — `PermissionService`
 
@@ -1238,7 +1257,7 @@ fork(source: SessionForkSource, boundary?: number, childSessionId?: SessionId):
 
 Types: [CreateSessionOptions](../core-data-structures/persistence.md) · [OutOfBandSessionEventType](../core-data-structures/session.md) · [Session](../core-data-structures/session.md) · [SessionEvent](../core-data-structures/core.md) · [SessionEventMap](../core-data-structures/session.md) · [SessionId](../core-data-structures/core.md) · [TurnTrigger](../core-data-structures/session.md)
 
-Source: [`packages/core/session/src/index.ts:605`](../../packages/core/session/src/index.ts)
+Source: [`packages/core/session/src/index.ts:610`](../../packages/core/session/src/index.ts)
 
 ## `ctx.sessionTitle` — `SessionTitleService`
 
diff --git a/docs/core-data-structures/core.md b/docs/core-data-structures/core.md
index 86809ddc8b..5380573c55 100644
--- a/docs/core-data-structures/core.md
+++ b/docs/core-data-structures/core.md
@@ -205,12 +205,46 @@ interface LlmModelContext {
 }
 ```
 
+Reasoning effort is another exact-route capability. The core brands identifiers but does not enumerate their values; each adapter owns the ordered set, display names, and optional deployment default.
+
+```ts type-equiv
+/** Adapter-owned identifier for one model's selectable reasoning effort. */
+type ReasoningEffortId = Branded<'ReasoningEffortId'>
+```
+
+```ts type-equiv
+/** Display metadata for one adapter-owned reasoning effort. */
+interface LlmReasoningEffortInfo {
+  /** Opaque stable value accepted by {@link GenerateOptions.reasoningEffort}. */
+  id: ReasoningEffortId
+  /** Human-readable effort name for selectors and diagnostics. */
+  name: string
+  /** Optional user-facing distinction from otherwise similar efforts. */
+  description?: string
+}
+```
+
+```ts type-equiv
+/** Selectable reasoning efforts for one exact provider/model route. */
+interface LlmModelReasoningInfo {
+  /** Supported efforts in adapter-preferred display order. */
+  efforts: readonly LlmReasoningEffortInfo[]
+  /**
+   * Adapter-configured default materialized into requests when callers omit
+   * an effort. Absence preserves the provider's own default.
+   */
+  defaultEffort?: ReasoningEffortId
+}
+```
+
 ```ts type-equiv
 /** A single model request, fully assembled. */
 interface GenerateOptions {
   /** Registered provider route selecting the adapter instance. */
   provider: string
   model: string
+  /** Adapter-owned reasoning effort selected for this exact model. */
+  reasoningEffort?: ReasoningEffortId
   /**
    * Ordered conversation messages, exactly as the provider sees them (after
    * the `system` slot). A loop-built request assembles them as
@@ -287,21 +321,23 @@ The model-facing `ToolSchema` is the wire shape; the registered `ToolDefinition`
 
 The loop builds each request from logged state. `EpochHeader` records call config, rendered prompt, authoritative returned tool order (configured by `toolOrder`, or lexicographic when unset), and session prefix through full `request/header` snapshots. Together with derived history, this makes the request reconstructable from the session log. See [session.md](session.md#the-request-header-event-requestheader) and the [reconstructability Agent Note](../../.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.md).
 
-`agent/request` receives a frozen call-config seed and may return a replacement to switch provider, model, or sampling. `agent/session-prefix` composes request-only prefix messages once per loop instance, and the header records the exact result used. Requests reaching `llm/stream` are deep-frozen, so mutation throws, and carry a process-local loop identity so observers do not confuse separately logged frozen auxiliary calls with conversation requests.
+`agent/request` receives a frozen call-config seed and may return a replacement to switch provider, model, reasoning effort, or sampling. The loop resolves the exact model capability after the waterfall, rejects unsupported explicit effort ids without clamping, materializes an adapter-configured default, and logs the effective value. `agent/session-prefix` composes request-only prefix messages once per loop instance, and the header records the exact result used. Requests reaching `llm/stream` are deep-frozen, so mutation throws, and carry a process-local loop identity so observers do not confuse separately logged frozen auxiliary calls with conversation requests.
 
 On the wire, a loop-built request reads in this order: the `system` slot (the rendered prompt assembly) → `messagePrefix` (the frozen session prefix) → the derived history — the boundary snapshot, whose tail is the newest `user/message` on a turn's first step and the previous step's tool results on later steps. The prefix never enters the derived history; its durable record is the header events, and the dev invariant recomputes exactly this equation against every loop-built request.
 
-FIXME(call-config-shape): revisit the exact definition of this type — which fields are genuinely epoch-level for cache purposes (`model` certainly; the sampling scalars sit here out of caution), and where provider-specific extras (reasoning options, extra body params) belong when an adapter needs them.
+FIXME(call-config-shape): revisit which remaining fields are genuinely epoch-level for cache purposes (`model` and the model-owned reasoning effort are explicit; the sampling scalars sit here out of caution).
 
 ```ts type-equiv
 /**
- * Provider + model + sampling scalars of one conversation's requests. Every field maps
- * 1:1 onto the same-named `GenerateOptions` field; the loop builds requests
- * from the logged header rather than accepting these per call.
+ * Provider, model, reasoning effort, and sampling scalars of one conversation's
+ * requests. Every field maps 1:1 onto the same-named `GenerateOptions` field;
+ * the loop builds requests from the logged header rather than accepting these
+ * per call.
  */
 interface LlmCallConfig {
   provider: string
   model: string
+  reasoningEffort?: ReasoningEffortId
   temperature?: number
   maxTokens?: number
   stop?: string[]
diff --git a/docs/core-data-structures/llm-streaming.md b/docs/core-data-structures/llm-streaming.md
index 257ce90cda..b3c285ebba 100644
--- a/docs/core-data-structures/llm-streaming.md
+++ b/docs/core-data-structures/llm-streaming.md
@@ -154,7 +154,7 @@ declare class BlockAssembler {
 
 ## The seam
 
-`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. The separate `resolveModelContext()` query exposes correctness-sensitive capacity for an exact route without making catalog membership authoritative; absence means unknown metadata, not invalid routing. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
+`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. The separate `resolveModelContext()` query exposes correctness-sensitive capacity, while `resolveModelReasoning()` exposes ordered model-owned effort ids and an optional deployment default; absence from either query means unavailable metadata or capability, not invalid catalog membership. The service validates and materializes reasoning through `resolveCallConfig()` at the final adapter boundary, so direct calls cannot bypass unsupported-effort rejection. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
 
 ```ts public-api
 /**
@@ -189,6 +189,17 @@ declare abstract class LlmAdapter {
     _provider: string,
     _model: string,
   ): Promise<LlmModelContext | undefined>;
+  /**
+   * Resolve selectable reasoning efforts for one exact model. Absence means
+   * the model has no selectable reasoning-effort capability.
+   * @param _provider - one provider route owned by this adapter.
+   * @param _model - exact model id passed to {@link GenerateOptions.model}.
+   * @returns adapter-owned effort metadata, or `undefined` when unsupported.
+   */
+  resolveModelReasoning(
+    _provider: string,
+    _model: string,
+  ): Promise<LlmModelReasoningInfo | undefined>;
   /**
    * Stream one model call as raw chunks. The only required method.
    * @param options - the fully-assembled request; implementations must honor `options.signal`.
diff --git a/docs/core-data-structures/session.md b/docs/core-data-structures/session.md
index ba1dcb98ba..eb530f7709 100644
--- a/docs/core-data-structures/session.md
+++ b/docs/core-data-structures/session.md
@@ -167,7 +167,7 @@ The request envelope — the `EpochHeader` (call config + rendered system prompt
  * canonical empty optional fields are absent.
  */
 interface EpochHeader {
-  /** The conversation's call configuration (provider, model, and sampling scalars). */
+  /** The conversation's call configuration (provider, model, reasoning effort, and sampling scalars). */
   config: LlmCallConfig
   /** Rendered system prompt text; absent for a system-less request. */
   system?: string
diff --git a/docs/event-producer-consumer.md b/docs/event-producer-consumer.md
index 179ad0dcee..5659a8c7bf 100644
--- a/docs/event-producer-consumer.md
+++ b/docs/event-producer-consumer.md
@@ -30,7 +30,7 @@ This matrix shows which packages dispatch each harness-owned event and which pac
 | `fs/observed` | `emit` | [`packages/fs/fs/src/index.ts:71`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`emit`) | [`fs-policy`](../packages/fs/fs-policy) |
 | `fs/write-intent` | `waterfall` | [`packages/fs/fs/src/index.ts:54`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`waterfall`) | [`fs-policy`](../packages/fs/fs-policy) |
 | `goal/changed` | `emit` | [`packages/goal/goal/src/types.ts:167`](../packages/goal/goal/src/types.ts) | [`goal`](../packages/goal/goal) (`emit`) | [`goal-session`](../packages/goal/goal-session) |
-| `llm/stream` | `waterfall` | [`packages/llm/llm/src/index.ts:52`](../packages/llm/llm/src/index.ts) | [`llm`](../packages/llm/llm) (`waterfall`) | [`agent-loop`](../packages/core/agent-loop), [`llm`](../packages/llm/llm), [`llm-replay`](../packages/support/llm-replay), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`session-title`](../packages/session-title/session-title) |
+| `llm/stream` | `waterfall` | [`packages/llm/llm/src/index.ts:54`](../packages/llm/llm/src/index.ts) | [`llm`](../packages/llm/llm) (`waterfall`) | [`agent-loop`](../packages/core/agent-loop), [`llm`](../packages/llm/llm), [`llm-replay`](../packages/support/llm-replay), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`session-title`](../packages/session-title/session-title) |
 | `session/created` | `emit` | [`packages/core/session/src/index.ts:79`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`compact`](../packages/compact/compact), [`goal`](../packages/goal/goal), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`llm-retry`](../packages/llm/llm-retry), `runtime`, [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`user-approval`](../packages/ui/user-approval) |
 | `session/disposed` | `emit` | [`packages/core/session/src/index.ts:89`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`agent-loop`](../packages/core/agent-loop), `runtime`, [`session-persistence`](../packages/session-persistence/session-persistence), [`session-title`](../packages/session-title/session-title) |
 | `session/event` | `emit` | [`packages/core/session/src/index.ts:101`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`acp`](../packages/ui/acp), [`cli-demo`](../packages/examples/cli-demo), [`compact`](../packages/compact/compact), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), `runtime`, [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-title`](../packages/session-title/session-title), [`token-meter`](../packages/llm/token-meter), [`tui`](../packages/ui/tui), [`user-approval`](../packages/ui/user-approval), [`workspace-context`](../packages/context/workspace-context) |
diff --git a/docs/user/develop/practice/llm-adapter.i18n.yaml b/docs/user/develop/practice/llm-adapter.i18n.yaml
index 8735e8d5a6..c179b693f8 100644
--- a/docs/user/develop/practice/llm-adapter.i18n.yaml
+++ b/docs/user/develop/practice/llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-llm-adapter.md: 3e83289b8072ef231f83c0fa3cfe3260547b42fa
-llm-adapter.zh.md: 92fcf9b22f4bb356ada4c46f9a03ef0cc2d159da
+llm-adapter.md: f133a6ea748fd7d9bfa6a7adaa35b1cccbe40c8e
+llm-adapter.zh.md: 3a4fd9a516126d1f9fa8675fdc87be10c85f3489
diff --git a/docs/user/develop/practice/llm-adapter.md b/docs/user/develop/practice/llm-adapter.md
index 3e83289b80..f133a6ea74 100644
--- a/docs/user/develop/practice/llm-adapter.md
+++ b/docs/user/develop/practice/llm-adapter.md
@@ -110,7 +110,9 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 ## GenerateOptions
 
-`stream()` receives the exported `GenerateOptions` type. It includes the model, conversation history, system prompt, tool schemas, generation parameters, stop sequences, and abort signal; treat the TypeScript type exported by `@deepseek-ai/dsh-llm` as authoritative. Map supported fields to the provider API. If the provider cannot honor a field, throw `LlmError` with a stable code instead of silently dropping it.
+`stream()` receives the exported `GenerateOptions` type. It includes the model, adapter-owned reasoning-effort id, conversation history, system prompt, tool schemas, generation parameters, stop sequences, and abort signal; treat the TypeScript type exported by `@deepseek-ai/dsh-llm` as authoritative. Map supported fields to the provider API. If the provider cannot honor a field, throw `LlmError` with a stable code instead of silently dropping it.
+
+Override `resolveModelReasoning(provider, model)` when an exact model exposes selectable reasoning strengths. Return ordered opaque ids and display names plus an optional configured default; do not promote provider names into a core enum. The service validates the metadata and rejects unsupported explicit values before `stream()`. Returning `undefined` means that model has no selectable reasoning-effort capability.
 
 ## Register an adapter
 
diff --git a/docs/user/develop/practice/llm-adapter.zh.md b/docs/user/develop/practice/llm-adapter.zh.md
index 92fcf9b22f..3a4fd9a516 100644
--- a/docs/user/develop/practice/llm-adapter.zh.md
+++ b/docs/user/develop/practice/llm-adapter.zh.md
@@ -110,7 +110,9 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 ## GenerateOptions
 
-`stream()` 接收仓库导出的 `GenerateOptions`。它包含模型名、对话历史、系统提示词、tool schema、生成参数、停止序列和中止信号；完整字段以 `@deepseek-ai/dsh-llm` 导出的 TypeScript 类型为准。适配器必须将支持的字段映射到具体 API；无法支持的字段应抛出带稳定 code 的 `LlmError`，不能静默丢弃。
+`stream()` 接收仓库导出的 `GenerateOptions`。它包含模型名、由适配器持有的推理强度 ID、对话历史、系统提示词、tool schema、生成参数、停止序列和中止信号；完整字段以 `@deepseek-ai/dsh-llm` 导出的 TypeScript 类型为准。适配器必须将支持的字段映射到具体 API；无法支持的字段应抛出带稳定 code 的 `LlmError`，不能静默丢弃。
+
+当某个具体模型提供可选推理强度时，请覆写 `resolveModelReasoning(provider, model)`。返回有序的不透明 ID、展示名称，以及可选的配置默认值；不要将提供方使用的等级名称提升为核心枚举。服务会校验元数据，并在调用 `stream()` 前拒绝显式指定但不受支持的值。返回 `undefined` 表示该模型没有可选的推理强度能力。
 
 ## 注册适配器
 
diff --git a/examples/headless-agent/tests/fixtures/cli-mock-llm.ts b/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
index 5238e67374..23b5e18fc9 100644
--- a/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
+++ b/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
@@ -1,8 +1,28 @@
 import type { Context } from 'cordis'
-import { CallId, LlmAdapter, type GenerateOptions, type StreamChunk } from '@deepseek-ai/dsh-llm'
+import {
+  CallId,
+  LlmAdapter,
+  ReasoningEffortId,
+  type GenerateOptions,
+  type LlmModelReasoningInfo,
+  type StreamChunk,
+} from '@deepseek-ai/dsh-llm'
+
+const HIGH = ReasoningEffortId('high')
+const MAX = ReasoningEffortId('max')
 
 /** Keyless headless-agent adapter: one real bash call followed by a final answer. */
 class CliMockAdapter extends LlmAdapter {
+  override async resolveModelReasoning(): Promise<LlmModelReasoningInfo> {
+    return {
+      efforts: [
+        { id: HIGH, name: 'High' },
+        { id: MAX, name: 'Max' },
+      ],
+      defaultEffort: HIGH,
+    }
+  }
+
   async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
     const toolResult = options.messages.at(-1)?.content.find(block => block.type === 'tool-result')
     if (toolResult === undefined) {
@@ -34,4 +54,8 @@ export const inject = ['llm']
 /** Register the keyless `cli-mock` adapter. */
 export function apply(ctx: Context): void {
   ctx.llm.registerAdapter(['cli-mock'], new CliMockAdapter())
+  ctx.on('agent/request', async (_agent, _turn, step, _config, _signal, next) => {
+    const config = await next()
+    return step === 2 ? { ...config, reasoningEffort: MAX } : config
+  })
 }
diff --git a/examples/headless-agent/tests/headless.snapshot.ts b/examples/headless-agent/tests/headless.snapshot.ts
index 1626489bf5..70c1fa8bfe 100644
--- a/examples/headless-agent/tests/headless.snapshot.ts
+++ b/examples/headless-agent/tests/headless.snapshot.ts
@@ -28,6 +28,7 @@ const ralphScenarioDir = join(snapshotsDir, 'ralph-loop')
 const ralphConfigPath = fileURLToPath(new URL('../ralph.cordis.snapshot.yml', import.meta.url))
 const binScript = fileURLToPath(new URL('../../../packages/examples/cli-demo/src/bin.ts', import.meta.url))
 const tsconfigPath = fileURLToPath(new URL('../../../tsconfig.json', import.meta.url))
+const reasoningConfigPath = fileURLToPath(new URL('./fixtures/cli.cordis.yml', import.meta.url))
 const refreshing = process.env.DSH_SNAPSHOT === 'refresh'
 
 interface JsonObject {
@@ -124,6 +125,46 @@ async function persistedLogs(cwd: string): Promise<PersistedLog[]> {
 }
 
 describe('headless stream-json snapshots', () => {
+  it('logs the model default and a dynamic next-step reasoning effort', async () => {
+    const result = await runLoaderSmoke({
+      label: 'reasoning effort headless stream-json snapshot',
+      tempDirPrefix: 'headless-snapshot-reasoning-effort-',
+      binScript,
+      configPath: reasoningConfigPath,
+      binArgs: ['--config', reasoningConfigPath, '--output-format', 'stream-json', 'prove dynamic reasoning effort'],
+      tsconfigPath,
+    })
+
+    expect(result.stderr).toBe('')
+    const headers = parseJsonl(result.stdout)
+      .map(record => record.event)
+      .filter((event): event is JsonObject => (
+        event !== null
+        && typeof event === 'object'
+        && !Array.isArray(event)
+        && 'type' in event
+        && event.type === 'request/header'
+      ))
+      .map((event) => {
+        const data = event.data as JsonObject
+        return (data.header as JsonObject).config
+      })
+    expect(headers).toMatchInlineSnapshot(`
+      [
+        {
+          "model": "cli-mock",
+          "provider": "cli-mock",
+          "reasoningEffort": "high",
+        },
+        {
+          "model": "cli-mock",
+          "provider": "cli-mock",
+          "reasoningEffort": "max",
+        },
+      ]
+    `)
+  }, LOADER_SMOKE_TEST_TIMEOUT_MS)
+
   it('replays the advanced toolchain through the one-shot app', async () => {
     const prompt = await scenarioPrompt(advancedScenarioDir, 'advanced-toolchain')
     const fixtureFiles = [
diff --git a/packages/cordis/tool-cordis/src/api-catalog.ts b/packages/cordis/tool-cordis/src/api-catalog.ts
index ffc1670f4e..604a490405 100644
--- a/packages/cordis/tool-cordis/src/api-catalog.ts
+++ b/packages/cordis/tool-cordis/src/api-catalog.ts
@@ -344,6 +344,14 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
         signature: 'async resolveModelContext( provider: string, model: string, ): Promise<LlmModelContext | undefined>',
         jsDoc: '/**\n * Resolve context capacity from the adapter that owns one exact route.\n * This query is independent of the advisory model catalog: an unlisted model\n * may return metadata, while `undefined` never rejects later routing.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @returns detached context metadata, or `undefined` when the adapter has none.\n */',
       },
+      {
+        signature: 'async resolveModelReasoning( provider: string, model: string, ): Promise<LlmModelReasoningInfo | undefined>',
+        jsDoc: '/**\n * Resolve selectable reasoning efforts from the adapter that owns one exact\n * route. Metadata is validated and detached; an absent result means an\n * effort selector is unsupported for that model.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @returns detached reasoning metadata, or `undefined` when unsupported.\n */',
+      },
+      {
+        signature: 'async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>',
+        jsDoc: '/**\n * Validate a conversation call config against its exact model capability and\n * materialize an adapter-configured default. Unsupported explicit efforts\n * reject before provider I/O; no clamping or aliasing is performed.\n * @param config - provider/model route and optional request controls.\n * @returns a detached config only when a default must be materialized.\n */',
+      },
       {
         signature: 'stream(options: GenerateOptions): AsyncIterable<StreamChunk>',
         jsDoc: '/**\n * Stream one model call as raw chunks (token-level deltas). Throws\n * `LlmError` with code `NO_ADAPTER` if no adapter is registered for\n * `options.provider`. Replay state is retained only when the same adapter\n * instance owns its historical provider and the target provider. Final\n * adapter selection, dispatch, and iteration failures retain their original\n * Error identity and are tagged in a call-local scope for narrow agent-loop\n * request recovery; middleware and nested-call failures remain untagged for\n * the outer call.\n * @param options - the full request; `options.provider` selects the adapter.\n * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.\n */',
@@ -1455,7 +1463,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'GenerateOptions',
-    declaration: 'export interface GenerateOptions {\n    provider: string;\n    model: string;\n    messages: Message[];\n    system?: string;\n    tools?: ToolSchema[];\n    temperature?: number;\n    maxTokens?: number;\n    stop?: string[];\n    signal?: AbortSignal;\n    sessionId?: Branded<\'SessionId\'>;\n    purpose?: \'compaction\' | \'session-title\';\n}',
+    declaration: 'export interface GenerateOptions {\n    provider: string;\n    model: string;\n    reasoningEffort?: ReasoningEffortId;\n    messages: Message[];\n    system?: string;\n    tools?: ToolSchema[];\n    temperature?: number;\n    maxTokens?: number;\n    stop?: string[];\n    signal?: AbortSignal;\n    sessionId?: Branded<\'SessionId\'>;\n    purpose?: \'compaction\' | \'session-title\';\n}',
   },
   {
     name: 'GenericCallView',
@@ -1527,7 +1535,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'LlmCallConfig',
-    declaration: 'export interface LlmCallConfig {\n    provider: string;\n    model: string;\n    temperature?: number;\n    maxTokens?: number;\n    stop?: string[];\n}',
+    declaration: 'export interface LlmCallConfig {\n    provider: string;\n    model: string;\n    reasoningEffort?: ReasoningEffortId;\n    temperature?: number;\n    maxTokens?: number;\n    stop?: string[];\n}',
   },
   {
     name: 'LlmFailure',
@@ -1541,10 +1549,18 @@ export const TYPE_API: readonly TypeApiEntry[] = [
     name: 'LlmModelInfo',
     declaration: 'export interface LlmModelInfo {\n    provider: string;\n    id: string;\n    name: string;\n    description?: string;\n}',
   },
+  {
+    name: 'LlmModelReasoningInfo',
+    declaration: 'export interface LlmModelReasoningInfo {\n    efforts: readonly LlmReasoningEffortInfo[];\n    defaultEffort?: ReasoningEffortId;\n}',
+  },
   {
     name: 'LlmProviderInfo',
     declaration: 'export interface LlmProviderInfo {\n    id: string;\n    name: string;\n}',
   },
+  {
+    name: 'LlmReasoningEffortInfo',
+    declaration: 'export interface LlmReasoningEffortInfo {\n    id: ReasoningEffortId;\n    name: string;\n    description?: string;\n}',
+  },
   {
     name: 'Message',
     declaration: 'export interface Message {\n    role: \'system\' | \'user\' | \'assistant\';\n    content: ContentBlock[];\n    provenance?: AssistantProvenance;\n}',
@@ -1689,6 +1705,10 @@ export const TYPE_API: readonly TypeApiEntry[] = [
     name: 'ReasoningBlock',
     declaration: 'export interface ReasoningBlock {\n    type: \'reasoning\';\n    text: string;\n}',
   },
+  {
+    name: 'ReasoningEffortId',
+    declaration: 'export type ReasoningEffortId = Branded<\'ReasoningEffortId\'>;',
+  },
   {
     name: 'ResumeAgentOptions',
     declaration: 'export interface ResumeAgentOptions {\n    readonly resumeSessionId: SessionId;\n    readonly agentOptions?: AgentOptions;\n    readonly signal?: AbortSignal;\n    readonly setup?: (agentCtx: Context) => Promise<void> | void;\n}',
diff --git a/packages/core/agent-loop/README.md b/packages/core/agent-loop/README.md
index 79fbbd7824..0390d54f94 100644
--- a/packages/core/agent-loop/README.md
+++ b/packages/core/agent-loop/README.md
@@ -60,6 +60,8 @@ The driver owns one agent for its lifetime and runs inside `ctx.agents.withIniti
 
 Every provider call that reaches a successful finish appends exactly one `assistant/message` completion anchor, including content-less calls and `max-tokens` finishes. A successful `agent/step-result` stores its transformed content; a rejected result records empty content before the original failure continues. The anchor retains exact chunk provenance (`[]` for a stream with no chunks) and usage when available, while empty content stays out of derived message history.
 
+After `agent/request` returns a provider/model call config, the loop asks `ctx.llm.resolveCallConfig()` to validate any adapter-owned reasoning effort and materialize its configured default. The effective config is logged in the full `request/header` before dispatch, so a listener can change effort between steps without hidden request drift. A route with no registered adapter preserves the proposed config so an `llm/stream` listener can own and short-circuit it; unhandled terminal dispatch still fails with `NO_ADAPTER`. A new loop instance restores the last effort only when its initial provider/model route exactly matches the logged route; a route change discards that opaque model-owned ID and resolves the new model independently.
+
 Plugin failure ends the current turn, not the loop. Only final adapter dispatch/iteration failures and terminal in-band error or aborted finishes enter `agent/request-error`; middleware, result processing, tools, and `agent/post-step` remain ordinary turn failures. Recovery receives the exact live error, immutable provider facts, and immutable prior failures after the failed step closes. A retry rebuilds from the durable log in a new numbered step, success clears the consecutive history, and exhaustion records the structured failure once on `turn/end`. AgentLoop privately owns one cancellation holder whose explicit signal spans prompt policy, assembly, every step, model and tool work, recovery, continuation, and terminal stop; it retires the holder immediately before publishing `turn/end`, while the driver may remain `running` through the durability flush. An effective `cancel()` emits the typed runtime-only `user | parent` cause before clearing pending work and cooperatively aborting the holder; notification failures cannot veto cancellation, work queued by a notification observer is cleared, work queued by a later abort observer belongs to the next turn, and idle cancellation emits nothing. Durable `turn/end` remains coarse `aborted`; undispatched model tool calls receive synthetic `tool/call` and `ABORTED_BEFORE_DISPATCH` result pairs. Disposal wins terminal classification, and work that ignores the signal must settle before quiescence. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns the lifecycle and race contract. Terminal continuation stops remain authoritative through turn close and durability flush.
 
 Within a step, exclusive calls form barriers; parallel-safe calls use a bounded rolling pool and are reclassified before start. Only dispatch/body overlaps. Policy, durable results, and result context remain model-ordered. Abort stops new calls, drains started results, then drains accepted batch context before the turn closes through the normal abort path.
diff --git a/packages/core/agent-loop/src/loop.ts b/packages/core/agent-loop/src/loop.ts
index 1cfd913b77..dae3cf0efc 100644
--- a/packages/core/agent-loop/src/loop.ts
+++ b/packages/core/agent-loop/src/loop.ts
@@ -621,19 +621,43 @@ async function runStep(
 
   // Seed the first request from agent options and later requests from the logged header;
   // detach and freeze so listeners must return an attributable replacement.
-  const seedConfig: LlmCallConfig = deepFreeze(structuredClone(transmission.loggedHeader
-    // eslint-disable-next-line @typescript-eslint/no-non-null-assertion -- loggedHeader ⟹ a snapshot is in the log
-    ? session.requestHeader()!.config
-    : { provider: options.provider ?? '', model: options.model ?? '' }))
+  const loggedConfig = session.requestHeader()?.config
+  const initialProvider = options.provider ?? ''
+  const initialModel = options.model ?? ''
+  const initialConfig: LlmCallConfig = {
+    provider: initialProvider,
+    model: initialModel,
+    ...loggedConfig?.provider === initialProvider
+      && loggedConfig.model === initialModel
+      && loggedConfig.reasoningEffort !== undefined
+      ? { reasoningEffort: loggedConfig.reasoningEffort }
+      : {},
+  }
+  const seedConfig: LlmCallConfig = deepFreeze(structuredClone(
+    transmission.loggedHeader
+      // eslint-disable-next-line @typescript-eslint/no-non-null-assertion -- loggedHeader ⟹ a snapshot is in the log
+      ? session.requestHeader()!.config
+      : initialConfig,
+  ))
 
   // Listener replacements are recorded in the request header before dispatch.
-  const config = await events.waterfall(
+  const proposedConfig = await events.waterfall(
     'agent/request', turn, step, seedConfig, signal, () => Promise.resolve(seedConfig),
   )
   interruptionCheckpoint(signal)
-  if (!config.provider || !config.model) {
+  if (!proposedConfig.provider || !proposedConfig.model) {
     throw new Error(`agent "${agent.id}" has no provider/model: set AgentOptions.provider and AgentOptions.model or supply both via the agent/request waterfall`)
   }
+  let config: LlmCallConfig
+  try {
+    config = await ctx.llm.resolveCallConfig(proposedConfig)
+  } catch (error: unknown) {
+    // A waterfall listener may own and short-circuit a route with no adapter.
+    // Terminal dispatch still raises NO_ADAPTER when no listener handles it.
+    if (!(error instanceof LlmError) || error.code !== 'NO_ADAPTER') throw error
+    config = proposedConfig
+  }
+  interruptionCheckpoint(signal)
 
   // eslint-disable-next-line @typescript-eslint/no-non-null-assertion -- runTurn composes the prefix before every runStep call
   const sessionPrefix = transmission.sessionPrefix!
@@ -651,6 +675,9 @@ async function runStep(
   const request: GenerateOptions = markAgentLoopRequest(deepFreeze({
     provider: header.config.provider,
     model: header.config.model,
+    ...header.config.reasoningEffort !== undefined
+      ? { reasoningEffort: header.config.reasoningEffort }
+      : {},
     messages: [...header.messagePrefix ?? [], ...boundaryMessages],
     ...header.system !== undefined ? { system: header.system } : {},
     ...header.tools !== undefined ? { tools: header.tools } : {},
diff --git a/packages/core/agent-loop/tests/mock-adapter.ts b/packages/core/agent-loop/tests/mock-adapter.ts
index 4aaff72415..8b47a9880d 100644
--- a/packages/core/agent-loop/tests/mock-adapter.ts
+++ b/packages/core/agent-loop/tests/mock-adapter.ts
@@ -1,4 +1,4 @@
-import type { GenerateOptions, StreamChunk } from '@deepseek-ai/dsh-llm'
+import type { GenerateOptions, LlmModelReasoningInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
 import { CallId, LlmAdapter } from '@deepseek-ai/dsh-llm'
 
 /** Helpers to write scripted responses tersely. */
@@ -64,10 +64,20 @@ export function toolCallResponse(rawCallId: string, name: string, args: object,
 export class MockAdapter extends LlmAdapter {
   requests: GenerateOptions[] = []
 
-  constructor(private script: (StreamChunk[] | ((options: GenerateOptions) => StreamChunk[]) | 'hang')[]) {
+  constructor(
+    private script: (StreamChunk[] | ((options: GenerateOptions) => StreamChunk[]) | 'hang')[],
+    private readonly reasoning?: LlmModelReasoningInfo,
+  ) {
     super()
   }
 
+  override resolveModelReasoning(
+    _provider: string,
+    _model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    return Promise.resolve(this.reasoning)
+  }
+
   async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
     this.requests.push(options)
     const entry = this.script.shift()
diff --git a/packages/core/agent-loop/tests/request-reconstruction.spec.ts b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
index 63c59618e3..781ee908a7 100644
--- a/packages/core/agent-loop/tests/request-reconstruction.spec.ts
+++ b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
@@ -7,7 +7,7 @@
 
 import { describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
-import LlmService from '@deepseek-ai/dsh-llm'
+import LlmService, { LlmError, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type { GenerateOptions } from '@deepseek-ai/dsh-llm'
 import SessionStore, { Session, SessionId, foldRequestHeader } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
@@ -104,6 +104,81 @@ describe('request stability across the loop', () => {
     expectPrefixExtension(adapter.requests[0]!, adapter.requests[1]!)
   })
 
+  it('logs adapter defaults, supports per-turn effort changes, and restores the effective value', async () => {
+    const reasoning = {
+      efforts: [
+        { id: ReasoningEffortId('high'), name: 'High' },
+        { id: ReasoningEffortId('max'), name: 'Max' },
+      ],
+      defaultEffort: ReasoningEffortId('high'),
+    }
+    const adapter = new MockAdapter([textResponse('one'), textResponse('two')], reasoning)
+    const ctx = await harness(adapter)
+    const agent = ctx.agentLoop.create(SessionId('effort'), { provider: 'mock', model: 'mock' })
+    ctx.on('agent/request', async (_agent, turn, _step, _config, _signal, next) => {
+      const config = await next()
+      return turn === 2 ? { ...config, reasoningEffort: ReasoningEffortId('max') } : config
+    })
+
+    send(agent, 'first')
+    await waitForIdle(ctx, agent)
+    send(agent, 'second')
+    await waitForIdle(ctx, agent)
+
+    expect(adapter.requests.map(request => request.reasoningEffort)).toEqual([
+      ReasoningEffortId('high'),
+      ReasoningEffortId('max'),
+    ])
+    const headers = agent.session.events.filter(event => event.type === 'request/header')
+    expect(headers.map(event => event.data.header.config.reasoningEffort)).toEqual([
+      ReasoningEffortId('high'),
+      ReasoningEffortId('max'),
+    ])
+    expect(headers.map(event => event.data.reason)).toEqual(['initial', 'change'])
+
+    const resumedAdapter = new MockAdapter([textResponse('three')], reasoning)
+    const resumedCtx = await harness(resumedAdapter)
+    const resumedHandle = await resumedCtx.agents.create({
+      sessionId: SessionId('effort-resumed'),
+      seed: structuredClone(agent.session.events),
+      agentOptions: { provider: 'mock', model: 'mock' },
+    })
+    send(resumedHandle.agent, 'third')
+    await waitForIdle(resumedCtx, resumedHandle.agent)
+
+    expect(resumedAdapter.requests[0]?.reasoningEffort).toBe(ReasoningEffortId('max'))
+    const resumedHeaders = resumedHandle.agent.session.events.filter(event => event.type === 'request/header')
+    expect(resumedHeaders.at(-1)?.data.header.config.reasoningEffort).toBe(ReasoningEffortId('max'))
+    expect(resumedHeaders.at(-1)?.data.reason).toBe('resume')
+  })
+
+  it.each(['plain error', 'LLM error'] as const)(
+    'does not swallow a %s from reasoning resolution',
+    async (kind) => {
+      const failure = kind === 'plain error'
+        ? new Error('reasoning metadata failed')
+        : new LlmError('unsupported effort', 'UNSUPPORTED_REASONING_EFFORT')
+      const adapter = new class extends MockAdapter {
+        override resolveModelReasoning(): Promise<never> {
+          return Promise.reject(failure)
+        }
+      }([])
+      const ctx = await harness(adapter)
+      const errors: Error[] = []
+      ctx.on('agent/error', (_agent, _turn, _step, error) => void errors.push(error))
+      const agent = ctx.agentLoop.create(SessionId(`reasoning-${kind}`), {
+        provider: 'mock',
+        model: 'mock',
+      })
+
+      send(agent, 'go')
+      await waitForIdle(ctx, agent)
+
+      expect(errors).toContain(failure)
+      expect(adapter.requests).toHaveLength(0)
+    },
+  )
+
   it('a compaction replace rewrites the resend, and the log explains it', async () => {
     const adapter = new MockAdapter([textResponse('one'), textResponse('two')])
     const ctx = await harness(adapter)
@@ -301,6 +376,7 @@ describe('request stability across the loop', () => {
       const firstChunk = events.find(e => e.type === 'assistant/chunk' && e.seq > stepStart.seq)!
       const header = foldRequestHeader(events.slice(0, firstChunk.seq))!
       expect(request.model).toBe(header.config.model)
+      expect(request.reasoningEffort).toBe(header.config.reasoningEffort)
       expect(request.system).toEqual(header.system)
       expect(structuredClone(request.tools ?? [])).toEqual(structuredClone(header.tools ?? []))
       expect(request.temperature).toBe(header.config.temperature)
diff --git a/packages/core/session/src/index.ts b/packages/core/session/src/index.ts
index 9c5831bb18..4f287b1517 100644
--- a/packages/core/session/src/index.ts
+++ b/packages/core/session/src/index.ts
@@ -181,6 +181,11 @@ function assertCurrentLlmShape(event: Record<string, unknown>, index: number): v
     const header = record['header']
     const config = typeof header === 'object' && header !== null ? (header as Record<string, unknown>)['config'] : undefined
     if (!hasProviderModel(config)) throw new Error(`seed request/header at index ${index} lacks provider/model`)
+    const reasoningEffort = (config as Record<string, unknown>)['reasoningEffort']
+    if (reasoningEffort !== undefined
+      && (typeof reasoningEffort !== 'string' || reasoningEffort.length === 0)) {
+      throw new Error(`seed request/header at index ${index} has an invalid reasoningEffort`)
+    }
   }
   if (event['type'] === 'assistant/message' && !hasProviderModel(record['provenance'])) {
     throw new Error(`seed assistant/message at index ${index} lacks provider/model provenance`)
diff --git a/packages/core/session/src/types.ts b/packages/core/session/src/types.ts
index 8b3a1e8cb6..4bc1c7eb04 100644
--- a/packages/core/session/src/types.ts
+++ b/packages/core/session/src/types.ts
@@ -156,7 +156,7 @@ export interface TodoItem {
  * canonical empty optional fields are absent.
  */
 export interface EpochHeader {
-  /** The conversation's call configuration (provider, model, and sampling scalars). */
+  /** The conversation's call configuration (provider, model, reasoning effort, and sampling scalars). */
   config: LlmCallConfig
   /** Rendered system prompt text; absent for a system-less request. */
   system?: string
diff --git a/packages/core/session/tests/request-header.spec.ts b/packages/core/session/tests/request-header.spec.ts
index b185cf6e56..e2fd4cc781 100644
--- a/packages/core/session/tests/request-header.spec.ts
+++ b/packages/core/session/tests/request-header.spec.ts
@@ -4,6 +4,7 @@ import { describe, expect, it } from 'vitest'
 import { Session, SessionId, canonicalHeader, foldRequestHeader, headerEquals } from '@deepseek-ai/dsh-session'
 import type { EpochHeader, SessionEvent } from '@deepseek-ai/dsh-session'
 import type { Message, ToolSchema } from '@deepseek-ai/dsh-llm'
+import { ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 
 const CONFIG = { provider: 'mock', model: 'm' }
 
@@ -29,6 +30,10 @@ describe('headerEquals', () => {
   it('compares every canonical field and preserves tool order', () => {
     expect(headerEquals(base, structuredClone(base))).toBe(true)
     expect(headerEquals(base, { ...base, config: { provider: 'mock', model: 'other' } })).toBe(false)
+    expect(headerEquals(base, {
+      ...base,
+      config: { ...base.config, reasoningEffort: ReasoningEffortId('high') },
+    })).toBe(false)
     expect(headerEquals(base, { ...base, system: 'other' })).toBe(false)
     expect(headerEquals(base, { ...base, messagePrefix: [msg('other')] })).toBe(false)
     expect(headerEquals(base, { ...base, tools: [] })).toBe(false)
diff --git a/packages/core/session/tests/session.spec.ts b/packages/core/session/tests/session.spec.ts
index d880153dd3..880a5df42f 100644
--- a/packages/core/session/tests/session.spec.ts
+++ b/packages/core/session/tests/session.spec.ts
@@ -1,6 +1,6 @@
 import { describe, expect, expectTypeOf, it, vi } from 'vitest'
 import { Context } from 'cordis'
-import { CallId } from '@deepseek-ai/dsh-llm'
+import { CallId, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import SessionStore, {
   displayPromptContent,
   findLastMessageTurnEnd,
@@ -118,6 +118,11 @@ describe('Session', () => {
   })
 
   it('renders context and steering messages as plain user content', () => {
+    expect(displayPromptContent({
+      content: [{ type: 'text', text: 'plain prompt' }],
+      source: { kind: 'user' },
+    })).toEqual([{ type: 'text', text: 'plain prompt' }])
+
     const session = new Session(SessionId('s2'))
     session.append('context/message', {
       content: [{ type: 'text', text: 'file changed: a.ts' }],
@@ -228,6 +233,35 @@ describe('Session', () => {
       .toEqual([unrelatedPrimitiveData])
   })
 
+  it('round-trips a non-empty reasoning effort and rejects invalid durable values', () => {
+    const valid = {
+      type: 'request/header',
+      seq: 0,
+      time: 1,
+      data: {
+        header: {
+          config: {
+            provider: 'mock',
+            model: 'model',
+            reasoningEffort: ReasoningEffortId('adapter-owned'),
+          },
+        },
+        reason: 'initial',
+      },
+    } as const
+    expect(new Session(SessionId('reasoning-effort'), [valid]).events[0])
+      .toEqual(valid)
+
+    for (const reasoningEffort of ['', 1]) {
+      const invalid = structuredClone(valid) as unknown as SessionEvent
+      if (invalid.type !== 'request/header') throw new Error('test fixture must be a request header')
+      const config = invalid.data.header.config as unknown as Record<string, unknown>
+      config.reasoningEffort = reasoningEffort
+      expect(() => new Session(SessionId('invalid-reasoning-effort'), [invalid]))
+        .toThrow('seed request/header at index 0 has an invalid reasoningEffort')
+    }
+  })
+
   it('isolates the log from mutation through a derived message (append-only contract)', () => {
     const session = new Session(SessionId('s4'))
     session.append('user/message', { content: [{ type: 'text', text: 'original' }], source: { kind: 'user' } }, { surfaceOp: 'append' })
diff --git a/packages/llm/llm-deepseek/README.md b/packages/llm/llm-deepseek/README.md
index 5a9a1bfdbc..13e889847f 100644
--- a/packages/llm/llm-deepseek/README.md
+++ b/packages/llm/llm-deepseek/README.md
@@ -15,7 +15,7 @@ The package root exposes the Cordis plugin contract and `DeepSeekAdapter`; wire
     apiKey: !!js process.env.DEEPSEEK_API_KEY    # or rely on the env fallback
     baseURL: !!js process.env.DEEPSEEK_BASE_URL  # default: https://api.deepseek.com
     thinking: enabled        # optional; provider default is enabled
-    reasoningEffort: high    # optional; high | max — omitted ⇒ not sent
+    reasoningEffort: high    # optional; high | max — omitted ⇒ high
     streamIdleTimeoutMs: 300000 # optional; positive finite Node timer delay; five-minute default
     defaultContextWindow: 256000 # optional positive-integer fallback for models without an exact value
     models:                  # optional; defaults to V4 Flash and V4 Pro
@@ -30,9 +30,9 @@ The plugin registers the single provider route `deepseek`. A request selects it
 
 `contextWindow` is optional per configured model and is not exposed through the advisory catalog. `ctx.llm.resolveModelContext('deepseek', model)` returns an exact model value first, then `defaultContextWindow` for an entry without capacity or an unlisted pass-through id. When neither value exists it returns `undefined` without invalidating routing. Pressure-sensitive plugins therefore get deployment-owned capacity without treating the model selector as authoritative. Registering another adapter for `deepseek` throws `LlmError('DUPLICATE_ADAPTER')`.
 
-`reasoningEffort` is **omitted by default** — when unset, the `reasoning_effort` wire field is not sent and the server applies its own default for the model. The only accepted values are `high` and `max` (DeepSeek's official effort levels). It is meaningful only with thinking enabled (the provider default).
+`ctx.llm.resolveModelReasoning('deepseek', model)` returns the ordered `high` and `max` efforts for every pass-through model while thinking is enabled. `reasoningEffort` selects the deployment default and falls back to `high` when omitted. `agent/request` can replace it on each conversation step; the resolved value is logged in `request/header` and serialized as the official top-level `reasoning_effort` field. An unsupported value fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O.
 
-`thinking`/`reasoningEffort` are adapter-level request defaults serialized as the official top-level `thinking: {type}` / `reasoning_effort` wire fields. They live in adapter config (not `GenerateOptions`) to keep the core vocabulary provider-neutral. A request with `GenerateOptions.purpose: 'session-title'` forces thinking disabled and omits `reasoning_effort`, reserving its bounded output for visible title text without changing conversation or compaction defaults.
+`thinking: disabled` removes the reasoning capability and omits `reasoning_effort`; combining it with a configured default fails plugin loading, and a per-request effort fails as unsupported. A request with `GenerateOptions.purpose: 'session-title'` also forces thinking disabled and omits the already-resolved effort, reserving its bounded output for visible title text without changing conversation or compaction defaults.
 
 `streamIdleTimeoutMs` bounds each outstanding provider read, including the initial `fetch`, without counting time the consumer spends between chunks. One stable abort signal reaches the request and body reader for the whole call; expiry stops the transport and throws `LlmError('TIMEOUT')`, while an earlier caller abort throws `LlmError('ABORTED')`. The adapter makes exactly one provider request per `stream()` call; agent-level retry is a separate plugin policy.
 
@@ -79,7 +79,7 @@ Reasoning, text, and raw-string tool arguments are translated into harness chunk
 
 #### Token effect
 
-Generated tokens follow provider thinking and effort settings plus the request's `maxTokens`; only loop-retained blocks affect later input.
+Generated tokens follow the request's logged reasoning effort and `maxTokens`; only loop-retained blocks affect later input.
 
 #### KV Cache effect
 
diff --git a/packages/llm/llm-deepseek/src/adapter.ts b/packages/llm/llm-deepseek/src/adapter.ts
index 64faa3b725..d8ca534c89 100644
--- a/packages/llm/llm-deepseek/src/adapter.ts
+++ b/packages/llm/llm-deepseek/src/adapter.ts
@@ -5,11 +5,12 @@
  * @module dsh-llm-deepseek/adapter
  */
 
-import { attributionHeaders, CONTEXT_WINDOW_EXCEEDED_CODE, isContextWindowExceededError, isQuotaExceededError, LlmAdapter, LlmError, ProviderRequestId, QUOTA_EXCEEDED_CODE } from '@deepseek-ai/dsh-llm'
+import { attributionHeaders, CONTEXT_WINDOW_EXCEEDED_CODE, isContextWindowExceededError, isQuotaExceededError, LlmAdapter, LlmError, ProviderRequestId, QUOTA_EXCEEDED_CODE, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type {
   GenerateOptions,
   LlmModelContext,
   LlmModelInfo,
+  LlmModelReasoningInfo,
   LlmProviderInfo,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
@@ -51,6 +52,12 @@ export interface DeepSeekAdapterOptions {
 /** Default maximum idle interval while an adapter stream read is outstanding. */
 export const DEFAULT_STREAM_IDLE_TIMEOUT_MS = 300_000
 const STREAM_IDLE_TIMEOUT_CODE = 'LLM_STREAM_IDLE_TIMEOUT'
+const HIGH_REASONING_EFFORT = ReasoningEffortId('high')
+const MAX_REASONING_EFFORT = ReasoningEffortId('max')
+const REASONING_EFFORTS = [
+  { id: HIGH_REASONING_EFFORT, name: 'High' },
+  { id: MAX_REASONING_EFFORT, name: 'Max' },
+] as const
 
 function providerRetryAfterMs(value: string | null): number | undefined {
   if (value === null) return undefined
@@ -98,6 +105,9 @@ export class DeepSeekAdapter extends LlmAdapter {
 
   constructor(private readonly options: DeepSeekAdapterOptions) {
     super()
+    if (options.defaults?.thinking === 'disabled' && options.defaults.reasoningEffort !== undefined) {
+      throw new Error('llm-deepseek: reasoningEffort cannot be configured when thinking is disabled')
+    }
     if (options.defaultContextWindow !== undefined
       && (!Number.isInteger(options.defaultContextWindow) || options.defaultContextWindow <= 0)) {
       throw new Error('llm-deepseek: defaultContextWindow must be a positive integer')
@@ -134,6 +144,19 @@ export class DeepSeekAdapter extends LlmAdapter {
     return Promise.resolve(contextWindow === undefined ? undefined : { contextWindow })
   }
 
+  override resolveModelReasoning(
+    _provider: string,
+    _model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    if (this.options.defaults?.thinking === 'disabled') return Promise.resolve(undefined)
+    return Promise.resolve({
+      efforts: REASONING_EFFORTS,
+      defaultEffort: this.options.defaults?.reasoningEffort === 'max'
+        ? MAX_REASONING_EFFORT
+        : HIGH_REASONING_EFFORT,
+    })
+  }
+
   async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
     const consumer = new AbortController()
     const upstream = options.signal === undefined
diff --git a/packages/llm/llm-deepseek/src/index.ts b/packages/llm/llm-deepseek/src/index.ts
index 66828fc954..1dad768440 100644
--- a/packages/llm/llm-deepseek/src/index.ts
+++ b/packages/llm/llm-deepseek/src/index.ts
@@ -28,8 +28,9 @@ const DEFAULT_MODELS: DeepSeekCatalogModel[] = [
 /**
  * Plugin config, validated by the same-named schemastery schema. Every field
  * is optional in yml: credentials/endpoint fall back to the environment (a
- * missing API key fails plugin load, not the first call), and omitted
- * thinking fields send nothing on the wire, so the provider default applies.
+ * missing API key fails plugin load, not the first call), omitted thinking
+ * mode uses the provider default, and omitted reasoning effort resolves to
+ * `high`.
  */
 export interface Config {
   /** API key; falls back to $DEEPSEEK_API_KEY. Required one way or the other. */
@@ -38,7 +39,7 @@ export interface Config {
   baseURL?: string
   /** Thinking-mode default for every request (provider default: enabled). */
   thinking?: 'enabled' | 'disabled'
-  /** Thinking effort (only meaningful with thinking enabled). */
+  /** Default thinking effort when thinking is enabled (default `high`). */
   reasoningEffort?: 'high' | 'max'
   /** Positive context capacity used when the selected model has no exact value. */
   defaultContextWindow?: number
@@ -94,6 +95,9 @@ function resolveModels(models: readonly DeepSeekCatalogModel[] | undefined): Dee
 }
 
 export function apply(ctx: Context, config: Config): void {
+  if (config.thinking === 'disabled' && config.reasoningEffort !== undefined) {
+    throw new Error('llm-deepseek: reasoningEffort cannot be configured when thinking is disabled')
+  }
   const apiKey = config.apiKey ?? process.env.DEEPSEEK_API_KEY
   if (apiKey === undefined || apiKey.length === 0) {
     throw new Error('llm-deepseek: an API key is required (Config.apiKey or $DEEPSEEK_API_KEY)')
diff --git a/packages/llm/llm-deepseek/src/serialize.ts b/packages/llm/llm-deepseek/src/serialize.ts
index bf9515c942..70d51c8549 100644
--- a/packages/llm/llm-deepseek/src/serialize.ts
+++ b/packages/llm/llm-deepseek/src/serialize.ts
@@ -6,6 +6,7 @@
  * @module dsh-llm-deepseek/serialize
  */
 
+import { LlmError } from '@deepseek-ai/dsh-llm'
 import type { ContentBlock, GenerateOptions, Message } from '@deepseek-ai/dsh-llm'
 import type { WireMessage, WireRequest, WireTool } from './types.ts'
 
@@ -15,6 +16,17 @@ export interface RequestDefaults {
   reasoningEffort?: 'high' | 'max' | undefined
 }
 
+/** Validate the adapter-owned effort before assigning its narrower wire type. */
+function reasoningEffort(options: GenerateOptions): 'high' | 'max' | undefined {
+  const effort = options.reasoningEffort
+  if (effort === undefined) return undefined
+  if (effort === 'high' || effort === 'max') return effort as 'high' | 'max'
+  throw new LlmError(
+    `DeepSeek does not support reasoning effort "${effort}"`,
+    'UNSUPPORTED_REASONING_EFFORT',
+  )
+}
+
 /** Join the text blocks of a message (used for user/tool-result content). */
 function flattenText(blocks: ContentBlock[]): string {
   return blocks
@@ -121,7 +133,9 @@ export function serializeRequest(options: GenerateOptions, defaults: RequestDefa
   // A short title budget must produce visible text; conversation and
   // compaction calls continue to inherit the adapter's thinking defaults.
   const thinking = options.purpose === 'session-title' ? 'disabled' : defaults.thinking
-  const reasoningEffort = options.purpose === 'session-title' ? undefined : defaults.reasoningEffort
+  const resolvedReasoningEffort = options.purpose === 'session-title'
+    ? undefined
+    : reasoningEffort(options)
 
   return {
     model: options.model,
@@ -129,7 +143,7 @@ export function serializeRequest(options: GenerateOptions, defaults: RequestDefa
     stream: true,
     stream_options: { include_usage: true },
     ...thinking !== undefined ? { thinking: { type: thinking } } : {},
-    ...reasoningEffort !== undefined ? { reasoning_effort: reasoningEffort } : {},
+    ...resolvedReasoningEffort !== undefined ? { reasoning_effort: resolvedReasoningEffort } : {},
     ...tools !== undefined && tools.length > 0 ? { tools } : {},
     ...options.temperature !== undefined ? { temperature: options.temperature } : {},
     ...options.maxTokens !== undefined ? { max_tokens: options.maxTokens } : {},
diff --git a/packages/llm/llm-deepseek/tests/adapter.e2e.ts b/packages/llm/llm-deepseek/tests/adapter.e2e.ts
index e02476eaef..8897d500d5 100644
--- a/packages/llm/llm-deepseek/tests/adapter.e2e.ts
+++ b/packages/llm/llm-deepseek/tests/adapter.e2e.ts
@@ -1,6 +1,6 @@
 import { afterEach, describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
-import LlmService, { CallId } from '@deepseek-ai/dsh-llm'
+import LlmService, { CallId, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type { Message, ToolSchema } from '@deepseek-ai/dsh-llm'
 import * as LlmDeepSeek from '@deepseek-ai/dsh-llm-deepseek'
 import type { Config } from '@deepseek-ai/dsh-llm-deepseek'
@@ -80,11 +80,12 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-deepseek e2e (real API)', ()
   it.each(['high', 'max'] as const)(
     'pro + thinking enabled (effort %s): tool-call round trip with reasoning passback',
     async (effort) => {
-      const ctx = await harness(PRO, { thinking: 'enabled', reasoningEffort: effort })
+      const ctx = await harness(PRO, { thinking: 'enabled' })
 
       // Turn 1: the model must call the tool (and think before it).
       const first = await assemble(ctx,{
         model: PRO,
+        reasoningEffort: ReasoningEffortId(effort),
         messages: ask('What is the weather in Paris right now? Use the get_weather tool.'),
         tools: [weatherTool],
         maxTokens: 2000,
@@ -99,6 +100,7 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-deepseek e2e (real API)', ()
       // block in history (the official thinking+tools passback rule).
       const second = await assemble(ctx,{
         model: PRO,
+        reasoningEffort: ReasoningEffortId(effort),
         messages: [
           ...ask('What is the weather in Paris right now? Use the get_weather tool.'),
           { role: 'assistant', content: first.message.content },
diff --git a/packages/llm/llm-deepseek/tests/adapter.spec.ts b/packages/llm/llm-deepseek/tests/adapter.spec.ts
index 145017ea3d..7d3d8f9f08 100644
--- a/packages/llm/llm-deepseek/tests/adapter.spec.ts
+++ b/packages/llm/llm-deepseek/tests/adapter.spec.ts
@@ -8,6 +8,7 @@ import LlmService, {
   LlmError,
   ProviderRequestId,
   QUOTA_EXCEEDED_CODE,
+  ReasoningEffortId,
   userAgent,
 } from '@deepseek-ai/dsh-llm'
 import { MAX_TIMER_DELAY_MS } from '@deepseek-ai/dsh-timeout'
@@ -120,6 +121,7 @@ describe('DeepSeekAdapter against a mock server', () => {
     // The wire request carried the auth header contents we configured.
     expect(server.requests[0]).toMatchObject({
       model: 'deepseek-v4-flash',
+      reasoning_effort: 'high',
       stream: true,
       stream_options: { include_usage: true },
     })
@@ -173,9 +175,35 @@ describe('DeepSeekAdapter against a mock server', () => {
     expect(server.headers[0]?.['x-deepseek-harness-compact']).toBe('1')
   })
 
-  it('forwards thinking config onto the wire', async () => {
+  it('forwards the configured reasoning default and a dynamic request override', async () => {
+    const server = await mockServer([
+      { kind: 'sse', events: textEvents },
+      { kind: 'sse', events: textEvents },
+    ])
+    const ctx = await harness(server.url, { thinking: 'enabled', reasoningEffort: 'max' })
+
+    await assemble(ctx,{
+      model: 'deepseek-v4-flash',
+      messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
+    })
+    await assemble(ctx,{
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('high'),
+      messages: [{ role: 'user', content: [{ type: 'text', text: 'hi again' }] }],
+    })
+    expect(server.requests[0]).toMatchObject({
+      thinking: { type: 'enabled' },
+      reasoning_effort: 'max',
+    })
+    expect(server.requests[1]).toMatchObject({
+      thinking: { type: 'enabled' },
+      reasoning_effort: 'high',
+    })
+  })
+
+  it('omits reasoning capability and effort when thinking is disabled', async () => {
     const server = await mockServer([{ kind: 'sse', events: textEvents }])
-    const ctx = await harness(server.url, { thinking: 'disabled', reasoningEffort: 'high' })
+    const ctx = await harness(server.url, { thinking: 'disabled' })
 
     await assemble(ctx,{
       model: 'deepseek-v4-flash',
@@ -183,8 +211,22 @@ describe('DeepSeekAdapter against a mock server', () => {
     })
     expect(server.requests[0]).toMatchObject({
       thinking: { type: 'disabled' },
-      reasoning_effort: 'high',
     })
+    expect(server.requests[0]).not.toHaveProperty('reasoning_effort')
+    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+      .resolves.toBeUndefined()
+  })
+
+  it('rejects a per-request effort before I/O when thinking is disabled', async () => {
+    const server = await mockServer([])
+    const ctx = await harness(server.url, { thinking: 'disabled' })
+
+    await expect(assemble(ctx, {
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('high'),
+      messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
+    })).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+    expect(server.requests).toHaveLength(0)
   })
 
   it.each([
@@ -533,6 +575,34 @@ describe('plugin registration and config', () => {
     ])
     await expect(ctx.llm.resolveModelContext('deepseek', 'deepseek-v4-flash'))
       .resolves.toEqual({ contextWindow: 128_000 })
+    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+      .resolves.toEqual({
+        efforts: [
+          { id: ReasoningEffortId('high'), name: 'High' },
+          { id: ReasoningEffortId('max'), name: 'Max' },
+        ],
+        defaultEffort: ReasoningEffortId('high'),
+      })
+  })
+
+  it('rejects a configured reasoning effort when thinking is disabled', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    await expect(ctx.plugin(LlmDeepSeek, {
+      apiKey: 'k',
+      baseURL: 'http://127.0.0.1:1',
+      thinking: 'disabled',
+      reasoningEffort: 'high',
+    })).rejects.toThrow(/reasoningEffort cannot be configured/)
+    expect(ctx.llm.listProviders()).toEqual([])
+  })
+
+  it('rejects a disabled-thinking effort at the direct constructor boundary', () => {
+    expect(() => new DeepSeekAdapter({
+      apiKey: 'k',
+      baseURL: 'http://127.0.0.1:1',
+      defaults: { thinking: 'disabled', reasoningEffort: 'high' },
+    })).toThrow(/reasoningEffort cannot be configured/)
   })
 
   it('uses the default model catalog when apply is called directly', async () => {
diff --git a/packages/llm/llm-deepseek/tests/serialize.spec.ts b/packages/llm/llm-deepseek/tests/serialize.spec.ts
index 566c71b7f2..0f294a5e3a 100644
--- a/packages/llm/llm-deepseek/tests/serialize.spec.ts
+++ b/packages/llm/llm-deepseek/tests/serialize.spec.ts
@@ -1,5 +1,5 @@
 import { describe, expect, it } from 'vitest'
-import { CallId } from '@deepseek-ai/dsh-llm'
+import { CallId, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type { ContentBlock, GenerateOptions, Message } from '@deepseek-ai/dsh-llm'
 import { serializeMessages, serializeRequest } from '../src/serialize.ts'
 
@@ -174,15 +174,22 @@ describe('serializeRequest', () => {
     expect(wire.tools).toBeUndefined()
   })
 
-  it('applies adapter defaults for thinking and effort', () => {
-    const wire = serializeRequest(request({ messages: history }), { thinking: 'enabled', reasoningEffort: 'max' })
+  it('maps adapter-default thinking and the request reasoning effort', () => {
+    const wire = serializeRequest(
+      request({ messages: history, reasoningEffort: ReasoningEffortId('max') }),
+      { thinking: 'enabled', reasoningEffort: 'high' },
+    )
     expect(wire.thinking).toEqual({ type: 'enabled' })
     expect(wire.reasoning_effort).toBe('max')
   })
 
   it('disables thinking for session-title requests without changing adapter defaults', () => {
     const wire = serializeRequest(
-      request({ messages: history, purpose: 'session-title' }),
+      request({
+        messages: history,
+        purpose: 'session-title',
+        reasoningEffort: ReasoningEffortId('max'),
+      }),
       { thinking: 'enabled', reasoningEffort: 'max' },
     )
     expect(wire.thinking).toEqual({ type: 'disabled' })
@@ -194,6 +201,13 @@ describe('serializeRequest', () => {
     expect(wire.thinking).toBeUndefined()
     expect(wire.reasoning_effort).toBeUndefined()
   })
+
+  it('rejects an effort outside the DeepSeek capability', () => {
+    expect(() => serializeRequest(request({
+      messages: history,
+      reasoningEffort: ReasoningEffortId('medium'),
+    }))).toThrow(expect.objectContaining({ code: 'UNSUPPORTED_REASONING_EFFORT' }))
+  })
 })
 
 describe('review fixes: assistant content shapes', () => {
diff --git a/packages/llm/llm-pi-ai/README.md b/packages/llm/llm-pi-ai/README.md
index 8a6736f112..d32f06a9a1 100644
--- a/packages/llm/llm-pi-ai/README.md
+++ b/packages/llm/llm-pi-ai/README.md
@@ -30,6 +30,8 @@ Each provider name must exist in pi-ai's installed catalog and may appear only o
 
 The adapter exposes each configured provider's installed pi-ai models through `ctx.llm.listModels(provider)`. This is provider-neutral selector metadata derived from `getModels(provider)`; request-time resolution still performs the authoritative catalog lookup, so discovery does not create a second model registry. `ctx.llm.resolveModelContext(provider, model)` performs the same exact descriptor lookup and returns its context window, keeping capacity metadata on the route-owning adapter rather than a consuming plugin.
 
+`ctx.llm.resolveModelReasoning(provider, model)` uses pi-ai's `getSupportedThinkingLevels(model)` and returns that model's ordered levels after filtering the separate `off` control. The Harness exposes the canonical pi-ai level as an opaque ID; provider/model wire spellings remain inside pi-ai's `thinkingLevelMap`. A non-reasoning model returns `undefined`. The profile `reasoning` value is the deployment default when configured; omitting it preserves the provider default. Per-request `GenerateOptions.reasoningEffort` takes precedence, and any explicit value absent from the exact model capability fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O instead of being clamped.
+
 Supported profile fields are `provider`, `apiKey`, `baseURL`, `headers`, `reasoning`, `thinkingBudgets`, `cacheRetention`, `transport`, `timeoutMs`, `websocketConnectTimeoutMs`, and `streamIdleTimeoutMs`. The stream-idle interval is a positive finite Node timer delay, defaults to five minutes, and covers only an outstanding provider read, not consumer think time. Harness app attribution wins a conflicting configured header name.
 
 The adapter forces pi-ai's SDK `maxRetries` to zero so one `stream()` call makes one provider request. The removed profile fields `maxRetries` and `maxRetryDelayMs` fail load instead of silently multiplying or hiding the separately composed agent-level retry budget. Idle expiry aborts the SDK's stable request signal and surfaces `TIMEOUT`; an earlier caller abort remains `ABORTED`.
@@ -47,6 +49,7 @@ If a listener rewrites assembled assistant content, the loop drops replay state
 - pi-ai tool-call arguments are parsed objects; the harness stores raw JSON strings. The adapter parses input and re-stringifies output.
 - pi-ai reports failures as in-stream error events; these map to `finish {kind:'error'|'aborted', failure}` chunks. Provider-specific error text distinguishes terminal `QUOTA` from transient `RATE_LIMIT`, while text and usage signals evaluated against the resolved model's context window normalize overflow to `CONTEXT_WINDOW_EXCEEDED`.
 - pi-ai folds reasoning tokens into output usage; there is no separate reasoning count to map.
+- pi-ai may internally support an `off` thinking level, but the Harness reasoning-effort capability deliberately excludes mode changes.
 - `GenerateOptions.stop` is rejected with `UNSUPPORTED_OPTION` because pi-ai's common streaming surface cannot guarantee it across providers.
 
 ## App attribution
diff --git a/packages/llm/llm-pi-ai/src/adapter.ts b/packages/llm/llm-pi-ai/src/adapter.ts
index ed0fb9fae4..1dde9d9dc7 100644
--- a/packages/llm/llm-pi-ai/src/adapter.ts
+++ b/packages/llm/llm-pi-ai/src/adapter.ts
@@ -7,13 +7,27 @@
 import { streamSimple } from '@earendil-works/pi-ai/compat'
 import { getBuiltinModels } from '@earendil-works/pi-ai/providers/all'
 import type { BuiltinProvider } from '@earendil-works/pi-ai/providers/all'
+import { getSupportedThinkingLevels } from '@earendil-works/pi-ai'
 import type {
   Api,
   Model,
   SimpleStreamOptions,
+  ThinkingLevel,
 } from '@earendil-works/pi-ai'
-import { attributionHeaders, LlmAdapter, LlmError } from '@deepseek-ai/dsh-llm'
-import type { GenerateOptions, LlmModelContext, LlmModelInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
+import {
+  attributionHeaders,
+  LlmAdapter,
+  LlmError,
+  ReasoningEffortId,
+} from '@deepseek-ai/dsh-llm'
+import type {
+  GenerateOptions,
+  LlmModelContext,
+  LlmModelInfo,
+  LlmModelReasoningInfo,
+  ReasoningEffortId as ReasoningEffortIdType,
+  StreamChunk,
+} from '@deepseek-ai/dsh-llm'
 import { idleWatchdog, timeoutOf } from '@deepseek-ai/dsh-timeout'
 import { resolveProfiles } from './config.ts'
 import type { PiAiProviderProfile, ResolvedPiAiProviderProfile } from './config.ts'
@@ -39,10 +53,13 @@ function resolveModel(profile: PiAiProviderProfile, modelId: string): Model<Api>
 }
 
 /** Copy profile stream knobs into pi-ai's common option vocabulary. */
-function profileOptions(profile: PiAiProviderProfile): SimpleStreamOptions {
+function profileOptions(
+  profile: PiAiProviderProfile,
+  reasoning: ThinkingLevel | undefined,
+): SimpleStreamOptions {
   return {
     ...profile.apiKey === undefined ? {} : { apiKey: profile.apiKey },
-    ...profile.reasoning === undefined ? {} : { reasoning: profile.reasoning },
+    ...reasoning === undefined ? {} : { reasoning },
     ...profile.thinkingBudgets === undefined ? {} : { thinkingBudgets: profile.thinkingBudgets },
     ...profile.cacheRetention === undefined ? {} : { cacheRetention: profile.cacheRetention },
     ...profile.transport === undefined ? {} : { transport: profile.transport },
@@ -53,6 +70,25 @@ function profileOptions(profile: PiAiProviderProfile): SimpleStreamOptions {
   }
 }
 
+/** Selectable pi-ai levels exclude the separate on/off control. */
+function supportedReasoningLevels(model: Model<Api>): ThinkingLevel[] {
+  return getSupportedThinkingLevels(model).filter((level): level is ThinkingLevel => level !== 'off')
+}
+
+/** Validate an explicit Harness/profile effort without invoking pi-ai's clamp. */
+function resolveReasoningLevel(
+  model: Model<Api>,
+  effort: ReasoningEffortIdType | ThinkingLevel | undefined,
+): ThinkingLevel | undefined {
+  if (effort === undefined) return undefined
+  const supported = supportedReasoningLevels(model)
+  if (supported.some(level => level === effort)) return effort as ThinkingLevel
+  throw new LlmError(
+    `pi-ai provider "${model.provider}" model "${model.id}" does not support reasoning effort "${effort}"`,
+    'UNSUPPORTED_REASONING_EFFORT',
+  )
+}
+
 /** Merge deployment headers while removing case-insensitive attribution collisions. */
 function requestHeaders(headers: Readonly<Record<string, string>> | undefined): Record<string, string> {
   const attribution = attributionHeaders()
@@ -103,6 +139,37 @@ export class PiAiAdapter extends LlmAdapter {
     }))
   }
 
+  override resolveModelReasoning(
+    provider: string,
+    model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    const profile = this.profiles.get(provider)
+    if (profile === undefined) {
+      return Promise.reject(new LlmError(
+        `pi-ai adapter does not own provider "${provider}"`,
+        'NO_ADAPTER',
+      ))
+    }
+    return Promise.resolve().then(() => {
+      const resolvedModel = resolveModel(profile, model)
+      const levels = supportedReasoningLevels(resolvedModel)
+      if (levels.length === 0) {
+        resolveReasoningLevel(resolvedModel, profile.reasoning)
+        return undefined
+      }
+      const defaultLevel = resolveReasoningLevel(resolvedModel, profile.reasoning)
+      return {
+        efforts: levels.map(level => ({
+          id: ReasoningEffortId(level),
+          name: `${level.charAt(0).toUpperCase()}${level.slice(1)}`,
+        })),
+        ...defaultLevel === undefined
+          ? {}
+          : { defaultEffort: ReasoningEffortId(defaultLevel) },
+      }
+    })
+  }
+
   async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
     if (options.stop !== undefined) {
       throw new LlmError('llm-pi-ai does not support GenerateOptions.stop', 'UNSUPPORTED_OPTION')
@@ -112,6 +179,10 @@ export class PiAiAdapter extends LlmAdapter {
       throw new LlmError(`pi-ai adapter does not own provider "${options.provider}"`, 'NO_ADAPTER')
     }
     const model = resolveModel(profile, options.model)
+    const reasoning = resolveReasoningLevel(
+      model,
+      options.reasoningEffort ?? profile.reasoning,
+    )
 
     const consumer = new AbortController()
     const upstream = options.signal === undefined
@@ -122,7 +193,7 @@ export class PiAiAdapter extends LlmAdapter {
 
     try {
       const events = streamSimple(model, toPiContext(options), {
-        ...profileOptions(profile),
+        ...profileOptions(profile, reasoning),
         ...options.temperature === undefined ? {} : { temperature: options.temperature },
         ...options.maxTokens === undefined ? {} : { maxTokens: options.maxTokens },
         ...options.sessionId === undefined ? {} : { sessionId: String(options.sessionId) },
diff --git a/packages/llm/llm-pi-ai/tests/adapter.e2e.ts b/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
index 77d2cc81dd..e91a951d85 100644
--- a/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
+++ b/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
@@ -1,6 +1,6 @@
 import { afterEach, describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
-import LlmService, { CallId } from '@deepseek-ai/dsh-llm'
+import LlmService, { CallId, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type { Message, ToolSchema } from '@deepseek-ai/dsh-llm'
 import * as LlmPiAi from '@deepseek-ai/dsh-llm-pi-ai'
 import type { PiAiProviderProfile } from '@deepseek-ai/dsh-llm-pi-ai'
@@ -9,7 +9,7 @@ import { assemble, type AssembledResult } from './assemble.ts'
 
 /**
  * Real-API e2e for the pi-ai-backed adapter: V4 Flash + V4 Pro with provider
- * defaults and representative high/xhigh reasoning. Mirrors the native
+ * defaults and representative high/max reasoning. Mirrors the native
  * adapter's StreamChunk contract and exercises a replayed tool follow-up.
  * Key-gated.
  */
@@ -75,9 +75,10 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-pi-ai e2e (real API)', () =>
   })
 
   it.each([FLASH, PRO])('%s + reasoning high: reasoning blocks present', async (model) => {
-    const ctx = await harness(model, { reasoning: 'high' })
+    const ctx = await harness(model)
     const result = await assemble(ctx,{
       model,
+      reasoningEffort: ReasoningEffortId('high'),
       messages: ask('Which is larger, 9.11 or 9.8? Answer with just the number.'),
       maxTokens: 2000,
     })
@@ -86,11 +87,12 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-pi-ai e2e (real API)', () =>
     expect(textOf(result)).toContain('9.8')
   })
 
-  it('pro + reasoning xhigh (wire max): tool-call round trip', async () => {
-    const ctx = await harness(PRO, { reasoning: 'xhigh' })
+  it('pro + reasoning max: tool-call round trip', async () => {
+    const ctx = await harness(PRO)
 
     const first = await assemble(ctx,{
       model: PRO,
+      reasoningEffort: ReasoningEffortId('max'),
       messages: ask('What is the weather in Paris right now? Use the get_weather tool.'),
       tools: [weatherTool],
       maxTokens: 2000,
@@ -103,6 +105,7 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-pi-ai e2e (real API)', () =>
 
     const second = await assemble(ctx,{
       model: PRO,
+      reasoningEffort: ReasoningEffortId('max'),
       messages: [
         ...ask('What is the weather in Paris right now? Use the get_weather tool.'),
         first.message,
diff --git a/packages/llm/llm-pi-ai/tests/adapter.spec.ts b/packages/llm/llm-pi-ai/tests/adapter.spec.ts
index f37b07f624..0bf8d34da7 100644
--- a/packages/llm/llm-pi-ai/tests/adapter.spec.ts
+++ b/packages/llm/llm-pi-ai/tests/adapter.spec.ts
@@ -2,7 +2,7 @@ import { createServer } from 'node:http'
 import type { IncomingMessage, Server, ServerResponse } from 'node:http'
 import { afterEach, describe, expect, it, vi } from 'vitest'
 import { Context } from 'cordis'
-import LlmService, { CONTEXT_WINDOW_EXCEEDED_CODE, LlmError, userAgent } from '@deepseek-ai/dsh-llm'
+import LlmService, { CONTEXT_WINDOW_EXCEEDED_CODE, LlmError, ReasoningEffortId, userAgent } from '@deepseek-ai/dsh-llm'
 import * as LlmPiAi from '@deepseek-ai/dsh-llm-pi-ai'
 import { PiAiAdapter } from '@deepseek-ai/dsh-llm-pi-ai'
 import { MAX_TIMER_DELAY_MS } from '@deepseek-ai/dsh-timeout'
@@ -124,7 +124,7 @@ describe('PiAiAdapter provider routing', () => {
   it('forwards common stream options and profile reasoning', async () => {
     const server = await mockServer([{ events: textEvents }])
     const ctx = await harness(server.url, {
-      reasoning: 'xhigh',
+      reasoning: 'max',
       cacheRetention: 'none',
       transport: 'sse',
       timeoutMs: 5000,
@@ -148,6 +148,25 @@ describe('PiAiAdapter provider routing', () => {
     })
   })
 
+  it('uses a dynamic request effort and rejects unsupported efforts before network I/O', async () => {
+    const server = await mockServer([{ events: textEvents }])
+    const ctx = await harness(server.url, { reasoning: 'max' })
+
+    await assemble(ctx, {
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('high'),
+      messages: [],
+    })
+    expect(server.requests[0]).toMatchObject({ reasoning_effort: 'high' })
+
+    await expect(assemble(ctx, {
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('xhigh'),
+      messages: [],
+    })).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+    expect(server.requests).toHaveLength(1)
+  })
+
   it('preserves omitted profile options when constructing the adapter directly', async () => {
     const server = await mockServer([{ events: textEvents }])
     const ctx = new Context()
@@ -327,6 +346,51 @@ describe('provider profile lifecycle', () => {
     expect(typeof context?.contextWindow).toBe('number')
   })
 
+  it('exposes model-specific reasoning levels without off or an invented provider default', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    await ctx.plugin(LlmPiAi, {
+      providers: [{ provider: 'deepseek' }, { provider: 'openai' }],
+    })
+
+    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+      .resolves.toEqual({
+        efforts: [
+          { id: ReasoningEffortId('high'), name: 'High' },
+          { id: ReasoningEffortId('max'), name: 'Max' },
+        ],
+      })
+    const extended = await ctx.llm.resolveModelReasoning('openai', 'gpt-5.6-sol')
+    expect(extended?.efforts.map(effort => effort.id)).toEqual([
+      ReasoningEffortId('minimal'),
+      ReasoningEffortId('low'),
+      ReasoningEffortId('medium'),
+      ReasoningEffortId('high'),
+      ReasoningEffortId('xhigh'),
+      ReasoningEffortId('max'),
+    ])
+    await expect(ctx.llm.resolveModelReasoning('openai', 'gpt-4.1'))
+      .resolves.toBeUndefined()
+  })
+
+  it('uses a supported profile reasoning value as the model default and rejects an unsupported one', async () => {
+    const supported = new Context()
+    await supported.plugin(LlmService)
+    await supported.plugin(LlmPiAi, {
+      providers: [{ provider: 'deepseek', reasoning: 'max' }],
+    })
+    await expect(supported.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+      .resolves.toMatchObject({ defaultEffort: ReasoningEffortId('max') })
+
+    const unsupported = new Context()
+    await unsupported.plugin(LlmService)
+    await unsupported.plugin(LlmPiAi, {
+      providers: [{ provider: 'deepseek', reasoning: 'medium' }],
+    })
+    await expect(unsupported.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+      .rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+  })
+
   it('accepts absent credentials for pi-ai ambient authentication', async () => {
     vi.stubEnv('DEEPSEEK_API_KEY', 'ambient-key')
     const server = await mockServer([{ events: textEvents }])
@@ -378,6 +442,8 @@ describe('provider profile lifecycle', () => {
     await expect(adapter.listModels('anthropic')).rejects.toMatchObject({ code: 'NO_ADAPTER' })
     await expect(adapter.resolveModelContext('anthropic', 'claude-sonnet-4'))
       .rejects.toMatchObject({ code: 'NO_ADAPTER' })
+    await expect(adapter.resolveModelReasoning('anthropic', 'claude-sonnet-4'))
+      .rejects.toMatchObject({ code: 'NO_ADAPTER' })
     await expect(adapter.resolveModelContext('openai', 'not-a-catalog-model'))
       .rejects.toMatchObject({ code: 'UNKNOWN_MODEL' })
     await expect((async () => {
diff --git a/packages/llm/llm/README.md b/packages/llm/llm/README.md
index beac6d5e8e..d15ce7f0ac 100644
--- a/packages/llm/llm/README.md
+++ b/packages/llm/llm/README.md
@@ -12,6 +12,8 @@ An adapter registry plus a single streaming call surface, interceptable via a wa
 - `ctx.llm.listProviders(): LlmProviderInfo[]` Describe registered provider routes in registration order.
 - `ctx.llm.listModels(provider: string): Promise<LlmModelInfo[]>` Discover the models one registered provider currently advertises.
 - `ctx.llm.resolveModelContext(provider: string, model: string): Promise<LlmModelContext | undefined>` Resolve authoritative context capacity for one exact route from its owning adapter.
+- `ctx.llm.resolveModelReasoning(provider: string, model: string): Promise<LlmModelReasoningInfo | undefined>` Resolve ordered adapter-owned reasoning efforts and an optional deployment default for one exact route.
+- `ctx.llm.resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>` Validate an explicit effort and materialize an adapter-configured default without clamping.
 - `ctx.llm.stream(options: GenerateOptions): AsyncIterable<StreamChunk>` Stream one model call as raw chunks (token-level deltas). Consumers assemble the chunks into blocks/messages with `BlockAssembler`.
 
 `LlmService` preserves errors from final adapter selection, synchronous dispatch, iterator construction, and iteration, and binds their provenance to the exact stream handle returned for that model call. `isLlmAdapterFailure(stream, value)` reports only errors from that call's final adapter boundary; `llmFailureOf(stream, value)` returns the adjacent immutable `LlmFailure`. Nested model calls, `llm/stream` middleware, and downstream consumer failures remain unclassified for the outer call. Classification never replaces or mutates the adapter's original coded `Error`.
@@ -20,6 +22,8 @@ Provider and model metadata is a discovery surface, not a routing whitelist. `re
 
 Context capacity is a separate correctness query, not a catalog decoration or global LLM setting. `resolveModelContext()` asks the adapter that owns the exact provider/model route; an adapter can describe an unlisted dynamic model, and `undefined` means only that capacity is unavailable. Invalid returned capacity fails with `INVALID_MODEL_CONTEXT`.
 
+Reasoning effort is also an exact-route capability, but its identifiers are opaque adapter-owned strings rather than a core enum. `resolveModelReasoning()` validates and detaches the ordered display metadata; `undefined` means the model has no selectable effort. `resolveCallConfig()` accepts only an exact advertised identifier, materializes `defaultEffort` when present, and otherwise preserves the provider default. Invalid capability metadata fails with `INVALID_MODEL_REASONING`; an unsupported explicit or configured effort fails with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+
 ### Events
 
 | Event | Mode | Purpose |
@@ -28,7 +32,7 @@ Context capacity is a separate correctness query, not a catalog decoration or gl
 
 ### Extension points
 
-- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, and `resolveModelContext()` when exact capacity is known; the defaults use the route id as its name, advertise no models, and return no capacity.
+- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, `resolveModelContext()` when exact capacity is known, and `resolveModelReasoning()` when a model exposes selectable efforts; the defaults use the route id as its name, advertise no models, and return neither capacity nor reasoning metadata.
 - Wrap `llm/stream` via `ctx.on()` waterfall listeners for caching, logging, or routing. A wrapper that retries after emitting a chunk has no durable attempt boundary; shipped agent retry policy therefore uses `agent/request-error` instead.
 
 ### Content-block vocabulary (`types.ts`)
@@ -39,7 +43,7 @@ Streaming is a raw chunk protocol (`block-start`, `text-delta`, `reasoning-delta
 
 ### Call configuration (`call-config.ts`)
 
-`LlmCallConfig` is the provider + model + sampling scalars of one conversation's requests (`provider`, `model`, `temperature`, `maxTokens`, `stop` — each mapping 1:1 onto the same-named `GenerateOptions` field). It is per-conversation state recorded in the session log as part of the request header (see the dsh-session `request/header` events), never a silently-adjustable per-call knob: the `agent/request` waterfall proposes a replacement and the loop logs a real change. `callConfigEquals(a, b)` is the field-wise real-change detector; `deepFreeze(value)` is the ownership helper the loop applies to every built request before dispatch (`llm/stream` listeners and adapters read, never rewrite). `markAgentLoopRequest()` gives that exact object process-local loop provenance, and `isAgentLoopRequest()` lets observers distinguish it from independently logged auxiliary calls that may also be frozen and session-associated. `GenerateOptions.purpose` classifies logged auxiliary compaction and session-title calls so adapters can apply purpose-specific transport policy without changing ordinary conversation requests.
+`LlmCallConfig` is the provider, model, optional adapter-owned reasoning effort, and sampling scalars of one conversation's requests (`provider`, `model`, `reasoningEffort`, `temperature`, `maxTokens`, `stop` — each mapping 1:1 onto the same-named `GenerateOptions` field). It is per-conversation state recorded in the session log as part of the request header (see the dsh-session `request/header` events), never a silently-adjustable per-call knob: the `agent/request` waterfall proposes a replacement, `resolveCallConfig()` validates and defaults it, and the loop logs the effective value before dispatch. `callConfigEquals(a, b)` is the field-wise real-change detector; `deepFreeze(value)` is the ownership helper the loop applies to every built request before dispatch (`llm/stream` listeners and adapters read, never rewrite). `markAgentLoopRequest()` gives that exact object process-local loop provenance, and `isAgentLoopRequest()` lets observers distinguish it from independently logged auxiliary calls that may also be frozen and session-associated. `GenerateOptions.purpose` classifies logged auxiliary compaction and session-title calls so adapters can apply purpose-specific transport policy without changing ordinary conversation requests.
 
 ### App attribution (`attribution.ts`)
 
@@ -61,7 +65,7 @@ Two adapters implement `LlmAdapter` on different internals: [`@deepseek-ai/dsh-l
 
 ## Model Experience
 
-None, as this adapter registry forwards an already assembled request without adding or changing any model-bound text, schema, or message.
+None, as the service adds no model-bound text, schema, or message; it only materializes and logs an adapter-configured reasoning effort.
 
 #### KV Cache effect
 
diff --git a/packages/llm/llm/src/brand.ts b/packages/llm/llm/src/brand.ts
index 259dc49bce..0c0190a325 100644
--- a/packages/llm/llm/src/brand.ts
+++ b/packages/llm/llm/src/brand.ts
@@ -38,3 +38,15 @@ export type ProviderRequestId = Branded<'ProviderRequestId'>
 export function ProviderRequestId(id: string): ProviderRequestId {
   return id as ProviderRequestId
 }
+
+/** Adapter-owned identifier for one model's selectable reasoning effort. */
+export type ReasoningEffortId = Branded<'ReasoningEffortId'>
+
+/**
+ * Brand an adapter-owned reasoning-effort identifier.
+ * @param id - the opaque identifier exposed by one model capability.
+ * @returns the same string, branded; no validation is performed.
+ */
+export function ReasoningEffortId(id: string): ReasoningEffortId {
+  return id as ReasoningEffortId
+}
diff --git a/packages/llm/llm/src/call-config.ts b/packages/llm/llm/src/call-config.ts
index fd6ecf9df4..c5247af143 100644
--- a/packages/llm/llm/src/call-config.ts
+++ b/packages/llm/llm/src/call-config.ts
@@ -1,24 +1,27 @@
 /**
  * Conversation call configuration and freeze utilities. Provider routing,
- * model, and sampling values are request-header state that can affect cache
- * reuse; request waterfalls replace them and the loop logs changed snapshots
- * instead of allowing silent per-call drift.
+ * model, reasoning effort, and sampling values are request-header state that
+ * can affect cache reuse; request waterfalls replace them and the loop logs
+ * changed snapshots instead of allowing silent per-call drift.
  * @module dsh-llm/call-config
  */
 
 import type { GenerateOptions } from './types.ts'
+import type { ReasoningEffortId } from './brand.ts'
 
 /** Process-local identities of request objects assembled by dsh-agent-loop. */
 const AGENT_LOOP_REQUESTS = new WeakSet<GenerateOptions>()
 
 /**
- * Provider + model + sampling scalars of one conversation's requests. Every field maps
- * 1:1 onto the same-named `GenerateOptions` field; the loop builds requests
- * from the logged header rather than accepting these per call.
+ * Provider, model, reasoning effort, and sampling scalars of one conversation's
+ * requests. Every field maps 1:1 onto the same-named `GenerateOptions` field;
+ * the loop builds requests from the logged header rather than accepting these
+ * per call.
  */
 export interface LlmCallConfig {
   provider: string
   model: string
+  reasoningEffort?: ReasoningEffortId
   temperature?: number
   maxTokens?: number
   stop?: string[]
@@ -33,7 +36,13 @@ export interface LlmCallConfig {
  * @returns whether every field (including the `stop` list, element-wise) matches.
  */
 export function callConfigEquals(a: LlmCallConfig, b: LlmCallConfig): boolean {
-  if (a.provider !== b.provider || a.model !== b.model || a.temperature !== b.temperature || a.maxTokens !== b.maxTokens) return false
+  if (
+    a.provider !== b.provider
+    || a.model !== b.model
+    || a.reasoningEffort !== b.reasoningEffort
+    || a.temperature !== b.temperature
+    || a.maxTokens !== b.maxTokens
+  ) return false
   if (a.stop === undefined || b.stop === undefined) return a.stop === b.stop
   return a.stop.length === b.stop.length && a.stop.every((s, i) => s === b.stop?.[i])
 }
diff --git a/packages/llm/llm/src/index.ts b/packages/llm/llm/src/index.ts
index 6765833e8c..e56c2d17a6 100644
--- a/packages/llm/llm/src/index.ts
+++ b/packages/llm/llm/src/index.ts
@@ -12,12 +12,14 @@ import type {
   LlmFailure,
   LlmModelContext,
   LlmModelInfo,
+  LlmModelReasoningInfo,
   LlmProviderInfo,
   Message,
   StreamChunk,
 } from './types.ts'
 import type { ProviderRequestId } from './brand.ts'
-import { deepFreeze } from './call-config.ts'
+import { callConfigEquals, deepFreeze } from './call-config.ts'
+import type { LlmCallConfig } from './call-config.ts'
 import { HarnessError } from './error.ts'
 import { bindAdapterFailureScope, markLlmAdapterFailure } from './adapter-failure.ts'
 import type { AdapterFailureScope } from './adapter-failure.ts'
@@ -144,6 +146,20 @@ export abstract class LlmAdapter {
     return Promise.resolve(undefined)
   }
 
+  /**
+   * Resolve selectable reasoning efforts for one exact model. Absence means
+   * the model has no selectable reasoning-effort capability.
+   * @param _provider - one provider route owned by this adapter.
+   * @param _model - exact model id passed to {@link GenerateOptions.model}.
+   * @returns adapter-owned effort metadata, or `undefined` when unsupported.
+   */
+  resolveModelReasoning(
+    _provider: string,
+    _model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    return Promise.resolve(undefined)
+  }
+
   /**
    * Stream one model call as raw chunks. The only required method.
    * @param options - the fully-assembled request; implementations must honor `options.signal`.
@@ -262,6 +278,90 @@ export class LlmService extends Service {
     return { contextWindow: context.contextWindow }
   }
 
+  /**
+   * Resolve selectable reasoning efforts from the adapter that owns one exact
+   * route. Metadata is validated and detached; an absent result means an
+   * effort selector is unsupported for that model.
+   * @param provider - registered provider route to inspect.
+   * @param model - exact model id passed to the adapter.
+   * @returns detached reasoning metadata, or `undefined` when unsupported.
+   */
+  async resolveModelReasoning(
+    provider: string,
+    model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    const reasoning = await this.registration(provider).adapter.resolveModelReasoning(provider, model)
+    if (reasoning === undefined) return undefined
+    if (reasoning.efforts.length === 0) {
+      throw new LlmError(
+        `adapter returned invalid reasoning metadata for provider "${provider}" model "${model}"`,
+        'INVALID_MODEL_REASONING',
+      )
+    }
+    const seen = new Set<string>()
+    const efforts = reasoning.efforts.map((effort) => {
+      if (
+        typeof effort.id !== 'string'
+        || effort.id.length === 0
+        || typeof effort.name !== 'string'
+        || effort.name.length === 0
+        || (effort.description !== undefined && typeof effort.description !== 'string')
+        || seen.has(effort.id)
+      ) {
+        throw new LlmError(
+          `adapter returned invalid or duplicate reasoning effort metadata for provider "${provider}" model "${model}"`,
+          'INVALID_MODEL_REASONING',
+        )
+      }
+      seen.add(effort.id)
+      return {
+        id: effort.id,
+        name: effort.name,
+        ...effort.description === undefined ? {} : { description: effort.description },
+      }
+    })
+    if (reasoning.defaultEffort !== undefined && !seen.has(reasoning.defaultEffort)) {
+      throw new LlmError(
+        `adapter returned an unknown default reasoning effort for provider "${provider}" model "${model}"`,
+        'INVALID_MODEL_REASONING',
+      )
+    }
+    return {
+      efforts,
+      ...reasoning.defaultEffort === undefined ? {} : { defaultEffort: reasoning.defaultEffort },
+    }
+  }
+
+  /**
+   * Validate a conversation call config against its exact model capability and
+   * materialize an adapter-configured default. Unsupported explicit efforts
+   * reject before provider I/O; no clamping or aliasing is performed.
+   * @param config - provider/model route and optional request controls.
+   * @returns a detached config only when a default must be materialized.
+   */
+  async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig> {
+    const reasoning = await this.resolveModelReasoning(config.provider, config.model)
+    const requested = config.reasoningEffort
+    if (reasoning === undefined) {
+      if (requested !== undefined) {
+        throw new LlmError(
+          `provider "${config.provider}" model "${config.model}" does not support reasoning effort "${requested}"`,
+          'UNSUPPORTED_REASONING_EFFORT',
+        )
+      }
+      return config
+    }
+    const effective = requested ?? reasoning.defaultEffort
+    if (effective === undefined) return config
+    if (!reasoning.efforts.some(effort => effort.id === effective)) {
+      throw new LlmError(
+        `provider "${config.provider}" model "${config.model}" does not support reasoning effort "${effective}"`,
+        'UNSUPPORTED_REASONING_EFFORT',
+      )
+    }
+    return requested === effective ? config : { ...config, reasoningEffort: effective }
+  }
+
   private registration(provider: string): { adapter: LlmAdapter; provider: LlmProviderInfo } {
     const registration = this.adapters.get(provider)
     if (!registration) throw new LlmError(`no adapter registered for provider "${provider}"`, 'NO_ADAPTER')
@@ -298,8 +398,14 @@ export class LlmService extends Service {
   ): AsyncGenerator<StreamChunk> {
     let iterator: AsyncIterator<StreamChunk>
     try {
-      const adapter = this.registration(options.provider).adapter
-      const stream = adapter.stream(this.forAdapter(options, adapter))
+      const resolvedConfig = await this.resolveCallConfig(options)
+      const resolvedOptions = callConfigEquals(options, resolvedConfig)
+        ? options
+        : Object.isFrozen(options)
+          ? deepFreeze({ ...options, ...resolvedConfig })
+          : { ...options, ...resolvedConfig }
+      const adapter = this.registration(resolvedOptions.provider).adapter
+      const stream = adapter.stream(this.forAdapter(resolvedOptions, adapter))
       iterator = stream[Symbol.asyncIterator]()
     } catch (error: unknown) {
       throw markLlmAdapterFailure(failures, error)
diff --git a/packages/llm/llm/src/types.ts b/packages/llm/llm/src/types.ts
index f9f97d532a..7882985288 100644
--- a/packages/llm/llm/src/types.ts
+++ b/packages/llm/llm/src/types.ts
@@ -5,7 +5,7 @@
  */
 
 import type { Branded } from '@deepseek-ai/dsh-brand'
-import type { CallId, ProviderRequestId } from './brand.ts'
+import type { CallId, ProviderRequestId, ReasoningEffortId } from './brand.ts'
 
 /** Serializable provider-boundary facts; policy decides whether they are retryable. */
 export interface LlmFailure {
@@ -161,6 +161,27 @@ export interface LlmModelContext {
   contextWindow: number
 }
 
+/** Display metadata for one adapter-owned reasoning effort. */
+export interface LlmReasoningEffortInfo {
+  /** Opaque stable value accepted by {@link GenerateOptions.reasoningEffort}. */
+  id: ReasoningEffortId
+  /** Human-readable effort name for selectors and diagnostics. */
+  name: string
+  /** Optional user-facing distinction from otherwise similar efforts. */
+  description?: string
+}
+
+/** Selectable reasoning efforts for one exact provider/model route. */
+export interface LlmModelReasoningInfo {
+  /** Supported efforts in adapter-preferred display order. */
+  efforts: readonly LlmReasoningEffortInfo[]
+  /**
+   * Adapter-configured default materialized into requests when callers omit
+   * an effort. Absence preserves the provider's own default.
+   */
+  defaultEffort?: ReasoningEffortId
+}
+
 /**
  * Raw streaming protocol emitted by adapters.
  * Block indexes correlate interleaved deltas, and `block-end` carries the
@@ -201,6 +222,8 @@ export interface GenerateOptions {
   /** Registered provider route selecting the adapter instance. */
   provider: string
   model: string
+  /** Adapter-owned reasoning effort selected for this exact model. */
+  reasoningEffort?: ReasoningEffortId
   /**
    * Ordered conversation messages, exactly as the provider sees them (after
    * the `system` slot). A loop-built request assembles them as
diff --git a/packages/llm/llm/tests/call-config.spec.ts b/packages/llm/llm/tests/call-config.spec.ts
index a5426e815f..2237d4639f 100644
--- a/packages/llm/llm/tests/call-config.spec.ts
+++ b/packages/llm/llm/tests/call-config.spec.ts
@@ -6,6 +6,7 @@
 
 import { describe, expect, it } from 'vitest'
 import { callConfigEquals, deepFreeze, isAgentLoopRequest, markAgentLoopRequest } from '../src/call-config.ts'
+import { ReasoningEffortId } from '../src/brand.ts'
 import type { GenerateOptions } from '../src/types.ts'
 
 describe('callConfigEquals', () => {
@@ -14,6 +15,11 @@ describe('callConfigEquals', () => {
     expect(callConfigEquals(base, base)).toBe(true)
     expect(callConfigEquals(base, { provider: 'x', model: 'm' })).toBe(false)
     expect(callConfigEquals(base, { provider: 'p', model: 'x' })).toBe(false)
+    expect(callConfigEquals({ ...base, reasoningEffort: ReasoningEffortId('high') }, base)).toBe(false)
+    expect(callConfigEquals(
+      { ...base, reasoningEffort: ReasoningEffortId('high') },
+      { ...base, reasoningEffort: ReasoningEffortId('high') },
+    )).toBe(true)
     expect(callConfigEquals({ ...base, temperature: 0.5 }, base)).toBe(false)
     expect(callConfigEquals({ ...base, maxTokens: 1 }, { ...base, maxTokens: 2 })).toBe(false)
     expect(callConfigEquals({ ...base, stop: ['a'] }, base)).toBe(false)
diff --git a/packages/llm/llm/tests/service.spec.ts b/packages/llm/llm/tests/service.spec.ts
index 90be1ffcb0..59684ac8b8 100644
--- a/packages/llm/llm/tests/service.spec.ts
+++ b/packages/llm/llm/tests/service.spec.ts
@@ -11,9 +11,15 @@ import LlmService, {
   LlmError,
   llmFailureOf,
   ProviderRequestId,
+  ReasoningEffortId,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
-import type { LlmModelContext, LlmModelInfo, LlmProviderInfo } from '@deepseek-ai/dsh-llm'
+import type {
+  LlmModelContext,
+  LlmModelInfo,
+  LlmModelReasoningInfo,
+  LlmProviderInfo,
+} from '@deepseek-ai/dsh-llm'
 
 class ScriptedAdapter extends LlmAdapter {
   constructor(private script: StreamChunk[]) {
@@ -49,6 +55,7 @@ class CatalogAdapter extends ScriptedAdapter {
     private readonly provider: LlmProviderInfo,
     private readonly models: readonly LlmModelInfo[],
     private readonly contexts: Readonly<Record<string, LlmModelContext>> = {},
+    private readonly reasoning: Readonly<Record<string, LlmModelReasoningInfo>> = {},
   ) {
     super(SCRIPT)
   }
@@ -67,6 +74,13 @@ class CatalogAdapter extends ScriptedAdapter {
   ): Promise<LlmModelContext | undefined> {
     return Promise.resolve(this.contexts[model])
   }
+
+  override resolveModelReasoning(
+    _provider: string,
+    model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    return Promise.resolve(this.reasoning[model])
+  }
 }
 
 const SCRIPT: StreamChunk[] = [
@@ -674,6 +688,117 @@ describe('LlmService', () => {
     await expect(ctx.llm.resolveModelContext('route', 'other')).resolves.toBeUndefined()
   })
 
+  it('resolves detached adapter-owned reasoning metadata and materializes its default', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const source = {
+      efforts: [
+        { id: ReasoningEffortId('standard'), name: 'Standard' },
+        { id: ReasoningEffortId('ultra'), name: 'Ultra', description: 'Largest budget' },
+      ],
+      defaultEffort: ReasoningEffortId('standard'),
+    }
+    ctx.llm.registerAdapter(['route'], new CatalogAdapter(
+      { id: 'route', name: 'Route' },
+      [],
+      {},
+      { model: source },
+    ))
+
+    const resolved = await ctx.llm.resolveModelReasoning('route', 'model')
+    expect(resolved).toEqual(source)
+    source.efforts[0]!.name = 'mutated'
+    expect(resolved?.efforts[0]?.name).toBe('Standard')
+    await expect(ctx.llm.resolveCallConfig({ provider: 'route', model: 'model' })).resolves.toEqual({
+      provider: 'route',
+      model: 'model',
+      reasoningEffort: ReasoningEffortId('standard'),
+    })
+    const explicit = { provider: 'route', model: 'model', reasoningEffort: ReasoningEffortId('ultra') }
+    await expect(ctx.llm.resolveCallConfig(explicit)).resolves.toBe(explicit)
+  })
+
+  it.each([
+    [{ efforts: [] }, 'empty effort list'],
+    [{ efforts: [{ id: '', name: 'Empty' }] }, 'empty id'],
+    [{ efforts: [{ id: 'valid', name: '' }] }, 'empty name'],
+    [{ efforts: [{ id: 'valid', name: 'Valid', description: 1 }] }, 'non-string description'],
+    [{ efforts: [{ id: 'same', name: 'One' }, { id: 'same', name: 'Two' }] }, 'duplicate id'],
+    [{ efforts: [{ id: 'valid', name: 'Valid' }], defaultEffort: 'other' }, 'unknown default'],
+  ] as const)('rejects invalid model reasoning metadata (%s: %s)', async (metadata, _label) => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    ctx.llm.registerAdapter(['route'], new CatalogAdapter(
+      { id: 'route', name: 'Route' },
+      [],
+      {},
+      { model: metadata as unknown as LlmModelReasoningInfo },
+    ))
+    await expect(ctx.llm.resolveModelReasoning('route', 'model'))
+      .rejects.toMatchObject({ code: 'INVALID_MODEL_REASONING' })
+  })
+
+  it('rejects unsupported reasoning efforts without clamping', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    ctx.llm.registerAdapter(['route'], new CatalogAdapter(
+      { id: 'route', name: 'Route' },
+      [],
+      {},
+      { model: { efforts: [{ id: ReasoningEffortId('ultra'), name: 'Ultra' }] } },
+    ))
+
+    await expect(ctx.llm.resolveCallConfig({
+      provider: 'route',
+      model: 'model',
+      reasoningEffort: ReasoningEffortId('standard'),
+    })).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+    await expect(ctx.llm.resolveCallConfig({
+      provider: 'route',
+      model: 'plain',
+      reasoningEffort: ReasoningEffortId('standard'),
+    })).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+  })
+
+  it('resolves reasoning defaults at the final adapter boundary after routing middleware', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const adapter = new class extends RecordingAdapter {
+      override resolveModelReasoning(
+        _provider: string,
+        _model: string,
+      ): Promise<LlmModelReasoningInfo> {
+        return Promise.resolve({
+          efforts: [{ id: ReasoningEffortId('standard'), name: 'Standard' }],
+          defaultEffort: ReasoningEffortId('standard'),
+        })
+      }
+    }(SCRIPT)
+    ctx.llm.registerAdapter(['routed'], adapter)
+    const disposeRouting = ctx.on('llm/stream', (options, next) => {
+      options.provider = 'routed'
+      return next()
+    })
+
+    for await (const _chunk of ctx.llm.stream({
+      provider: 'initial',
+      model: 'model',
+      messages: [],
+    })) { /* drain */ }
+
+    expect(adapter.lastOptions?.reasoningEffort).toBe(ReasoningEffortId('standard'))
+    disposeRouting()
+
+    const frozenRequest: GenerateOptions = Object.freeze({
+      provider: 'routed',
+      model: 'model',
+      messages: [],
+    })
+    for await (const _chunk of ctx.llm.stream(frozenRequest)) { /* drain */ }
+    expect(adapter.lastOptions?.reasoningEffort).toBe(ReasoningEffortId('standard'))
+    expect(Object.isFrozen(adapter.lastOptions)).toBe(true)
+  })
+
   it.each([0, -1, 1.5, Number.NaN])(
     'rejects invalid adapter model context %s',
     async (contextWindow) => {
diff --git a/scripts/gen-cordis-catalog.ts b/scripts/gen-cordis-catalog.ts
index 8318d59996..88f7ebcf4c 100644
--- a/scripts/gen-cordis-catalog.ts
+++ b/scripts/gen-cordis-catalog.ts
@@ -38,6 +38,7 @@ export const LINK_MAP: Record<string, string> = {
   HookContext: 'core.md',
   LlmCallConfig: 'core.md',
   LlmModelContext: 'core.md',
+  LlmModelReasoningInfo: 'core.md',
   LlmFailure: 'llm-streaming.md',
   LlmModelInfo: 'core.md',
   LlmProviderInfo: 'core.md',
diff --git a/scripts/type-equiv.manifest.json b/scripts/type-equiv.manifest.json
index 89a0778827..f94dba52c4 100644
--- a/scripts/type-equiv.manifest.json
+++ b/scripts/type-equiv.manifest.json
@@ -46,6 +46,21 @@
       "symbol": "LlmModelContext",
       "source": "packages/llm/llm/src/types.ts"
     },
+    {
+      "doc": "docs/core-data-structures/core.md",
+      "symbol": "ReasoningEffortId",
+      "source": "packages/llm/llm/src/brand.ts"
+    },
+    {
+      "doc": "docs/core-data-structures/core.md",
+      "symbol": "LlmReasoningEffortInfo",
+      "source": "packages/llm/llm/src/types.ts"
+    },
+    {
+      "doc": "docs/core-data-structures/core.md",
+      "symbol": "LlmModelReasoningInfo",
+      "source": "packages/llm/llm/src/types.ts"
+    },
     {
       "doc": "docs/core-data-structures/core.md",
       "symbol": "GenerateOptions",

From 1bfca8612835e9c246ebc1c3c2a7e180ed3ac0fd Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sat, 25 Jul 2026 08:32:38 +0800
Subject: [PATCH 19/36] feat(tui): select model reasoning effort

---
 ...cated-full-screen-tui-front-door.i18n.yaml |   4 +-
 ...17-dedicated-full-screen-tui-front-door.md |   6 +-
 ...dedicated-full-screen-tui-front-door.zh.md |   6 +-
 docs/config-catalog.md                        |   2 +-
 docs/cordis-catalog/services.md               |   2 +-
 examples/tui-agent/README.md                  |   2 +-
 .../tests/fixtures/tui-scripted-llm.ts        |  32 +++-
 .../tui-agent/tests/tui-keyless-smoke.e2e.ts  |   3 +-
 packages/core/agent/README.md                 |   2 +-
 packages/core/agent/src/llm-target.ts         |  25 ++-
 packages/core/agent/tests/llm-target.spec.ts  |  23 ++-
 packages/ui/tui/README.md                     |  10 +-
 packages/ui/tui/src/index.ts                  | 154 ++++++++++++++---
 packages/ui/tui/tests/harness.ts              |  15 +-
 .../model-effort-switching.expected.txt       |  42 +++++
 .../snapshots/model-selector.expected.txt     |  38 ++---
 .../snapshots/model-switching.expected.txt    |  12 +-
 .../status-diagnostics-narrow.expected.txt    |   6 +-
 .../snapshots/status-diagnostics.expected.txt |  62 +++----
 packages/ui/tui/tests/tui.snapshot.ts         |  24 ++-
 packages/ui/tui/tests/tui.spec.ts             | 159 +++++++++++++++++-
 21 files changed, 497 insertions(+), 132 deletions(-)
 create mode 100644 packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt

diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
index 8d6c7be831..56b2e3536d 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-17-dedicated-full-screen-tui-front-door.md: ecfda138593fc2b98ac42929acc586b11e437ee2
-2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 6b8cc63f7657672a6da542e2033d765b54bd4f07
+2026-07-17-dedicated-full-screen-tui-front-door.md: 5ab38d2f9dc97d22ab35b42b04e6da68210690eb
+2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 8aee779d10370efd1bef832dbcaebdc25b34a595
diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
index ecfda13859..5ab38d2f9d 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
@@ -22,9 +22,9 @@ The selected front door receives the exact generated or resumed `SessionId` used
 
 The TUI rebuilds the transcript from the active `session.surface` and reprojects it whenever an event carries a `surfaceOp`, so resumed and compacted history matches the model-visible conversation. It renders Markdown text and reasoning, token totals, the latest `todo/write` plan, and tool cards produced through each tool definition's `presentCall` and `presentResult` methods. Long card bodies retain a configurable head/tail preview with the hidden-line count; one terminal control expands or collapses every card. Pending chunks and tool calls update the same components that completed events settle.
 
-Editor input calls `agent.send()` while idle and `agent.steer()` while a turn is running. Cancellation, reasoning visibility, tool-card expansion, redraw, transcript clearing, and exit are terminal-only controls. The idle footer derives context occupancy from `tokenMeter` and shows the selected model; during a run, elapsed activity and the Escape interrupt hint replace that summary. `/status` remains available in either state and appends a detailed terminal-only snapshot: session identity and timestamps, selected model and reasoning visibility, lifecycle counts folded from the event log, the same deduplicated usage buckets and KV-cache rate as the footer, and context use from `tokenMeter` plus the selected model's advertised capacity. The plugin registers the shared `userInteraction` provider and presents queued questions in a wide bottom-left keyboard panel with batch progress, numbered options, and aligned descriptions; agent behavior and answer logging remain owned by their existing services.
+Editor input calls `agent.send()` while idle and `agent.steer()` while a turn is running. Cancellation, reasoning visibility, tool-card expansion, redraw, transcript clearing, and exit are terminal-only controls. The idle footer derives context occupancy from `tokenMeter` and shows the selected model and explicit reasoning effort; during a run, elapsed activity and the Escape interrupt hint replace that summary. `/status` remains available in either state and appends a detailed terminal-only snapshot: session identity and timestamps, selected model, reasoning effort/default state and reasoning visibility, lifecycle counts folded from the event log, the same deduplicated usage buckets and KV-cache rate as the footer, and context use from `tokenMeter` plus the selected model's advertised capacity. The plugin registers the shared `userInteraction` provider and presents queued questions in a wide bottom-left keyboard panel with batch progress, numbered options, and aligned descriptions; agent behavior and answer logging remain owned by their existing services.
 
-The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model pair per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
+The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. Each model row owns the adapter-advertised reasoning-effort order and default: Shift+Tab cycles that row's efforts, while a model without selectable metadata keeps default behavior. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model/reasoning-effort target per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
 
 ### Terminal ownership
 
@@ -49,4 +49,4 @@ The implemented [TUI terminal-state snapshot Agent Note](../testing/2026-07-18-t
 - The TUI carries a pi-tui dependency and a strict TTY requirement; non-TTY deployments use the Headless app or a structured protocol.
 - Session projection makes resume and compaction consistent with the durable conversation, but one configured session owns the transcript and editor.
 - Tool packages extend terminal cards through their existing presentation methods without adding tool-specific branches to the TUI.
-- Model selection uses adapter-advertised metadata without turning catalog membership into request validation; unused selections are not durable state.
+- Model and reasoning-effort selection use adapter-advertised metadata without turning catalog membership into request validation; unused selections are not durable state.
diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
index 6b8cc63f76..8aee779d10 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
@@ -22,9 +22,9 @@ DeepSeek Harness 将 [`@deepseek-ai/dsh-tui`](../../../../packages/ui/tui/README
 
 TUI 从活跃的 `session.surface` 重建 transcript（文本记录），并在事件携带 `surfaceOp` 时重新投影，因此恢复或压缩后的历史与模型可见会话保持一致。TUI 渲染 Markdown 文本与推理、token 用量、最新 `todo/write` 计划，以及各工具定义通过 `presentCall` 和 `presentResult` 方法生成的工具卡片。较长的工具卡片正文会保留可配置的头尾预览，并显示隐藏行数；一个终端控制可以展开或收起全部卡片。进行中的分片与工具调用会更新同一组组件，随后由完成事件收束状态。
 
-agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调用 `agent.steer()`。取消、推理显隐、工具卡片展开、重绘、清空 transcript 和退出都只是终端控制。空闲态页脚根据 `tokenMeter` 得出上下文占用率，并显示所选模型；agent 运行期间，该摘要会替换为带已用时长的活动指示和 Escape 中断提示。`/status` 在这两种状态下均可用，并会追加一份仅在终端显示的详细快照，其中包括会话标识与时间戳、所选模型及推理显隐状态、从事件日志归并得出的生命周期计数、与页脚一致的去重用量分项和 KV 缓存命中率，以及 `tokenMeter` 给出的上下文用量和所选模型公布的容量。插件注册共享的 `userInteraction` 提供方，在左下角宽幅键盘操作面板中呈现排队的问题，面板显示批次进度、带编号的选项和对齐的描述；agent 行为和答案日志仍由既有服务负责。
+agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调用 `agent.steer()`。取消、推理显隐、工具卡片展开、重绘、清空 transcript 和退出都只是终端控制。空闲态页脚根据 `tokenMeter` 得出上下文占用率，并显示所选模型和显式选定的推理强度；agent 运行期间，该摘要会替换为带已用时长的活动指示和 Escape 中断提示。`/status` 在这两种状态下均可用，并会追加一份仅在终端显示的详细快照，其中包括会话标识与时间戳、所选模型、推理强度（或默认状态）及推理显隐状态、从事件日志归并得出的生命周期计数、与页脚一致的去重用量分项和 KV 缓存命中率，以及 `tokenMeter` 给出的上下文用量和所选模型公布的容量。插件注册共享的 `userInteraction` 提供方，在左下角宽幅键盘操作面板中呈现排队的问题，面板显示批次进度、带编号的选项和对齐的描述；agent 行为和答案日志仍由既有服务负责。
 
-`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个 step 快照一次同一个提供方/模型字段组合，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
+`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。每个模型行都持有适配器公布的推理强度顺序和默认值：按 Shift+Tab 可循环切换该行的推理强度；没有可选元数据的模型保持默认行为。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个步骤快照一次同一个提供方/模型/推理强度目标，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
 
 ### 终端所有权
 
@@ -49,4 +49,4 @@ agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调
 - TUI 会引入 pi-tui 依赖并严格要求 TTY；非 TTY 部署使用 Headless app 或结构化协议。
 - 会话投影使恢复和压缩与持久会话保持一致，但只有一个已配置会话拥有 transcript 和编辑器。
 - 工具包通过既有呈现方法扩展终端卡片，无需在 TUI 中增加工具专用分支。
-- 模型选择使用适配器提供的目录元数据，但不会把目录成员关系变成请求校验；未使用的选择不属于持久化状态。
+- 模型和推理强度选择使用适配器公布的元数据，但不会把目录成员关系变成请求校验；未使用的选择不属于持久化状态。
diff --git a/docs/config-catalog.md b/docs/config-catalog.md
index 0ad683d2f4..90da953d66 100644
--- a/docs/config-catalog.md
+++ b/docs/config-catalog.md
@@ -1632,7 +1632,7 @@ export interface TuiConfig {
 }
 ```
 
-Source: [`packages/ui/tui/src/index.ts:270`](../packages/ui/tui/src/index.ts)
+Source: [`packages/ui/tui/src/index.ts:273`](../packages/ui/tui/src/index.ts)
 
 ## `@deepseek-ai/dsh-tui-demo`
 
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index 537365a477..b4ba1a6698 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -1728,7 +1728,7 @@ The concrete provider retains pi-tui, focus, and terminal lifecycle state. Plugi
 abstract openOverlay(request: TuiOverlayRequest): TuiOverlaySession
 ```
 
-Source: [`packages/ui/tui/src/index.ts:150`](../../packages/ui/tui/src/index.ts)
+Source: [`packages/ui/tui/src/index.ts:153`](../../packages/ui/tui/src/index.ts)
 
 ## `ctx.userInteraction` — `UserInteractionService`
 
diff --git a/examples/tui-agent/README.md b/examples/tui-agent/README.md
index 2e87df0a27..106f1412a2 100644
--- a/examples/tui-agent/README.md
+++ b/examples/tui-agent/README.md
@@ -17,7 +17,7 @@ Type a coding task. The agent works through the `read`/`write`/`edit` filesystem
 
 The `todo_write` task tracker is opt-in and not in the shipped config: add `@deepseek-ai/dsh-tool-todo` to `cordis.yml` (or a personal-config overlay under `~/.dsh`) to expose it. Once loaded, the model records a whole-list plan to the session log and the TUI renders it.
 
-The TUI renders Markdown history, reasoning, tool-owned terminal/diff/generic cards, token totals, and — when `todo_write` is loaded — the latest plan. Long tool bodies keep a head/tail preview; Ctrl+O expands or collapses every card. Enter submits or steers while the agent runs, Ctrl+R toggles reasoning, Escape cancels, and `/help` lists commands. `/plan` selects plan mode for the next step; `/plan <message>` also submits the message into that step, while `/plan off` selects the default mode without model input. `/status` expands the current session's identity, activity counts, exact token/cache buckets, context use, and timestamps without interrupting a running turn. `/model` opens a keyboard selector for the current provider catalog; use Up/Down and Enter, or `/model <model>` and `/model <provider>/<model>` for direct selection. `ask_user_question` opens a wide bottom-left keyboard panel with batch progress and numbered options.
+The TUI renders Markdown history, reasoning, tool-owned terminal/diff/generic cards, token totals, and — when `todo_write` is loaded — the latest plan. Long tool bodies keep a head/tail preview; Ctrl+O expands or collapses every card. Enter submits or steers while the agent runs, Ctrl+R toggles reasoning, Escape cancels, and `/help` lists commands. `/plan` selects plan mode for the next step; `/plan <message>` also submits the message into that step, while `/plan off` selects the default mode without model input. `/status` expands the current session's identity, activity counts, exact token/cache buckets, context use, and timestamps without interrupting a running turn. `/model` opens a keyboard selector for the current provider catalog; use Up/Down to focus a model, Shift+Tab to cycle its advertised reasoning efforts, and Enter to select, or use `/model <model>` and `/model <provider>/<model>` for direct selection. `ask_user_question` opens a wide bottom-left keyboard panel with batch progress and numbered options.
 
 ### Resuming a prior session
 
diff --git a/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts b/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
index c3bdbfd1b1..e668689ca9 100644
--- a/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
+++ b/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
@@ -1,6 +1,12 @@
 import type { Context } from 'cordis'
-import type { GenerateOptions, LlmModelContext, LlmModelInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
-import { CallId, LlmAdapter } from '@deepseek-ai/dsh-llm'
+import type {
+  GenerateOptions,
+  LlmModelContext,
+  LlmModelInfo,
+  LlmModelReasoningInfo,
+  StreamChunk,
+} from '@deepseek-ai/dsh-llm'
+import { CallId, LlmAdapter, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 
 const CONTROL_PROBE = '\u001b]2;MODEL_CONTROLLED\u0007\u001b[999CMODEL_CURSOR\u009b31mMODEL_C1'
 const INITIAL_TEXT = `I need one decision before I continue. ${CONTROL_PROBE}`
@@ -39,6 +45,20 @@ class ScriptedTuiAdapter extends LlmAdapter {
     return Promise.resolve({ contextWindow: 128_000 })
   }
 
+  override resolveModelReasoning(
+    _provider: string,
+    model: string,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    if (model !== 'tui-scripted-model-pro') return Promise.resolve(undefined)
+    return Promise.resolve({
+      efforts: [
+        { id: ReasoningEffortId('high'), name: 'High' },
+        { id: ReasoningEffortId('max'), name: 'Max' },
+      ],
+      defaultEffort: ReasoningEffortId('high'),
+    })
+  }
+
   override async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
     // The session-title provider's auxiliary request carries no tool schemas,
     // unlike every agent turn; answer it with a fixed title so the PTY test can
@@ -47,8 +67,12 @@ class ScriptedTuiAdapter extends LlmAdapter {
       for (const chunk of textChunks(TITLE_TEXT)) yield chunk
       return
     }
-    if (options.model !== 'tui-scripted-model-pro' || !options.system?.includes('tui-scripted-model-pro')) {
-      throw new Error('the scripted TUI request did not apply the selected model to routing and prompt variables')
+    if (
+      options.model !== 'tui-scripted-model-pro'
+      || !options.system?.includes('tui-scripted-model-pro')
+      || options.reasoningEffort !== ReasoningEffortId('max')
+    ) {
+      throw new Error('the scripted TUI request did not apply the selected model and reasoning effort')
     }
     const lastMessage = options.messages.at(-1)
     const lastText = (lastMessage?.content ?? [])
diff --git a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
index 43ac327d8b..951cce98c5 100644
--- a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
+++ b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
@@ -104,7 +104,7 @@ function smoke(overrides: Partial<TuiPtySmokeOptions> & { label: string }): Prom
 // other route (see fixtures/tui-scripted-llm.ts).
 const SELECT_PRO_MODEL = [
   { waitFor: 'scripted TUI ready.', send: '/model\r' },
-  { waitFor: 'Select model', send: '\x1b[B\r' },
+  { waitFor: 'Select model', send: '\x1b[B\x1b[Z\r' },
 ] as const
 
 describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => {
@@ -157,6 +157,7 @@ describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => {
       ],
     })
     expect(output).toContain('I need one decision before I continue.')
+    expect(output).toContain('Reasoning effort: Max.')
     expect(output).toContain('Entering plan mode (applies from the next step). Use /plan off to leave.')
     expect(output).toContain('Leaving plan mode (applies from the next step).')
     expect(output).toContain('Default mode confirmed.')
diff --git a/packages/core/agent/README.md b/packages/core/agent/README.md
index 511ce32235..75a16039cd 100644
--- a/packages/core/agent/README.md
+++ b/packages/core/agent/README.md
@@ -10,7 +10,7 @@ Tracks live agents and carries the initiating Agent through asynchronous driver
 
 ### Public API
 
-The scoped-registration surface: `Agent.ctx` is the agent's scope context (`dsh-scope`, key = the agent) — register tools/sections/variables/listeners through it for that agent alone, all unwound on disposal. `agentEvents(ctx, agent)` is the fused dispatcher for ordinary agent-subject operations (carrier + injected subject in one move); its notification mode invokes every listener and contains both synchronous throws and returned-promise rejections. The registry lifecycle pair reuses one stable routing carrier. `assembleContextFor(agent)` builds the per-agent assembly context (`agent` + `scope` together). `installAgentLlmTarget(agentCtx, target)` snapshots a mutable provider/model selection during prompt assembly and applies that pair to both prompt variables and request routing for one step. `CreateAgentOptions.setup(agentCtx)` and `ResumeAgentOptions.setup(agentCtx)` compose a fresh or resumed agent's scoped world while both objects remain unpublished. Setup is trusted, composition-only same-process code: drive the agent only after creation resolves.
+The scoped-registration surface: `Agent.ctx` is the agent's scope context (`dsh-scope`, key = the agent) — register tools/sections/variables/listeners through it for that agent alone, all unwound on disposal. `agentEvents(ctx, agent)` is the fused dispatcher for ordinary agent-subject operations (carrier + injected subject in one move); its notification mode invokes every listener and contains both synchronous throws and returned-promise rejections. The registry lifecycle pair reuses one stable routing carrier. `assembleContextFor(agent)` builds the per-agent assembly context (`agent` + `scope` together). `installAgentLlmTarget(agentCtx, target)` snapshots a mutable provider/model/reasoning-effort selection during prompt assembly, applies the route to prompt variables, and applies the complete target to request routing for one step; an absent selected effort clears an inherited effort so the target uses adapter/provider defaults. `CreateAgentOptions.setup(agentCtx)` and `ResumeAgentOptions.setup(agentCtx)` compose a fresh or resumed agent's scoped world while both objects remain unpublished. Setup is trusted, composition-only same-process code: drive the agent only after creation resolves.
 
 - `ctx.agents.register(agent: Agent): () => void` — record an **already-constructed** agent. Disposed with the calling fiber.
 - Advanced ordered lifecycle: `enter(agent, owner): () => void` enforces `agent.id === agent.session.id`, performs the authoritative ID collision check, and inserts without announcing; `owner` explicitly records the live creator-agent relation (or `undefined` for a root), independently of durable session lineage. `announce(agent)` emits `agent/created` exactly once. A detach requested synchronously by a creation listener is deferred until that dispatch unwinds, and every detach checks the captured entry object, so a stale capability cannot delete a later same-ID replacement. The async factory uses this split; ordinary plugins use `register()`.
diff --git a/packages/core/agent/src/llm-target.ts b/packages/core/agent/src/llm-target.ts
index 18287a3ff5..e0492a71d2 100644
--- a/packages/core/agent/src/llm-target.ts
+++ b/packages/core/agent/src/llm-target.ts
@@ -1,17 +1,19 @@
 /**
- * Agent-scoped provider/model target snapshot shared by interactive front doors.
+ * Agent-scoped LLM target snapshot shared by interactive front doors.
  * @module @deepseek-ai/dsh-agent/llm-target
  */
 
 import type { Context } from 'cordis'
-import type { LlmCallConfig } from '@deepseek-ai/dsh-llm'
+import type { LlmCallConfig, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 
-/** Complete provider/model route selected for one live agent. */
+/** Complete provider/model route and optional reasoning effort selected for one live agent. */
 export interface AgentLlmTarget {
   /** Registered provider route. */
   provider: string
   /** Provider-owned model id. */
   model: string
+  /** Adapter-owned reasoning effort, or provider/default behavior when absent. */
+  reasoningEffort?: ReasoningEffortId
 }
 
 /** Mutable selection plus the target captured for the current step. */
@@ -24,9 +26,11 @@ export interface AgentLlmTargetRef {
 
 /**
  * Couple one mutable target to agent-scoped prompt assembly and request routing.
- * Prompt assembly snapshots the selected pair before delegating, then applies
- * both prompt variables and request config to that snapshot so a concurrent
- * switch takes effect on a later step instead of splitting the two surfaces.
+ * Prompt assembly snapshots the selected target before delegating, then applies
+ * its route to prompt variables and its route/effort to request config so a
+ * concurrent switch takes effect on a later step instead of splitting the two
+ * surfaces. An absent selected effort clears any inherited effort so a model
+ * switch can restore that target's provider/default behavior.
  *
  * @param agentCtx - The target agent's scoped context.
  * @param target - Mutable selection owned by the calling front door.
@@ -52,10 +56,15 @@ export function installAgentLlmTarget(agentCtx: Context, target: AgentLlmTargetR
     async (_agent, _turn, _step, _config, _signal, next): Promise<LlmCallConfig> => {
       const resolved = await next()
       const selected = target.assembled
-      return selected === undefined ? resolved : {
-        ...resolved,
+      if (selected === undefined) return resolved
+      const { reasoningEffort: _inheritedEffort, ...withoutInheritedEffort } = resolved
+      return {
+        ...withoutInheritedEffort,
         provider: selected.provider,
         model: selected.model,
+        ...selected.reasoningEffort === undefined
+          ? {}
+          : { reasoningEffort: selected.reasoningEffort },
       }
     },
   )
diff --git a/packages/core/agent/tests/llm-target.spec.ts b/packages/core/agent/tests/llm-target.spec.ts
index fa4ef2b459..10d688d6ed 100644
--- a/packages/core/agent/tests/llm-target.spec.ts
+++ b/packages/core/agent/tests/llm-target.spec.ts
@@ -7,7 +7,7 @@ import {
   type Agent,
   type AgentLlmTargetRef,
 } from '../src/index.ts'
-import type { LlmCallConfig } from '@deepseek-ai/dsh-llm'
+import { ReasoningEffortId, type LlmCallConfig } from '@deepseek-ai/dsh-llm'
 
 describe('installAgentLlmTarget()', () => {
   it('snapshots prompt variables and request routing together, then disposes both listeners', async () => {
@@ -24,16 +24,31 @@ describe('installAgentLlmTarget()', () => {
       'agent/request', 1, 0, seed, signal, () => Promise.resolve(seed),
     )).resolves.toBe(seed)
 
-    target.current = { provider: 'alpha', model: 'a1' }
+    target.current = {
+      provider: 'alpha',
+      model: 'a1',
+      reasoningEffort: ReasoningEffortId('high'),
+    }
     expect((await ctx.systemPrompt.assemble()).variables).toMatchObject({ provider: 'alpha', model: 'a1' })
     target.current = { provider: 'beta', model: 'b1' }
     await expect(agentEvents(ctx, agent).waterfall(
       'agent/request', 1, 0, seed, signal, () => Promise.resolve(seed),
-    )).resolves.toEqual({ provider: 'alpha', model: 'a1', temperature: 0.2 })
+    )).resolves.toEqual({
+      provider: 'alpha',
+      model: 'a1',
+      reasoningEffort: ReasoningEffortId('high'),
+      temperature: 0.2,
+    })
 
     expect((await ctx.systemPrompt.assemble()).variables).toMatchObject({ provider: 'beta', model: 'b1' })
+    const inherited: LlmCallConfig = {
+      provider: 'alpha',
+      model: 'a1',
+      reasoningEffort: ReasoningEffortId('max'),
+      temperature: 0.2,
+    }
     await expect(agentEvents(ctx, agent).waterfall(
-      'agent/request', 1, 1, seed, signal, () => Promise.resolve(seed),
+      'agent/request', 1, 1, inherited, signal, () => Promise.resolve(inherited),
     )).resolves.toEqual({ provider: 'beta', model: 'b1', temperature: 0.2 })
 
     dispose()
diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md
index 616544779f..3326c548eb 100644
--- a/packages/ui/tui/README.md
+++ b/packages/ui/tui/README.md
@@ -10,7 +10,7 @@ This package owns interactive terminal presentation and input only. It injects `
 
 After terminal startup succeeds, the package provides the terminal-local `ctx.tui` extension service. A plugin that injects it can call `openOverlay()` with a component factory and constrained layout options; the host exposes the viewport, semantic theme, display-text escaping, redraw, close, and a lifetime signal, but not the pi-tui tree, terminal, focus controller, or overlay handle. Plugin overlays, the model selector, and user questions share one FIFO modal queue. Each request is an effect of the calling plugin fiber, so unload removes queued work or closes visible work before cleanup settles; terminal shutdown unloads dependents before stopping pi-tui. Overlay state is not logged or replayed. Component code is trusted and may render ANSI styling, but must pass untrusted text through `host.display()`. The [interactive-extension Agent Note](../../../.agents/notes/implemented/architecture/2026-07-22-tui-interactive-extension-service.md) owns the boundary and rejected alternatives.
 
-The TUI rebuilds resumed history from the active session surface, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the latest `todo/write` plan above the editor, and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes `<session title> — <configured title>`. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelContext()` for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode and the current model with reasoning state; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. Surface replacement events rebuild the transcript so compacted history does not reappear.
+The TUI rebuilds resumed history from the active session surface, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the latest `todo/write` plan above the editor, and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes `<session title> — <configured title>`. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelContext()` for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode plus the current model and any explicitly selected reasoning effort; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. Surface replacement events rebuild the transcript so compacted history does not reappear.
 
 An embedding may provide `TuiRuntime.formatCwd` when its logical workspace label differs from the session's host directory. The override changes only the footer label; tools continue to use the session `cwd`.
 
@@ -22,13 +22,13 @@ When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers
 
 While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/reasoning`, `/tools`, `/redraw`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool cards collapse long bodies into a configurable head/tail preview; Ctrl+O toggles every card between its preview and full output. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle.
 
-`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Enter selects, and Escape closes it. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same pair through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
+`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape closes it. Models without selectable effort metadata ignore Shift+Tab; the selector does not synthesize `off`, clamp a value, or transfer an effort between models. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
 
 `/reload` (EXPERIMENTAL, dev-only) re-reads every file-backed loader config tree and applies the diff to the running app — the HMR watcher's config path, invoked manually; it needs the cordis Loader in the context and degrades to a warning without one, runs only while the agent is idle, and refuses re-entry while a reload is in flight. Module-source hot reload remains watcher-owned. When a `skills` service is mounted, `/skill:<name> [instructions]` loads that skill's instructions into the conversation as a user turn; autocomplete lists the model-invocable skills, and any skill (including a model-disabled one) is loadable by its exact name.
 
 The footer sums the session's reported usage as `↑<uncached input> ↓<output>`, followed by `cache <rate>%` once any input has been billed — the share of billed prompt tokens (uncached input plus cache reads and writes) served from the provider cache, rounded to a percent. It also compares token-meter pressure with `ctx.llm.resolveModelContext()` for the current route (omitting the context share when the adapter has no capacity metadata) and shows the current model and tool-card mode; the right side clips first when the footer is narrow.
 
-`/status` adds a point-in-time diagnostics card to the transcript and remains available while the agent runs. It reports the session id, title, working directory, selected provider/model, reasoning-block visibility, agent state, event/turn/step/tool-call counts, exact input/output/cache token buckets, KV-cache hit rate, token-meter context use and capacity, creation time, and latest event time. Missing titles, models, cache input, or context capacity are labeled instead of inferred. The card is terminal-only and does not duplicate the compact footer.
+`/status` adds a point-in-time diagnostics card to the transcript and remains available while the agent runs. It reports the session id, title, working directory, selected provider/model, selected reasoning effort or default behavior, reasoning-block visibility, agent state, event/turn/step/tool-call counts, exact input/output/cache token buckets, KV-cache hit rate, token-meter context use and capacity, creation time, and latest event time. Missing titles, models, cache input, or context capacity are labeled instead of inferred. The card is terminal-only and does not duplicate the compact footer.
 
 `/resume` opens a full-viewport keyboard selector over the current workspace instead of a centered dialog. Its focused search field starts immediately after the search glyph and emits pi-tui's cursor marker, so terminal IME composition remains anchored inside the field. Candidates are sorted by last logged activity and searchable by log-backed title or session id; each row reports current/live/persisted state, last turn outcome, recent provider/model, and durable goal phase when present. Up/Down and Page Up/Page Down navigate, Enter resumes, Escape clears a non-empty search before a second Escape cancels, and Ctrl+C cancels directly. The current session, a session already live in this runtime, an unreadable log, a mismatched cwd, or a session whose logged provider has no current adapter remains visible but disabled. Selection repeats those checks and requires the current agent to be idle before flushing the current session. The TUI then stops the terminal UI and calls the optional host-owned `TuiRuntime.handoffResume`; where `process.execve` is available, the shipped `dsh` host disposes the app and replaces its process. Resume restores the same `SessionId`, transcript, title, todos, and durable goal; goal activation remains disarmed and the TUI asks for human confirmation or `/goal resume`.
 
@@ -47,7 +47,7 @@ The footer sums the session's reported usage as `↑<uncached input> ↓<output>
 | `maxResumeOptions` | `8` | Visible sessions in the resume selector |
 | `questionDialogWidth` | `200` | Question-panel width in columns, clamped to the terminal |
 | `questionDialogMaxHeight` | `20` | Question-panel maximum rows |
-| `modelDialogWidth` | `72` | Model-selector width in columns |
+| `modelDialogWidth` | `76` | Model-selector width in columns |
 | `modelDialogMaxHeight` | `20` | Model-selector maximum rows |
 | `fileSearchMaxResults` | `20` | Maximum file and directory candidates shown for one `@` query |
 | `fileSearchMaxEntries` | `10000` | Maximum paths retained in the bounded workspace index used by bare fuzzy queries |
@@ -114,7 +114,7 @@ The fixed instruction is part of the stable system-prompt prefix and is reusable
 
 #### What the model sees
 
-The `/model` command text and keyboard-selector input are not logged or sent. New steps receive the selected provider/model pair in both prompt variables and request routing.
+The `/model` command text and keyboard-selector input are not logged or sent. New steps receive the selected provider/model route in prompt variables and the selected provider/model/reasoning-effort target in request routing.
 
 #### Token effect
 
diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts
index de0ef02f7e..21db41789c 100644
--- a/packages/ui/tui/src/index.ts
+++ b/packages/ui/tui/src/index.ts
@@ -31,6 +31,7 @@ import {
   type EditorTheme,
   type Focusable,
   type MarkdownTheme,
+  type SelectItem,
   type SelectListTheme,
   type SlashCommand,
   type Terminal,
@@ -53,6 +54,8 @@ import { assertNever, errorChain } from '@deepseek-ai/dsh-llm'
 import type {
   ContentBlock,
   LlmModelInfo,
+  LlmModelReasoningInfo,
+  ReasoningEffortId,
   StreamChunk,
   TokenUsage,
 } from '@deepseek-ai/dsh-llm'
@@ -233,7 +236,7 @@ const maxModelOptionsSchema = z.number().step(1).min(1).default(8)
 const maxResumeOptionsSchema = z.number().step(1).min(1).default(8)
 const questionDialogWidthSchema = z.number().step(1).min(20).default(200)
 const questionDialogMaxHeightSchema = z.number().step(1).min(6).default(20)
-const modelDialogWidthSchema = z.number().step(1).min(20).default(72)
+const modelDialogWidthSchema = z.number().step(1).min(20).default(76)
 const modelDialogMaxHeightSchema = z.number().step(1).min(6).default(20)
 const fileSearchMaxResultsSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_RESULTS)
 const fileSearchMaxEntriesSchema = z.number().step(1).min(1).default(DEFAULT_FILE_SEARCH_MAX_ENTRIES)
@@ -356,7 +359,7 @@ export function resolveTuiConfig(config: TuiConfig | undefined): ResolvedTuiConf
     maxResumeOptions: config?.maxResumeOptions ?? 8,
     questionDialogWidth: config?.questionDialogWidth ?? 200,
     questionDialogMaxHeight: config?.questionDialogMaxHeight ?? 20,
-    modelDialogWidth: config?.modelDialogWidth ?? 72,
+    modelDialogWidth: config?.modelDialogWidth ?? 76,
     modelDialogMaxHeight: config?.modelDialogMaxHeight ?? 20,
     fileSearchMaxResults: config?.fileSearchMaxResults ?? DEFAULT_FILE_SEARCH_MAX_RESULTS,
     fileSearchMaxEntries: config?.fileSearchMaxEntries ?? DEFAULT_FILE_SEARCH_MAX_ENTRIES,
@@ -580,15 +583,34 @@ function textBlocks(content: readonly ContentBlock[], type: 'text' | 'reasoning'
 interface ModelChoice extends AgentLlmTarget {
   modelName: string
   description?: string
+  reasoning?: LlmModelReasoningInfo
 }
 
 function targetLabel(target: AgentLlmTarget): string {
   return `${target.provider}/${target.model}`
 }
 
+function compactTargetLabel(target: AgentLlmTarget): string {
+  return `${target.model}${target.reasoningEffort === undefined ? '' : ` ${target.reasoningEffort}`}`
+}
+
+function targetReasoningLabel(choice: ModelChoice, effort: ReasoningEffortId | undefined): string | undefined {
+  if (effort === undefined) return choice.reasoning === undefined ? undefined : 'provider default'
+  return choice.reasoning?.efforts.find(candidate => candidate.id === effort)?.name ?? effort
+}
+
 function initialTarget(agent: Agent): AgentLlmTarget | undefined {
   const logged = agent.session.requestHeader()?.config
-  if (logged !== undefined) return { provider: logged.provider, model: logged.model }
+  if (logged !== undefined) {
+    if (logged.reasoningEffort === undefined) {
+      return { provider: logged.provider, model: logged.model }
+    }
+    return {
+      provider: logged.provider,
+      model: logged.model,
+      reasoningEffort: logged.reasoningEffort,
+    }
+  }
   if (agent.options.provider === undefined || agent.options.model === undefined) return undefined
   return { provider: agent.options.provider, model: agent.options.model }
 }
@@ -607,11 +629,15 @@ async function readModelChoices(
     ) {
       models.push({ provider: provider.id, id: current.model, name: current.model })
     }
-    return models.map((model): ModelChoice => ({
-      provider: provider.id,
-      model: model.id,
-      modelName: model.name,
-      ...model.description === undefined ? {} : { description: model.description },
+    return Promise.all(models.map(async (model): Promise<ModelChoice> => {
+      const reasoning = await ctx.llm.resolveModelReasoning(provider.id, model.id)
+      return {
+        provider: provider.id,
+        model: model.id,
+        modelName: model.name,
+        ...model.description === undefined ? {} : { description: model.description },
+        ...reasoning === undefined ? {} : { reasoning },
+      }
     }))
   }))
   return groups.flat()
@@ -1226,6 +1252,10 @@ function renderDialog(
 
 class ModelDialog implements Component {
   private readonly list: SelectList
+  private readonly items: Map<string, SelectItem>
+  private readonly choices: Map<string, ModelChoice>
+  private readonly efforts: Map<string, ReasoningEffortId | undefined>
+  private readonly currentValue: string | undefined
 
   constructor(
     choices: readonly ModelChoice[],
@@ -1235,15 +1265,27 @@ class ModelDialog implements Component {
     done: (choice: ModelChoice) => void,
     cancel: () => void,
   ) {
-    this.list = new SelectList(choices.map(choice => ({
-      value: targetLabel(choice),
-      label: displayText(targetLabel(choice)),
-      description: [
-        displayText(choice.modelName),
-        ...choice.description === undefined ? [] : [displayText(choice.description)],
-        ...current?.provider === choice.provider && current.model === choice.model ? ['current'] : [],
-      ].join(' — '),
-    })), maxVisible, dialogSelectTheme(palette))
+    this.items = new Map()
+    this.choices = new Map()
+    this.efforts = new Map()
+    this.currentValue = current === undefined ? undefined : targetLabel(current)
+    for (const choice of choices) {
+      const value = targetLabel(choice)
+      const isCurrent = current?.provider === choice.provider && current.model === choice.model
+      this.choices.set(value, choice)
+      this.efforts.set(
+        value,
+        isCurrent
+          ? current.reasoningEffort ?? choice.reasoning?.defaultEffort
+          : choice.reasoning?.defaultEffort,
+      )
+      this.items.set(value, {
+        value,
+        label: displayText(value),
+        description: this.describeChoice(choice, isCurrent),
+      })
+    }
+    this.list = new SelectList([...this.items.values()], maxVisible, dialogSelectTheme(palette))
     const currentIndex = current === undefined
       ? 0
       : choices.findIndex(choice => choice.provider === current.provider && choice.model === current.model)
@@ -1252,17 +1294,57 @@ class ModelDialog implements Component {
       const selected = choices.find(choice => targetLabel(choice) === item.value)
       /* v8 ignore next -- SelectList only returns values built from `choices`. */
       if (selected === undefined) return
-      done(selected)
+      const effort = this.efforts.get(item.value)
+      done({
+        ...selected,
+        ...effort === undefined ? {} : { reasoningEffort: effort },
+      })
     }
     this.list.onCancel = cancel
   }
 
+  private describeChoice(choice: ModelChoice, isCurrent: boolean): string {
+    const selectedEffort = this.efforts.get(targetLabel(choice))
+    const effort = choice.reasoning?.efforts.find(candidate => candidate.id === selectedEffort)
+    const effortLabel = selectedEffort === undefined
+      ? choice.reasoning === undefined ? undefined : 'provider default'
+      : effort?.name ?? selectedEffort
+    return [
+      displayText(choice.modelName),
+      ...choice.description === undefined ? [] : [displayText(choice.description)],
+      ...effortLabel === undefined ? [] : [displayText(effortLabel)],
+      ...isCurrent ? ['current'] : [],
+    ].join(' — ')
+  }
+
+  private cycleReasoningEffort(): void {
+    const selectedItem = this.list.getSelectedItem()
+    /* v8 ignore next -- the dialog is opened only for a non-empty catalog. */
+    if (selectedItem === null) return
+    const choice = this.choices.get(selectedItem.value)
+    if (choice?.reasoning === undefined) return
+    const current = this.efforts.get(selectedItem.value)
+    const currentIndex = choice.reasoning.efforts.findIndex(effort => effort.id === current)
+    const next = choice.reasoning.efforts[(currentIndex + 1) % choice.reasoning.efforts.length]
+    /* v8 ignore next -- validated reasoning metadata always carries at least one effort. */
+    if (next === undefined) return
+    this.efforts.set(selectedItem.value, next.id)
+    const item = this.items.get(selectedItem.value)
+    /* v8 ignore next -- items and choices are constructed from the same values. */
+    if (item === undefined) return
+    item.description = this.describeChoice(choice, selectedItem.value === this.currentValue)
+  }
+
   invalidate(): void {
     this.list.invalidate()
   }
 
   handleInput(data: string): void {
-    this.list.handleInput(data)
+    if (matchesKey(data, Key.shift(Key.tab))) {
+      this.cycleReasoningEffort()
+    } else {
+      this.list.handleInput(data)
+    }
     this.invalidate()
   }
 
@@ -1271,7 +1353,7 @@ class ModelDialog implements Component {
     return renderDialog('Select model', [
       ...this.list.render(innerWidth),
       '',
-      this.palette.dim('↑/↓ navigate • Enter select • Esc cancel'),
+      this.palette.dim('↑/↓ navigate • Shift+Tab reasoning • Enter select • Esc cancel'),
     ], width, this.palette)
   }
 }
@@ -1938,7 +2020,7 @@ export function createTuiChat(
     () => sessionTitle ?? config.welcome,
     palette,
     resolved.color && resolved.truecolor,
-    () => target.current?.model,
+    () => target.current === undefined ? undefined : compactTargetLabel(target.current),
   )
   const footer = new FooterComponent(
     agent,
@@ -1946,7 +2028,7 @@ export function createTuiChat(
     () => toolsExpanded,
     () => tokens,
     runtime.formatCwd,
-    () => target.current?.model,
+    () => target.current === undefined ? undefined : compactTargetLabel(target.current),
     () => contextWindow === undefined
       ? undefined
       : Math.min(100, Math.round(ctx.tokenMeter.measure(agent.session).totalTokens / contextWindow * 100)),
@@ -2037,13 +2119,26 @@ export function createTuiChat(
   resolveContextWindow(target.current)
 
   const selectModel = (selected: ModelChoice): void => {
-    if (target.current?.provider === selected.provider && target.current.model === selected.model) {
-      appendNotice(`Model is already ${targetLabel(selected)}.`)
+    const sameRoute = target.current?.provider === selected.provider && target.current.model === selected.model
+    const reasoningEffort = selected.reasoningEffort
+      ?? (sameRoute ? target.current?.reasoningEffort ?? selected.reasoning?.defaultEffort : selected.reasoning?.defaultEffort)
+    if (sameRoute && target.current?.reasoningEffort === reasoningEffort) {
+      const reasoning = targetReasoningLabel(selected, reasoningEffort)
+      appendNotice(`Model is already ${targetLabel(selected)}${reasoning === undefined ? '' : ` with reasoning effort ${displayText(reasoning)}`}.`)
       return
     }
-    target.current = { provider: selected.provider, model: selected.model }
+    target.current = {
+      provider: selected.provider,
+      model: selected.model,
+      ...reasoningEffort === undefined ? {} : { reasoningEffort },
+    }
     resolveContextWindow(target.current)
-    appendNotice(`Model selected: ${targetLabel(selected)}. New steps will use it.`)
+    const reasoning = targetReasoningLabel(selected, reasoningEffort)
+    appendNotice([
+      `Model selected: ${targetLabel(selected)}.`,
+      ...reasoning === undefined ? [] : [`Reasoning effort: ${displayText(reasoning)}.`],
+      'New steps will use it.',
+    ].join(' '))
   }
 
   const showModelSelector = (choices: readonly ModelChoice[]): void => {
@@ -2633,12 +2728,17 @@ export function createTuiChat(
     const steps = events.filter(event => event.type === 'step/start').length
     const toolCalls = events.filter(event => event.type === 'tool/call').length
     const model = target.current === undefined ? 'unset' : displayText(targetLabel(target.current))
+    const effort = target.current === undefined
+      ? 'unset'
+      : target.current.reasoningEffort === undefined
+        ? 'default'
+        : displayText(target.current.reasoningEffort)
     const groups: readonly (readonly StatusCardRow[])[] = [
       [
         ['Session', displayText(agent.session.id)],
         ['Title', displayText(sessionTitle ?? 'untitled')],
         ['Directory', displayText(cwd)],
-        ['Model', `${model} ${palette.dim(`(reasoning ${showReasoning ? 'shown' : 'hidden'})`)}`],
+        ['Model', `${model} ${palette.dim(`(effort ${effort}; reasoning blocks ${showReasoning ? 'shown' : 'hidden'})`)}`],
       ],
       [
         ['Agent', [
diff --git a/packages/ui/tui/tests/harness.ts b/packages/ui/tui/tests/harness.ts
index 37eee99735..e42610644c 100644
--- a/packages/ui/tui/tests/harness.ts
+++ b/packages/ui/tui/tests/harness.ts
@@ -8,7 +8,13 @@ import AgentRegistry, {
   type AgentStatus,
   type SendOptions,
 } from '@deepseek-ai/dsh-agent'
-import type { ContentBlock, LlmModelContext, LlmModelInfo, LlmProviderInfo } from '@deepseek-ai/dsh-llm'
+import type {
+  ContentBlock,
+  LlmModelContext,
+  LlmModelInfo,
+  LlmModelReasoningInfo,
+  LlmProviderInfo,
+} from '@deepseek-ai/dsh-llm'
 import CommandService from '@deepseek-ai/dsh-commands'
 import SessionStore, { SessionId, type Session, type SessionHeader } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
@@ -48,6 +54,10 @@ export interface TuiHarnessOptions {
     models: LlmModelInfo[]
     listModels?: (provider: string) => Promise<LlmModelInfo[]>
     resolveModelContext?: (provider: string, model: string) => Promise<LlmModelContext | undefined>
+    resolveModelReasoning?: (
+      provider: string,
+      model: string,
+    ) => Promise<LlmModelReasoningInfo | undefined>
   }
   /** Provide a fake `sessionPersistence` service so resume surfaces can list sessions. */
   sessionPersistence?: {
@@ -122,6 +132,9 @@ export async function createTuiTestHarness<TerminalType extends Terminal, Exit e
         return catalog.resolveModelContext?.(provider, model)
           ?? Promise.resolve({ contextWindow: options.contextWindow ?? 128_000 })
       },
+      resolveModelReasoning(provider: string, model: string) {
+        return catalog.resolveModelReasoning?.(provider, model) ?? Promise.resolve(undefined)
+      },
     } as never)
   }
   if (ctx.get('systemPrompt') === undefined) await ctx.plugin(SystemPrompt)
diff --git a/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
new file mode 100644
index 0000000000..a7db8ccba0
--- /dev/null
+++ b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
@@ -0,0 +1,42 @@
+terminal 92x32 buffer=normal length=32 base=0 viewport=0
+lifecycle started=1 stopped=0 progress=inactive
+title "DSH snapshot"
+cursor hidden column=92 viewportRow=15 bufferRow=15
+buffer
+0| " DEEPSEEK HARNESS"
+  style 1-8 fg=bright-blue bold
+  style 10-16 bold
+1| " Snapshot agent ready."
+  style 1-21 fg=bright-black
+2| " deepseek-v4-flash  •  main-session"
+  style 1-34 dim
+3| "────────────────────────────────────────────────────────────────────────────────────────────"
+  style 0-91 dim
+4| "                                                                                            "
+  style 1-1 inverse
+5| "────────────────────────────────────────────────────────────────────────────────────────────"
+  style 0-91 dim
+6| "deepseek-v4-flash  /workspace/project  ↑0 ↓0                     0% context  tools:collapsed"
+  style 0-43 dim
+  style 65-91 dim
+7-12| <blank>
+13| "        ╭ Select model ────────────────────────────────────────────────────────────╮        "
+  style 8-83 fg=bright-blue
+14| "        │   deepseek/deepseek-v4-flash      DeepSeek V4 Flash — High — current     │        "
+  style 8-8 fg=bright-blue
+  style 38-77 fg=bright-black
+  style 83-83 fg=bright-blue
+15| "        │ → deepseek/deepseek-v4-pro        DeepSeek V4 Pro — Max                  │        "
+  style 8-8 fg=bright-blue
+  style 10-64 fg=bright-blue inverse
+  style 83-83 fg=bright-blue
+16| "        │                                                                          │        "
+  style 8-8 fg=bright-blue
+  style 83-83 fg=bright-blue
+17| "        │ ↑/↓ navigate • Shift+Tab reasoning • Enter select • Esc cancel           │        "
+  style 8-8 fg=bright-blue
+  style 10-71 dim
+  style 83-83 fg=bright-blue
+18| "        ╰──────────────────────────────────────────────────────────────────────────╯        "
+  style 8-83 fg=bright-blue
+19-31| <blank>
diff --git a/packages/ui/tui/tests/snapshots/model-selector.expected.txt b/packages/ui/tui/tests/snapshots/model-selector.expected.txt
index f10c99d03a..1caba208d6 100644
--- a/packages/ui/tui/tests/snapshots/model-selector.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-selector.expected.txt
@@ -20,23 +20,23 @@ buffer
   style 0-43 dim
   style 65-91 dim
 7-12| <blank>
-13| "          ╭ Select model ────────────────────────────────────────────────────────╮          "
-  style 10-81 fg=bright-blue
-14| "          │ → deepseek/deepseek-v4-flash      DeepSeek V4 Flash — current        │          "
-  style 10-10 fg=bright-blue
-  style 12-72 fg=bright-blue inverse
-  style 81-81 fg=bright-blue
-15| "          │   deepseek/deepseek-v4-pro        DeepSeek V4 Pro                    │          "
-  style 10-10 fg=bright-blue
-  style 38-60 fg=bright-black
-  style 81-81 fg=bright-blue
-16| "          │                                                                      │          "
-  style 10-10 fg=bright-blue
-  style 81-81 fg=bright-blue
-17| "          │ ↑/↓ navigate • Enter select • Esc cancel                             │          "
-  style 10-10 fg=bright-blue
-  style 12-51 dim
-  style 81-81 fg=bright-blue
-18| "          ╰──────────────────────────────────────────────────────────────────────╯          "
-  style 10-81 fg=bright-blue
+13| "        ╭ Select model ────────────────────────────────────────────────────────────╮        "
+  style 8-83 fg=bright-blue
+14| "        │ → deepseek/deepseek-v4-flash      DeepSeek V4 Flash — High — current     │        "
+  style 8-8 fg=bright-blue
+  style 10-77 fg=bright-blue inverse
+  style 83-83 fg=bright-blue
+15| "        │   deepseek/deepseek-v4-pro        DeepSeek V4 Pro — High                 │        "
+  style 8-8 fg=bright-blue
+  style 36-65 fg=bright-black
+  style 83-83 fg=bright-blue
+16| "        │                                                                          │        "
+  style 8-8 fg=bright-blue
+  style 83-83 fg=bright-blue
+17| "        │ ↑/↓ navigate • Shift+Tab reasoning • Enter select • Esc cancel           │        "
+  style 8-8 fg=bright-blue
+  style 10-71 dim
+  style 83-83 fg=bright-blue
+18| "        ╰──────────────────────────────────────────────────────────────────────────╯        "
+  style 8-83 fg=bright-blue
 19-31| <blank>
diff --git a/packages/ui/tui/tests/snapshots/model-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
index 5800901f08..70b61f1022 100644
--- a/packages/ui/tui/tests/snapshots/model-switching.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
@@ -8,18 +8,18 @@ buffer
   style 10-16 bold
 1| " Snapshot agent ready."
   style 1-21 fg=bright-black
-2| " deepseek-v4-pro  •  main-session"
-  style 1-32 dim
+2| " deepseek-v4-pro max  •  main-session"
+  style 1-36 dim
 3| <blank>
-4| " Model selected: deepseek/deepseek-v4-pro. New steps will use it.                           "
-  style 1-64 fg=bright-black
+4| " Model selected: deepseek/deepseek-v4-pro. Reasoning effort: Max. New steps will use it.    "
+  style 1-87 fg=bright-black
 5| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
 6| "                                                                                            "
   style 1-1 inverse
 7| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
-8| "deepseek-v4-pro  /workspace/project  ↑0 ↓0                       0% context  tools:collapsed"
-  style 0-41 dim
+8| "deepseek-v4-pro max  /workspace/project  ↑0 ↓0                   0% context  tools:collapsed"
+  style 0-45 dim
   style 65-91 dim
 9-31| <blank>
diff --git a/packages/ui/tui/tests/snapshots/status-diagnostics-narrow.expected.txt b/packages/ui/tui/tests/snapshots/status-diagnostics-narrow.expected.txt
index de46d59543..4937592cc7 100644
--- a/packages/ui/tui/tests/snapshots/status-diagnostics-narrow.expected.txt
+++ b/packages/ui/tui/tests/snapshots/status-diagnostics-narrow.expected.txt
@@ -41,13 +41,13 @@ buffer
   style 0-0 dim
   style 3-12 fg=bright-black
   style 55-55 dim
-16| "│  Model:      deepseek/deepseek-v4-pro (reasoning     │"
+16| "│  Model:      deepseek/deepseek-v4-pro (effort        │"
   style 0-0 dim
   style 3-12 fg=bright-black
   style 40-55 dim
-17| "│              shown)                                  │"
+17| "│              default; reasoning blocks shown)        │"
   style 0-0 dim
-  style 15-20 dim
+  style 15-46 dim
   style 55-55 dim
 18| "│                                                      │"
   style 0-0 dim
diff --git a/packages/ui/tui/tests/snapshots/status-diagnostics.expected.txt b/packages/ui/tui/tests/snapshots/status-diagnostics.expected.txt
index 73d98953f6..915d4e58ef 100644
--- a/packages/ui/tui/tests/snapshots/status-diagnostics.expected.txt
+++ b/packages/ui/tui/tests/snapshots/status-diagnostics.expected.txt
@@ -25,68 +25,68 @@ buffer
   style 1-9 fg=bright-magenta bold
 10| " Session inspected.                                                                         "
 11| <blank>
-12| "╭─ Session status ─────────────────────────────────────────────────╮"
+12| "╭─ Session status ───────────────────────────────────────────────────────────────╮"
   style 0-2 dim
   style 3-16 fg=bright-blue bold
-  style 17-67 dim
-13| "│  Session:    main-session                                        │"
+  style 17-81 dim
+13| "│  Session:    main-session                                                      │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-14| "│  Title:      Inspect session diagnostics                         │"
+  style 81-81 dim
+14| "│  Title:      Inspect session diagnostics                                       │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-15| "│  Directory:  /workspace/project                                  │"
+  style 81-81 dim
+15| "│  Directory:  /workspace/project                                                │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-16| "│  Model:      deepseek/deepseek-v4-pro (reasoning shown)          │"
+  style 81-81 dim
+16| "│  Model:      deepseek/deepseek-v4-pro (effort default; reasoning blocks shown) │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 40-56 dim
-  style 67-67 dim
-17| "│                                                                  │"
+  style 40-79 dim
+  style 81-81 dim
+17| "│                                                                                │"
   style 0-0 dim
-  style 67-67 dim
-18| "│  Agent:      idle · 6 events · 1 turn · 1 step · 1 tool call     │"
+  style 81-81 dim
+18| "│  Agent:      idle · 6 events · 1 turn · 1 step · 1 tool call                   │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-19| "│                                                                  │"
+  style 81-81 dim
+19| "│                                                                                │"
   style 0-0 dim
-  style 67-67 dim
-20| "│  Tokens:     1,250 input + 340 output                            │"
+  style 81-81 dim
+20| "│  Tokens:     1,250 input + 340 output                                          │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-21| "│  KV cache:   [███████████░░░░░] 67% hit (3,000 read + 250 write) │"
+  style 81-81 dim
+21| "│  KV cache:   [███████████░░░░░] 67% hit (3,000 read + 250 write)               │"
   style 0-0 dim
   style 3-12 fg=bright-black
   style 15-15 dim
   style 16-26 fg=bright-blue
   style 27-32 dim
-  style 67-67 dim
-22| "│  Context:    [█████░░░░░░░░░░░] 33% used (42,000 / 128,000)      │"
+  style 81-81 dim
+22| "│  Context:    [█████░░░░░░░░░░░] 33% used (42,000 / 128,000)                    │"
   style 0-0 dim
   style 3-12 fg=bright-black
   style 15-15 dim
   style 16-20 fg=bright-blue
   style 21-32 dim
-  style 67-67 dim
-23| "│                                                                  │"
+  style 81-81 dim
+23| "│                                                                                │"
   style 0-0 dim
-  style 67-67 dim
-24| "│  Created:    2026-07-22 09:10:11 UTC                             │"
+  style 81-81 dim
+24| "│  Created:    2026-07-22 09:10:11 UTC                                           │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-25| "│  Active:     2026-07-22 09:10:11 UTC                             │"
+  style 81-81 dim
+25| "│  Active:     2026-07-22 09:10:11 UTC                                           │"
   style 0-0 dim
   style 3-12 fg=bright-black
-  style 67-67 dim
-26| "╰──────────────────────────────────────────────────────────────────╯"
-  style 0-67 dim
+  style 81-81 dim
+26| "╰────────────────────────────────────────────────────────────────────────────────╯"
+  style 0-81 dim
 27| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
 28| "                                                                                            "
diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts
index d837841286..9e30d3525e 100644
--- a/packages/ui/tui/tests/tui.snapshot.ts
+++ b/packages/ui/tui/tests/tui.snapshot.ts
@@ -5,7 +5,7 @@ import { fileURLToPath } from 'node:url'
 import { afterAll, describe, expect, it, vi } from 'vitest'
 import type { Context } from 'cordis'
 import { agentEvents } from '@deepseek-ai/dsh-agent'
-import { CallId, type ContentBlock } from '@deepseek-ai/dsh-llm'
+import { CallId, ReasoningEffortId, type ContentBlock } from '@deepseek-ai/dsh-llm'
 import type {} from '@deepseek-ai/dsh-llm-retry'
 import { SessionId, type JsonValue, type Session } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
@@ -45,6 +45,7 @@ const CHECKPOINTS = [
   'surface-after-compaction-narrow',
   'surface-after-compaction-wide',
   'model-selector',
+  'model-effort-switching',
   'model-switching',
   'errors-and-help',
   'disposed-terminal',
@@ -632,7 +633,22 @@ describe('TUI terminal-state snapshots', () => {
   })
 
   it('pins the model selector and selection notice', async () => {
-    const harness = await setupSnapshot({}, { columns: 92, rows: 32 })
+    const harness = await setupSnapshot({
+      catalog: {
+        providers: [{ id: 'deepseek', name: 'DeepSeek' }],
+        models: [
+          { provider: 'deepseek', id: 'deepseek-v4-flash', name: 'DeepSeek V4 Flash' },
+          { provider: 'deepseek', id: 'deepseek-v4-pro', name: 'DeepSeek V4 Pro' },
+        ],
+        resolveModelReasoning: () => Promise.resolve({
+          efforts: [
+            { id: ReasoningEffortId('high'), name: 'High' },
+            { id: ReasoningEffortId('max'), name: 'Max' },
+          ],
+          defaultEffort: ReasoningEffortId('high'),
+        }),
+      },
+    }, { columns: 92, rows: 32 })
     await renderAfter(harness, () => {
       harness.terminal.send('/model')
       harness.terminal.send('\r')
@@ -640,6 +656,10 @@ describe('TUI terminal-state snapshots', () => {
     await checkpoint('model-selector', harness.terminal, { includeScrollback: true })
     await renderAfter(harness, () => {
       harness.terminal.send('\x1b[B')
+      harness.terminal.send('\x1b[Z')
+    })
+    await checkpoint('model-effort-switching', harness.terminal, { includeScrollback: true })
+    await renderAfter(harness, () => {
       harness.terminal.send('\r')
     })
     await checkpoint('model-switching', harness.terminal, { includeScrollback: true })
diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts
index 0fea8b2750..4ada432060 100644
--- a/packages/ui/tui/tests/tui.spec.ts
+++ b/packages/ui/tui/tests/tui.spec.ts
@@ -5,7 +5,7 @@ import { describe, expect, it, vi } from 'vitest'
 import { Context } from 'cordis'
 import { CombinedAutocompleteProvider, type Terminal } from '@earendil-works/pi-tui'
 import AgentRegistry, { agentEvents, assembleContextFor, AgentMessageId, type Agent } from '@deepseek-ai/dsh-agent'
-import { type LlmCallConfig } from '@deepseek-ai/dsh-llm'
+import { ReasoningEffortId, type LlmCallConfig } from '@deepseek-ai/dsh-llm'
 import { GOAL_CHANGE_VERSION, GoalId, renderGoalChange, type GoalSnapshotChangeMeta } from '@deepseek-ai/dsh-goal'
 import CommandService, { type CommandInvocation } from '@deepseek-ai/dsh-commands'
 import SessionStore, { SessionId, type JsonValue, type SessionEvent, type SessionHeader, type TurnEndReason } from '@deepseek-ai/dsh-session'
@@ -157,7 +157,7 @@ describe('TUI config', () => {
       maxResumeOptions: 8,
       questionDialogWidth: 200,
       questionDialogMaxHeight: 20,
-      modelDialogWidth: 72,
+      modelDialogWidth: 76,
       modelDialogMaxHeight: 20,
       fileSearchMaxResults: 20,
       fileSearchMaxEntries: 10_000,
@@ -1712,7 +1712,8 @@ describe('pi-tui chat lifecycle and transcript', () => {
     expect(result.terminal.output).toContain('main-session')
     expect(result.terminal.output).toContain('Inspect status \\x1b]2;unsafe\\x07')
     expect(result.terminal.output).toContain('/workspace/status')
-    expect(result.terminal.output).toContain('deepseek/deepseek-v4-pro (reasoning hidden)')
+    expect(result.terminal.output).toContain('deepseek/deepseek-v4-pro (effort default; reasoning blocks')
+    expect(result.terminal.output).toContain('hidden)')
     expect(result.terminal.output).toContain('running · 6 events · 1 turn · 1 step · 2 tool calls')
     expect(result.terminal.output).toContain('1,250 input + 340 output')
     expect(result.terminal.output).toContain('[███████████░░░░░] 67% hit (3,000 read + 250 write)')
@@ -1748,7 +1749,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
     await tick()
 
     expect(result.terminal.output).toContain('untitled')
-    expect(result.terminal.output).toContain('unset (reasoning shown)')
+    expect(result.terminal.output).toContain('unset (effort unset; reasoning blocks shown)')
     expect(result.terminal.output).toContain('idle · 0 events · 0 turns · 0 steps · 0 tool calls')
     expect(result.terminal.output).toContain('n/a (0 read + 0 write)')
     expect(result.terminal.output).toContain('7 used · capacity unknown')
@@ -2301,6 +2302,35 @@ describe('pi-tui chat lifecycle and transcript', () => {
         resolveModelContext: (provider, model) => provider === 'alpha' && model === 'a1'
           ? initialContext.promise
           : Promise.resolve({ contextWindow: 200 }),
+        resolveModelReasoning: (provider, model) => {
+          if (model === 'a1') {
+            return Promise.resolve({
+              efforts: [
+                { id: ReasoningEffortId('low'), name: 'Low' },
+                { id: ReasoningEffortId('high'), name: 'High' },
+              ],
+              defaultEffort: ReasoningEffortId('low'),
+            })
+          }
+          if (model === 'b1') {
+            return Promise.resolve({
+              efforts: [
+                { id: ReasoningEffortId('high'), name: 'High' },
+                { id: ReasoningEffortId('max'), name: 'Max' },
+              ],
+              defaultEffort: ReasoningEffortId('high'),
+            })
+          }
+          if (provider === 'alpha' && model === 'shared') {
+            return Promise.resolve({
+              efforts: [
+                { id: ReasoningEffortId('standard'), name: 'Standard' },
+                { id: ReasoningEffortId('ultra'), name: 'Ultra' },
+              ],
+            })
+          }
+          return Promise.resolve(undefined)
+        },
       },
     })
 
@@ -2325,6 +2355,62 @@ describe('pi-tui chat lifecycle and transcript', () => {
     result.terminal.send('\x1b')
     await tick()
 
+    const providerDefaultOutput = result.terminal.output.length
+    result.terminal.send('/model alpha/shared')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Reasoning effort: provider default.')
+    })
+    result.terminal.send('/model')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Select model')
+    })
+    expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Alpha Shared — provider default')
+    result.terminal.send('\x1b[Z')
+    await tick()
+    expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Alpha Shared — Standard')
+    result.terminal.send('\r')
+    await tick()
+    expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Reasoning effort: Standard.')
+
+    const nonReasoningOutput = result.terminal.output.length
+    result.terminal.send('/model')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(nonReasoningOutput)).toContain('Select model')
+    })
+    result.terminal.send('\x1b[B')
+    result.terminal.send('\x1b[B')
+    result.terminal.send('\x1b[Z')
+    result.terminal.send('\r')
+    await tick()
+    expect(result.terminal.output.slice(nonReasoningOutput)).toContain('Model selected: beta/shared.')
+    result.terminal.send('/model beta/shared')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(nonReasoningOutput)).toContain('Model is already beta/shared.')
+    })
+    await result.ctx.systemPrompt.assemble(assembleContextFor(result.agent))
+    result.terminal.send('/model alpha/a1')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(nonReasoningOutput)).toContain('Reasoning effort: Low.')
+    })
+    const inheritedEffort: LlmCallConfig = {
+      provider: 'alpha',
+      model: 'a1',
+      reasoningEffort: ReasoningEffortId('max'),
+    }
+    await expect(agentEvents(result.ctx, result.agent).waterfall(
+      'agent/request',
+      0,
+      0,
+      inheritedEffort,
+      new AbortController().signal,
+      () => Promise.resolve(inheritedEffort),
+    )).resolves.toEqual({ provider: 'beta', model: 'shared' })
+
     result.agent.status = 'running'
     const runningSelectorOutput = result.terminal.output.length
     result.terminal.send('/model')
@@ -2333,13 +2419,18 @@ describe('pi-tui chat lifecycle and transcript', () => {
       const output = result.terminal.output.slice(runningSelectorOutput)
       expect(output).toContain('Select model')
       expect(output).toContain('alpha/a1')
-      expect(output).toContain('Alpha One — Fast — current')
+      expect(output).toContain('Alpha One — Fast — Low — current')
+      expect(output).toContain('Beta One — High')
     })
     result.terminal.send('\x1b[B')
     result.terminal.send('\x1b[B')
+    result.terminal.send('\x1b[Z')
+    await tick()
+    expect(result.terminal.output).toContain('Beta One — Max')
     result.terminal.send('\r')
     await tick()
     expect(result.terminal.output).toContain('Model selected: beta/b1')
+    expect(result.terminal.output).toContain('Reasoning effort: Max.')
     expect(result.agent.sent).toEqual([])
     expect(result.agent.steered).toEqual([])
     initialContext.resolve({ contextWindow: 100 })
@@ -2358,8 +2449,12 @@ describe('pi-tui chat lifecycle and transcript', () => {
     result.agent.status = 'idle'
     result.ctx.emit('agent/status', result.agent, 'idle')
     await tick()
-    expect(result.terminal.output).toContain('b1  ')
+    expect(result.terminal.output).toContain('b1 max  ')
     expect(result.terminal.output).toContain('25% context  tools:collapsed')
+    result.terminal.send('/status')
+    result.terminal.send('\r')
+    await tick()
+    expect(result.terminal.output).toContain('beta/b1 (effort max; reasoning blocks shown)')
 
     const assembly = await result.ctx.systemPrompt.assemble(assembleContextFor(result.agent))
     expect(assembly.variables).toMatchObject({ provider: 'beta', model: 'b1' })
@@ -2367,7 +2462,12 @@ describe('pi-tui chat lifecycle and transcript', () => {
     const request = await agentEvents(result.ctx, result.agent).waterfall(
       'agent/request', 1, 0, seed, new AbortController().signal, () => Promise.resolve(seed),
     )
-    expect(request).toEqual({ provider: 'beta', model: 'b1', temperature: 0.2 })
+    expect(request).toEqual({
+      provider: 'beta',
+      model: 'b1',
+      reasoningEffort: ReasoningEffortId('max'),
+      temperature: 0.2,
+    })
     await dispose(result)
   })
 
@@ -2377,7 +2477,13 @@ describe('pi-tui chat lifecycle and transcript', () => {
       catalog: { providers: [{ id: 'beta', name: 'Beta' }], models: [] },
       beforeMount(session) {
         session.append('request/header', {
-          header: { config: { provider: 'beta', model: 'private' } },
+          header: {
+            config: {
+              provider: 'beta',
+              model: 'private',
+              reasoningEffort: ReasoningEffortId('ultra'),
+            },
+          },
           reason: 'initial',
         })
       },
@@ -2387,9 +2493,30 @@ describe('pi-tui chat lifecycle and transcript', () => {
     await tick()
     expect(resumed.terminal.output).toContain('Select model')
     expect(resumed.terminal.output).toContain('beta/private')
-    expect(resumed.terminal.output).toContain('private — current')
+    expect(resumed.terminal.output).toContain('private — ultra — current')
+    resumed.terminal.send('\x1b')
+    await tick()
+    resumed.terminal.send('/model beta/private')
+    resumed.terminal.send('\r')
+    await tick()
+    expect(resumed.terminal.output).toContain('with reasoning effort ultra')
     await dispose(resumed)
 
+    const resumedDefault = await setup({
+      catalog: {
+        providers: [{ id: 'alpha', name: 'Alpha' }],
+        models: [{ provider: 'alpha', id: 'default', name: 'Default Model' }],
+      },
+      beforeMount(session) {
+        session.append('request/header', {
+          header: { config: { provider: 'alpha', model: 'default' } },
+          reason: 'initial',
+        })
+      },
+    })
+    expect(resumedDefault.terminal.output).toContain('default  •  main-session')
+    await dispose(resumedDefault)
+
     const unset = await setup({
       agentOptions: {},
       catalog: {
@@ -2437,6 +2564,20 @@ describe('pi-tui chat lifecycle and transcript', () => {
     })
     expect(failed.terminal.output).toContain('Could not resolve model context: capacity offline')
     await dispose(failed)
+
+    const reasoningFailed = await setup({
+      catalog: {
+        providers: [{ id: 'deepseek', name: 'DeepSeek' }],
+        models: [{ provider: 'deepseek', id: 'model-1', name: 'Model One' }],
+        resolveModelReasoning: () => Promise.reject(new Error('reasoning metadata offline')),
+      },
+    })
+    reasoningFailed.terminal.send('/model')
+    reasoningFailed.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(reasoningFailed.terminal.output).toContain('Could not read the model catalog: reasoning metadata offline')
+    })
+    await dispose(reasoningFailed)
   })
 
   it('does not render a model catalog that resolves after TUI disposal', async () => {

From baea5018e57525e3ebfb2582eb98e4193f90e69a Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sat, 25 Jul 2026 22:59:45 +0800
Subject: [PATCH 20/36] fix(llm): bind reasoning resolution to adapter
 lifecycle

---
 ...ed-reasoning-effort-capabilities.i18n.yaml |   4 +-
 ...ter-owned-reasoning-effort-capabilities.md |   4 +-
 ...-owned-reasoning-effort-capabilities.zh.md |   4 +-
 docs/architecture.i18n.yaml                   |   4 +-
 docs/architecture.md                          |   4 +-
 docs/architecture.zh.md                       |   4 +-
 docs/cookbook/adding-an-llm-adapter.i18n.yaml |   4 +-
 docs/cookbook/adding-an-llm-adapter.md        |   2 +-
 docs/cookbook/adding-an-llm-adapter.zh.md     |   2 +-
 docs/cordis-catalog/services.md               |  33 +++--
 docs/core-data-structures/core.i18n.yaml      |   4 +-
 docs/core-data-structures/core.md             |   2 +-
 docs/core-data-structures/core.zh.md          |   2 +-
 .../llm-streaming.i18n.yaml                   |   4 +-
 docs/core-data-structures/llm-streaming.md    |  21 ++-
 docs/core-data-structures/llm-streaming.zh.md |  21 ++-
 .../develop/practice/llm-adapter.i18n.yaml    |   4 +-
 docs/user/develop/practice/llm-adapter.md     |   2 +-
 docs/user/develop/practice/llm-adapter.zh.md  |   2 +-
 .../cordis/tool-cordis/src/api-catalog.ts     |  20 ++-
 packages/core/agent-loop/README.md            |   2 +-
 packages/core/agent-loop/src/loop.ts          |   8 +-
 .../tests/request-reconstruction.spec.ts      |  84 +++++++++++-
 packages/llm/llm-deepseek/src/adapter.ts      |   1 +
 packages/llm/llm-pi-ai/src/adapter.ts         |   1 +
 packages/llm/llm/README.md                    |  11 +-
 packages/llm/llm/src/index.ts                 | 123 +++++++++++++++---
 packages/llm/llm/tests/service.spec.ts        | 109 ++++++++++++++++
 scripts/gen-cordis-catalog.ts                 |   1 +
 scripts/type-equiv.manifest.json              |  10 ++
 30 files changed, 429 insertions(+), 68 deletions(-)

diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
index 489f9845f6..b0a7acb852 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-24-adapter-owned-reasoning-effort-capabilities.md: 806e808ab18d617003f757200f0cdee5853476f5
-2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 9be8b6f4f7bcbe063ff90492ac44b9590860f67f
+2026-07-24-adapter-owned-reasoning-effort-capabilities.md: 04aebf5fa61d896ac16b29578a39c9b0e6e38cf4
+2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: a1c89417562276fd70ac5e5c1be17c9e639eb3ed
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
index 806e808ab1..04aebf5fa6 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
@@ -12,7 +12,7 @@ Reasoning strength was adapter configuration only, so a conversation could not d
 
 `dsh-llm` represents a reasoning effort as the opaque branded `ReasoningEffortId`. An adapter's `resolveModelReasoning(provider, model)` returns a non-empty ordered list of ids with display metadata and may name one configured default. The core validates metadata, requires an explicit or configured effort to appear exactly in that list, and never clamps or aliases a value.
 
-`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop resolves the post-`agent/request` config before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id. The terminal `LlmService` adapter boundary repeats resolution for direct calls that do not pass through the loop.
+`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop prepares the post-`agent/request` config under the active turn signal before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. The prepared call retains the exact adapter registration across asynchronous capability resolution, durable header logging, and dispatch; direct `LlmService.stream()` calls likewise capture their final registration before awaiting resolution. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id.
 
 The native DeepSeek adapter advertises `high` and `max`, defaults to configured effort or `high`, and exposes no effort capability while thinking is disabled. The pi-ai adapter derives each exact model's list from `getSupportedThinkingLevels()`, excludes `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai.
 
@@ -30,4 +30,4 @@ The native DeepSeek adapter advertises `high` and `max`, defaults to configured
 
 Clients can query one exact route and render the adapter's order and names without knowing a global enum. Adapter configuration remains the deployment-default owner, while `agent/request` can replace the effective effort on each step. Invalid metadata fails with `INVALID_MODEL_REASONING`, and unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
 
-The capability query is asynchronous and exact-model resolution may fail for adapters backed by authoritative catalogs. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, and resume behavior; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
+The capability query is asynchronous and exact-model resolution may fail for adapters backed by authoritative catalogs. Its optional signal is the caller's cancellation boundary; an asynchronous adapter must settle promptly after abort so loop disposal can reach quiescence. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, resume behavior, HMR registration ownership, and cancellation; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
index 9be8b6f4f7..a1c8941756 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
@@ -12,7 +12,7 @@ Status: implemented
 
 `dsh-llm` 使用不透明的品牌类型 `ReasoningEffortId` 表示推理强度。适配器的 `resolveModelReasoning(provider, model)` 返回非空的有序 ID 列表及其展示元数据，并可指定一个由配置确定的默认值。核心会校验元数据，要求显式指定或配置指定的推理强度与列表中的某个 ID 完全一致，且绝不自动调整或为值提供别名。
 
-`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在 `agent/request` 处理完成后、写入 `request/header` 前解析配置，因此默认值和动态变更只有成为持久化事实后才对模型可见。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。最终的 `LlmService` 适配器边界会再次执行解析，以覆盖未经过主循环的直接调用。
+`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在活跃轮次信号的控制下准备 `agent/request` 处理完成后的配置，再写入 `request/header`，因此默认值和动态变更只有成为持久化事实后才对模型可见。准备完成的调用在异步能力解析、请求头持久记录和分派全程保留同一项确切的适配器注册；直接调用 `LlmService.stream()` 时，也会在等待解析前捕获最终的适配器注册。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。
 
 原生 DeepSeek 适配器声明 `high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`；禁用思考时不暴露推理强度能力。pi-ai 适配器通过 `getSupportedThinkingLevels()` 按具体模型推导等级列表，排除 `off`，在 profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。
 
@@ -30,4 +30,4 @@ Status: implemented
 
 客户端可以查询一条确切路由，并按适配器给出的顺序和名称渲染等级，而无需了解全局枚举。适配器配置仍负责提供部署默认值，`agent/request` 则可以在每个步骤替换实际生效的推理强度。元数据无效时抛出 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
 
-能力查询采用异步方式；对于由权威目录支持的适配器，确切模型解析可能失败。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录和恢复行为提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
+能力查询采用异步方式；对于由权威目录支持的适配器，确切模型解析可能失败。可选信号构成调用方的取消边界；异步适配器必须在信号中止后迅速完成结算，使主循环的资源释放达到完全停稳。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录、恢复行为、HMR（热模块替换）期间的注册所有权和取消提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
diff --git a/docs/architecture.i18n.yaml b/docs/architecture.i18n.yaml
index f1d0d49bfb..06b7abc7f9 100644
--- a/docs/architecture.i18n.yaml
+++ b/docs/architecture.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-architecture.md: 124f1caadb3f8557698965ccb3d17587b7b9ed8a
-architecture.zh.md: af6fdee935f665809dfd35218a0398118ed5dba9
+architecture.md: 88b3a1d133c6e4b28ae39b379678a3202124979f
+architecture.zh.md: ccc29ddfb1c7e80f550fda6fb3f759be1bef1946
diff --git a/docs/architecture.md b/docs/architecture.md
index 124f1caadb..88b3a1d133 100644
--- a/docs/architecture.md
+++ b/docs/architecture.md
@@ -91,7 +91,7 @@ forever:
       agent/pre-step
       snapshot the derived messages (the reconstruction boundary)
       'step/start'
-      agent/request -> resolve reasoning/default -> log request/header -> checkpoint -> llm/stream (frozen)
+      agent/request -> prepare reasoning/default under turn signal -> log request/header -> checkpoint -> llm/stream (frozen, registration-bound)
       on final adapter-path or terminal in-band failure:
         'step/end'
         agent/request-error(original error, failure facts, immutable prior failures, signal)
@@ -125,7 +125,7 @@ Pruning precedes summaries; overflow retries require durable progress. Bounded r
 
 Adapter failures close the step before `agent/request-error` with exact `Error`, `LlmFailure`, and history. Retries open steps; success clears history; exhaustion stores failure on `turn/end`. Failed chunks commit nothing.
 
-Other failures use `agent/error`. Cancellation and disposal beat recovery; undispatched tools get synthetic `tool/call`/`ABORTED_BEFORE_DISPATCH` pairs. The signal retires before `turn/end`. Effective `cancel()` emits its cause, clears queues, and aborts; observers cannot veto, idle calls emit nothing, and durability records `aborted`. Disposal awaits quiescence ([decision](../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)).
+Other failures use `agent/error`. Cancellation and disposal beat recovery; the turn signal also cancels asynchronous model-capability preparation before any request header is committed, and undispatched tools get synthetic `tool/call`/`ABORTED_BEFORE_DISPATCH` pairs. The signal retires before `turn/end`. Effective `cancel()` emits its cause, clears queues, and aborts; observers cannot veto, idle calls emit nothing, and durability records `aborted`. Disposal awaits quiescence ([decision](../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)).
 
 Session events are turn-enclosed; reload closes an interrupted tail with a synthetic `interrupted` turn end. Post-close failures use `agent/error`. Each turn has one [TurnEndReason](core-data-structures/session.md#why-a-turn-ended-turnendreasonmap).
 
diff --git a/docs/architecture.zh.md b/docs/architecture.zh.md
index af6fdee935..ccc29ddfb1 100644
--- a/docs/architecture.zh.md
+++ b/docs/architecture.zh.md
@@ -91,7 +91,7 @@ forever:
       agent/pre-step
       snapshot the derived messages (the reconstruction boundary)
       'step/start'
-      agent/request -> resolve reasoning/default -> log request/header -> checkpoint -> llm/stream (frozen)
+      agent/request -> prepare reasoning/default under turn signal -> log request/header -> checkpoint -> llm/stream (frozen, registration-bound)
       on final adapter-path or terminal in-band failure:
         'step/end'
         agent/request-error(original error, failure facts, immutable prior failures, signal)
@@ -125,7 +125,7 @@ forever:
 
 适配器故障会先关闭步骤，再进入 `agent/request-error`；该事件会收到准确的 `Error`、`LlmFailure` 和历史记录。重试会开启步骤；成功会清除历史记录；重试耗尽后，故障存入 `turn/end`。失败分片不会提交任何内容。
 
-其他故障使用 `agent/error`。取消和资源释放均优先于恢复；尚未分派的工具会得到合成的 `tool/call`/`ABORTED_BEFORE_DISPATCH` 对。信号会在 `turn/end` 前失效。实际生效的 `cancel()` 会发出原因、清空队列并中止；观察方不能否决该操作，空闲状态下的调用不发出任何事件，持久化会记录 `aborted`。dispose（资源释放）会等待系统停稳（[决策](../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)）。
+其他故障使用 `agent/error`。取消和资源释放均优先于恢复；轮次信号还会在提交任何请求头之前取消异步模型能力准备，尚未分派的工具会得到合成的 `tool/call`/`ABORTED_BEFORE_DISPATCH` 对。信号会在 `turn/end` 前失效。实际生效的 `cancel()` 会发出原因、清空队列并中止；观察方不能否决该操作，空闲状态下的调用不发出任何事件，持久化会记录 `aborted`。dispose（资源释放）会等待系统停稳（[决策](../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md)）。
 
 会话事件均位于轮次边界内；重新加载会用合成的 `interrupted` 轮次结束事件闭合中断的日志尾部。关闭后的故障使用 `agent/error`。每个轮次有一个 [TurnEndReason](core-data-structures/session.md#why-a-turn-ended-turnendreasonmap)。
 
diff --git a/docs/cookbook/adding-an-llm-adapter.i18n.yaml b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
index 15c62ae966..85aa0bc8f5 100644
--- a/docs/cookbook/adding-an-llm-adapter.i18n.yaml
+++ b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-adding-an-llm-adapter.md: 3adae89f360cd81e264912d74bd86f5a06cd42f8
-adding-an-llm-adapter.zh.md: 80329da5e7c89bfb5ee2e2aa5b9931a3ef0e83f6
+adding-an-llm-adapter.md: 76bcfdd6638ef8b40977a5e1fe4f678648c043cd
+adding-an-llm-adapter.zh.md: 13ae388bb0c6f29ab6269f364b849cb1694e2ef2
diff --git a/docs/cookbook/adding-an-llm-adapter.md b/docs/cookbook/adding-an-llm-adapter.md
index 3adae89f36..76bcfdd663 100644
--- a/docs/cookbook/adding-an-llm-adapter.md
+++ b/docs/cookbook/adding-an-llm-adapter.md
@@ -32,7 +32,7 @@ Registration is effect-based (HMR-safe); one adapter per provider route — dupl
 - A `GenerateOptions` field your provider cannot honor (e.g. a `stop` list on a provider without stop sequences): throw `LlmError(..., 'UNSUPPORTED')` rather than silently dropping it.
 - If the provider requires response ids, signatures, or other native metadata on follow-up calls, emit the minimal lossless-JSON projection as `finish.replayState`. Validate it when rebuilding history. `LlmService` passes it only when the historical provider route and target provider route are currently owned by the exact same adapter instance; your adapter decides whether same-model, cross-model, or cross-provider restoration is legal. Never infer native replay from provider/model names alone when state is absent.
 
-Provider-specific thinking-mode toggles remain in the adapter's Config. Selectable reasoning strength uses the provider-neutral capability seam: return ordered opaque ids from `resolveModelReasoning()`, declare a configured `defaultEffort` only when one exists, and map `GenerateOptions.reasoningEffort` to the provider wire value. Do not expose provider wire spellings, clamp unsupported values, or include an `off` mode as an effort.
+Provider-specific thinking-mode toggles remain in the adapter's Config. Selectable reasoning strength uses the provider-neutral capability seam: return ordered opaque ids from `resolveModelReasoning()`, declare a configured `defaultEffort` only when one exists, honor the resolver's optional `AbortSignal`, and map `GenerateOptions.reasoningEffort` to the provider wire value. Do not expose provider wire spellings, clamp unsupported values, or include an `off` mode as an effort.
 
 ## Structure that worked
 
diff --git a/docs/cookbook/adding-an-llm-adapter.zh.md b/docs/cookbook/adding-an-llm-adapter.zh.md
index 80329da5e7..13ae388bb0 100644
--- a/docs/cookbook/adding-an-llm-adapter.zh.md
+++ b/docs/cookbook/adding-an-llm-adapter.zh.md
@@ -32,7 +32,7 @@ export function apply(ctx: Context, config: Config) {
 - 如果 `GenerateOptions` 中某个字段你的提供方无法支持（例如提供方不支持 stop sequences 时收到 `stop` 列表）：抛出 `LlmError(..., 'UNSUPPORTED')`，而非静默丢弃。
 - 如果提供方在后续调用中需要响应 ID、签名或其他原生元数据，请将其最小无损 JSON 投影作为 `finish.replayState` 发出。重建历史时验证该状态。只有历史提供方路由和目标提供方路由当前由完全相同的适配器实例拥有时，`LlmService` 才会传递该状态；由适配器决定同模型、跨模型或跨提供方恢复是否合法。状态缺失时，切勿仅根据提供方/模型名称推断原生回放。
 
-提供方特有的 thinking 模式开关仍放在适配器的 Config 中。可选的推理强度使用提供方无关的能力 seam：`resolveModelReasoning()` 返回有序的不透明 ID；仅当存在配置指定的默认值时才声明 `defaultEffort`；并将 `GenerateOptions.reasoningEffort` 映射为提供方协议值。不得暴露提供方协议值的具体拼写、自动调整不支持的值，也不得把 `off` 模式列为推理强度。
+提供方特有的 thinking 模式开关仍放在适配器的 Config 中。可选的推理强度使用提供方无关的能力 seam：`resolveModelReasoning()` 返回有序的不透明 ID；仅当存在配置指定的默认值时才声明 `defaultEffort`；响应传给解析器的可选 `AbortSignal`；并将 `GenerateOptions.reasoningEffort` 映射为提供方协议值。不得暴露提供方协议值的具体拼写、自动调整不支持的值，也不得把 `off` 模式列为推理强度。
 
 ## 经验证有效的结构
 
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index 025e78b4e5..db108e6ed2 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -745,37 +745,52 @@ async resolveModelContext( provider: string, model: string, ): Promise<LlmModelC
  * effort selector is unsupported for that model.
  * @param provider - registered provider route to inspect.
  * @param model - exact model id passed to the adapter.
+ * @param signal - optional cancellation for adapter-owned asynchronous lookup.
  * @returns detached reasoning metadata, or `undefined` when unsupported.
  */
-async resolveModelReasoning( provider: string, model: string, ): Promise<LlmModelReasoningInfo | undefined>
+async resolveModelReasoning( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmModelReasoningInfo | undefined>
 
 /**
  * Validate a conversation call config against its exact model capability and
  * materialize an adapter-configured default. Unsupported explicit efforts
- * reject before provider I/O; no clamping or aliasing is performed.
+ * reject before provider I/O; no clamping or aliasing is performed. This
+ * standalone query does not bind a later dispatch; use {@link prepareCall}
+ * when logging and streaming must share one adapter registration.
  * @param config - provider/model route and optional request controls.
+ * @param signal - optional cancellation for adapter-owned capability lookup.
  * @returns a detached config only when a default must be materialized.
  */
-async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>
+async resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>
+
+/**
+ * Resolve one call under its current adapter registration. The returned
+ * one-shot handle keeps that registration across header logging and dispatch,
+ * so HMR cannot combine one adapter's capability result with another adapter.
+ * @param config - provider/model route and optional request controls.
+ * @param signal - optional cancellation for adapter-owned capability lookup.
+ * @returns a prepared config and its registration-bound stream entry point.
+ */
+async prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall>
 
 /**
  * Stream one model call as raw chunks (token-level deltas). Throws
  * `LlmError` with code `NO_ADAPTER` if no adapter is registered for
  * `options.provider`. Replay state is retained only when the same adapter
  * instance owns its historical provider and the target provider. Final
- * adapter selection, dispatch, and iteration failures retain their original
- * Error identity and are tagged in a call-local scope for narrow agent-loop
- * request recovery; middleware and nested-call failures remain untagged for
- * the outer call.
+ * adapter selection remains fixed through asynchronous reasoning resolution
+ * and dispatch. Selection, dispatch, and iteration failures retain their
+ * original Error identity and are tagged in a call-local scope for narrow
+ * agent-loop request recovery; middleware and nested-call failures remain
+ * untagged for the outer call.
  * @param options - the full request; `options.provider` selects the adapter.
  * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.
  */
 stream(options: GenerateOptions): AsyncIterable<StreamChunk>
 ```
 
-Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmCallConfig](../core-data-structures/core.md) · [LlmModelContext](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmModelReasoningInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
+Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmCallConfig](../core-data-structures/core.md) · [LlmModelContext](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmModelReasoningInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [PreparedLlmCall](../core-data-structures/llm-streaming.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
 
-Source: [`packages/llm/llm/src/index.ts:175`](../../packages/llm/llm/src/index.ts)
+Source: [`packages/llm/llm/src/index.ts:192`](../../packages/llm/llm/src/index.ts)
 
 ## `ctx.permission` — `PermissionService`
 
diff --git a/docs/core-data-structures/core.i18n.yaml b/docs/core-data-structures/core.i18n.yaml
index afa9317d44..ba95f5d56c 100644
--- a/docs/core-data-structures/core.i18n.yaml
+++ b/docs/core-data-structures/core.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-core.md: 7938bd2766735303153b741b7442b698067d9089
-core.zh.md: 7c5298659cd5e9b004969ee4e626be4fad62dc18
+core.md: d60268a163580558b09ae0af7b15b78232436a4b
+core.zh.md: b81c6a90ab7f74580b0f6599bd78868771f3e339
diff --git a/docs/core-data-structures/core.md b/docs/core-data-structures/core.md
index 7938bd2766..d60268a163 100644
--- a/docs/core-data-structures/core.md
+++ b/docs/core-data-structures/core.md
@@ -325,7 +325,7 @@ The model-facing `ToolSchema` is the wire shape; the registered `ToolDefinition`
 
 The loop builds each request from logged state. `EpochHeader` records call config, rendered prompt, authoritative returned tool order (configured by `toolOrder`, or lexicographic when unset), and session prefix through full `request/header` snapshots. Together with derived history, this makes the request reconstructable from the session log. See [session.md](session.md#the-request-header-event-requestheader) and the [reconstructability Agent Note](../../.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.md).
 
-`agent/request` receives a frozen call-config seed and may return a replacement to switch provider, model, reasoning effort, or sampling. The loop resolves the exact model capability after the waterfall, rejects unsupported explicit effort ids without clamping, materializes an adapter-configured default, and logs the effective value. `agent/session-prefix` composes request-only prefix messages once per loop instance, and the header records the exact result used. Requests reaching `llm/stream` are deep-frozen, so mutation throws, and carry a process-local loop identity so observers do not confuse separately logged frozen auxiliary calls with conversation requests.
+`agent/request` receives a frozen call-config seed and may return a replacement to switch provider, model, reasoning effort, or sampling. After the waterfall, the loop prepares the exact model capability under the turn signal, rejects unsupported explicit effort ids without clamping, materializes an adapter-configured default, and logs the effective value. The prepared call keeps one adapter registration through dispatch. `agent/session-prefix` composes request-only prefix messages once per loop instance, and the header records the exact result used. Requests reaching `llm/stream` are deep-frozen, so mutation throws, and carry a process-local loop identity so observers do not confuse separately logged frozen auxiliary calls with conversation requests.
 
 On the wire, a loop-built request reads in this order: the `system` slot (the rendered prompt assembly) → `messagePrefix` (the frozen session prefix) → the derived history — the boundary snapshot, whose tail is the newest `user/message` on a turn's first step and the previous step's tool results on later steps. The prefix never enters the derived history; its durable record is the header events, and the dev invariant recomputes exactly this equation against every loop-built request.
 
diff --git a/docs/core-data-structures/core.zh.md b/docs/core-data-structures/core.zh.md
index 7c5298659c..b81c6a90ab 100644
--- a/docs/core-data-structures/core.zh.md
+++ b/docs/core-data-structures/core.zh.md
@@ -331,7 +331,7 @@ interface ToolSchema {
 
 循环从已记录状态构建每个请求。`EpochHeader` 通过完整的 `request/header` 快照记录调用配置、渲染后的提示词、权威返回工具顺序（由 `toolOrder` 配置；未配置时按字典序）以及会话前缀。结合派生历史，请求便可由会话日志重建。见 [session.md](session.md#the-request-header-event-requestheader) 与[可重建性 Agent Note（agent 决策记录）](../../.agents/notes/implemented/architecture/2026-07-05-reconstructable-requests.md)。
 
-`agent/request` 接收冻结的调用配置种子，并可返回替代值以切换提供方、模型、推理强度或采样参数。循环在 waterfall 之后解析确切模型的能力，拒绝显式指定但不受支持的推理强度 ID（不自动调整），填入适配器配置的默认值，并记录最终生效值。`agent/session-prefix` 为每个循环实例组合一次仅用于请求的 prefix 消息，header 记录实际使用的确切结果。到达 `llm/stream` 的请求会被深度冻结，因此变更会抛异常；请求还携带进程本地循环标识，使观察者不会把单独记录的冻结辅助调用误认成对话请求。
+`agent/request` 接收冻结的调用配置种子，并可返回替代值以切换提供方、模型、推理强度或采样参数。waterfall 结束后，循环会在轮次信号控制下完成确切模型的能力准备，拒绝显式指定但不受支持的推理强度 ID（不自动调整），填入适配器配置的默认值，并记录最终生效值。准备完成的调用直至分派完成始终持有同一项适配器注册。`agent/session-prefix` 为每个循环实例组合一次仅用于请求的 prefix 消息，header 记录实际使用的确切结果。到达 `llm/stream` 的请求会被深度冻结，因此变更会抛异常；请求还携带进程本地循环标识，使观察者不会把单独记录的冻结辅助调用误认成对话请求。
 
 在协议格式上，循环构建的请求按此顺序读取：`system` 槽位（渲染后的提示词组装）→ `messagePrefix`（冻结的会话前缀）→ 派生历史——边界快照，其尾部在轮次首步是最新的 `user/message`，在后续步骤是上一步的工具结果。前缀从不进入派生历史；它的持久记录是 header 事件，开发不变式针对每个循环构建的请求精确重算此等式。
 
diff --git a/docs/core-data-structures/llm-streaming.i18n.yaml b/docs/core-data-structures/llm-streaming.i18n.yaml
index 214627ce54..6029b11adc 100644
--- a/docs/core-data-structures/llm-streaming.i18n.yaml
+++ b/docs/core-data-structures/llm-streaming.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-llm-streaming.md: eec706d3b440833a78d7f5c24e716809fe961ffd
-llm-streaming.zh.md: ce93bb905234b6369389c7d9615ed621f224bc6d
+llm-streaming.md: 5155e680bfcc253d06742c7d3adcddc633a01c52
+llm-streaming.zh.md: 9b928253065da99c74f749b905375d790fab46c9
diff --git a/docs/core-data-structures/llm-streaming.md b/docs/core-data-structures/llm-streaming.md
index eec706d3b4..5155e680bf 100644
--- a/docs/core-data-structures/llm-streaming.md
+++ b/docs/core-data-structures/llm-streaming.md
@@ -157,7 +157,23 @@ declare class BlockAssembler {
 
 ## The seam
 
-`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. The separate `resolveModelContext()` query exposes correctness-sensitive capacity, while `resolveModelReasoning()` exposes ordered model-owned effort ids and an optional deployment default; absence from either query means unavailable metadata or capability, not invalid catalog membership. The service validates and materializes reasoning through `resolveCallConfig()` at the final adapter boundary, so direct calls cannot bypass unsupported-effort rejection. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
+`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. The separate `resolveModelContext()` query exposes correctness-sensitive capacity, while `resolveModelReasoning()` exposes ordered model-owned effort ids and an optional deployment default; absence from either query means unavailable metadata or capability, not invalid catalog membership. A reasoning resolver receives optional cancellation and must settle promptly after abort. The service validates and materializes reasoning through `resolveCallConfig()` at the final adapter boundary, so direct calls cannot bypass unsupported-effort rejection; direct dispatch captures one registration before awaiting that resolution. The agent loop instead uses `prepareCall()` to keep the same registration across capability resolution, durable header logging, and dispatch. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
+
+```ts type-equiv
+/** One model call whose config and adapter registration were resolved together. */
+interface PreparedLlmCall {
+  /** Detached, deep-frozen config with any adapter-owned default materialized. */
+  readonly config: LlmCallConfig
+  /**
+   * Dispatch this call once through the registration captured during
+   * preparation. The request's call-config fields must match {@link config};
+   * reuse or mismatch fails with `INVALID_PREPARED_CALL`.
+   * @param options - fully assembled request carrying the prepared config.
+   * @returns the chunk stream, including the `llm/stream` waterfall.
+   */
+  stream(options: GenerateOptions): AsyncIterable<StreamChunk>
+}
+```
 
 ```ts public-api
 /**
@@ -197,11 +213,14 @@ declare abstract class LlmAdapter {
    * the model has no selectable reasoning-effort capability.
    * @param _provider - one provider route owned by this adapter.
    * @param _model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; implementations
+   *   must settle promptly after it aborts.
    * @returns adapter-owned effort metadata, or `undefined` when unsupported.
    */
   resolveModelReasoning(
     _provider: string,
     _model: string,
+    _signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined>;
   /**
    * Stream one model call as raw chunks. The only required method.
diff --git a/docs/core-data-structures/llm-streaming.zh.md b/docs/core-data-structures/llm-streaming.zh.md
index ce93bb9052..9b92825306 100644
--- a/docs/core-data-structures/llm-streaming.zh.md
+++ b/docs/core-data-structures/llm-streaming.zh.md
@@ -157,7 +157,23 @@ declare class BlockAssembler {
 
 ## seam
 
-`LlmAdapter` 是提供方 seam：创建子类、实现 `stream()`，再用 `ctx.llm.registerAdapter(providers, adapter)` 注册一个适配器实例。`GenerateOptions.provider` 选择已注册适配器；`GenerateOptions.model` 会传给该适配器，无需在生命周期启动时注册。重复提供方路由会原子失败。可选的 `providerInfo()` 与异步 `listModels()` 方法为 `LlmService.listProviders()` / `listModels()` 提供分离的 selector 元数据。该目录仅供参考，不是请求白名单：适配器仍是权威，并可接受未列出的模型 id。单独的 `resolveModelContext()` 查询暴露对正确性敏感的容量信息，`resolveModelReasoning()` 则暴露由模型持有的有序推理强度 ID 和可选的部署默认值；任一查询返回缺失都表示元数据或能力不可用，而不表示目录成员关系无效。服务通过最终适配器边界的 `resolveCallConfig()` 校验推理强度并填入默认值，因此直接调用也无法绕过对不支持推理强度的拒绝。适配器查找发生在 `llm/stream` waterfall（瀑布式事件）的终端 continuation，因此 listener 可以在查找前短路调用，或路由一个可变的一次性请求。`block-start` / `block-end` 的 `index` 关联与 assembler 共同意味着适配器只需 emit 格式正确的分片——块重组不是每个适配器各自的问题。消费方 surface（`ctx.llm.stream()`）与 `llm/stream` waterfall 见 [architecture.md § 内容块与流式传输](../architecture.md#content-blocks-and-streaming-dsh-llm)。
+`LlmAdapter` 是提供方 seam：创建子类、实现 `stream()`，再用 `ctx.llm.registerAdapter(providers, adapter)` 注册一个适配器实例。`GenerateOptions.provider` 选择已注册适配器；`GenerateOptions.model` 会传给该适配器，无需在生命周期启动时注册。重复提供方路由会原子失败。可选的 `providerInfo()` 与异步 `listModels()` 方法为 `LlmService.listProviders()` / `listModels()` 提供分离的 selector 元数据。该目录仅供参考，不是请求白名单：适配器仍是权威，并可接受未列出的模型 id。单独的 `resolveModelContext()` 查询暴露对正确性敏感的容量信息，`resolveModelReasoning()` 则暴露由模型持有的有序推理强度 ID 和可选的部署默认值；任一查询返回缺失都表示元数据或能力不可用，而不表示目录成员关系无效。推理能力解析器会接收可选的取消信号，并且必须在信号中止后迅速完成结算。服务通过最终适配器边界的 `resolveCallConfig()` 校验推理强度并填入默认值，因此直接调用也无法绕过对不支持推理强度的拒绝；直接分派会在等待解析前捕获一项适配器注册。agent loop 则使用 `prepareCall()`，使能力解析、请求头持久记录和分派全程使用同一项注册。适配器查找发生在 `llm/stream` waterfall（瀑布式事件）的终端 continuation，因此 listener 可以在查找前短路调用，或路由一个可变的一次性请求。`block-start` / `block-end` 的 `index` 关联与 assembler 共同意味着适配器只需 emit 格式正确的分片——块重组不是每个适配器各自的问题。消费方 surface（`ctx.llm.stream()`）与 `llm/stream` waterfall 见 [architecture.md § 内容块与流式传输](../architecture.md#content-blocks-and-streaming-dsh-llm)。
+
+```ts type-equiv
+/** One model call whose config and adapter registration were resolved together. */
+interface PreparedLlmCall {
+  /** Detached, deep-frozen config with any adapter-owned default materialized. */
+  readonly config: LlmCallConfig
+  /**
+   * Dispatch this call once through the registration captured during
+   * preparation. The request's call-config fields must match {@link config};
+   * reuse or mismatch fails with `INVALID_PREPARED_CALL`.
+   * @param options - fully assembled request carrying the prepared config.
+   * @returns the chunk stream, including the `llm/stream` waterfall.
+   */
+  stream(options: GenerateOptions): AsyncIterable<StreamChunk>
+}
+```
 
 ```ts public-api
 /**
@@ -197,11 +213,14 @@ declare abstract class LlmAdapter {
    * the model has no selectable reasoning-effort capability.
    * @param _provider - one provider route owned by this adapter.
    * @param _model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; implementations
+   *   must settle promptly after it aborts.
    * @returns adapter-owned effort metadata, or `undefined` when unsupported.
    */
   resolveModelReasoning(
     _provider: string,
     _model: string,
+    _signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined>;
   /**
    * Stream one model call as raw chunks. The only required method.
diff --git a/docs/user/develop/practice/llm-adapter.i18n.yaml b/docs/user/develop/practice/llm-adapter.i18n.yaml
index c179b693f8..05d30a8b62 100644
--- a/docs/user/develop/practice/llm-adapter.i18n.yaml
+++ b/docs/user/develop/practice/llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-llm-adapter.md: f133a6ea748fd7d9bfa6a7adaa35b1cccbe40c8e
-llm-adapter.zh.md: 3a4fd9a516126d1f9fa8675fdc87be10c85f3489
+llm-adapter.md: 8b8201099e2b78385301daf14605be53436353c7
+llm-adapter.zh.md: 735b7a4babf3d31652374e91cbce794cdd580725
diff --git a/docs/user/develop/practice/llm-adapter.md b/docs/user/develop/practice/llm-adapter.md
index f133a6ea74..8b8201099e 100644
--- a/docs/user/develop/practice/llm-adapter.md
+++ b/docs/user/develop/practice/llm-adapter.md
@@ -112,7 +112,7 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 `stream()` receives the exported `GenerateOptions` type. It includes the model, adapter-owned reasoning-effort id, conversation history, system prompt, tool schemas, generation parameters, stop sequences, and abort signal; treat the TypeScript type exported by `@deepseek-ai/dsh-llm` as authoritative. Map supported fields to the provider API. If the provider cannot honor a field, throw `LlmError` with a stable code instead of silently dropping it.
 
-Override `resolveModelReasoning(provider, model)` when an exact model exposes selectable reasoning strengths. Return ordered opaque ids and display names plus an optional configured default; do not promote provider names into a core enum. The service validates the metadata and rejects unsupported explicit values before `stream()`. Returning `undefined` means that model has no selectable reasoning-effort capability.
+Override `resolveModelReasoning(provider, model, signal?)` when an exact model exposes selectable reasoning strengths. Return ordered opaque ids and display names plus an optional configured default; do not promote provider names into a core enum. Honor the optional signal for asynchronous lookup so cancellation and disposal reach quiescence. The service validates the metadata and rejects unsupported explicit values before `stream()`. Returning `undefined` means that model has no selectable reasoning-effort capability.
 
 ## Register an adapter
 
diff --git a/docs/user/develop/practice/llm-adapter.zh.md b/docs/user/develop/practice/llm-adapter.zh.md
index 3a4fd9a516..735b7a4bab 100644
--- a/docs/user/develop/practice/llm-adapter.zh.md
+++ b/docs/user/develop/practice/llm-adapter.zh.md
@@ -112,7 +112,7 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 `stream()` 接收仓库导出的 `GenerateOptions`。它包含模型名、由适配器持有的推理强度 ID、对话历史、系统提示词、tool schema、生成参数、停止序列和中止信号；完整字段以 `@deepseek-ai/dsh-llm` 导出的 TypeScript 类型为准。适配器必须将支持的字段映射到具体 API；无法支持的字段应抛出带稳定 code 的 `LlmError`，不能静默丢弃。
 
-当某个具体模型提供可选推理强度时，请覆写 `resolveModelReasoning(provider, model)`。返回有序的不透明 ID、展示名称，以及可选的配置默认值；不要将提供方使用的等级名称提升为核心枚举。服务会校验元数据，并在调用 `stream()` 前拒绝显式指定但不受支持的值。返回 `undefined` 表示该模型没有可选的推理强度能力。
+当某个具体模型提供可选推理强度时，请覆写 `resolveModelReasoning(provider, model, signal?)`。返回有序的不透明 ID、展示名称，以及可选的配置默认值；不要将提供方使用的等级名称提升为核心枚举。异步查询必须响应这个可选信号，让取消和资源释放都能达到完全停稳。服务会校验元数据，并在调用 `stream()` 前拒绝显式指定但不受支持的值。返回 `undefined` 表示该模型没有可选的推理强度能力。
 
 ## 注册适配器
 
diff --git a/packages/cordis/tool-cordis/src/api-catalog.ts b/packages/cordis/tool-cordis/src/api-catalog.ts
index 7e1fd88926..5286e76d06 100644
--- a/packages/cordis/tool-cordis/src/api-catalog.ts
+++ b/packages/cordis/tool-cordis/src/api-catalog.ts
@@ -385,16 +385,20 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
         jsDoc: '/**\n * Resolve context capacity from the adapter that owns one exact route.\n * This query is independent of the advisory model catalog: an unlisted model\n * may return metadata, while `undefined` never rejects later routing.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @returns detached context metadata, or `undefined` when the adapter has none.\n */',
       },
       {
-        signature: 'async resolveModelReasoning( provider: string, model: string, ): Promise<LlmModelReasoningInfo | undefined>',
-        jsDoc: '/**\n * Resolve selectable reasoning efforts from the adapter that owns one exact\n * route. Metadata is validated and detached; an absent result means an\n * effort selector is unsupported for that model.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @returns detached reasoning metadata, or `undefined` when unsupported.\n */',
+        signature: 'async resolveModelReasoning( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmModelReasoningInfo | undefined>',
+        jsDoc: '/**\n * Resolve selectable reasoning efforts from the adapter that owns one exact\n * route. Metadata is validated and detached; an absent result means an\n * effort selector is unsupported for that model.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @param signal - optional cancellation for adapter-owned asynchronous lookup.\n * @returns detached reasoning metadata, or `undefined` when unsupported.\n */',
       },
       {
-        signature: 'async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>',
-        jsDoc: '/**\n * Validate a conversation call config against its exact model capability and\n * materialize an adapter-configured default. Unsupported explicit efforts\n * reject before provider I/O; no clamping or aliasing is performed.\n * @param config - provider/model route and optional request controls.\n * @returns a detached config only when a default must be materialized.\n */',
+        signature: 'async resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>',
+        jsDoc: '/**\n * Validate a conversation call config against its exact model capability and\n * materialize an adapter-configured default. Unsupported explicit efforts\n * reject before provider I/O; no clamping or aliasing is performed. This\n * standalone query does not bind a later dispatch; use {@link prepareCall}\n * when logging and streaming must share one adapter registration.\n * @param config - provider/model route and optional request controls.\n * @param signal - optional cancellation for adapter-owned capability lookup.\n * @returns a detached config only when a default must be materialized.\n */',
+      },
+      {
+        signature: 'async prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall>',
+        jsDoc: '/**\n * Resolve one call under its current adapter registration. The returned\n * one-shot handle keeps that registration across header logging and dispatch,\n * so HMR cannot combine one adapter\'s capability result with another adapter.\n * @param config - provider/model route and optional request controls.\n * @param signal - optional cancellation for adapter-owned capability lookup.\n * @returns a prepared config and its registration-bound stream entry point.\n */',
       },
       {
         signature: 'stream(options: GenerateOptions): AsyncIterable<StreamChunk>',
-        jsDoc: '/**\n * Stream one model call as raw chunks (token-level deltas). Throws\n * `LlmError` with code `NO_ADAPTER` if no adapter is registered for\n * `options.provider`. Replay state is retained only when the same adapter\n * instance owns its historical provider and the target provider. Final\n * adapter selection, dispatch, and iteration failures retain their original\n * Error identity and are tagged in a call-local scope for narrow agent-loop\n * request recovery; middleware and nested-call failures remain untagged for\n * the outer call.\n * @param options - the full request; `options.provider` selects the adapter.\n * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.\n */',
+        jsDoc: '/**\n * Stream one model call as raw chunks (token-level deltas). Throws\n * `LlmError` with code `NO_ADAPTER` if no adapter is registered for\n * `options.provider`. Replay state is retained only when the same adapter\n * instance owns its historical provider and the target provider. Final\n * adapter selection remains fixed through asynchronous reasoning resolution\n * and dispatch. Selection, dispatch, and iteration failures retain their\n * original Error identity and are tagged in a call-local scope for narrow\n * agent-loop request recovery; middleware and nested-call failures remain\n * untagged for the outer call.\n * @param options - the full request; `options.provider` selects the adapter.\n * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.\n */',
       },
     ],
   },
@@ -1702,7 +1706,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'LlmAdapter',
-    declaration: 'export abstract class LlmAdapter {\n    providerInfo(provider: string): LlmProviderInfo;\n    listModels(_provider: string): Promise<readonly LlmModelInfo[]>;\n    resolveModelContext(_provider: string, _model: string): Promise<LlmModelContext | undefined>;\n    resolveModelReasoning(_provider: string, _model: string): Promise<LlmModelReasoningInfo | undefined>;\n    abstract stream(options: GenerateOptions): AsyncIterable<StreamChunk>;\n}',
+    declaration: 'export abstract class LlmAdapter {\n    providerInfo(provider: string): LlmProviderInfo;\n    listModels(_provider: string): Promise<readonly LlmModelInfo[]>;\n    resolveModelContext(_provider: string, _model: string): Promise<LlmModelContext | undefined>;\n    resolveModelReasoning(_provider: string, _model: string, _signal?: AbortSignal): Promise<LlmModelReasoningInfo | undefined>;\n    abstract stream(options: GenerateOptions): AsyncIterable<StreamChunk>;\n}',
   },
   {
     name: 'LlmCallConfig',
@@ -1756,6 +1760,10 @@ export const TYPE_API: readonly TypeApiEntry[] = [
     name: 'OutOfBandSessionEventType',
     declaration: 'export type OutOfBandSessionEventType = Exclude<Extract<SessionEventType, keyof OutOfBandSessionEventMap>, SurfaceEventType>;',
   },
+  {
+    name: 'PreparedLlmCall',
+    declaration: 'export interface PreparedLlmCall {\n    readonly config: LlmCallConfig;\n    stream(options: GenerateOptions): AsyncIterable<StreamChunk>;\n}',
+  },
   {
     name: 'PreparedReferencedMessage',
     declaration: 'export interface PreparedReferencedMessage {\n    content: ContentBlock[];\n    contexts: HookContext[];\n}',
diff --git a/packages/core/agent-loop/README.md b/packages/core/agent-loop/README.md
index f7387c2309..d02ea91527 100644
--- a/packages/core/agent-loop/README.md
+++ b/packages/core/agent-loop/README.md
@@ -60,7 +60,7 @@ The driver owns one agent for its lifetime and runs inside `ctx.agents.withIniti
 
 Every provider call that reaches a successful finish appends exactly one `assistant/message` completion anchor, including content-less calls and `max-tokens` finishes. A successful `agent/step-result` stores its transformed content; a rejected result records empty content before the original failure continues. The anchor retains exact chunk provenance (`[]` for a stream with no chunks) and usage when available, while empty content stays out of derived message history.
 
-After `agent/request` returns a provider/model call config, the loop asks `ctx.llm.resolveCallConfig()` to validate any adapter-owned reasoning effort and materialize its configured default. The effective config is logged in the full `request/header` before dispatch, so a listener can change effort between steps without hidden request drift. A route with no registered adapter preserves the proposed config so an `llm/stream` listener can own and short-circuit it; unhandled terminal dispatch still fails with `NO_ADAPTER`. A new loop instance restores the last effort only when its initial provider/model route exactly matches the logged route; a route change discards that opaque model-owned ID and resolves the new model independently.
+After `agent/request` returns a provider/model call config, the loop asks `ctx.llm.prepareCall()` to validate any adapter-owned reasoning effort and materialize its configured default under the active turn signal. The prepared call retains the exact adapter registration across this asynchronous resolution, `request/header` logging, and terminal dispatch, so HMR cannot mix one adapter's capability result with another adapter's request. The effective config is logged before dispatch, so a listener can change effort between steps without hidden request drift. A route with no registered adapter preserves the proposed config so an `llm/stream` listener can own and short-circuit it; unhandled terminal dispatch still fails with `NO_ADAPTER`. A new loop instance restores the last effort only when its initial provider/model route exactly matches the logged route; a route change discards that opaque model-owned ID and resolves the new model independently.
 
 Plugin failure ends the current turn, not the loop. Only final adapter dispatch/iteration failures and terminal in-band error or aborted finishes enter `agent/request-error`; middleware, result processing, tools, and `agent/post-step` remain ordinary turn failures. Recovery receives the exact live error, immutable provider facts, and immutable prior failures after the failed step closes. A retry rebuilds from the durable log in a new numbered step, success clears the consecutive history, and exhaustion records the structured failure once on `turn/end`. AgentLoop privately owns one cancellation holder whose explicit signal spans prompt policy, assembly, every step, model and tool work, recovery, continuation, and terminal stop; it retires the holder immediately before publishing `turn/end`, while the driver may remain `running` through the durability flush. An effective `cancel()` emits the typed runtime-only `user | parent` cause before clearing pending work and cooperatively aborting the holder; notification failures cannot veto cancellation, work queued by a notification observer is cleared, work queued by a later abort observer belongs to the next turn, and idle cancellation emits nothing. Durable `turn/end` remains coarse `aborted`; undispatched model tool calls receive synthetic `tool/call` and `ABORTED_BEFORE_DISPATCH` result pairs. Disposal wins terminal classification, and work that ignores the signal must settle before quiescence. The [explicit-cancellation decision](../../../.agents/notes/implemented/architecture/2026-07-16-explicit-turn-cancellation.md) owns the lifecycle and race contract. Terminal continuation stops remain authoritative through turn close and durability flush.
 
diff --git a/packages/core/agent-loop/src/loop.ts b/packages/core/agent-loop/src/loop.ts
index b2cdefb19e..2484cd5793 100644
--- a/packages/core/agent-loop/src/loop.ts
+++ b/packages/core/agent-loop/src/loop.ts
@@ -7,7 +7,7 @@
 
 import { randomUUID } from 'node:crypto'
 import type { Context } from 'cordis'
-import type { ContentBlock, FinishReason, GenerateOptions, LlmCallConfig, LlmFailure, Message } from '@deepseek-ai/dsh-llm'
+import type { ContentBlock, FinishReason, GenerateOptions, LlmCallConfig, LlmFailure, Message, PreparedLlmCall } from '@deepseek-ai/dsh-llm'
 import { isDeepStrictEqual } from 'node:util'
 import { BlockAssembler, HarnessError, LlmError, assertNever, deepFreeze, errorChain, llmFailureOf, markAgentLoopRequest } from '@deepseek-ai/dsh-llm'
 import { agentEvents, agentInterruptReasonOf, assembleContextFor, AgentMessageId } from '@deepseek-ai/dsh-agent'
@@ -689,8 +689,10 @@ async function runStep(
     throw new Error(`agent "${agent.id}" has no provider/model: set AgentOptions.provider and AgentOptions.model or supply both via the agent/request waterfall`)
   }
   let config: LlmCallConfig
+  let preparedCall: PreparedLlmCall | undefined
   try {
-    config = await ctx.llm.resolveCallConfig(proposedConfig)
+    preparedCall = await ctx.llm.prepareCall(proposedConfig, signal)
+    config = preparedCall.config
   } catch (error: unknown) {
     // A waterfall listener may own and short-circuit a route with no adapter.
     // Terminal dispatch still raises NO_ADAPTER when no listener handles it.
@@ -731,7 +733,7 @@ async function runStep(
   // --- Model call (streaming-first; raw chunks are the replay record) ---
   const assembler = new BlockAssembler()
   const chunkSeqs: number[] = []
-  const stream = ctx.llm.stream(request)
+  const stream = preparedCall?.stream(request) ?? ctx.llm.stream(request)
   try {
     for await (const chunk of stream) {
       interruptionCheckpoint(signal)
diff --git a/packages/core/agent-loop/tests/request-reconstruction.spec.ts b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
index 40b26e5a3e..1c101289f1 100644
--- a/packages/core/agent-loop/tests/request-reconstruction.spec.ts
+++ b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
@@ -8,7 +8,7 @@
 import { describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
 import LlmService, { LlmError, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
-import type { GenerateOptions } from '@deepseek-ai/dsh-llm'
+import type { GenerateOptions, LlmModelReasoningInfo } from '@deepseek-ai/dsh-llm'
 import SessionStore, { Session, SessionId, foldRequestHeader } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
 import ToolRegistry, { defineContentToolFixture } from '@deepseek-ai/dsh-tools'
@@ -152,6 +152,88 @@ describe('request stability across the loop', () => {
     expect(resumedHeaders.at(-1)?.data.reason).toBe('resume')
   })
 
+  it('keeps reasoning resolution, request logging, and dispatch on one adapter registration', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    await ctx.plugin(SessionStore)
+    await ctx.plugin(SystemPrompt, { persona: 'stable base' })
+    await ctx.plugin(ToolRegistry)
+    await ctx.plugin(AgentRegistry)
+    await ctx.plugin(AgentLoop, { agents: [] })
+    const started = Promise.withResolvers<undefined>()
+    const reasoning = Promise.withResolvers<LlmModelReasoningInfo>()
+    const first = new class extends MockAdapter {
+      override resolveModelReasoning(
+        _provider: string,
+        _model: string,
+        _signal?: AbortSignal,
+      ): typeof reasoning.promise {
+        started.resolve(undefined)
+        return reasoning.promise
+      }
+    }([textResponse('first')])
+    const second = new MockAdapter([textResponse('second')], {
+      efforts: [{ id: ReasoningEffortId('max'), name: 'Max' }],
+      defaultEffort: ReasoningEffortId('max'),
+    })
+    const disposeFirst = ctx.llm.registerAdapter(['mock'], first)
+    const agent = ctx.agentLoop.create(SessionId('effort-hmr'), { provider: 'mock', model: 'mock' })
+
+    send(agent, 'go')
+    await started.promise
+    disposeFirst()
+    ctx.llm.registerAdapter(['mock'], second)
+    reasoning.resolve({
+      efforts: [{ id: ReasoningEffortId('high'), name: 'High' }],
+      defaultEffort: ReasoningEffortId('high'),
+    })
+    await waitForIdle(ctx, agent)
+
+    expect(first.requests.map(request => request.reasoningEffort)).toEqual([
+      ReasoningEffortId('high'),
+    ])
+    expect(second.requests).toHaveLength(0)
+    const headers = agent.session.events.filter(event => event.type === 'request/header')
+    expect(headers.at(-1)?.data.header.config.reasoningEffort).toBe(ReasoningEffortId('high'))
+  })
+
+  it('aborts a blocked reasoning lookup before quiescent disposal completes', async () => {
+    const started = Promise.withResolvers<AbortSignal>()
+    const adapter = new class extends MockAdapter {
+      override resolveModelReasoning(
+        _provider: string,
+        _model: string,
+        signal?: AbortSignal,
+      ): Promise<never> {
+        if (signal === undefined) return Promise.reject(new Error('missing reasoning signal'))
+        started.resolve(signal)
+        return new Promise((_resolve, reject) => {
+          if (signal.aborted) {
+            reject(signal.reason instanceof Error ? signal.reason : new Error('reasoning aborted'))
+            return
+          }
+          signal.addEventListener('abort', () => {
+            reject(signal.reason instanceof Error ? signal.reason : new Error('reasoning aborted'))
+          }, { once: true })
+        })
+      }
+    }([])
+    const ctx = await harness(adapter)
+    const handle = await ctx.agents.create({
+      sessionId: SessionId('reasoning-dispose'),
+      agentOptions: { provider: 'mock', model: 'mock' },
+    })
+
+    send(handle.agent, 'go')
+    const signal = await started.promise
+    await handle.dispose()
+
+    expect(signal.aborted).toBe(true)
+    expect(handle.agent.status).toBe('disposed')
+    expect(adapter.requests).toHaveLength(0)
+    expect(handle.agent.session.events.some(event => event.type === 'request/header')).toBe(false)
+  })
+
   it.each(['plain error', 'LLM error'] as const)(
     'does not swallow a %s from reasoning resolution',
     async (kind) => {
diff --git a/packages/llm/llm-deepseek/src/adapter.ts b/packages/llm/llm-deepseek/src/adapter.ts
index d8ca534c89..8eabdc499f 100644
--- a/packages/llm/llm-deepseek/src/adapter.ts
+++ b/packages/llm/llm-deepseek/src/adapter.ts
@@ -147,6 +147,7 @@ export class DeepSeekAdapter extends LlmAdapter {
   override resolveModelReasoning(
     _provider: string,
     _model: string,
+    _signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined> {
     if (this.options.defaults?.thinking === 'disabled') return Promise.resolve(undefined)
     return Promise.resolve({
diff --git a/packages/llm/llm-pi-ai/src/adapter.ts b/packages/llm/llm-pi-ai/src/adapter.ts
index 1dde9d9dc7..de91f2daa7 100644
--- a/packages/llm/llm-pi-ai/src/adapter.ts
+++ b/packages/llm/llm-pi-ai/src/adapter.ts
@@ -142,6 +142,7 @@ export class PiAiAdapter extends LlmAdapter {
   override resolveModelReasoning(
     provider: string,
     model: string,
+    _signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined> {
     const profile = this.profiles.get(provider)
     if (profile === undefined) {
diff --git a/packages/llm/llm/README.md b/packages/llm/llm/README.md
index cf561b9032..2a13e386f2 100644
--- a/packages/llm/llm/README.md
+++ b/packages/llm/llm/README.md
@@ -12,8 +12,9 @@ An adapter registry plus a single streaming call surface, interceptable via a wa
 - `ctx.llm.listProviders(): LlmProviderInfo[]` Describe registered provider routes in registration order.
 - `ctx.llm.listModels(provider: string): Promise<LlmModelInfo[]>` Discover the models one registered provider currently advertises.
 - `ctx.llm.resolveModelContext(provider: string, model: string): Promise<LlmModelContext | undefined>` Resolve authoritative context capacity for one exact route from its owning adapter.
-- `ctx.llm.resolveModelReasoning(provider: string, model: string): Promise<LlmModelReasoningInfo | undefined>` Resolve ordered adapter-owned reasoning efforts and an optional deployment default for one exact route.
-- `ctx.llm.resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig>` Validate an explicit effort and materialize an adapter-configured default without clamping.
+- `ctx.llm.resolveModelReasoning(provider: string, model: string, signal?: AbortSignal): Promise<LlmModelReasoningInfo | undefined>` Resolve ordered adapter-owned reasoning efforts and an optional deployment default for one exact route, with optional cancellation for asynchronous adapters.
+- `ctx.llm.resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>` Validate an explicit effort and materialize an adapter-configured default without clamping.
+- `ctx.llm.prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall>` Resolve a config and capture its current adapter registration as one cancellable, one-shot call.
 - `ctx.llm.stream(options: GenerateOptions): AsyncIterable<StreamChunk>` Stream one model call as raw chunks (token-level deltas). Consumers assemble the chunks into blocks/messages with `BlockAssembler`.
 
 `LlmService` preserves errors from final adapter selection, synchronous dispatch, iterator construction, and iteration, and binds their provenance to the exact stream handle returned for that model call. `isLlmAdapterFailure(stream, value)` reports only errors from that call's final adapter boundary; `llmFailureOf(stream, value)` returns the adjacent immutable `LlmFailure`. Nested model calls, `llm/stream` middleware, and downstream consumer failures remain unclassified for the outer call. Classification never replaces or mutates the adapter's original coded `Error`.
@@ -22,7 +23,7 @@ Provider and model metadata is a discovery surface, not a routing whitelist. `re
 
 Context capacity is a separate correctness query, not a catalog decoration or global LLM setting. `resolveModelContext()` asks the adapter that owns the exact provider/model route; an adapter can describe an unlisted dynamic model, and `undefined` means only that capacity is unavailable. Invalid returned capacity fails with `INVALID_MODEL_CONTEXT`.
 
-Reasoning effort is also an exact-route capability, but its identifiers are opaque adapter-owned strings rather than a core enum. `resolveModelReasoning()` validates and detaches the ordered display metadata; `undefined` means the model has no selectable effort. `resolveCallConfig()` accepts only an exact advertised identifier, materializes `defaultEffort` when present, and otherwise preserves the provider default. Invalid capability metadata fails with `INVALID_MODEL_REASONING`; an unsupported explicit or configured effort fails with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+Reasoning effort is also an exact-route capability, but its identifiers are opaque adapter-owned strings rather than a core enum. `resolveModelReasoning()` validates and detaches the ordered display metadata; `undefined` means the model has no selectable effort. `resolveCallConfig()` accepts only an exact advertised identifier, materializes `defaultEffort` when present, and otherwise preserves the provider default. Asynchronous resolvers receive the caller's signal and must settle promptly after cancellation. `prepareCall()` additionally retains the exact adapter registration through header logging and terminal dispatch, so HMR cannot combine one adapter's capability result with another adapter's request; reusing its one-shot handle or changing its call-config fields fails with `INVALID_PREPARED_CALL`. Invalid capability metadata fails with `INVALID_MODEL_REASONING`; an unsupported explicit or configured effort fails with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
 
 ### Events
 
@@ -32,7 +33,7 @@ Reasoning effort is also an exact-route capability, but its identifiers are opaq
 
 ### Extension points
 
-- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, `resolveModelContext()` when exact capacity is known, and `resolveModelReasoning()` when a model exposes selectable efforts; the defaults use the route id as its name, advertise no models, and return neither capacity nor reasoning metadata.
+- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, `resolveModelContext()` when exact capacity is known, and `resolveModelReasoning()` when a model exposes selectable efforts; an asynchronous reasoning resolver must honor its optional cancellation signal. The defaults use the route id as its name, advertise no models, and return neither capacity nor reasoning metadata.
 - Wrap `llm/stream` via `ctx.on()` waterfall listeners for caching, logging, or routing. A wrapper that retries after emitting a chunk has no durable attempt boundary; shipped agent retry policy therefore uses `agent/request-error` instead.
 
 ### Content-block vocabulary (`types.ts`)
@@ -43,7 +44,7 @@ Streaming is a raw chunk protocol (`block-start`, `text-delta`, `reasoning-delta
 
 ### Call configuration (`call-config.ts`)
 
-`LlmCallConfig` is the provider, model, optional adapter-owned reasoning effort, and sampling scalars of one conversation's requests (`provider`, `model`, `reasoningEffort`, `temperature`, `maxTokens`, `stop` — each mapping 1:1 onto the same-named `GenerateOptions` field). It is per-conversation state recorded in the session log as part of the request header (see the dsh-session `request/header` events), never a silently-adjustable per-call knob: the `agent/request` waterfall proposes a replacement, `resolveCallConfig()` validates and defaults it, and the loop logs the effective value before dispatch. `callConfigEquals(a, b)` is the field-wise real-change detector; `deepFreeze(value)` is the ownership helper the loop applies to every built request before dispatch (`llm/stream` listeners and adapters read, never rewrite). `markAgentLoopRequest()` gives that exact object process-local loop provenance, and `isAgentLoopRequest()` lets observers distinguish it from independently logged auxiliary calls that may also be frozen and session-associated. `GenerateOptions.purpose` classifies logged auxiliary compaction and session-title calls so adapters can apply purpose-specific transport policy without changing ordinary conversation requests.
+`LlmCallConfig` is the provider, model, optional adapter-owned reasoning effort, and sampling scalars of one conversation's requests (`provider`, `model`, `reasoningEffort`, `temperature`, `maxTokens`, `stop` — each mapping 1:1 onto the same-named `GenerateOptions` field). It is per-conversation state recorded in the session log as part of the request header (see the dsh-session `request/header` events), never a silently-adjustable per-call knob: the `agent/request` waterfall proposes a replacement, `prepareCall()` validates and defaults it under the turn signal, and the loop logs the effective value before using the prepared call's registration-bound stream. `callConfigEquals(a, b)` is the field-wise real-change detector; `deepFreeze(value)` is the ownership helper the loop applies to every built request before dispatch (`llm/stream` listeners and adapters read, never rewrite). `markAgentLoopRequest()` gives that exact object process-local loop provenance, and `isAgentLoopRequest()` lets observers distinguish it from independently logged auxiliary calls that may also be frozen and session-associated. `GenerateOptions.purpose` classifies logged auxiliary compaction and session-title calls so adapters can apply purpose-specific transport policy without changing ordinary conversation requests.
 
 ### App attribution (`attribution.ts`)
 
diff --git a/packages/llm/llm/src/index.ts b/packages/llm/llm/src/index.ts
index e56c2d17a6..824b960232 100644
--- a/packages/llm/llm/src/index.ts
+++ b/packages/llm/llm/src/index.ts
@@ -105,6 +105,20 @@ export class LlmError extends HarnessError {
   }
 }
 
+/** One model call whose config and adapter registration were resolved together. */
+export interface PreparedLlmCall {
+  /** Detached, deep-frozen config with any adapter-owned default materialized. */
+  readonly config: LlmCallConfig
+  /**
+   * Dispatch this call once through the registration captured during
+   * preparation. The request's call-config fields must match {@link config};
+   * reuse or mismatch fails with `INVALID_PREPARED_CALL`.
+   * @param options - fully assembled request carrying the prepared config.
+   * @returns the chunk stream, including the `llm/stream` waterfall.
+   */
+  stream(options: GenerateOptions): AsyncIterable<StreamChunk>
+}
+
 /**
  * Provider-wire adapter for the harness message and stream vocabulary. Register implementations
  * with `ctx.llm.registerAdapter(providers, adapter)`. Every provider HTTP request must include
@@ -151,11 +165,14 @@ export abstract class LlmAdapter {
    * the model has no selectable reasoning-effort capability.
    * @param _provider - one provider route owned by this adapter.
    * @param _model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; implementations
+   *   must settle promptly after it aborts.
    * @returns adapter-owned effort metadata, or `undefined` when unsupported.
    */
   resolveModelReasoning(
     _provider: string,
     _model: string,
+    _signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined> {
     return Promise.resolve(undefined)
   }
@@ -173,7 +190,7 @@ export abstract class LlmAdapter {
  * surface, interceptable via the `llm/stream` waterfall.
  */
 export class LlmService extends Service {
-  private adapters = new Map<string, { adapter: LlmAdapter; provider: LlmProviderInfo }>()
+  private adapters = new Map<string, AdapterRegistration>()
 
   constructor(ctx: Context) {
     super(ctx, 'llm')
@@ -191,7 +208,7 @@ export class LlmService extends Service {
     const dispose = this.ctx.effect(function* (this: LlmService) {
       if (providers.length === 0) throw new LlmError('an adapter must register at least one provider', 'INVALID_ADAPTER')
       const unique = new Set<string>()
-      const registrations: { adapter: LlmAdapter; provider: LlmProviderInfo }[] = []
+      const registrations: AdapterRegistration[] = []
       for (const provider of providers) {
         if (provider.length === 0) throw new LlmError('adapter provider names must be non-empty', 'INVALID_ADAPTER')
         if (unique.has(provider) || this.adapters.has(provider)) {
@@ -284,13 +301,24 @@ export class LlmService extends Service {
    * effort selector is unsupported for that model.
    * @param provider - registered provider route to inspect.
    * @param model - exact model id passed to the adapter.
+   * @param signal - optional cancellation for adapter-owned asynchronous lookup.
    * @returns detached reasoning metadata, or `undefined` when unsupported.
    */
   async resolveModelReasoning(
     provider: string,
     model: string,
+    signal?: AbortSignal,
   ): Promise<LlmModelReasoningInfo | undefined> {
-    const reasoning = await this.registration(provider).adapter.resolveModelReasoning(provider, model)
+    return this.resolveModelReasoningFor(this.registration(provider), model, signal)
+  }
+
+  private async resolveModelReasoningFor(
+    registration: AdapterRegistration,
+    model: string,
+    signal?: AbortSignal,
+  ): Promise<LlmModelReasoningInfo | undefined> {
+    const provider = registration.provider.id
+    const reasoning = await registration.adapter.resolveModelReasoning(provider, model, signal)
     if (reasoning === undefined) return undefined
     if (reasoning.efforts.length === 0) {
       throw new LlmError(
@@ -335,12 +363,23 @@ export class LlmService extends Service {
   /**
    * Validate a conversation call config against its exact model capability and
    * materialize an adapter-configured default. Unsupported explicit efforts
-   * reject before provider I/O; no clamping or aliasing is performed.
+   * reject before provider I/O; no clamping or aliasing is performed. This
+   * standalone query does not bind a later dispatch; use {@link prepareCall}
+   * when logging and streaming must share one adapter registration.
    * @param config - provider/model route and optional request controls.
+   * @param signal - optional cancellation for adapter-owned capability lookup.
    * @returns a detached config only when a default must be materialized.
    */
-  async resolveCallConfig(config: LlmCallConfig): Promise<LlmCallConfig> {
-    const reasoning = await this.resolveModelReasoning(config.provider, config.model)
+  async resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig> {
+    return this.resolveCallConfigFor(this.registration(config.provider), config, signal)
+  }
+
+  private async resolveCallConfigFor(
+    registration: AdapterRegistration,
+    config: LlmCallConfig,
+    signal?: AbortSignal,
+  ): Promise<LlmCallConfig> {
+    const reasoning = await this.resolveModelReasoningFor(registration, config.model, signal)
     const requested = config.reasoningEffort
     if (reasoning === undefined) {
       if (requested !== undefined) {
@@ -362,7 +401,33 @@ export class LlmService extends Service {
     return requested === effective ? config : { ...config, reasoningEffort: effective }
   }
 
-  private registration(provider: string): { adapter: LlmAdapter; provider: LlmProviderInfo } {
+  /**
+   * Resolve one call under its current adapter registration. The returned
+   * one-shot handle keeps that registration across header logging and dispatch,
+   * so HMR cannot combine one adapter's capability result with another adapter.
+   * @param config - provider/model route and optional request controls.
+   * @param signal - optional cancellation for adapter-owned capability lookup.
+   * @returns a prepared config and its registration-bound stream entry point.
+   */
+  async prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall> {
+    const registration = this.registration(config.provider)
+    const resolvedConfig = deepFreeze(structuredClone(
+      await this.resolveCallConfigFor(registration, config, signal),
+    ))
+    let dispatched = false
+    return Object.freeze({
+      config: resolvedConfig,
+      stream: (options: GenerateOptions): AsyncIterable<StreamChunk> => {
+        if (dispatched) {
+          throw new LlmError('a prepared LLM call can only be dispatched once', 'INVALID_PREPARED_CALL')
+        }
+        dispatched = true
+        return this.streamWithRegistration(options, { registration, config: resolvedConfig })
+      },
+    })
+  }
+
+  private registration(provider: string): AdapterRegistration {
     const registration = this.adapters.get(provider)
     if (!registration) throw new LlmError(`no adapter registered for provider "${provider}"`, 'NO_ADAPTER')
     return registration
@@ -395,16 +460,26 @@ export class LlmService extends Service {
   private async * adapterStream(
     options: GenerateOptions,
     failures: AdapterFailureScope,
+    prepared?: { registration: AdapterRegistration; config: LlmCallConfig },
   ): AsyncGenerator<StreamChunk> {
     let iterator: AsyncIterator<StreamChunk>
     try {
-      const resolvedConfig = await this.resolveCallConfig(options)
-      const resolvedOptions = callConfigEquals(options, resolvedConfig)
+      const registration = prepared?.registration ?? this.registration(options.provider)
+      const resolvedConfig = prepared === undefined
+        ? await this.resolveCallConfigFor(registration, options, options.signal)
+        : prepared.config
+      if (prepared !== undefined && !callConfigEquals(options, resolvedConfig)) {
+        throw new LlmError(
+          'prepared LLM call config changed before adapter dispatch',
+          'INVALID_PREPARED_CALL',
+        )
+      }
+      const resolvedOptions = prepared !== undefined || callConfigEquals(options, resolvedConfig)
         ? options
         : Object.isFrozen(options)
           ? deepFreeze({ ...options, ...resolvedConfig })
           : { ...options, ...resolvedConfig }
-      const adapter = this.registration(resolvedOptions.provider).adapter
+      const adapter = registration.adapter
       const stream = adapter.stream(this.forAdapter(resolvedOptions, adapter))
       iterator = stream[Symbol.asyncIterator]()
     } catch (error: unknown) {
@@ -445,18 +520,36 @@ export class LlmService extends Service {
    * `LlmError` with code `NO_ADAPTER` if no adapter is registered for
    * `options.provider`. Replay state is retained only when the same adapter
    * instance owns its historical provider and the target provider. Final
-   * adapter selection, dispatch, and iteration failures retain their original
-   * Error identity and are tagged in a call-local scope for narrow agent-loop
-   * request recovery; middleware and nested-call failures remain untagged for
-   * the outer call.
+   * adapter selection remains fixed through asynchronous reasoning resolution
+   * and dispatch. Selection, dispatch, and iteration failures retain their
+   * original Error identity and are tagged in a call-local scope for narrow
+   * agent-loop request recovery; middleware and nested-call failures remain
+   * untagged for the outer call.
    * @param options - the full request; `options.provider` selects the adapter.
    * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.
    */
   stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
+    return this.streamWithRegistration(options)
+  }
+
+  private streamWithRegistration(
+    options: GenerateOptions,
+    prepared?: { registration: AdapterRegistration; config: LlmCallConfig },
+  ): AsyncIterable<StreamChunk> {
     const failures: AdapterFailureScope = new WeakMap<Error, LlmFailure>()
-    const stream = this.ctx.waterfall(this, 'llm/stream', options, () => this.adapterStream(options, failures))
+    const stream = this.ctx.waterfall(
+      this,
+      'llm/stream',
+      options,
+      () => this.adapterStream(options, failures, prepared),
+    )
     return bindAdapterFailureScope(stream, failures)
   }
 }
 
+interface AdapterRegistration {
+  readonly adapter: LlmAdapter
+  readonly provider: LlmProviderInfo
+}
+
 export default LlmService
diff --git a/packages/llm/llm/tests/service.spec.ts b/packages/llm/llm/tests/service.spec.ts
index 59684ac8b8..cf6636aa5d 100644
--- a/packages/llm/llm/tests/service.spec.ts
+++ b/packages/llm/llm/tests/service.spec.ts
@@ -799,6 +799,115 @@ describe('LlmService', () => {
     expect(Object.isFrozen(adapter.lastOptions)).toBe(true)
   })
 
+  it('pins one adapter registration across asynchronous reasoning resolution and dispatch', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const started = Promise.withResolvers<undefined>()
+    const reasoning = Promise.withResolvers<LlmModelReasoningInfo>()
+    const first = new class extends RecordingAdapter {
+      override resolveModelReasoning(
+        _provider: string,
+        _model: string,
+        _signal?: AbortSignal,
+      ): Promise<LlmModelReasoningInfo> {
+        started.resolve(undefined)
+        return reasoning.promise
+      }
+    }(SCRIPT)
+    const disposeFirst = ctx.llm.registerAdapter(['route'], first)
+    const draining = (async () => {
+      for await (const _chunk of ctx.llm.stream({
+        provider: 'route',
+        model: 'model',
+        messages: [],
+      })) { /* drain */ }
+    })()
+
+    await started.promise
+    disposeFirst()
+    const second = new RecordingAdapter(SCRIPT)
+    ctx.llm.registerAdapter(['route'], second)
+    reasoning.resolve({
+      efforts: [{ id: ReasoningEffortId('high'), name: 'High' }],
+      defaultEffort: ReasoningEffortId('high'),
+    })
+    await draining
+
+    expect(first.lastOptions?.reasoningEffort).toBe(ReasoningEffortId('high'))
+    expect(second.lastOptions).toBeUndefined()
+  })
+
+  it('prepares a one-shot registration-bound call and rejects config drift', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const adapter = new CatalogAdapter(
+      { id: 'route', name: 'Route' },
+      [],
+      {},
+      {
+        model: {
+          efforts: [{ id: ReasoningEffortId('high'), name: 'High' }],
+          defaultEffort: ReasoningEffortId('high'),
+        },
+      },
+    )
+    ctx.llm.registerAdapter(['route'], adapter)
+    const prepared = await ctx.llm.prepareCall({ provider: 'route', model: 'model' })
+    expect(Object.isFrozen(prepared.config)).toBe(true)
+    const stream = prepared.stream({
+      ...prepared.config,
+      model: 'other',
+      messages: [],
+    })
+
+    await expect((async () => {
+      for await (const _chunk of stream) { /* drain */ }
+    })()).rejects.toMatchObject({ code: 'INVALID_PREPARED_CALL' })
+    expect(() => prepared.stream({
+      ...prepared.config,
+      messages: [],
+    })).toThrow(expect.objectContaining({ code: 'INVALID_PREPARED_CALL' }))
+  })
+
+  it('passes cancellation through reasoning capability resolution', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const started = Promise.withResolvers<undefined>()
+    const adapter = new class extends ScriptedAdapter {
+      override resolveModelReasoning(
+        _provider: string,
+        _model: string,
+        signal?: AbortSignal,
+      ): Promise<LlmModelReasoningInfo | undefined> {
+        started.resolve(undefined)
+        return new Promise((_resolve, reject) => {
+          if (signal === undefined) {
+            reject(new Error('missing reasoning signal'))
+            return
+          }
+          if (signal.aborted) {
+            reject(signal.reason instanceof Error ? signal.reason : new Error('reasoning aborted'))
+            return
+          }
+          signal.addEventListener('abort', () => {
+            reject(signal.reason instanceof Error ? signal.reason : new Error('reasoning aborted'))
+          }, { once: true })
+        })
+      }
+    }(SCRIPT)
+    ctx.llm.registerAdapter(['route'], adapter)
+    const controller = new AbortController()
+    const resolving = ctx.llm.resolveCallConfig(
+      { provider: 'route', model: 'model' },
+      controller.signal,
+    )
+
+    await started.promise
+    const reason = new Error('cancel reasoning')
+    controller.abort(reason)
+    await expect(resolving).rejects.toBe(reason)
+  })
+
   it.each([0, -1, 1.5, Number.NaN])(
     'rejects invalid adapter model context %s',
     async (contextWindow) => {
diff --git a/scripts/gen-cordis-catalog.ts b/scripts/gen-cordis-catalog.ts
index dbfc344fc9..30b3ff7518 100644
--- a/scripts/gen-cordis-catalog.ts
+++ b/scripts/gen-cordis-catalog.ts
@@ -93,6 +93,7 @@ export const LINK_MAP: Record<string, string> = {
   CommandResult: 'commands.md',
   CommandSurface: 'commands.md',
   LlmAdapter: 'llm-streaming.md',
+  PreparedLlmCall: 'llm-streaming.md',
   LlmService: 'llm-streaming.md',
   StreamChunk: 'llm-streaming.md',
   CreateSessionOptions: 'persistence.md',
diff --git a/scripts/type-equiv.manifest.json b/scripts/type-equiv.manifest.json
index 6cad65be75..c003b4b199 100644
--- a/scripts/type-equiv.manifest.json
+++ b/scripts/type-equiv.manifest.json
@@ -307,6 +307,11 @@
       "source": "packages/llm/llm/src/assembler.ts",
       "projection": "public-api"
     },
+    {
+      "doc": "docs/core-data-structures/llm-streaming.md",
+      "symbol": "PreparedLlmCall",
+      "source": "packages/llm/llm/src/index.ts"
+    },
     {
       "doc": "docs/core-data-structures/llm-streaming.md",
       "symbol": "LlmAdapter",
@@ -1465,6 +1470,11 @@
       "source": "packages/llm/llm/src/assembler.ts",
       "projection": "public-api"
     },
+    {
+      "doc": "docs/core-data-structures/llm-streaming.zh.md",
+      "symbol": "PreparedLlmCall",
+      "source": "packages/llm/llm/src/index.ts"
+    },
     {
       "doc": "docs/core-data-structures/llm-streaming.zh.md",
       "symbol": "LlmAdapter",

From 99805b63409d1d8d6e08e07c817bdaac5441c657 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 04:30:58 +0800
Subject: [PATCH 21/36] fix(gui): merge adaptations to master's apiproxy
 defaults and deleted test hook

The origin/master merge introduced a required workspaceRoot on
ApiProxyDefaults and deleted the ui-conversation test hook.ts helper.
Add workspaceRoot to the new todo-projection api-proxy test and bind
the todo-panel spec's selector hook via bindSnapshotSelector directly,
matching the sibling specs.
---
 packages/client/ui-conversation/tests/todo-panel.spec.tsx | 4 ++--
 packages/host/apiproxy/tests/api-proxy-view.spec.ts       | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
index 8971bba2cd..5261e1d9e4 100644
--- a/packages/client/ui-conversation/tests/todo-panel.spec.tsx
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -6,7 +6,7 @@
  */
 import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
 import { afterEach, describe, expect, it, vi } from 'vitest'
-import { hookOf } from './hook.ts'
+import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react'
 import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
 import type { TodoItem, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
 import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
@@ -19,7 +19,7 @@ afterEach(cleanup)
 
 function sessionWith(todos: readonly TodoItem[]) {
   const store = createSnapshotStore<{ todos: readonly TodoItem[] }>({ todos })
-  return { store, useSession: hookOf(store) as unknown as UseSession }
+  return { store, useSession: bindSnapshotSelector(store) as unknown as UseSession }
 }
 
 const LIST: TodoItem[] = [
diff --git a/packages/host/apiproxy/tests/api-proxy-view.spec.ts b/packages/host/apiproxy/tests/api-proxy-view.spec.ts
index f30babc296..4263c53cea 100644
--- a/packages/host/apiproxy/tests/api-proxy-view.spec.ts
+++ b/packages/host/apiproxy/tests/api-proxy-view.spec.ts
@@ -156,7 +156,7 @@ describe('mux live view computation', () => {
 
   it('tail page carries the full-log todo projection; older pages and todo-less sessions omit it', async () => {
     const { ctx } = await harness()
-    const api = createApiProxy(ctx, { provider: 'p', model: 'm', cwd: '/tmp' })
+    const api = createApiProxy(ctx, { provider: 'p', model: 'm', cwd: '/tmp', workspaceRoot: '/tmp' })
     const session = ctx.sessions.create()
     ctx.agents.register({ id: session.id, session, status: 'idle', ctx } as Agent)
     // Superseded write early in the log, latest write later; enough messages to page.

From e1ee763e995bfa73f12b9c1facc089cccc603472 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 04:48:35 +0800
Subject: [PATCH 22/36] test(snapshot): tighten todo_write item schema in
 master's new acp snapshots

The origin/master merge added the session-query-spill and
escalation-approved acp scenarios, whose pinned tool-schemas still
carried additionalProperties: true on the todo_write item schema. This
PR tightens that to false (model-visible via the request header), so
re-record it in the two new expected outputs. session-sandbox-root,
escalation-rejected, and fs-escalation-approved compare against the
escalation-approved pinned header and pass once it is fixed.
---
 .../snapshots/escalation-approved/tool-schemas.expected.json    | 2 +-
 .../snapshots/session-query-spill/tool-schemas.expected.json    | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/examples/acp-agent/tests/snapshots/escalation-approved/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/escalation-approved/tool-schemas.expected.json
index d4973bfea4..01ac777a42 100644
--- a/examples/acp-agent/tests/snapshots/escalation-approved/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/escalation-approved/tool-schemas.expected.json
@@ -288,7 +288,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",
diff --git a/examples/acp-agent/tests/snapshots/session-query-spill/tool-schemas.expected.json b/examples/acp-agent/tests/snapshots/session-query-spill/tool-schemas.expected.json
index dde0ba0d7a..beb93c6b53 100644
--- a/examples/acp-agent/tests/snapshots/session-query-spill/tool-schemas.expected.json
+++ b/examples/acp-agent/tests/snapshots/session-query-spill/tool-schemas.expected.json
@@ -492,7 +492,7 @@
             "description": "The COMPLETE task list, replacing any previous list.",
             "items": {
               "type": "object",
-              "additionalProperties": true,
+              "additionalProperties": false,
               "properties": {
                 "content": {
                   "type": "string",

From 73e7e27799235480b9971427d74038fa192a83ca Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sun, 26 Jul 2026 13:07:27 +0800
Subject: [PATCH 23/36] refactor(llm): resolve model metadata together

---
 ...el-context-and-compaction-policy.i18n.yaml |   4 +-
 ...ted-model-context-and-compaction-policy.md |   4 +-
 ...-model-context-and-compaction-policy.zh.md |   4 +-
 ...ed-reasoning-effort-capabilities.i18n.yaml |   4 +-
 ...ter-owned-reasoning-effort-capabilities.md |  12 +-
 ...-owned-reasoning-effort-capabilities.zh.md |  12 +-
 docs/config-catalog.md                        |   6 +-
 docs/cookbook/adding-an-llm-adapter.i18n.yaml |   4 +-
 docs/cookbook/adding-an-llm-adapter.md        |   2 +-
 docs/cookbook/adding-an-llm-adapter.zh.md     |   2 +-
 docs/cordis-catalog/events.md                 |   2 +-
 docs/cordis-catalog/services.md               |  26 ++--
 docs/core-data-structures/core.i18n.yaml      |   4 +-
 docs/core-data-structures/core.md             |  12 +-
 docs/core-data-structures/core.zh.md          |  12 +-
 .../llm-streaming.i18n.yaml                   |   4 +-
 docs/core-data-structures/llm-streaming.md    |  35 ++---
 docs/core-data-structures/llm-streaming.zh.md |  35 ++---
 docs/event-producer-consumer.md               |   2 +-
 .../develop/practice/llm-adapter.i18n.yaml    |   4 +-
 docs/user/develop/practice/llm-adapter.md     |   2 +-
 docs/user/develop/practice/llm-adapter.zh.md  |   2 +-
 .../tests/fixtures/cli-mock-llm.ts            |  23 +--
 .../headless-agent/tests/headless.snapshot.ts |   2 +-
 .../tests/fixtures/tui-scripted-llm.ts        |  35 +++--
 packages/compact/compact-basic/src/index.ts   |   2 +-
 .../compact-basic/tests/compact-basic.spec.ts |  32 ++++-
 .../tests/compact-loop-repro.spec.ts          |  20 ++-
 .../cordis/tool-cordis/src/api-catalog.ts     |  16 +--
 .../core/agent-loop/tests/mock-adapter.ts     |  17 ++-
 .../tests/request-reconstruction.spec.ts      |  25 ++--
 packages/llm/llm-deepseek/README.md           |   4 +-
 packages/llm/llm-deepseek/src/adapter.ts      |  56 ++++----
 .../llm/llm-deepseek/tests/adapter.spec.ts    |  51 ++++---
 packages/llm/llm-pi-ai/README.md              |   6 +-
 packages/llm/llm-pi-ai/src/adapter.ts         |  74 ++++------
 packages/llm/llm-pi-ai/src/config.ts          |   6 +-
 packages/llm/llm-pi-ai/tests/adapter.e2e.ts   |  15 +-
 packages/llm/llm-pi-ai/tests/adapter.spec.ts  |  67 ++++++---
 packages/llm/llm/README.md                    |   9 +-
 packages/llm/llm/src/index.ts                 | 132 +++++++++---------
 packages/llm/llm/src/types.ts                 |   8 ++
 packages/llm/llm/tests/service.spec.ts        | 108 +++++++++-----
 packages/llm/token-meter/README.md            |   2 +-
 packages/support/llm-replay/src/index.ts      |  18 ++-
 .../llm-replay/tests/llm-replay.spec.ts       |  10 +-
 packages/ui/tui/README.md                     |   6 +-
 packages/ui/tui/src/index.ts                  |   6 +-
 packages/ui/tui/tests/harness.ts              |  28 ++--
 .../model-effort-switching.expected.txt       |   2 +-
 .../snapshots/model-switching.expected.txt    |   6 +-
 packages/ui/tui/tests/tui.snapshot.ts         |  17 ++-
 packages/ui/tui/tests/tui.spec.ts             |  59 +++++---
 scripts/gen-cordis-catalog.ts                 |   1 +
 scripts/type-equiv.manifest.json              |  10 ++
 55 files changed, 607 insertions(+), 460 deletions(-)

diff --git a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.i18n.yaml
index 6eeebc9848..cb487f67a2 100644
--- a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-20-routed-model-context-and-compaction-policy.md: b637ba24d4ba5fc25c8cdd515a821ee97883a326
-2026-07-20-routed-model-context-and-compaction-policy.zh.md: 084e762ec29ddc0aecb0bf422c147b9d3122726b
+2026-07-20-routed-model-context-and-compaction-policy.md: b00c744e4ac79983a1f492d710e7b1542278c4f4
+2026-07-20-routed-model-context-and-compaction-policy.zh.md: 88e26aa7a99e8deb434d9a8fd80784be410a999a
diff --git a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md
index b637ba24d4..b00c744e4a 100644
--- a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md
+++ b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.md
@@ -14,9 +14,9 @@ Neither obvious configuration owner is sufficient. Compact-basic is optional and
 
 ### Adapters own exact-route capacity
 
-`LlmAdapter.resolveModelContext(provider, model)` optionally returns `LlmModelContext` for one exact route. `LlmService.resolveModelContext()` selects the registered route owner, validates a positive integer `contextWindow`, and returns a detached value. The query is independent of `listModels()`: an unlisted dynamic model may have capacity metadata, and `undefined` means only that the adapter cannot describe capacity.
+`LlmAdapter.resolveModel(provider, model, signal?)` returns aggregate metadata for one exact route, with optional `LlmModelContext` under its `context` field. `LlmService.resolveModelInfo()` selects the registered route owner, validates a positive integer `contextWindow`, and returns detached metadata. The query is independent of `listModels()`: an unlisted dynamic model may have capacity metadata, and an absent `context` means only that the adapter cannot describe capacity.
 
-The hand-rolled DeepSeek adapter accepts optional `contextWindow` on each configured model plus an adapter-wide `defaultContextWindow`. Exact model capacity wins; an entry without capacity and an unlisted pass-through id inherit the adapter default, or return `undefined` when it is absent. The two built-in model entries each publish an exact 128,000-token capacity. The pi-ai adapter resolves capacity from the same catalog descriptor that authoritatively resolves the request model.
+The hand-rolled DeepSeek adapter accepts optional `contextWindow` on each configured model plus an adapter-wide `defaultContextWindow`. Exact model capacity wins; an entry without capacity and an unlisted pass-through id inherit the adapter default, or omit `context` when it is absent. The two built-in model entries each publish an exact 128,000-token capacity. The pi-ai adapter resolves capacity from the same catalog descriptor that authoritatively resolves the request model.
 
 ### Token measurement remains model-agnostic
 
diff --git a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.zh.md b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.zh.md
index 084e762ec2..88e26aa7a9 100644
--- a/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-20-routed-model-context-and-compaction-policy.zh.md
@@ -14,9 +14,9 @@ Status: implemented
 
 ### 适配器拥有精确路由容量
 
-`LlmAdapter.resolveModelContext(provider, model)` 可以为一条精确路由返回 `LlmModelContext`。`LlmService.resolveModelContext()` 选择已注册的路由所属方，验证 `contextWindow` 为正整数，并返回分离值。该查询独立于 `listModels()`：不在目录中的动态模型也可以拥有容量元数据，而 `undefined` 只表示适配器无法描述容量。
+`LlmAdapter.resolveModel(provider, model, signal?)` 返回一条精确路由的聚合元数据，其中可选的 `LlmModelContext` 位于 `context` 字段下。`LlmService.resolveModelInfo()` 选择已注册的路由所属方，验证 `contextWindow` 为正整数，并返回分离的元数据。该查询独立于 `listModels()`：不在目录中的动态模型也可以拥有容量元数据，而缺少 `context` 只表示适配器无法描述容量。
 
-手写 DeepSeek 适配器允许每个已配置模型提供可选 `contextWindow`，并支持适配器级 `defaultContextWindow`。精确模型容量优先；未提供容量的模型项与未列出的透传 id 会继承适配器默认值，若默认值也不存在则返回 `undefined`。两个内置模型项都公开精确的 128,000 token 容量。pi-ai 适配器从同一个目录描述符解析容量，该描述符也用于权威解析请求模型。
+手写 DeepSeek 适配器允许每个已配置模型提供可选 `contextWindow`，并支持适配器级 `defaultContextWindow`。精确模型容量优先；未提供容量的模型项与未列出的透传 id 会继承适配器默认值，若默认值也不存在则省略 `context`。两个内置模型项都公开精确的 128,000 token 容量。pi-ai 适配器从同一个目录描述符解析容量，该描述符也用于权威解析请求模型。
 
 ### Token 计量保持模型无关
 
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
index b0a7acb852..c3d5b45684 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-24-adapter-owned-reasoning-effort-capabilities.md: 04aebf5fa61d896ac16b29578a39c9b0e6e38cf4
-2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: a1c89417562276fd70ac5e5c1be17c9e639eb3ed
+2026-07-24-adapter-owned-reasoning-effort-capabilities.md: ffd633a3876c96f312fb274159f3f8422f8811ab
+2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 05c9941ad9c0906ab5400b5a215b92381fba3119
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
index 04aebf5fa6..ffd633a387 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
@@ -10,11 +10,11 @@ Reasoning strength was adapter configuration only, so a conversation could not d
 
 ## Decision
 
-`dsh-llm` represents a reasoning effort as the opaque branded `ReasoningEffortId`. An adapter's `resolveModelReasoning(provider, model)` returns a non-empty ordered list of ids with display metadata and may name one configured default. The core validates metadata, requires an explicit or configured effort to appear exactly in that list, and never clamps or aliases a value.
+`dsh-llm` represents a reasoning effort as the opaque branded `ReasoningEffortId`. One adapter-owned `resolveModel(provider, model, signal?)` query returns `LlmResolvedModelInfo`: exact model identity plus optional context and reasoning metadata. `LlmService.resolveModelInfo()` validates and detaches that aggregate. When present, `reasoning.efforts` is a non-empty ordered list of ids with display metadata and may name one configured default. The core requires an explicit or configured effort to appear exactly in that list and never clamps or aliases a value.
 
-`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop prepares the post-`agent/request` config under the active turn signal before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. The prepared call retains the exact adapter registration across asynchronous capability resolution, durable header logging, and dispatch; direct `LlmService.stream()` calls likewise capture their final registration before awaiting resolution. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id.
+`LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop prepares the post-`agent/request` config under the active turn signal before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. The prepared call retains the exact adapter registration across asynchronous exact-model resolution, durable header logging, and dispatch; direct `LlmService.stream()` calls likewise capture their final registration before awaiting resolution. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id.
 
-The native DeepSeek adapter advertises `high` and `max`, defaults to configured effort or `high`, and exposes no effort capability while thinking is disabled. The pi-ai adapter derives each exact model's list from `getSupportedThinkingLevels()`, excludes `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai.
+The native DeepSeek adapter advertises `high` and `max`, defaults to configured effort or `high`, and exposes no effort capability while thinking is disabled. The pi-ai adapter publishes each exact model's `getSupportedThinkingLevels()` result unchanged, including `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai. Its common stream options represent `off` by omitting `reasoning`, as required by pi-ai's own API.
 
 ## Alternatives considered
 
@@ -24,10 +24,10 @@ The native DeepSeek adapter advertises `high` and `max`, defaults to configured
 
 **Clamp unsupported levels.** Rejected because a silent substitution makes the user's selected control differ from the logged request intent and hides stale deployment configuration.
 
-**Include `off` as an effort.** Rejected because disabling reasoning is a mode capability with different request and output semantics, not a reasoning-strength level.
+**Normalize every adapter to a core-owned level list or remove `off`.** Rejected because the selectable vocabulary belongs to the exact model capability. A client can render an adapter's `off` option without requiring every adapter to expose it.
 
 ## Consequences
 
-Clients can query one exact route and render the adapter's order and names without knowing a global enum. Adapter configuration remains the deployment-default owner, while `agent/request` can replace the effective effort on each step. Invalid metadata fails with `INVALID_MODEL_REASONING`, and unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+Clients can query one exact route once and render its identity, context capacity, and adapter-owned reasoning choices without knowing a global enum. Adapter configuration remains the deployment-default owner, while `agent/request` can replace the effective effort on each step. Invalid exact identity, context, or reasoning metadata fails with `INVALID_MODEL_INFO`, `INVALID_MODEL_CONTEXT`, or `INVALID_MODEL_REASONING`; unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
 
-The capability query is asynchronous and exact-model resolution may fail for adapters backed by authoritative catalogs. Its optional signal is the caller's cancellation boundary; an asynchronous adapter must settle promptly after abort so loop disposal can reach quiescence. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, resume behavior, HMR registration ownership, and cancellation; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
+The aggregate exact-model query is asynchronous and may fail for adapters backed by authoritative catalogs. Its optional signal is the caller's cancellation boundary; an asynchronous adapter must settle promptly after abort so loop disposal can reach quiescence. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, resume behavior, HMR registration ownership, and cancellation; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
index a1c8941756..05c9941ad9 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
@@ -10,11 +10,11 @@ Status: implemented
 
 ## 决策
 
-`dsh-llm` 使用不透明的品牌类型 `ReasoningEffortId` 表示推理强度。适配器的 `resolveModelReasoning(provider, model)` 返回非空的有序 ID 列表及其展示元数据，并可指定一个由配置确定的默认值。核心会校验元数据，要求显式指定或配置指定的推理强度与列表中的某个 ID 完全一致，且绝不自动调整或为值提供别名。
+`dsh-llm` 使用不透明的品牌类型 `ReasoningEffortId` 表示推理强度。由适配器持有的单次 `resolveModel(provider, model, signal?)` 查询返回 `LlmResolvedModelInfo`，其中包含确切模型身份以及可选的上下文和推理元数据。`LlmService.resolveModelInfo()` 会校验该聚合结果并返回分离值。`reasoning.efforts` 存在时，是包含展示元数据的非空有序 ID 列表，并可指定一个由配置确定的默认值。核心要求显式指定或配置指定的推理强度与列表中的某个 ID 完全一致，且绝不自动调整或为值提供别名。
 
-`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在活跃轮次信号的控制下准备 `agent/request` 处理完成后的配置，再写入 `request/header`，因此默认值和动态变更只有成为持久化事实后才对模型可见。准备完成的调用在异步能力解析、请求头持久记录和分派全程保留同一项确切的适配器注册；直接调用 `LlmService.stream()` 时，也会在等待解析前捕获最终的适配器注册。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。
+`LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在活跃轮次信号的控制下准备 `agent/request` 处理完成后的配置，再写入 `request/header`，因此默认值和动态变更只有成为持久化事实后才对模型可见。准备完成的调用在异步确切模型解析、请求头持久记录和分派全程保留同一项确切的适配器注册；直接调用 `LlmService.stream()` 时，也会在等待解析前捕获最终的适配器注册。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。
 
-原生 DeepSeek 适配器声明 `high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`；禁用思考时不暴露推理强度能力。pi-ai 适配器通过 `getSupportedThinkingLevels()` 按具体模型推导等级列表，排除 `off`，在 profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。
+原生 DeepSeek 适配器声明 `high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`；禁用思考时不暴露推理强度能力。pi-ai 适配器原样发布每个确切模型的 `getSupportedThinkingLevels()` 结果，其中包括 `off`；profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。按照 pi-ai 自身 API 的要求，其通用流选项通过省略 `reasoning` 来表示 `off`。
 
 ## 备选方案
 
@@ -24,10 +24,10 @@ Status: implemented
 
 **自动调整不支持的等级。** 不予采纳：静默替换会导致用户选定的控制项与日志记录的请求意图不一致，还会掩盖陈旧的部署配置。
 
-**将 `off` 列为推理强度。** 不予采纳：禁用推理属于具有不同请求和输出语义的模式能力，而不是推理强度等级。
+**将每个适配器规范化为核心持有的等级列表，或移除 `off`。** 不予采纳：可选值集合属于确切模型的能力。客户端可以渲染某个适配器的 `off` 选项，而无需要求所有适配器都暴露该选项。
 
 ## 影响
 
-客户端可以查询一条确切路由，并按适配器给出的顺序和名称渲染等级，而无需了解全局枚举。适配器配置仍负责提供部署默认值，`agent/request` 则可以在每个步骤替换实际生效的推理强度。元数据无效时抛出 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
+客户端只需查询一次确切路由，即可渲染其身份、上下文容量和由适配器持有的推理选项，而无需了解全局枚举。适配器配置仍负责提供部署默认值，`agent/request` 则可以在每个步骤替换实际生效的推理强度。确切身份、上下文或推理元数据无效时，分别抛出 `INVALID_MODEL_INFO`、`INVALID_MODEL_CONTEXT` 或 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
 
-能力查询采用异步方式；对于由权威目录支持的适配器，确切模型解析可能失败。可选信号构成调用方的取消边界；异步适配器必须在信号中止后迅速完成结算，使主循环的资源释放达到完全停稳。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录、恢复行为、HMR（热模块替换）期间的注册所有权和取消提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
+确切模型元数据的聚合查询采用异步方式，并且对于由权威目录支持的适配器可能失败。可选信号构成调用方的取消边界；异步适配器必须在信号中止后迅速完成结算，使主循环的资源释放达到完全停稳。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录、恢复行为、HMR（热模块替换）期间的注册所有权和取消提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
diff --git a/docs/config-catalog.md b/docs/config-catalog.md
index 4abf1e9d28..5ff9656020 100644
--- a/docs/config-catalog.md
+++ b/docs/config-catalog.md
@@ -623,7 +623,7 @@ export interface PiAiProviderProfile {
   /** Provider request headers; Harness attribution wins reserved names. */
   headers?: Record<string, string>
   /** Provider-neutral pi-ai reasoning level. */
-  reasoning?: ThinkingLevel
+  reasoning?: ModelThinkingLevel
   /** Token budgets used by reasoning providers that support them. */
   thinkingBudgets?: ThinkingBudgets
   /** Prompt-cache retention preference. */
@@ -639,7 +639,7 @@ export interface PiAiProviderProfile {
 }
 ```
 
-Depends on: `CacheRetention` (`@earendil-works/pi-ai`) · `ThinkingBudgets` (`@earendil-works/pi-ai`) · `ThinkingLevel` (`@earendil-works/pi-ai`) · `Transport` (`@earendil-works/pi-ai`)
+Depends on: `CacheRetention` (`@earendil-works/pi-ai`) · `ModelThinkingLevel` (`@earendil-works/pi-ai`) · `ThinkingBudgets` (`@earendil-works/pi-ai`) · `Transport` (`@earendil-works/pi-ai`)
 
 Source: [`packages/llm/llm-pi-ai/src/config.ts:48`](../packages/llm/llm-pi-ai/src/config.ts)
 
@@ -687,7 +687,7 @@ export interface ReplayModelConfig {
 }
 ```
 
-Source: [`packages/support/llm-replay/src/index.ts:392`](../packages/support/llm-replay/src/index.ts)
+Source: [`packages/support/llm-replay/src/index.ts:400`](../packages/support/llm-replay/src/index.ts)
 
 ## `@deepseek-ai/dsh-llm-retry`
 
diff --git a/docs/cookbook/adding-an-llm-adapter.i18n.yaml b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
index 85aa0bc8f5..ca8c27ddae 100644
--- a/docs/cookbook/adding-an-llm-adapter.i18n.yaml
+++ b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-adding-an-llm-adapter.md: 76bcfdd6638ef8b40977a5e1fe4f678648c043cd
-adding-an-llm-adapter.zh.md: 13ae388bb0c6f29ab6269f364b849cb1694e2ef2
+adding-an-llm-adapter.md: 255dce40e5c10a4c94e34b358ad71f0283c1227f
+adding-an-llm-adapter.zh.md: 67d85f96326474f6acb1b604b2ed31eba18eab4f
diff --git a/docs/cookbook/adding-an-llm-adapter.md b/docs/cookbook/adding-an-llm-adapter.md
index 76bcfdd663..255dce40e5 100644
--- a/docs/cookbook/adding-an-llm-adapter.md
+++ b/docs/cookbook/adding-an-llm-adapter.md
@@ -32,7 +32,7 @@ Registration is effect-based (HMR-safe); one adapter per provider route — dupl
 - A `GenerateOptions` field your provider cannot honor (e.g. a `stop` list on a provider without stop sequences): throw `LlmError(..., 'UNSUPPORTED')` rather than silently dropping it.
 - If the provider requires response ids, signatures, or other native metadata on follow-up calls, emit the minimal lossless-JSON projection as `finish.replayState`. Validate it when rebuilding history. `LlmService` passes it only when the historical provider route and target provider route are currently owned by the exact same adapter instance; your adapter decides whether same-model, cross-model, or cross-provider restoration is legal. Never infer native replay from provider/model names alone when state is absent.
 
-Provider-specific thinking-mode toggles remain in the adapter's Config. Selectable reasoning strength uses the provider-neutral capability seam: return ordered opaque ids from `resolveModelReasoning()`, declare a configured `defaultEffort` only when one exists, honor the resolver's optional `AbortSignal`, and map `GenerateOptions.reasoningEffort` to the provider wire value. Do not expose provider wire spellings, clamp unsupported values, or include an `off` mode as an effort.
+Provider-specific thinking-mode toggles remain in the adapter's Config. Exact model metadata uses one provider-neutral capability seam: implement `resolveModel()` with provider/model identity and optional `context` and `reasoning` fields, declare a configured `defaultEffort` only when one exists, and honor the resolver's optional `AbortSignal`. Reasoning efforts are ordered opaque ids mapped to provider requests by the adapter. Preserve the adapter's authoritative selectable list—including `off` when its upstream capability API returns it—without exposing final wire spellings or clamping unsupported values.
 
 ## Structure that worked
 
diff --git a/docs/cookbook/adding-an-llm-adapter.zh.md b/docs/cookbook/adding-an-llm-adapter.zh.md
index 13ae388bb0..67d85f9632 100644
--- a/docs/cookbook/adding-an-llm-adapter.zh.md
+++ b/docs/cookbook/adding-an-llm-adapter.zh.md
@@ -32,7 +32,7 @@ export function apply(ctx: Context, config: Config) {
 - 如果 `GenerateOptions` 中某个字段你的提供方无法支持（例如提供方不支持 stop sequences 时收到 `stop` 列表）：抛出 `LlmError(..., 'UNSUPPORTED')`，而非静默丢弃。
 - 如果提供方在后续调用中需要响应 ID、签名或其他原生元数据，请将其最小无损 JSON 投影作为 `finish.replayState` 发出。重建历史时验证该状态。只有历史提供方路由和目标提供方路由当前由完全相同的适配器实例拥有时，`LlmService` 才会传递该状态；由适配器决定同模型、跨模型或跨提供方恢复是否合法。状态缺失时，切勿仅根据提供方/模型名称推断原生回放。
 
-提供方特有的 thinking 模式开关仍放在适配器的 Config 中。可选的推理强度使用提供方无关的能力 seam：`resolveModelReasoning()` 返回有序的不透明 ID；仅当存在配置指定的默认值时才声明 `defaultEffort`；响应传给解析器的可选 `AbortSignal`；并将 `GenerateOptions.reasoningEffort` 映射为提供方协议值。不得暴露提供方协议值的具体拼写、自动调整不支持的值，也不得把 `off` 模式列为推理强度。
+提供方特有的 thinking 模式开关仍放在适配器的 Config 中。确切模型元数据使用一处提供方无关的能力 seam：实现 `resolveModel()`，返回提供方/模型身份以及可选的 `context` 和 `reasoning` 字段；仅当存在配置指定的默认值时才声明 `defaultEffort`；响应传给解析器的可选 `AbortSignal`。推理强度是由适配器映射到提供方请求的有序不透明 ID。请保留适配器给出的权威可选列表，包括其上游能力 API 返回的 `off`，但不要暴露最终协议值的具体拼写，也不要自动调整不支持的值。
 
 ## 经验证有效的结构
 
diff --git a/docs/cordis-catalog/events.md b/docs/cordis-catalog/events.md
index 01312bc0f8..dbfdec75a8 100644
--- a/docs/cordis-catalog/events.md
+++ b/docs/cordis-catalog/events.md
@@ -616,7 +616,7 @@ Waterfall around every streaming model call (retry, replay, routing). Bound to t
 
 Types: [GenerateOptions](../core-data-structures/core.md) · [LlmService](../core-data-structures/llm-streaming.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
 
-Source: [`packages/llm/llm/src/index.ts:54`](../../packages/llm/llm/src/index.ts)
+Source: [`packages/llm/llm/src/index.ts:53`](../../packages/llm/llm/src/index.ts)
 
 ## `session/*`
 
diff --git a/docs/cordis-catalog/services.md b/docs/cordis-catalog/services.md
index db108e6ed2..098a628d4f 100644
--- a/docs/cordis-catalog/services.md
+++ b/docs/cordis-catalog/services.md
@@ -730,25 +730,15 @@ listProviders(): LlmProviderInfo[]
 async listModels(provider: string): Promise<LlmModelInfo[]>
 
 /**
- * Resolve context capacity from the adapter that owns one exact route.
- * This query is independent of the advisory model catalog: an unlisted model
- * may return metadata, while `undefined` never rejects later routing.
- * @param provider - registered provider route to inspect.
- * @param model - exact model id passed to the adapter.
- * @returns detached context metadata, or `undefined` when the adapter has none.
- */
-async resolveModelContext( provider: string, model: string, ): Promise<LlmModelContext | undefined>
-
-/**
- * Resolve selectable reasoning efforts from the adapter that owns one exact
- * route. Metadata is validated and detached; an absent result means an
- * effort selector is unsupported for that model.
+ * Resolve and validate all metadata from the adapter that owns one exact
+ * route. The result is detached from adapter-owned objects; catalog
+ * membership remains advisory and does not control request routing.
  * @param provider - registered provider route to inspect.
  * @param model - exact model id passed to the adapter.
  * @param signal - optional cancellation for adapter-owned asynchronous lookup.
- * @returns detached reasoning metadata, or `undefined` when unsupported.
+ * @returns exact model identity plus available context and reasoning metadata.
  */
-async resolveModelReasoning( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmModelReasoningInfo | undefined>
+async resolveModelInfo( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmResolvedModelInfo>
 
 /**
  * Validate a conversation call config against its exact model capability and
@@ -777,7 +767,7 @@ async prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<Prepared
  * `LlmError` with code `NO_ADAPTER` if no adapter is registered for
  * `options.provider`. Replay state is retained only when the same adapter
  * instance owns its historical provider and the target provider. Final
- * adapter selection remains fixed through asynchronous reasoning resolution
+ * adapter selection remains fixed through asynchronous exact-model resolution
  * and dispatch. Selection, dispatch, and iteration failures retain their
  * original Error identity and are tagged in a call-local scope for narrow
  * agent-loop request recovery; middleware and nested-call failures remain
@@ -788,9 +778,9 @@ async prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<Prepared
 stream(options: GenerateOptions): AsyncIterable<StreamChunk>
 ```
 
-Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmCallConfig](../core-data-structures/core.md) · [LlmModelContext](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmModelReasoningInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [PreparedLlmCall](../core-data-structures/llm-streaming.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
+Types: [GenerateOptions](../core-data-structures/core.md) · [LlmAdapter](../core-data-structures/llm-streaming.md) · [LlmCallConfig](../core-data-structures/core.md) · [LlmModelInfo](../core-data-structures/core.md) · [LlmProviderInfo](../core-data-structures/core.md) · [LlmResolvedModelInfo](../core-data-structures/core.md) · [PreparedLlmCall](../core-data-structures/llm-streaming.md) · [StreamChunk](../core-data-structures/llm-streaming.md)
 
-Source: [`packages/llm/llm/src/index.ts:192`](../../packages/llm/llm/src/index.ts)
+Source: [`packages/llm/llm/src/index.ts:177`](../../packages/llm/llm/src/index.ts)
 
 ## `ctx.permission` — `PermissionService`
 
diff --git a/docs/core-data-structures/core.i18n.yaml b/docs/core-data-structures/core.i18n.yaml
index ba95f5d56c..9b17460460 100644
--- a/docs/core-data-structures/core.i18n.yaml
+++ b/docs/core-data-structures/core.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-core.md: d60268a163580558b09ae0af7b15b78232436a4b
-core.zh.md: b81c6a90ab7f74580b0f6599bd78868771f3e339
+core.md: 86aea325f5401d728a9b4aa147d78db9163a32c1
+core.zh.md: d09fb48419c14959eefb5a1df1c593b36c188cf5
diff --git a/docs/core-data-structures/core.md b/docs/core-data-structures/core.md
index d60268a163..86aea325f5 100644
--- a/docs/core-data-structures/core.md
+++ b/docs/core-data-structures/core.md
@@ -199,7 +199,7 @@ interface LlmModelInfo {
 }
 ```
 
-Correctness-sensitive model capacity is queried separately from the advisory catalog and is owned by the adapter serving the exact route.
+Correctness-sensitive metadata is resolved separately from the advisory catalog and is owned by the adapter serving the exact route. Context capacity and reasoning choices share one exact-model result so consumers do not repeat authoritative model resolution.
 
 ```ts type-equiv
 /** Provider-owned context capacity for one exact provider/model route. */
@@ -241,6 +241,16 @@ interface LlmModelReasoningInfo {
 }
 ```
 
+```ts type-equiv
+/** Exact-route model metadata resolved by its owning adapter. */
+interface LlmResolvedModelInfo extends LlmModelInfo {
+  /** Provider-owned context capacity when known. */
+  context?: LlmModelContext
+  /** Adapter-owned selectable reasoning levels when exposed. */
+  reasoning?: LlmModelReasoningInfo
+}
+```
+
 ```ts type-equiv
 /** A single model request, fully assembled. */
 interface GenerateOptions {
diff --git a/docs/core-data-structures/core.zh.md b/docs/core-data-structures/core.zh.md
index b81c6a90ab..d09fb48419 100644
--- a/docs/core-data-structures/core.zh.md
+++ b/docs/core-data-structures/core.zh.md
@@ -205,7 +205,7 @@ interface LlmModelInfo {
 }
 ```
 
-对正确性敏感的模型容量与参考目录分开查询，并归服务该确切路由的适配器所有。
+对正确性敏感的元数据与参考目录分开解析，并归服务该确切路由的适配器所有。上下文容量和推理选项共用同一个确切模型结果，消费方因而无需重复执行权威模型解析。
 
 ```ts type-equiv
 /** Provider-owned context capacity for one exact provider/model route. */
@@ -247,6 +247,16 @@ interface LlmModelReasoningInfo {
 }
 ```
 
+```ts type-equiv
+/** Exact-route model metadata resolved by its owning adapter. */
+interface LlmResolvedModelInfo extends LlmModelInfo {
+  /** Provider-owned context capacity when known. */
+  context?: LlmModelContext
+  /** Adapter-owned selectable reasoning levels when exposed. */
+  reasoning?: LlmModelReasoningInfo
+}
+```
+
 ```ts type-equiv
 /** A single model request, fully assembled. */
 interface GenerateOptions {
diff --git a/docs/core-data-structures/llm-streaming.i18n.yaml b/docs/core-data-structures/llm-streaming.i18n.yaml
index 6029b11adc..df55d0f641 100644
--- a/docs/core-data-structures/llm-streaming.i18n.yaml
+++ b/docs/core-data-structures/llm-streaming.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-llm-streaming.md: 5155e680bfcc253d06742c7d3adcddc633a01c52
-llm-streaming.zh.md: 9b928253065da99c74f749b905375d790fab46c9
+llm-streaming.md: ffe520eca12af0fa2d483b9f231c943107d74649
+llm-streaming.zh.md: d4d41040ceaa7e2b2ca08c47ae83fefbaba3f7d4
diff --git a/docs/core-data-structures/llm-streaming.md b/docs/core-data-structures/llm-streaming.md
index 5155e680bf..ffe520eca1 100644
--- a/docs/core-data-structures/llm-streaming.md
+++ b/docs/core-data-structures/llm-streaming.md
@@ -157,7 +157,7 @@ declare class BlockAssembler {
 
 ## The seam
 
-`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. The separate `resolveModelContext()` query exposes correctness-sensitive capacity, while `resolveModelReasoning()` exposes ordered model-owned effort ids and an optional deployment default; absence from either query means unavailable metadata or capability, not invalid catalog membership. A reasoning resolver receives optional cancellation and must settle promptly after abort. The service validates and materializes reasoning through `resolveCallConfig()` at the final adapter boundary, so direct calls cannot bypass unsupported-effort rejection; direct dispatch captures one registration before awaiting that resolution. The agent loop instead uses `prepareCall()` to keep the same registration across capability resolution, durable header logging, and dispatch. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
+`LlmAdapter` is the provider seam: subclass, implement `stream()`, and register one adapter instance with `ctx.llm.registerAdapter(providers, adapter)`. `GenerateOptions.provider` selects the registered adapter; `GenerateOptions.model` is passed to that adapter and need not be registered at lifecycle start. Duplicate provider routes fail atomically. Optional `providerInfo()` and asynchronous `listModels()` methods feed `LlmService.listProviders()` / `listModels()` with detached selector metadata. That catalog is advisory rather than a request whitelist: the adapter remains authoritative and may accept unlisted model ids. One asynchronous `resolveModel()` query returns exact model identity plus optional correctness-sensitive context capacity and ordered model-owned reasoning ids with an optional deployment default; absent fields mean unavailable metadata or capability, not invalid catalog membership. The resolver receives optional cancellation and must settle promptly after abort. `LlmService.resolveModelInfo()` validates and detaches the aggregate. The service validates and materializes reasoning through `resolveCallConfig()` at the final adapter boundary, so direct calls cannot bypass unsupported-effort rejection; direct dispatch captures one registration before awaiting that resolution. The agent loop instead uses `prepareCall()` to keep the same registration across model resolution, durable header logging, and dispatch. Adapter lookup happens at the terminal continuation of the `llm/stream` waterfall, so a listener may short-circuit the call or route a mutable one-shot request before lookup. The `block-start` / `block-end` `index` correlation and the assembler together mean an adapter only has to emit well-formed chunks — block reassembly is not each adapter's problem. The consumer surface (`ctx.llm.stream()`) and the `llm/stream` waterfall are described in [architecture.md § Content blocks and streaming](../architecture.md#content-blocks-and-streaming-dsh-llm).
 
 ```ts type-equiv
 /** One model call whose config and adapter registration were resolved together. */
@@ -198,30 +198,19 @@ declare abstract class LlmAdapter {
    */
   listModels(_provider: string): Promise<readonly LlmModelInfo[]>;
   /**
-   * Resolve context capacity for one model accepted by this adapter. Absence
-   * means the adapter does not know the capacity, not that routing is invalid.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @returns provider-owned context metadata, or `undefined` when unavailable.
+   * Resolve all metadata available for one exact model. This query is
+   * independent of the advisory catalog and does not validate request routing.
+   * @param provider - one provider route owned by this adapter.
+   * @param model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; asynchronous
+   *   implementations must settle promptly after it aborts.
+   * @returns provider/model identity plus any context and reasoning metadata.
    */
-  resolveModelContext(
-    _provider: string,
-    _model: string,
-  ): Promise<LlmModelContext | undefined>;
-  /**
-   * Resolve selectable reasoning efforts for one exact model. Absence means
-   * the model has no selectable reasoning-effort capability.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @param _signal - cancellation for this exact-model lookup; implementations
-   *   must settle promptly after it aborts.
-   * @returns adapter-owned effort metadata, or `undefined` when unsupported.
-   */
-  resolveModelReasoning(
-    _provider: string,
-    _model: string,
+  resolveModel(
+    provider: string,
+    model: string,
     _signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined>;
+  ): Promise<LlmResolvedModelInfo>;
   /**
    * Stream one model call as raw chunks. The only required method.
    * @param options - the fully-assembled request; implementations must honor `options.signal`.
diff --git a/docs/core-data-structures/llm-streaming.zh.md b/docs/core-data-structures/llm-streaming.zh.md
index 9b92825306..d4d41040ce 100644
--- a/docs/core-data-structures/llm-streaming.zh.md
+++ b/docs/core-data-structures/llm-streaming.zh.md
@@ -157,7 +157,7 @@ declare class BlockAssembler {
 
 ## seam
 
-`LlmAdapter` 是提供方 seam：创建子类、实现 `stream()`，再用 `ctx.llm.registerAdapter(providers, adapter)` 注册一个适配器实例。`GenerateOptions.provider` 选择已注册适配器；`GenerateOptions.model` 会传给该适配器，无需在生命周期启动时注册。重复提供方路由会原子失败。可选的 `providerInfo()` 与异步 `listModels()` 方法为 `LlmService.listProviders()` / `listModels()` 提供分离的 selector 元数据。该目录仅供参考，不是请求白名单：适配器仍是权威，并可接受未列出的模型 id。单独的 `resolveModelContext()` 查询暴露对正确性敏感的容量信息，`resolveModelReasoning()` 则暴露由模型持有的有序推理强度 ID 和可选的部署默认值；任一查询返回缺失都表示元数据或能力不可用，而不表示目录成员关系无效。推理能力解析器会接收可选的取消信号，并且必须在信号中止后迅速完成结算。服务通过最终适配器边界的 `resolveCallConfig()` 校验推理强度并填入默认值，因此直接调用也无法绕过对不支持推理强度的拒绝；直接分派会在等待解析前捕获一项适配器注册。agent loop 则使用 `prepareCall()`，使能力解析、请求头持久记录和分派全程使用同一项注册。适配器查找发生在 `llm/stream` waterfall（瀑布式事件）的终端 continuation，因此 listener 可以在查找前短路调用，或路由一个可变的一次性请求。`block-start` / `block-end` 的 `index` 关联与 assembler 共同意味着适配器只需 emit 格式正确的分片——块重组不是每个适配器各自的问题。消费方 surface（`ctx.llm.stream()`）与 `llm/stream` waterfall 见 [architecture.md § 内容块与流式传输](../architecture.md#content-blocks-and-streaming-dsh-llm)。
+`LlmAdapter` 是提供方 seam：创建子类、实现 `stream()`，再用 `ctx.llm.registerAdapter(providers, adapter)` 注册一个适配器实例。`GenerateOptions.provider` 选择已注册适配器；`GenerateOptions.model` 会传给该适配器，无需在生命周期启动时注册。重复提供方路由会原子失败。可选的 `providerInfo()` 与异步 `listModels()` 方法为 `LlmService.listProviders()` / `listModels()` 提供分离的 selector 元数据。该目录仅供参考，不是请求白名单：适配器仍是权威，并可接受未列出的模型 id。单次异步 `resolveModel()` 查询返回确切模型身份，以及可选的对正确性敏感的上下文容量、由模型持有的有序推理强度 ID 和部署默认值；字段缺失表示元数据或能力不可用，而不表示目录成员关系无效。解析器会接收可选的取消信号，并且必须在信号中止后迅速完成结算。`LlmService.resolveModelInfo()` 会校验聚合结果并返回分离值。服务通过最终适配器边界的 `resolveCallConfig()` 校验推理强度并填入默认值，因此直接调用也无法绕过对不支持推理强度的拒绝；直接分派会在等待解析前捕获一项适配器注册。agent loop 则使用 `prepareCall()`，使模型解析、请求头持久记录和分派全程使用同一项注册。适配器查找发生在 `llm/stream` waterfall（瀑布式事件）的终端 continuation，因此 listener 可以在查找前短路调用，或路由一个可变的一次性请求。`block-start` / `block-end` 的 `index` 关联与 assembler 共同意味着适配器只需 emit 格式正确的分片——块重组不是每个适配器各自的问题。消费方 surface（`ctx.llm.stream()`）与 `llm/stream` waterfall 见 [architecture.md § 内容块与流式传输](../architecture.md#content-blocks-and-streaming-dsh-llm)。
 
 ```ts type-equiv
 /** One model call whose config and adapter registration were resolved together. */
@@ -198,30 +198,19 @@ declare abstract class LlmAdapter {
    */
   listModels(_provider: string): Promise<readonly LlmModelInfo[]>;
   /**
-   * Resolve context capacity for one model accepted by this adapter. Absence
-   * means the adapter does not know the capacity, not that routing is invalid.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @returns provider-owned context metadata, or `undefined` when unavailable.
+   * Resolve all metadata available for one exact model. This query is
+   * independent of the advisory catalog and does not validate request routing.
+   * @param provider - one provider route owned by this adapter.
+   * @param model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; asynchronous
+   *   implementations must settle promptly after it aborts.
+   * @returns provider/model identity plus any context and reasoning metadata.
    */
-  resolveModelContext(
-    _provider: string,
-    _model: string,
-  ): Promise<LlmModelContext | undefined>;
-  /**
-   * Resolve selectable reasoning efforts for one exact model. Absence means
-   * the model has no selectable reasoning-effort capability.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @param _signal - cancellation for this exact-model lookup; implementations
-   *   must settle promptly after it aborts.
-   * @returns adapter-owned effort metadata, or `undefined` when unsupported.
-   */
-  resolveModelReasoning(
-    _provider: string,
-    _model: string,
+  resolveModel(
+    provider: string,
+    model: string,
     _signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined>;
+  ): Promise<LlmResolvedModelInfo>;
   /**
    * Stream one model call as raw chunks. The only required method.
    * @param options - the fully-assembled request; implementations must honor `options.signal`.
diff --git a/docs/event-producer-consumer.md b/docs/event-producer-consumer.md
index 6d5fbeb8bb..db789010b4 100644
--- a/docs/event-producer-consumer.md
+++ b/docs/event-producer-consumer.md
@@ -33,7 +33,7 @@ This matrix shows which packages dispatch each harness-owned event and which pac
 | `fs/observed` | `emit` | [`packages/fs/fs/src/index.ts:71`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`emit`) | [`fs-policy`](../packages/fs/fs-policy) |
 | `fs/write-intent` | `waterfall` | [`packages/fs/fs/src/index.ts:54`](../packages/fs/fs/src/index.ts) | [`tool-fs`](../packages/fs/tool-fs) (`waterfall`) | [`fs-policy`](../packages/fs/fs-policy) |
 | `goal/changed` | `emit` | [`packages/goal/goal/src/types.ts:167`](../packages/goal/goal/src/types.ts) | [`goal`](../packages/goal/goal) (`emit`) | [`goal-session`](../packages/goal/goal-session) |
-| `llm/stream` | `waterfall` | [`packages/llm/llm/src/index.ts:54`](../packages/llm/llm/src/index.ts) | [`llm`](../packages/llm/llm) (`waterfall`) | [`agent-loop`](../packages/core/agent-loop), [`llm`](../packages/llm/llm), [`llm-replay`](../packages/support/llm-replay), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`session-title`](../packages/session-title/session-title) |
+| `llm/stream` | `waterfall` | [`packages/llm/llm/src/index.ts:53`](../packages/llm/llm/src/index.ts) | [`llm`](../packages/llm/llm) (`waterfall`) | [`agent-loop`](../packages/core/agent-loop), [`llm`](../packages/llm/llm), [`llm-replay`](../packages/support/llm-replay), [`session-checkpoint-policy`](../packages/session-persistence/session-checkpoint-policy), [`session-title`](../packages/session-title/session-title) |
 | `session/created` | `emit` | [`packages/core/session/src/index.ts:79`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | `apiproxy`, [`compact`](../packages/compact/compact), [`goal`](../packages/goal/goal), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`llm-retry`](../packages/llm/llm-retry), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`user-approval`](../packages/ui/user-approval) |
 | `session/disposed` | `emit` | [`packages/core/session/src/index.ts:89`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`agent-loop`](../packages/core/agent-loop), `apiproxy`, [`session-persistence`](../packages/session-persistence/session-persistence), [`session-title`](../packages/session-title/session-title) |
 | `session/event` | `emit` | [`packages/core/session/src/index.ts:101`](../packages/core/session/src/index.ts) | [`session`](../packages/core/session) (`events.dispatch`) | [`acp`](../packages/acp/acp), `apiproxy`, [`cli-demo`](../packages/examples/cli-demo), [`compact`](../packages/compact/compact), [`goal`](../packages/goal/goal), [`goal-session`](../packages/goal/goal-session), [`hook-protocol`](../packages/hooks/hook-protocol), [`jsonrpc`](../packages/ui/jsonrpc), [`session`](../packages/core/session), [`session-persistence`](../packages/session-persistence/session-persistence), [`session-title`](../packages/session-title/session-title), [`token-meter`](../packages/llm/token-meter), [`tui`](../packages/ui/tui), [`user-approval`](../packages/ui/user-approval), [`workspace`](../packages/workspace/workspace), [`workspace-context`](../packages/context/workspace-context) |
diff --git a/docs/user/develop/practice/llm-adapter.i18n.yaml b/docs/user/develop/practice/llm-adapter.i18n.yaml
index 05d30a8b62..402e3537a7 100644
--- a/docs/user/develop/practice/llm-adapter.i18n.yaml
+++ b/docs/user/develop/practice/llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-llm-adapter.md: 8b8201099e2b78385301daf14605be53436353c7
-llm-adapter.zh.md: 735b7a4babf3d31652374e91cbce794cdd580725
+llm-adapter.md: 3bf005647d3e7908f1d1553374266c1d6c12c2e6
+llm-adapter.zh.md: b967d112ed6879b11486ab7aedf64653089d93de
diff --git a/docs/user/develop/practice/llm-adapter.md b/docs/user/develop/practice/llm-adapter.md
index 8b8201099e..3bf005647d 100644
--- a/docs/user/develop/practice/llm-adapter.md
+++ b/docs/user/develop/practice/llm-adapter.md
@@ -112,7 +112,7 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 `stream()` receives the exported `GenerateOptions` type. It includes the model, adapter-owned reasoning-effort id, conversation history, system prompt, tool schemas, generation parameters, stop sequences, and abort signal; treat the TypeScript type exported by `@deepseek-ai/dsh-llm` as authoritative. Map supported fields to the provider API. If the provider cannot honor a field, throw `LlmError` with a stable code instead of silently dropping it.
 
-Override `resolveModelReasoning(provider, model, signal?)` when an exact model exposes selectable reasoning strengths. Return ordered opaque ids and display names plus an optional configured default; do not promote provider names into a core enum. Honor the optional signal for asynchronous lookup so cancellation and disposal reach quiescence. The service validates the metadata and rejects unsupported explicit values before `stream()`. Returning `undefined` means that model has no selectable reasoning-effort capability.
+Override `resolveModel(provider, model, signal?)` to return exact provider/model identity plus optional `context` and `reasoning` metadata in one lookup. Reasoning metadata contains ordered opaque ids and display names plus an optional configured default; preserve the adapter's authoritative selectable list, including `off` when its upstream capability API returns it, instead of promoting those values into a core enum. Honor the optional signal for asynchronous lookup so cancellation and disposal reach quiescence. The service validates the aggregate and rejects unsupported explicit efforts before `stream()`; omitting `reasoning` means that model has no selectable reasoning-effort capability.
 
 ## Register an adapter
 
diff --git a/docs/user/develop/practice/llm-adapter.zh.md b/docs/user/develop/practice/llm-adapter.zh.md
index 735b7a4bab..b967d112ed 100644
--- a/docs/user/develop/practice/llm-adapter.zh.md
+++ b/docs/user/develop/practice/llm-adapter.zh.md
@@ -112,7 +112,7 @@ async function* exampleChunks(): AsyncIterable<StreamChunk> {
 
 `stream()` 接收仓库导出的 `GenerateOptions`。它包含模型名、由适配器持有的推理强度 ID、对话历史、系统提示词、tool schema、生成参数、停止序列和中止信号；完整字段以 `@deepseek-ai/dsh-llm` 导出的 TypeScript 类型为准。适配器必须将支持的字段映射到具体 API；无法支持的字段应抛出带稳定 code 的 `LlmError`，不能静默丢弃。
 
-当某个具体模型提供可选推理强度时，请覆写 `resolveModelReasoning(provider, model, signal?)`。返回有序的不透明 ID、展示名称，以及可选的配置默认值；不要将提供方使用的等级名称提升为核心枚举。异步查询必须响应这个可选信号，让取消和资源释放都能达到完全停稳。服务会校验元数据，并在调用 `stream()` 前拒绝显式指定但不受支持的值。返回 `undefined` 表示该模型没有可选的推理强度能力。
+请覆写 `resolveModel(provider, model, signal?)`，在一次查询中返回确切的提供方/模型身份以及可选的 `context` 和 `reasoning` 元数据。推理元数据包含有序的不透明 ID、展示名称，以及可选的配置默认值；请保留适配器给出的权威可选列表，包括其上游能力 API 返回的 `off`，而不要将这些值提升为核心枚举。异步查询必须响应这个可选信号，让取消和资源释放都能达到完全停稳。服务会校验聚合结果，并在调用 `stream()` 前拒绝显式指定但不受支持的推理强度；省略 `reasoning` 表示该模型没有可选的推理强度能力。
 
 ## 注册适配器
 
diff --git a/examples/headless-agent/tests/fixtures/cli-mock-llm.ts b/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
index 23b5e18fc9..0f0b02c1d9 100644
--- a/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
+++ b/examples/headless-agent/tests/fixtures/cli-mock-llm.ts
@@ -4,22 +4,27 @@ import {
   LlmAdapter,
   ReasoningEffortId,
   type GenerateOptions,
-  type LlmModelReasoningInfo,
+  type LlmResolvedModelInfo,
   type StreamChunk,
 } from '@deepseek-ai/dsh-llm'
 
 const HIGH = ReasoningEffortId('high')
-const MAX = ReasoningEffortId('max')
+const OFF = ReasoningEffortId('off')
 
 /** Keyless headless-agent adapter: one real bash call followed by a final answer. */
 class CliMockAdapter extends LlmAdapter {
-  override async resolveModelReasoning(): Promise<LlmModelReasoningInfo> {
+  override async resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
     return {
-      efforts: [
-        { id: HIGH, name: 'High' },
-        { id: MAX, name: 'Max' },
-      ],
-      defaultEffort: HIGH,
+      provider,
+      id: model,
+      name: model,
+      reasoning: {
+        efforts: [
+          { id: OFF, name: 'Off' },
+          { id: HIGH, name: 'High' },
+        ],
+        defaultEffort: HIGH,
+      },
     }
   }
 
@@ -56,6 +61,6 @@ export function apply(ctx: Context): void {
   ctx.llm.registerAdapter(['cli-mock'], new CliMockAdapter())
   ctx.on('agent/request', async (_agent, _turn, step, _config, _signal, next) => {
     const config = await next()
-    return step === 2 ? { ...config, reasoningEffort: MAX } : config
+    return step === 2 ? { ...config, reasoningEffort: OFF } : config
   })
 }
diff --git a/examples/headless-agent/tests/headless.snapshot.ts b/examples/headless-agent/tests/headless.snapshot.ts
index 07feb3597c..f474cb5460 100644
--- a/examples/headless-agent/tests/headless.snapshot.ts
+++ b/examples/headless-agent/tests/headless.snapshot.ts
@@ -159,7 +159,7 @@ describe('headless stream-json snapshots', () => {
         {
           "model": "cli-mock",
           "provider": "cli-mock",
-          "reasoningEffort": "max",
+          "reasoningEffort": "off",
         },
       ]
     `)
diff --git a/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts b/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
index e668689ca9..d3e35b6ce5 100644
--- a/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
+++ b/examples/tui-agent/tests/fixtures/tui-scripted-llm.ts
@@ -1,9 +1,8 @@
 import type { Context } from 'cordis'
 import type {
   GenerateOptions,
-  LlmModelContext,
   LlmModelInfo,
-  LlmModelReasoningInfo,
+  LlmResolvedModelInfo,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
 import { CallId, LlmAdapter, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
@@ -41,21 +40,27 @@ class ScriptedTuiAdapter extends LlmAdapter {
     ])
   }
 
-  override resolveModelContext(_provider: string, _model: string): Promise<LlmModelContext> {
-    return Promise.resolve({ contextWindow: 128_000 })
-  }
-
-  override resolveModelReasoning(
-    _provider: string,
+  override resolveModel(
+    provider: string,
     model: string,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    if (model !== 'tui-scripted-model-pro') return Promise.resolve(undefined)
+  ): Promise<LlmResolvedModelInfo> {
     return Promise.resolve({
-      efforts: [
-        { id: ReasoningEffortId('high'), name: 'High' },
-        { id: ReasoningEffortId('max'), name: 'Max' },
-      ],
-      defaultEffort: ReasoningEffortId('high'),
+      provider,
+      id: model,
+      name: model === 'tui-scripted-model-pro' ? 'Scripted Pro' : 'Scripted Base',
+      context: { contextWindow: 128_000 },
+      ...model !== 'tui-scripted-model-pro'
+        ? {}
+        : {
+          reasoning: {
+            efforts: [
+              { id: ReasoningEffortId('off'), name: 'Off' },
+              { id: ReasoningEffortId('high'), name: 'High' },
+              { id: ReasoningEffortId('max'), name: 'Max' },
+            ],
+            defaultEffort: ReasoningEffortId('high'),
+          },
+        },
     })
   }
 
diff --git a/packages/compact/compact-basic/src/index.ts b/packages/compact/compact-basic/src/index.ts
index 10c64d10ca..f7ab37ccc4 100644
--- a/packages/compact/compact-basic/src/index.ts
+++ b/packages/compact/compact-basic/src/index.ts
@@ -272,7 +272,7 @@ export class BasicCompactService extends CompactService {
       return this.compactRegion(range.start, range.end, agent, signal)
     }
 
-    const context = await this.ctx.llm.resolveModelContext(target.provider, target.model)
+    const context = (await this.ctx.llm.resolveModelInfo(target.provider, target.model)).context
     const targetKey = `${target.provider}/${target.model}`
     if (context === undefined) {
       throw new TargetPressureConfigError(
diff --git a/packages/compact/compact-basic/tests/compact-basic.spec.ts b/packages/compact/compact-basic/tests/compact-basic.spec.ts
index 1db86d38e4..9433e70f32 100644
--- a/packages/compact/compact-basic/tests/compact-basic.spec.ts
+++ b/packages/compact/compact-basic/tests/compact-basic.spec.ts
@@ -16,7 +16,7 @@ import type {
   ContentBlock,
   GenerateOptions,
   LlmFailure,
-  LlmModelContext,
+  LlmResolvedModelInfo,
   Message,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
@@ -33,8 +33,13 @@ class ContextAdapter extends LlmAdapter {
     super()
   }
 
-  override resolveModelContext(): Promise<LlmModelContext> {
-    return Promise.resolve({ contextWindow: this.contextWindow })
+  override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      context: { contextWindow: this.contextWindow },
+    })
   }
 
   override async * stream(): AsyncIterable<StreamChunk> {
@@ -47,9 +52,14 @@ class RoutedContextAdapter extends LlmAdapter {
     super()
   }
 
-  override resolveModelContext(provider: string): Promise<LlmModelContext | undefined> {
+  override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
     const contextWindow = this.windows[provider]
-    return Promise.resolve(contextWindow === undefined ? undefined : { contextWindow })
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      ...contextWindow === undefined ? {} : { context: { contextWindow } },
+    })
   }
 
   override async * stream(): AsyncIterable<StreamChunk> {
@@ -485,7 +495,11 @@ describe('pressure measurement and retention', () => {
     void new LlmService(ctx)
     void new TokenMeterService(ctx)
     ctx.llm.registerAdapter(['unknown-context'], new ContextAdapter(1_000))
-    vi.spyOn(ctx.llm, 'resolveModelContext').mockResolvedValue(undefined)
+    vi.spyOn(ctx.llm, 'resolveModelInfo').mockImplementation((provider, model) => Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+    }))
     const compact = service(compactConfig, ctx)
     const session = conversation(4)
     session.append('request/header', {
@@ -1337,7 +1351,11 @@ describe('automatic listener and loader composition', () => {
     const ctx = createContext()
     const warnings: string[] = []
     ctx.logger.warn = ((message: string) => void warnings.push(message)) as typeof ctx.logger.warn
-    vi.spyOn(ctx.llm, 'resolveModelContext').mockResolvedValue(undefined)
+    vi.spyOn(ctx.llm, 'resolveModelInfo').mockImplementation((provider, model) => Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+    }))
     void new TestCompactService(ctx, {
       thresholdRatio: 0.5,
       retainTokens: 180,
diff --git a/packages/compact/compact-basic/tests/compact-loop-repro.spec.ts b/packages/compact/compact-basic/tests/compact-loop-repro.spec.ts
index 83c18f78b4..429bbac3e4 100644
--- a/packages/compact/compact-basic/tests/compact-loop-repro.spec.ts
+++ b/packages/compact/compact-basic/tests/compact-loop-repro.spec.ts
@@ -2,7 +2,7 @@ import { describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
 import { toolPairingBalancedAfter, toolPairingBalancedBefore } from '@deepseek-ai/dsh-compact'
 import { CONTEXT_WINDOW_EXCEEDED_CODE, LlmError } from '@deepseek-ai/dsh-llm'
-import type { ContentBlock, GenerateOptions, StreamChunk } from '@deepseek-ai/dsh-llm'
+import type { ContentBlock, GenerateOptions, LlmResolvedModelInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
 import { CallId, LlmAdapter } from '@deepseek-ai/dsh-llm'
 import { defineContentToolFixture } from '@deepseek-ai/dsh-tools'
 import type { Agent } from '@deepseek-ai/dsh-agent'
@@ -41,8 +41,13 @@ class StepwiseToolAdapter extends LlmAdapter {
     super()
   }
 
-  override resolveModelContext(): Promise<{ contextWindow: number }> {
-    return Promise.resolve({ contextWindow: 400 })
+  override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      context: { contextWindow: 400 },
+    })
   }
 
   async * stream(_options: GenerateOptions): AsyncIterable<StreamChunk> {
@@ -76,8 +81,13 @@ class OverflowRecoveryAdapter extends LlmAdapter {
     super()
   }
 
-  override resolveModelContext(): Promise<{ contextWindow: number }> {
-    return Promise.resolve({ contextWindow: 128 })
+  override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      context: { contextWindow: 128 },
+    })
   }
 
   override async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
diff --git a/packages/cordis/tool-cordis/src/api-catalog.ts b/packages/cordis/tool-cordis/src/api-catalog.ts
index 5286e76d06..c0eaadecd1 100644
--- a/packages/cordis/tool-cordis/src/api-catalog.ts
+++ b/packages/cordis/tool-cordis/src/api-catalog.ts
@@ -381,12 +381,8 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
         jsDoc: '/**\n * Discover models advertised by one registered provider. Catalog membership\n * is advisory and never changes routing or request validation.\n * @param provider - registered provider route to inspect.\n * @returns detached model metadata in adapter-preferred order.\n */',
       },
       {
-        signature: 'async resolveModelContext( provider: string, model: string, ): Promise<LlmModelContext | undefined>',
-        jsDoc: '/**\n * Resolve context capacity from the adapter that owns one exact route.\n * This query is independent of the advisory model catalog: an unlisted model\n * may return metadata, while `undefined` never rejects later routing.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @returns detached context metadata, or `undefined` when the adapter has none.\n */',
-      },
-      {
-        signature: 'async resolveModelReasoning( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmModelReasoningInfo | undefined>',
-        jsDoc: '/**\n * Resolve selectable reasoning efforts from the adapter that owns one exact\n * route. Metadata is validated and detached; an absent result means an\n * effort selector is unsupported for that model.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @param signal - optional cancellation for adapter-owned asynchronous lookup.\n * @returns detached reasoning metadata, or `undefined` when unsupported.\n */',
+        signature: 'async resolveModelInfo( provider: string, model: string, signal?: AbortSignal, ): Promise<LlmResolvedModelInfo>',
+        jsDoc: '/**\n * Resolve and validate all metadata from the adapter that owns one exact\n * route. The result is detached from adapter-owned objects; catalog\n * membership remains advisory and does not control request routing.\n * @param provider - registered provider route to inspect.\n * @param model - exact model id passed to the adapter.\n * @param signal - optional cancellation for adapter-owned asynchronous lookup.\n * @returns exact model identity plus available context and reasoning metadata.\n */',
       },
       {
         signature: 'async resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>',
@@ -398,7 +394,7 @@ export const SERVICE_API: readonly ServiceApiEntry[] = [
       },
       {
         signature: 'stream(options: GenerateOptions): AsyncIterable<StreamChunk>',
-        jsDoc: '/**\n * Stream one model call as raw chunks (token-level deltas). Throws\n * `LlmError` with code `NO_ADAPTER` if no adapter is registered for\n * `options.provider`. Replay state is retained only when the same adapter\n * instance owns its historical provider and the target provider. Final\n * adapter selection remains fixed through asynchronous reasoning resolution\n * and dispatch. Selection, dispatch, and iteration failures retain their\n * original Error identity and are tagged in a call-local scope for narrow\n * agent-loop request recovery; middleware and nested-call failures remain\n * untagged for the outer call.\n * @param options - the full request; `options.provider` selects the adapter.\n * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.\n */',
+        jsDoc: '/**\n * Stream one model call as raw chunks (token-level deltas). Throws\n * `LlmError` with code `NO_ADAPTER` if no adapter is registered for\n * `options.provider`. Replay state is retained only when the same adapter\n * instance owns its historical provider and the target provider. Final\n * adapter selection remains fixed through asynchronous exact-model resolution\n * and dispatch. Selection, dispatch, and iteration failures retain their\n * original Error identity and are tagged in a call-local scope for narrow\n * agent-loop request recovery; middleware and nested-call failures remain\n * untagged for the outer call.\n * @param options - the full request; `options.provider` selects the adapter.\n * @returns the chunk stream, possibly wrapped by `llm/stream` listeners.\n */',
       },
     ],
   },
@@ -1706,7 +1702,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'LlmAdapter',
-    declaration: 'export abstract class LlmAdapter {\n    providerInfo(provider: string): LlmProviderInfo;\n    listModels(_provider: string): Promise<readonly LlmModelInfo[]>;\n    resolveModelContext(_provider: string, _model: string): Promise<LlmModelContext | undefined>;\n    resolveModelReasoning(_provider: string, _model: string, _signal?: AbortSignal): Promise<LlmModelReasoningInfo | undefined>;\n    abstract stream(options: GenerateOptions): AsyncIterable<StreamChunk>;\n}',
+    declaration: 'export abstract class LlmAdapter {\n    providerInfo(provider: string): LlmProviderInfo;\n    listModels(_provider: string): Promise<readonly LlmModelInfo[]>;\n    resolveModel(provider: string, model: string, _signal?: AbortSignal): Promise<LlmResolvedModelInfo>;\n    abstract stream(options: GenerateOptions): AsyncIterable<StreamChunk>;\n}',
   },
   {
     name: 'LlmCallConfig',
@@ -1736,6 +1732,10 @@ export const TYPE_API: readonly TypeApiEntry[] = [
     name: 'LlmReasoningEffortInfo',
     declaration: 'export interface LlmReasoningEffortInfo {\n    id: ReasoningEffortId;\n    name: string;\n    description?: string;\n}',
   },
+  {
+    name: 'LlmResolvedModelInfo',
+    declaration: 'export interface LlmResolvedModelInfo extends LlmModelInfo {\n    context?: LlmModelContext;\n    reasoning?: LlmModelReasoningInfo;\n}',
+  },
   {
     name: 'Message',
     declaration: 'export interface Message {\n    role: \'system\' | \'user\' | \'assistant\';\n    content: ContentBlock[];\n    provenance?: AssistantProvenance;\n}',
diff --git a/packages/core/agent-loop/tests/mock-adapter.ts b/packages/core/agent-loop/tests/mock-adapter.ts
index 8b47a9880d..e754f1bd90 100644
--- a/packages/core/agent-loop/tests/mock-adapter.ts
+++ b/packages/core/agent-loop/tests/mock-adapter.ts
@@ -1,4 +1,4 @@
-import type { GenerateOptions, LlmModelReasoningInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
+import type { GenerateOptions, LlmModelReasoningInfo, LlmResolvedModelInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
 import { CallId, LlmAdapter } from '@deepseek-ai/dsh-llm'
 
 /** Helpers to write scripted responses tersely. */
@@ -71,11 +71,16 @@ export class MockAdapter extends LlmAdapter {
     super()
   }
 
-  override resolveModelReasoning(
-    _provider: string,
-    _model: string,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    return Promise.resolve(this.reasoning)
+  override resolveModel(
+    provider: string,
+    model: string,
+  ): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      ...this.reasoning === undefined ? {} : { reasoning: this.reasoning },
+    })
   }
 
   async * stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
diff --git a/packages/core/agent-loop/tests/request-reconstruction.spec.ts b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
index 1c101289f1..71fc9ea981 100644
--- a/packages/core/agent-loop/tests/request-reconstruction.spec.ts
+++ b/packages/core/agent-loop/tests/request-reconstruction.spec.ts
@@ -8,7 +8,7 @@
 import { describe, expect, it } from 'vitest'
 import { Context } from 'cordis'
 import LlmService, { LlmError, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
-import type { GenerateOptions, LlmModelReasoningInfo } from '@deepseek-ai/dsh-llm'
+import type { GenerateOptions, LlmModelReasoningInfo, LlmResolvedModelInfo } from '@deepseek-ai/dsh-llm'
 import SessionStore, { Session, SessionId, foldRequestHeader } from '@deepseek-ai/dsh-session'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
 import ToolRegistry, { defineContentToolFixture } from '@deepseek-ai/dsh-tools'
@@ -152,7 +152,7 @@ describe('request stability across the loop', () => {
     expect(resumedHeaders.at(-1)?.data.reason).toBe('resume')
   })
 
-  it('keeps reasoning resolution, request logging, and dispatch on one adapter registration', async () => {
+  it('keeps exact-model resolution, request logging, and dispatch on one adapter registration', async () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
     await ctx.plugin(SessionStore)
@@ -163,13 +163,18 @@ describe('request stability across the loop', () => {
     const started = Promise.withResolvers<undefined>()
     const reasoning = Promise.withResolvers<LlmModelReasoningInfo>()
     const first = new class extends MockAdapter {
-      override resolveModelReasoning(
-        _provider: string,
-        _model: string,
+      override async resolveModel(
+        provider: string,
+        model: string,
         _signal?: AbortSignal,
-      ): typeof reasoning.promise {
+      ): Promise<LlmResolvedModelInfo> {
         started.resolve(undefined)
-        return reasoning.promise
+        return {
+          provider,
+          id: model,
+          name: model,
+          reasoning: await reasoning.promise,
+        }
       }
     }([textResponse('first')])
     const second = new MockAdapter([textResponse('second')], {
@@ -200,7 +205,7 @@ describe('request stability across the loop', () => {
   it('aborts a blocked reasoning lookup before quiescent disposal completes', async () => {
     const started = Promise.withResolvers<AbortSignal>()
     const adapter = new class extends MockAdapter {
-      override resolveModelReasoning(
+      override resolveModel(
         _provider: string,
         _model: string,
         signal?: AbortSignal,
@@ -235,13 +240,13 @@ describe('request stability across the loop', () => {
   })
 
   it.each(['plain error', 'LLM error'] as const)(
-    'does not swallow a %s from reasoning resolution',
+    'does not swallow a %s from exact-model resolution',
     async (kind) => {
       const failure = kind === 'plain error'
         ? new Error('reasoning metadata failed')
         : new LlmError('unsupported effort', 'UNSUPPORTED_REASONING_EFFORT')
       const adapter = new class extends MockAdapter {
-        override resolveModelReasoning(): Promise<never> {
+        override resolveModel(): Promise<never> {
           return Promise.reject(failure)
         }
       }([])
diff --git a/packages/llm/llm-deepseek/README.md b/packages/llm/llm-deepseek/README.md
index 46e2823fc3..a92c48463d 100644
--- a/packages/llm/llm-deepseek/README.md
+++ b/packages/llm/llm-deepseek/README.md
@@ -28,9 +28,9 @@ The package root exposes the Cordis plugin contract and `DeepSeekAdapter`; wire
 
 The plugin registers the single provider route `deepseek`. A request selects it with `provider: deepseek`; its `model` is passed through as the wire `model` string, so changing DeepSeek models does not require lifecycle-time registration. Omitting `models` advertises `deepseek-v4-flash` and `deepseek-v4-pro`, each with a 128,000-token context window; an explicit list replaces those defaults, while `models: []` advertises none. Catalog entries are exposed through `ctx.llm.listModels('deepseek')` for UI selectors and deployment introspection, but remain advisory: unlisted model ids still pass through unchanged. An omitted entry name defaults to its id.
 
-`contextWindow` is optional per configured model and is not exposed through the advisory catalog. `ctx.llm.resolveModelContext('deepseek', model)` returns an exact model value first, then `defaultContextWindow` for an entry without capacity or an unlisted pass-through id. When neither value exists it returns `undefined` without invalidating routing. Pressure-sensitive plugins therefore get deployment-owned capacity without treating the model selector as authoritative. Registering another adapter for `deepseek` throws `LlmError('DUPLICATE_ADAPTER')`.
+`contextWindow` is optional per configured model and is not exposed through the advisory catalog. `ctx.llm.resolveModelInfo('deepseek', model).context` returns an exact model value first, then `defaultContextWindow` for an entry without capacity or an unlisted pass-through id. When neither value exists, `context` is absent without invalidating routing. Pressure-sensitive plugins therefore get deployment-owned capacity without treating the model selector as authoritative. Registering another adapter for `deepseek` throws `LlmError('DUPLICATE_ADAPTER')`.
 
-`ctx.llm.resolveModelReasoning('deepseek', model)` returns the ordered `high` and `max` efforts for every pass-through model while thinking is enabled. `reasoningEffort` selects the deployment default and falls back to `high` when omitted. `agent/request` can replace it on each conversation step; the resolved value is logged in `request/header` and serialized as the official top-level `reasoning_effort` field. An unsupported value fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O.
+The same exact-model result exposes ordered `high` and `max` efforts under `reasoning` for every pass-through model while thinking is enabled. `reasoningEffort` selects the deployment default and falls back to `high` when omitted. `agent/request` can replace it on each conversation step; the resolved value is logged in `request/header` and serialized as the official top-level `reasoning_effort` field. An unsupported value fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O.
 
 `thinking: disabled` removes the reasoning capability and omits `reasoning_effort`; combining it with a configured default fails plugin loading, and a per-request effort fails as unsupported. A request with `GenerateOptions.purpose: 'session-title'` also forces thinking disabled and omits the already-resolved effort, reserving its bounded output for visible title text without changing conversation or compaction defaults.
 
diff --git a/packages/llm/llm-deepseek/src/adapter.ts b/packages/llm/llm-deepseek/src/adapter.ts
index 8eabdc499f..3f6d50dbb8 100644
--- a/packages/llm/llm-deepseek/src/adapter.ts
+++ b/packages/llm/llm-deepseek/src/adapter.ts
@@ -8,10 +8,9 @@
 import { attributionHeaders, CONTEXT_WINDOW_EXCEEDED_CODE, isContextWindowExceededError, isQuotaExceededError, LlmAdapter, LlmError, ProviderRequestId, QUOTA_EXCEEDED_CODE, ReasoningEffortId } from '@deepseek-ai/dsh-llm'
 import type {
   GenerateOptions,
-  LlmModelContext,
   LlmModelInfo,
-  LlmModelReasoningInfo,
   LlmProviderInfo,
+  LlmResolvedModelInfo,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
 import { idleWatchdog, MAX_TIMER_DELAY_MS, timeoutOf } from '@deepseek-ai/dsh-timeout'
@@ -59,6 +58,15 @@ const REASONING_EFFORTS = [
   { id: MAX_REASONING_EFFORT, name: 'Max' },
 ] as const
 
+function modelInfo(provider: string, model: DeepSeekCatalogModel): LlmModelInfo {
+  return {
+    provider,
+    id: model.id,
+    name: model.name ?? model.id,
+    ...model.description === undefined ? {} : { description: model.description },
+  }
+}
+
 function providerRetryAfterMs(value: string | null): number | undefined {
   if (value === null) return undefined
   if (/^\d+$/.test(value)) {
@@ -127,34 +135,32 @@ export class DeepSeekAdapter extends LlmAdapter {
   }
 
   override listModels(provider: string): Promise<readonly LlmModelInfo[]> {
-    return Promise.resolve((this.options.models ?? []).map(model => ({
-      provider,
-      id: model.id,
-      name: model.name ?? model.id,
-      ...model.description === undefined ? {} : { description: model.description },
-    })))
+    return Promise.resolve((this.options.models ?? []).map(model => modelInfo(provider, model)))
   }
 
-  override resolveModelContext(
-    _provider: string,
+  override resolveModel(
+    provider: string,
     model: string,
-  ): Promise<LlmModelContext | undefined> {
-    const contextWindow = this.options.models?.find(entry => entry.id === model)?.contextWindow
-      ?? this.options.defaultContextWindow
-    return Promise.resolve(contextWindow === undefined ? undefined : { contextWindow })
-  }
-
-  override resolveModelReasoning(
-    _provider: string,
-    _model: string,
     _signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    if (this.options.defaults?.thinking === 'disabled') return Promise.resolve(undefined)
+  ): Promise<LlmResolvedModelInfo> {
+    const configured = this.options.models?.find(entry => entry.id === model)
+    const contextWindow = configured?.contextWindow
+      ?? this.options.defaultContextWindow
     return Promise.resolve({
-      efforts: REASONING_EFFORTS,
-      defaultEffort: this.options.defaults?.reasoningEffort === 'max'
-        ? MAX_REASONING_EFFORT
-        : HIGH_REASONING_EFFORT,
+      ...configured === undefined
+        ? { provider, id: model, name: model }
+        : modelInfo(provider, configured),
+      ...contextWindow === undefined ? {} : { context: { contextWindow } },
+      ...this.options.defaults?.thinking === 'disabled'
+        ? {}
+        : {
+          reasoning: {
+            efforts: REASONING_EFFORTS,
+            defaultEffort: this.options.defaults?.reasoningEffort === 'max'
+              ? MAX_REASONING_EFFORT
+              : HIGH_REASONING_EFFORT,
+          },
+        },
     })
   }
 
diff --git a/packages/llm/llm-deepseek/tests/adapter.spec.ts b/packages/llm/llm-deepseek/tests/adapter.spec.ts
index 7d3d8f9f08..9ac66ce5d1 100644
--- a/packages/llm/llm-deepseek/tests/adapter.spec.ts
+++ b/packages/llm/llm-deepseek/tests/adapter.spec.ts
@@ -213,8 +213,8 @@ describe('DeepSeekAdapter against a mock server', () => {
       thinking: { type: 'disabled' },
     })
     expect(server.requests[0]).not.toHaveProperty('reasoning_effort')
-    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
-      .resolves.toBeUndefined()
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
+      .resolves.not.toHaveProperty('reasoning')
   })
 
   it('rejects a per-request effort before I/O when thinking is disabled', async () => {
@@ -573,15 +573,19 @@ describe('plugin registration and config', () => {
       { provider: 'deepseek', id: 'deepseek-v4-flash', name: 'deepseek-v4-flash' },
       { provider: 'deepseek', id: 'deepseek-v4-pro', name: 'deepseek-v4-pro' },
     ])
-    await expect(ctx.llm.resolveModelContext('deepseek', 'deepseek-v4-flash'))
-      .resolves.toEqual({ contextWindow: 128_000 })
-    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
-      .resolves.toEqual({
-        efforts: [
-          { id: ReasoningEffortId('high'), name: 'High' },
-          { id: ReasoningEffortId('max'), name: 'Max' },
-        ],
-        defaultEffort: ReasoningEffortId('high'),
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
+      .resolves.toMatchObject({
+        provider: 'deepseek',
+        id: 'deepseek-v4-flash',
+        name: 'deepseek-v4-flash',
+        context: { contextWindow: 128_000 },
+        reasoning: {
+          efforts: [
+            { id: ReasoningEffortId('high'), name: 'High' },
+            { id: ReasoningEffortId('max'), name: 'Max' },
+          ],
+          defaultEffort: ReasoningEffortId('high'),
+        },
       })
   })
 
@@ -635,10 +639,15 @@ describe('plugin registration and config', () => {
       { provider: 'deepseek', id: 'private-fast', name: 'private-fast' },
       { provider: 'deepseek', id: 'private-reasoner', name: 'Private Reasoner', description: 'Higher reasoning budget' },
     ])
-    await expect(ctx.llm.resolveModelContext('deepseek', 'private-fast'))
-      .resolves.toEqual({ contextWindow: 32_000 })
-    await expect(ctx.llm.resolveModelContext('deepseek', 'arbitrary-unlisted'))
-      .resolves.toBeUndefined()
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'private-fast'))
+      .resolves.toMatchObject({ context: { contextWindow: 32_000 } })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'private-reasoner'))
+      .resolves.toMatchObject({
+        name: 'Private Reasoner',
+        description: 'Higher reasoning budget',
+      })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'arbitrary-unlisted'))
+      .resolves.not.toHaveProperty('context')
   })
 
   it('uses exact model capacity before the adapter-wide default', async () => {
@@ -654,12 +663,12 @@ describe('plugin registration and config', () => {
       ],
     })
 
-    await expect(ctx.llm.resolveModelContext('deepseek', 'inherits-default'))
-      .resolves.toEqual({ contextWindow: 256_000 })
-    await expect(ctx.llm.resolveModelContext('deepseek', 'exact-override'))
-      .resolves.toEqual({ contextWindow: 64_000 })
-    await expect(ctx.llm.resolveModelContext('deepseek', 'unlisted-pass-through'))
-      .resolves.toEqual({ contextWindow: 256_000 })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'inherits-default'))
+      .resolves.toMatchObject({ context: { contextWindow: 256_000 } })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'exact-override'))
+      .resolves.toMatchObject({ context: { contextWindow: 64_000 } })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'unlisted-pass-through'))
+      .resolves.toMatchObject({ context: { contextWindow: 256_000 } })
   })
 
   it('allows an explicit empty model catalog', async () => {
diff --git a/packages/llm/llm-pi-ai/README.md b/packages/llm/llm-pi-ai/README.md
index 60c8af6332..ae35052887 100644
--- a/packages/llm/llm-pi-ai/README.md
+++ b/packages/llm/llm-pi-ai/README.md
@@ -28,9 +28,9 @@ Configure credentials and deployment-specific transport settings per provider. O
 
 Each provider name must exist in pi-ai's installed catalog and may appear only once in this plugin instance. Registration with `ctx.llm` is atomic: a collision with any provider route already owned by another adapter fails plugin loading without registering the remaining routes. Model ids are not lifecycle config; an unknown model fails before any provider request with `LlmError('UNKNOWN_MODEL')`.
 
-The adapter exposes each configured provider's installed pi-ai models through `ctx.llm.listModels(provider)`. This is provider-neutral selector metadata derived from `getModels(provider)`; request-time resolution still performs the authoritative catalog lookup, so discovery does not create a second model registry. `ctx.llm.resolveModelContext(provider, model)` performs the same exact descriptor lookup and returns its context window, keeping capacity metadata on the route-owning adapter rather than a consuming plugin.
+The adapter exposes each configured provider's installed pi-ai models through `ctx.llm.listModels(provider)`. This is provider-neutral selector metadata derived from `getModels(provider)`; request-time resolution still performs the authoritative catalog lookup, so discovery does not create a second model registry. `ctx.llm.resolveModelInfo(provider, model)` performs that exact descriptor lookup once and returns its identity, context window, and selectable thinking levels, keeping authoritative metadata on the route-owning adapter rather than its consumers.
 
-`ctx.llm.resolveModelReasoning(provider, model)` uses pi-ai's `getSupportedThinkingLevels(model)` and returns that model's ordered levels after filtering the separate `off` control. The Harness exposes the canonical pi-ai level as an opaque ID; provider/model wire spellings remain inside pi-ai's `thinkingLevelMap`. A non-reasoning model returns `undefined`. The profile `reasoning` value is the deployment default when configured; omitting it preserves the provider default. Per-request `GenerateOptions.reasoningEffort` takes precedence, and any explicit value absent from the exact model capability fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O instead of being clamped.
+The `reasoning.efforts` list is pi-ai's ordered `getSupportedThinkingLevels(model)` result without filtering or normalization, including `off` and the model-specific availability of `xhigh` or `max`. The Harness exposes each canonical pi-ai level as an opaque ID; provider/model wire spellings remain inside pi-ai's `thinkingLevelMap`. A non-reasoning model therefore exposes pi-ai's `off` choice. The profile `reasoning` value, including `off`, is the deployment default when configured; omitting it preserves the provider default. Per-request `GenerateOptions.reasoningEffort` takes precedence, and any explicit value absent from the exact model capability fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O instead of being clamped. pi-ai's common stream options represent `off` by omitting `reasoning`.
 
 Supported profile fields are `provider`, `apiKey`, `baseURL`, `headers`, `reasoning`, `thinkingBudgets`, `cacheRetention`, `transport`, `timeoutMs`, `websocketConnectTimeoutMs`, and `streamIdleTimeoutMs`. The stream-idle interval is a positive finite Node timer delay, defaults to five minutes, and covers only an outstanding provider read, not consumer think time. Harness app attribution wins a conflicting configured header name.
 
@@ -49,7 +49,7 @@ If a listener rewrites assembled assistant content, the loop drops replay state
 - pi-ai tool-call arguments are parsed objects; the harness stores raw JSON strings. The adapter parses input and re-stringifies output.
 - pi-ai reports failures as in-stream error events; these map to `finish {kind:'error'|'aborted', failure}` chunks. Provider-specific error text distinguishes terminal `QUOTA` from transient `RATE_LIMIT`, while text and usage signals evaluated against the resolved model's context window normalize overflow to `CONTEXT_WINDOW_EXCEEDED`. A terminal `stop` whose message carries no content blocks maps to a `finish {kind:'error'}` with code `EMPTY_RESPONSE` (retried by default policy) instead of a successful empty message.
 - pi-ai folds reasoning tokens into output usage; there is no separate reasoning count to map.
-- pi-ai may internally support an `off` thinking level, but the Harness reasoning-effort capability deliberately excludes mode changes.
+- pi-ai's `off` thinking level crosses the Harness capability seam unchanged and becomes an omitted pi-ai common `reasoning` option at dispatch.
 - `GenerateOptions.stop` is rejected with `UNSUPPORTED_OPTION` because pi-ai's common streaming surface cannot guarantee it across providers.
 
 ## App attribution
diff --git a/packages/llm/llm-pi-ai/src/adapter.ts b/packages/llm/llm-pi-ai/src/adapter.ts
index de91f2daa7..5fb980d259 100644
--- a/packages/llm/llm-pi-ai/src/adapter.ts
+++ b/packages/llm/llm-pi-ai/src/adapter.ts
@@ -11,6 +11,7 @@ import { getSupportedThinkingLevels } from '@earendil-works/pi-ai'
 import type {
   Api,
   Model,
+  ModelThinkingLevel,
   SimpleStreamOptions,
   ThinkingLevel,
 } from '@earendil-works/pi-ai'
@@ -22,9 +23,8 @@ import {
 } from '@deepseek-ai/dsh-llm'
 import type {
   GenerateOptions,
-  LlmModelContext,
   LlmModelInfo,
-  LlmModelReasoningInfo,
+  LlmResolvedModelInfo,
   ReasoningEffortId as ReasoningEffortIdType,
   StreamChunk,
 } from '@deepseek-ai/dsh-llm'
@@ -44,7 +44,7 @@ export interface PiAiAdapterOptions {
  * Resolve a catalog model dynamically and apply only the configured endpoint
  * override, preserving the catalog's API/capability/compatibility metadata.
  */
-function resolveModel(profile: PiAiProviderProfile, modelId: string): Model<Api> {
+function resolvePiModel(profile: PiAiProviderProfile, modelId: string): Model<Api> {
   const model = getBuiltinModels(profile.provider as BuiltinProvider).find(candidate => candidate.id === modelId) as Model<Api> | undefined
   if (model === undefined) {
     throw new LlmError(`pi-ai provider "${profile.provider}" has no catalog model "${modelId}"`, 'UNKNOWN_MODEL')
@@ -55,11 +55,12 @@ function resolveModel(profile: PiAiProviderProfile, modelId: string): Model<Api>
 /** Copy profile stream knobs into pi-ai's common option vocabulary. */
 function profileOptions(
   profile: PiAiProviderProfile,
-  reasoning: ThinkingLevel | undefined,
+  reasoning: ModelThinkingLevel | undefined,
 ): SimpleStreamOptions {
+  const enabledReasoning: ThinkingLevel | undefined = reasoning === 'off' ? undefined : reasoning
   return {
     ...profile.apiKey === undefined ? {} : { apiKey: profile.apiKey },
-    ...reasoning === undefined ? {} : { reasoning },
+    ...enabledReasoning === undefined ? {} : { reasoning: enabledReasoning },
     ...profile.thinkingBudgets === undefined ? {} : { thinkingBudgets: profile.thinkingBudgets },
     ...profile.cacheRetention === undefined ? {} : { cacheRetention: profile.cacheRetention },
     ...profile.transport === undefined ? {} : { transport: profile.transport },
@@ -70,19 +71,14 @@ function profileOptions(
   }
 }
 
-/** Selectable pi-ai levels exclude the separate on/off control. */
-function supportedReasoningLevels(model: Model<Api>): ThinkingLevel[] {
-  return getSupportedThinkingLevels(model).filter((level): level is ThinkingLevel => level !== 'off')
-}
-
 /** Validate an explicit Harness/profile effort without invoking pi-ai's clamp. */
 function resolveReasoningLevel(
   model: Model<Api>,
-  effort: ReasoningEffortIdType | ThinkingLevel | undefined,
-): ThinkingLevel | undefined {
+  effort: ReasoningEffortIdType | ModelThinkingLevel | undefined,
+): ModelThinkingLevel | undefined {
   if (effort === undefined) return undefined
-  const supported = supportedReasoningLevels(model)
-  if (supported.some(level => level === effort)) return effort as ThinkingLevel
+  const supported = getSupportedThinkingLevels(model)
+  if (supported.some(level => level === effort)) return effort as ModelThinkingLevel
   throw new LlmError(
     `pi-ai provider "${model.provider}" model "${model.id}" does not support reasoning effort "${effort}"`,
     'UNSUPPORTED_REASONING_EFFORT',
@@ -123,27 +119,11 @@ export class PiAiAdapter extends LlmAdapter {
     })))
   }
 
-  override resolveModelContext(
-    provider: string,
-    model: string,
-  ): Promise<LlmModelContext | undefined> {
-    const profile = this.profiles.get(provider)
-    if (profile === undefined) {
-      return Promise.reject(new LlmError(
-        `pi-ai adapter does not own provider "${provider}"`,
-        'NO_ADAPTER',
-      ))
-    }
-    return Promise.resolve().then(() => ({
-      contextWindow: resolveModel(profile, model).contextWindow,
-    }))
-  }
-
-  override resolveModelReasoning(
+  override resolveModel(
     provider: string,
     model: string,
     _signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined> {
+  ): Promise<LlmResolvedModelInfo> {
     const profile = this.profiles.get(provider)
     if (profile === undefined) {
       return Promise.reject(new LlmError(
@@ -152,21 +132,23 @@ export class PiAiAdapter extends LlmAdapter {
       ))
     }
     return Promise.resolve().then(() => {
-      const resolvedModel = resolveModel(profile, model)
-      const levels = supportedReasoningLevels(resolvedModel)
-      if (levels.length === 0) {
-        resolveReasoningLevel(resolvedModel, profile.reasoning)
-        return undefined
-      }
+      const resolvedModel = resolvePiModel(profile, model)
+      const levels = getSupportedThinkingLevels(resolvedModel)
       const defaultLevel = resolveReasoningLevel(resolvedModel, profile.reasoning)
       return {
-        efforts: levels.map(level => ({
-          id: ReasoningEffortId(level),
-          name: `${level.charAt(0).toUpperCase()}${level.slice(1)}`,
-        })),
-        ...defaultLevel === undefined
-          ? {}
-          : { defaultEffort: ReasoningEffortId(defaultLevel) },
+        provider,
+        id: model,
+        name: resolvedModel.name,
+        context: { contextWindow: resolvedModel.contextWindow },
+        reasoning: {
+          efforts: levels.map(level => ({
+            id: ReasoningEffortId(level),
+            name: `${level.charAt(0).toUpperCase()}${level.slice(1)}`,
+          })),
+          ...defaultLevel === undefined
+            ? {}
+            : { defaultEffort: ReasoningEffortId(defaultLevel) },
+        },
       }
     })
   }
@@ -179,7 +161,7 @@ export class PiAiAdapter extends LlmAdapter {
     if (profile === undefined) {
       throw new LlmError(`pi-ai adapter does not own provider "${options.provider}"`, 'NO_ADAPTER')
     }
-    const model = resolveModel(profile, options.model)
+    const model = resolvePiModel(profile, options.model)
     const reasoning = resolveReasoningLevel(
       model,
       options.reasoningEffort ?? profile.reasoning,
diff --git a/packages/llm/llm-pi-ai/src/config.ts b/packages/llm/llm-pi-ai/src/config.ts
index 199463aaf6..bf9612d9d4 100644
--- a/packages/llm/llm-pi-ai/src/config.ts
+++ b/packages/llm/llm-pi-ai/src/config.ts
@@ -5,7 +5,7 @@
  */
 
 import { getBuiltinProviders } from '@earendil-works/pi-ai/providers/all'
-import type { CacheRetention, ThinkingBudgets, ThinkingLevel, Transport } from '@earendil-works/pi-ai'
+import type { CacheRetention, ModelThinkingLevel, ThinkingBudgets, Transport } from '@earendil-works/pi-ai'
 import z from 'schemastery'
 import { MAX_TIMER_DELAY_MS } from '@deepseek-ai/dsh-timeout'
 
@@ -23,7 +23,7 @@ export interface PiAiProviderProfile {
   /** Provider request headers; Harness attribution wins reserved names. */
   headers?: Record<string, string>
   /** Provider-neutral pi-ai reasoning level. */
-  reasoning?: ThinkingLevel
+  reasoning?: ModelThinkingLevel
   /** Token budgets used by reasoning providers that support them. */
   thinkingBudgets?: ThinkingBudgets
   /** Prompt-cache retention preference. */
@@ -62,7 +62,7 @@ const profile = z.object({
   apiKey: z.string(),
   baseURL: z.string(),
   headers: z.dict(z.string()),
-  reasoning: z.union(['minimal', 'low', 'medium', 'high', 'xhigh', 'max']),
+  reasoning: z.union(['off', 'minimal', 'low', 'medium', 'high', 'xhigh', 'max']),
   thinkingBudgets,
   cacheRetention: z.union(['none', 'short', 'long']),
   transport: z.union(['sse', 'websocket', 'websocket-cached', 'auto']),
diff --git a/packages/llm/llm-pi-ai/tests/adapter.e2e.ts b/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
index e91a951d85..be1e89de16 100644
--- a/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
+++ b/packages/llm/llm-pi-ai/tests/adapter.e2e.ts
@@ -9,7 +9,7 @@ import { assemble, type AssembledResult } from './assemble.ts'
 
 /**
  * Real-API e2e for the pi-ai-backed adapter: V4 Flash + V4 Pro with provider
- * defaults and representative high/max reasoning. Mirrors the native
+ * defaults and representative off/high/max reasoning. Mirrors the native
  * adapter's StreamChunk contract and exercises a replayed tool follow-up.
  * Key-gated.
  */
@@ -74,6 +74,19 @@ describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-pi-ai e2e (real API)', () =>
     expect(textOf(result).toLowerCase()).toContain('pong')
   })
 
+  it('flash + reasoning off: plain text without reasoning blocks', async () => {
+    const ctx = await harness(FLASH)
+    const result = await assemble(ctx,{
+      model: FLASH,
+      reasoningEffort: ReasoningEffortId('off'),
+      messages: ask('Reply with exactly the word: pong'),
+      maxTokens: 50,
+    })
+    expect(result.finish.kind).toBe('stop')
+    expect(result.message.content.some(block => block.type === 'reasoning')).toBe(false)
+    expect(textOf(result).toLowerCase()).toContain('pong')
+  })
+
   it.each([FLASH, PRO])('%s + reasoning high: reasoning blocks present', async (model) => {
     const ctx = await harness(model)
     const result = await assemble(ctx,{
diff --git a/packages/llm/llm-pi-ai/tests/adapter.spec.ts b/packages/llm/llm-pi-ai/tests/adapter.spec.ts
index 0bf8d34da7..cb5e60c343 100644
--- a/packages/llm/llm-pi-ai/tests/adapter.spec.ts
+++ b/packages/llm/llm-pi-ai/tests/adapter.spec.ts
@@ -149,7 +149,7 @@ describe('PiAiAdapter provider routing', () => {
   })
 
   it('uses a dynamic request effort and rejects unsupported efforts before network I/O', async () => {
-    const server = await mockServer([{ events: textEvents }])
+    const server = await mockServer([{ events: textEvents }, { events: textEvents }])
     const ctx = await harness(server.url, { reasoning: 'max' })
 
     await assemble(ctx, {
@@ -159,12 +159,20 @@ describe('PiAiAdapter provider routing', () => {
     })
     expect(server.requests[0]).toMatchObject({ reasoning_effort: 'high' })
 
+    await assemble(ctx, {
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('off'),
+      messages: [],
+    })
+    expect(server.requests[1]).toMatchObject({ thinking: { type: 'disabled' } })
+    expect(server.requests[1]).not.toHaveProperty('reasoning_effort')
+
     await expect(assemble(ctx, {
       model: 'deepseek-v4-flash',
       reasoningEffort: ReasoningEffortId('xhigh'),
       messages: [],
     })).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
-    expect(server.requests).toHaveLength(1)
+    expect(server.requests).toHaveLength(2)
   })
 
   it('preserves omitted profile options when constructing the adapter directly', async () => {
@@ -341,27 +349,30 @@ describe('provider profile lifecycle', () => {
       provider: 'openai', id: 'gpt-4.1', name: 'GPT-4.1',
     })
     expect(models.every(model => model.provider === 'openai')).toBe(true)
-    const context = await ctx.llm.resolveModelContext('openai', 'gpt-4.1')
-    expect(context).toBeDefined()
-    expect(typeof context?.contextWindow).toBe('number')
+    const info = await ctx.llm.resolveModelInfo('openai', 'gpt-4.1')
+    expect(typeof info.context?.contextWindow).toBe('number')
   })
 
-  it('exposes model-specific reasoning levels without off or an invented provider default', async () => {
+  it('exposes pi-ai model thinking levels verbatim without inventing a provider default', async () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
     await ctx.plugin(LlmPiAi, {
       providers: [{ provider: 'deepseek' }, { provider: 'openai' }],
     })
 
-    await expect(ctx.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
-      .resolves.toEqual({
-        efforts: [
-          { id: ReasoningEffortId('high'), name: 'High' },
-          { id: ReasoningEffortId('max'), name: 'Max' },
-        ],
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
+      .resolves.toMatchObject({
+        reasoning: {
+          efforts: [
+            { id: ReasoningEffortId('off'), name: 'Off' },
+            { id: ReasoningEffortId('high'), name: 'High' },
+            { id: ReasoningEffortId('max'), name: 'Max' },
+          ],
+        },
       })
-    const extended = await ctx.llm.resolveModelReasoning('openai', 'gpt-5.6-sol')
-    expect(extended?.efforts.map(effort => effort.id)).toEqual([
+    const extended = await ctx.llm.resolveModelInfo('openai', 'gpt-5.6-sol')
+    expect(extended.reasoning?.efforts.map(effort => effort.id)).toEqual([
+      ReasoningEffortId('off'),
       ReasoningEffortId('minimal'),
       ReasoningEffortId('low'),
       ReasoningEffortId('medium'),
@@ -369,8 +380,12 @@ describe('provider profile lifecycle', () => {
       ReasoningEffortId('xhigh'),
       ReasoningEffortId('max'),
     ])
-    await expect(ctx.llm.resolveModelReasoning('openai', 'gpt-4.1'))
-      .resolves.toBeUndefined()
+    await expect(ctx.llm.resolveModelInfo('openai', 'gpt-4.1'))
+      .resolves.toMatchObject({
+        reasoning: {
+          efforts: [{ id: ReasoningEffortId('off'), name: 'Off' }],
+        },
+      })
   })
 
   it('uses a supported profile reasoning value as the model default and rejects an unsupported one', async () => {
@@ -379,16 +394,24 @@ describe('provider profile lifecycle', () => {
     await supported.plugin(LlmPiAi, {
       providers: [{ provider: 'deepseek', reasoning: 'max' }],
     })
-    await expect(supported.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
-      .resolves.toMatchObject({ defaultEffort: ReasoningEffortId('max') })
+    await expect(supported.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
+      .resolves.toMatchObject({ reasoning: { defaultEffort: ReasoningEffortId('max') } })
 
     const unsupported = new Context()
     await unsupported.plugin(LlmService)
     await unsupported.plugin(LlmPiAi, {
       providers: [{ provider: 'deepseek', reasoning: 'medium' }],
     })
-    await expect(unsupported.llm.resolveModelReasoning('deepseek', 'deepseek-v4-flash'))
+    await expect(unsupported.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
       .rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+
+    const disabled = new Context()
+    await disabled.plugin(LlmService)
+    await disabled.plugin(LlmPiAi, {
+      providers: [{ provider: 'deepseek', reasoning: 'off' }],
+    })
+    await expect(disabled.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
+      .resolves.toMatchObject({ reasoning: { defaultEffort: ReasoningEffortId('off') } })
   })
 
   it('accepts absent credentials for pi-ai ambient authentication', async () => {
@@ -440,11 +463,9 @@ describe('provider profile lifecycle', () => {
   it('constructs the adapter directly and rejects routes it does not own', async () => {
     const adapter = new PiAiAdapter({ profiles: [{ provider: 'openai' }] })
     await expect(adapter.listModels('anthropic')).rejects.toMatchObject({ code: 'NO_ADAPTER' })
-    await expect(adapter.resolveModelContext('anthropic', 'claude-sonnet-4'))
+    await expect(adapter.resolveModel('anthropic', 'claude-sonnet-4'))
       .rejects.toMatchObject({ code: 'NO_ADAPTER' })
-    await expect(adapter.resolveModelReasoning('anthropic', 'claude-sonnet-4'))
-      .rejects.toMatchObject({ code: 'NO_ADAPTER' })
-    await expect(adapter.resolveModelContext('openai', 'not-a-catalog-model'))
+    await expect(adapter.resolveModel('openai', 'not-a-catalog-model'))
       .rejects.toMatchObject({ code: 'UNKNOWN_MODEL' })
     await expect((async () => {
       for await (const _chunk of adapter.stream({ provider: 'anthropic', model: 'claude-sonnet-4', messages: [] })) { /* drain */ }
diff --git a/packages/llm/llm/README.md b/packages/llm/llm/README.md
index 2a13e386f2..bec0283217 100644
--- a/packages/llm/llm/README.md
+++ b/packages/llm/llm/README.md
@@ -11,8 +11,7 @@ An adapter registry plus a single streaming call surface, interceptable via a wa
 - `ctx.llm.registerAdapter(providers: string[], adapter: LlmAdapter): () => void` Register one adapter instance for the given provider routes. Registration is all-or-nothing, and is disposed with the calling fiber.
 - `ctx.llm.listProviders(): LlmProviderInfo[]` Describe registered provider routes in registration order.
 - `ctx.llm.listModels(provider: string): Promise<LlmModelInfo[]>` Discover the models one registered provider currently advertises.
-- `ctx.llm.resolveModelContext(provider: string, model: string): Promise<LlmModelContext | undefined>` Resolve authoritative context capacity for one exact route from its owning adapter.
-- `ctx.llm.resolveModelReasoning(provider: string, model: string, signal?: AbortSignal): Promise<LlmModelReasoningInfo | undefined>` Resolve ordered adapter-owned reasoning efforts and an optional deployment default for one exact route, with optional cancellation for asynchronous adapters.
+- `ctx.llm.resolveModelInfo(provider: string, model: string, signal?: AbortSignal): Promise<LlmResolvedModelInfo>` Resolve validated exact-model identity plus available context and reasoning metadata from the owning adapter, with optional cancellation for asynchronous adapters.
 - `ctx.llm.resolveCallConfig(config: LlmCallConfig, signal?: AbortSignal): Promise<LlmCallConfig>` Validate an explicit effort and materialize an adapter-configured default without clamping.
 - `ctx.llm.prepareCall(config: LlmCallConfig, signal?: AbortSignal): Promise<PreparedLlmCall>` Resolve a config and capture its current adapter registration as one cancellable, one-shot call.
 - `ctx.llm.stream(options: GenerateOptions): AsyncIterable<StreamChunk>` Stream one model call as raw chunks (token-level deltas). Consumers assemble the chunks into blocks/messages with `BlockAssembler`.
@@ -21,9 +20,9 @@ An adapter registry plus a single streaming call surface, interceptable via a wa
 
 Provider and model metadata is a discovery surface, not a routing whitelist. `registerAdapter()` still owns provider exclusivity, while an adapter may accept model ids absent from `listModels()`; consumers must not reject a request because its model is unlisted. Returned metadata is detached and invalid or duplicate adapter entries fail with `INVALID_ADAPTER` or `INVALID_CATALOG`.
 
-Context capacity is a separate correctness query, not a catalog decoration or global LLM setting. `resolveModelContext()` asks the adapter that owns the exact provider/model route; an adapter can describe an unlisted dynamic model, and `undefined` means only that capacity is unavailable. Invalid returned capacity fails with `INVALID_MODEL_CONTEXT`.
+Exact-model metadata is a separate correctness query, not a catalog decoration or global LLM setting. `resolveModelInfo()` asks the adapter that owns the exact provider/model route once; an adapter can describe an unlisted dynamic model, and absent `context` or `reasoning` fields mean only that those capabilities are unavailable. Invalid identity, context, or reasoning metadata fails with `INVALID_MODEL_INFO`, `INVALID_MODEL_CONTEXT`, or `INVALID_MODEL_REASONING`.
 
-Reasoning effort is also an exact-route capability, but its identifiers are opaque adapter-owned strings rather than a core enum. `resolveModelReasoning()` validates and detaches the ordered display metadata; `undefined` means the model has no selectable effort. `resolveCallConfig()` accepts only an exact advertised identifier, materializes `defaultEffort` when present, and otherwise preserves the provider default. Asynchronous resolvers receive the caller's signal and must settle promptly after cancellation. `prepareCall()` additionally retains the exact adapter registration through header logging and terminal dispatch, so HMR cannot combine one adapter's capability result with another adapter's request; reusing its one-shot handle or changing its call-config fields fails with `INVALID_PREPARED_CALL`. Invalid capability metadata fails with `INVALID_MODEL_REASONING`; an unsupported explicit or configured effort fails with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+Reasoning identifiers are opaque adapter-owned strings rather than a core enum. An adapter publishes its ordered selectable list, including an `off` id when that model's capability API exposes one. `resolveCallConfig()` accepts only an exact advertised identifier, materializes `defaultEffort` when present, and otherwise preserves the provider default. Asynchronous model resolvers receive the caller's signal and must settle promptly after cancellation. `prepareCall()` additionally retains the exact adapter registration through header logging and terminal dispatch, so HMR cannot combine one adapter's capability result with another adapter's request; reusing its one-shot handle or changing its call-config fields fails with `INVALID_PREPARED_CALL`. An unsupported explicit or configured effort fails with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
 
 ### Events
 
@@ -33,7 +32,7 @@ Reasoning effort is also an exact-route capability, but its identifiers are opaq
 
 ### Extension points
 
-- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, `resolveModelContext()` when exact capacity is known, and `resolveModelReasoning()` when a model exposes selectable efforts; an asynchronous reasoning resolver must honor its optional cancellation signal. The defaults use the route id as its name, advertise no models, and return neither capacity nor reasoning metadata.
+- Subclass `LlmAdapter` and call `ctx.llm.registerAdapter(providers, adapter)` to add one or more provider routes. `GenerateOptions.provider` selects the adapter; `GenerateOptions.model` is adapter-owned and may be resolved dynamically. Override `providerInfo()` and asynchronous `listModels()` to expose selector metadata, then implement `resolveModel()` when exact identity, capacity, or selectable reasoning efforts are available; an asynchronous resolver must honor its optional cancellation signal. The defaults use the route and model ids as names, advertise no models, and return no capacity or reasoning metadata.
 - Wrap `llm/stream` via `ctx.on()` waterfall listeners for caching, logging, or routing. A wrapper that retries after emitting a chunk has no durable attempt boundary; shipped agent retry policy therefore uses `agent/request-error` instead.
 
 ### Content-block vocabulary (`types.ts`)
diff --git a/packages/llm/llm/src/index.ts b/packages/llm/llm/src/index.ts
index 824b960232..8e00fba59c 100644
--- a/packages/llm/llm/src/index.ts
+++ b/packages/llm/llm/src/index.ts
@@ -10,9 +10,8 @@ import { Context, Service } from 'cordis'
 import type {
   GenerateOptions,
   LlmFailure,
-  LlmModelContext,
   LlmModelInfo,
-  LlmModelReasoningInfo,
+  LlmResolvedModelInfo,
   LlmProviderInfo,
   Message,
   StreamChunk,
@@ -147,34 +146,20 @@ export abstract class LlmAdapter {
   }
 
   /**
-   * Resolve context capacity for one model accepted by this adapter. Absence
-   * means the adapter does not know the capacity, not that routing is invalid.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @returns provider-owned context metadata, or `undefined` when unavailable.
+   * Resolve all metadata available for one exact model. This query is
+   * independent of the advisory catalog and does not validate request routing.
+   * @param provider - one provider route owned by this adapter.
+   * @param model - exact model id passed to {@link GenerateOptions.model}.
+   * @param _signal - cancellation for this exact-model lookup; asynchronous
+   *   implementations must settle promptly after it aborts.
+   * @returns provider/model identity plus any context and reasoning metadata.
    */
-  resolveModelContext(
-    _provider: string,
-    _model: string,
-  ): Promise<LlmModelContext | undefined> {
-    return Promise.resolve(undefined)
-  }
-
-  /**
-   * Resolve selectable reasoning efforts for one exact model. Absence means
-   * the model has no selectable reasoning-effort capability.
-   * @param _provider - one provider route owned by this adapter.
-   * @param _model - exact model id passed to {@link GenerateOptions.model}.
-   * @param _signal - cancellation for this exact-model lookup; implementations
-   *   must settle promptly after it aborts.
-   * @returns adapter-owned effort metadata, or `undefined` when unsupported.
-   */
-  resolveModelReasoning(
-    _provider: string,
-    _model: string,
+  resolveModel(
+    provider: string,
+    model: string,
     _signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    return Promise.resolve(undefined)
+  ): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({ provider, id: model, name: model })
   }
 
   /**
@@ -273,53 +258,59 @@ export class LlmService extends Service {
   }
 
   /**
-   * Resolve context capacity from the adapter that owns one exact route.
-   * This query is independent of the advisory model catalog: an unlisted model
-   * may return metadata, while `undefined` never rejects later routing.
+   * Resolve and validate all metadata from the adapter that owns one exact
+   * route. The result is detached from adapter-owned objects; catalog
+   * membership remains advisory and does not control request routing.
    * @param provider - registered provider route to inspect.
    * @param model - exact model id passed to the adapter.
-   * @returns detached context metadata, or `undefined` when the adapter has none.
+   * @param signal - optional cancellation for adapter-owned asynchronous lookup.
+   * @returns exact model identity plus available context and reasoning metadata.
    */
-  async resolveModelContext(
+  async resolveModelInfo(
     provider: string,
     model: string,
-  ): Promise<LlmModelContext | undefined> {
-    const context = await this.registration(provider).adapter.resolveModelContext(provider, model)
-    if (context === undefined) return undefined
-    if (!Number.isInteger(context.contextWindow) || context.contextWindow <= 0) {
+    signal?: AbortSignal,
+  ): Promise<LlmResolvedModelInfo> {
+    return this.resolveModelInfoFor(this.registration(provider), model, signal)
+  }
+
+  private async resolveModelInfoFor(
+    registration: AdapterRegistration,
+    model: string,
+    signal?: AbortSignal,
+  ): Promise<LlmResolvedModelInfo> {
+    const provider = registration.provider.id
+    const resolved = await registration.adapter.resolveModel(provider, model, signal)
+    if (
+      typeof resolved.provider !== 'string'
+      || resolved.provider !== provider
+      || typeof resolved.id !== 'string'
+      || resolved.id !== model
+      || typeof resolved.name !== 'string'
+      || resolved.name.length === 0
+      || (resolved.description !== undefined && typeof resolved.description !== 'string')
+    ) {
+      throw new LlmError(
+        `adapter returned invalid exact model metadata for provider "${provider}" model "${model}"`,
+        'INVALID_MODEL_INFO',
+      )
+    }
+    const context = resolved.context
+    if (context !== undefined && (!Number.isInteger(context.contextWindow) || context.contextWindow <= 0)) {
       throw new LlmError(
         `adapter returned invalid context metadata for provider "${provider}" model "${model}"`,
         'INVALID_MODEL_CONTEXT',
       )
     }
-    return { contextWindow: context.contextWindow }
-  }
-
-  /**
-   * Resolve selectable reasoning efforts from the adapter that owns one exact
-   * route. Metadata is validated and detached; an absent result means an
-   * effort selector is unsupported for that model.
-   * @param provider - registered provider route to inspect.
-   * @param model - exact model id passed to the adapter.
-   * @param signal - optional cancellation for adapter-owned asynchronous lookup.
-   * @returns detached reasoning metadata, or `undefined` when unsupported.
-   */
-  async resolveModelReasoning(
-    provider: string,
-    model: string,
-    signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    return this.resolveModelReasoningFor(this.registration(provider), model, signal)
-  }
-
-  private async resolveModelReasoningFor(
-    registration: AdapterRegistration,
-    model: string,
-    signal?: AbortSignal,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    const provider = registration.provider.id
-    const reasoning = await registration.adapter.resolveModelReasoning(provider, model, signal)
-    if (reasoning === undefined) return undefined
+    const info: LlmResolvedModelInfo = {
+      provider,
+      id: model,
+      name: resolved.name,
+      ...resolved.description === undefined ? {} : { description: resolved.description },
+      ...context === undefined ? {} : { context: { contextWindow: context.contextWindow } },
+    }
+    const reasoning = resolved.reasoning
+    if (reasoning === undefined) return info
     if (reasoning.efforts.length === 0) {
       throw new LlmError(
         `adapter returned invalid reasoning metadata for provider "${provider}" model "${model}"`,
@@ -355,8 +346,11 @@ export class LlmService extends Service {
       )
     }
     return {
-      efforts,
-      ...reasoning.defaultEffort === undefined ? {} : { defaultEffort: reasoning.defaultEffort },
+      ...info,
+      reasoning: {
+        efforts,
+        ...reasoning.defaultEffort === undefined ? {} : { defaultEffort: reasoning.defaultEffort },
+      },
     }
   }
 
@@ -379,7 +373,7 @@ export class LlmService extends Service {
     config: LlmCallConfig,
     signal?: AbortSignal,
   ): Promise<LlmCallConfig> {
-    const reasoning = await this.resolveModelReasoningFor(registration, config.model, signal)
+    const reasoning = (await this.resolveModelInfoFor(registration, config.model, signal)).reasoning
     const requested = config.reasoningEffort
     if (reasoning === undefined) {
       if (requested !== undefined) {
@@ -520,7 +514,7 @@ export class LlmService extends Service {
    * `LlmError` with code `NO_ADAPTER` if no adapter is registered for
    * `options.provider`. Replay state is retained only when the same adapter
    * instance owns its historical provider and the target provider. Final
-   * adapter selection remains fixed through asynchronous reasoning resolution
+   * adapter selection remains fixed through asynchronous exact-model resolution
    * and dispatch. Selection, dispatch, and iteration failures retain their
    * original Error identity and are tagged in a call-local scope for narrow
    * agent-loop request recovery; middleware and nested-call failures remain
diff --git a/packages/llm/llm/src/types.ts b/packages/llm/llm/src/types.ts
index 7882985288..6ae8008bf6 100644
--- a/packages/llm/llm/src/types.ts
+++ b/packages/llm/llm/src/types.ts
@@ -182,6 +182,14 @@ export interface LlmModelReasoningInfo {
   defaultEffort?: ReasoningEffortId
 }
 
+/** Exact-route model metadata resolved by its owning adapter. */
+export interface LlmResolvedModelInfo extends LlmModelInfo {
+  /** Provider-owned context capacity when known. */
+  context?: LlmModelContext
+  /** Adapter-owned selectable reasoning levels when exposed. */
+  reasoning?: LlmModelReasoningInfo
+}
+
 /**
  * Raw streaming protocol emitted by adapters.
  * Block indexes correlate interleaved deltas, and `block-end` carries the
diff --git a/packages/llm/llm/tests/service.spec.ts b/packages/llm/llm/tests/service.spec.ts
index cf6636aa5d..20d8747d34 100644
--- a/packages/llm/llm/tests/service.spec.ts
+++ b/packages/llm/llm/tests/service.spec.ts
@@ -19,6 +19,7 @@ import type {
   LlmModelInfo,
   LlmModelReasoningInfo,
   LlmProviderInfo,
+  LlmResolvedModelInfo,
 } from '@deepseek-ai/dsh-llm'
 
 class ScriptedAdapter extends LlmAdapter {
@@ -68,18 +69,17 @@ class CatalogAdapter extends ScriptedAdapter {
     return Promise.resolve(this.models)
   }
 
-  override resolveModelContext(
-    _provider: string,
+  override resolveModel(
+    provider: string,
     model: string,
-  ): Promise<LlmModelContext | undefined> {
-    return Promise.resolve(this.contexts[model])
-  }
-
-  override resolveModelReasoning(
-    _provider: string,
-    model: string,
-  ): Promise<LlmModelReasoningInfo | undefined> {
-    return Promise.resolve(this.reasoning[model])
+  ): Promise<LlmResolvedModelInfo> {
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+      ...this.contexts[model] === undefined ? {} : { context: this.contexts[model] },
+      ...this.reasoning[model] === undefined ? {} : { reasoning: this.reasoning[model] },
+    })
   }
 }
 
@@ -667,8 +667,32 @@ describe('LlmService', () => {
     expect(ctx.llm.listProviders()).toEqual([{ id: 'plain', name: 'plain' }])
     await expect(ctx.llm.listModels('plain')).resolves.toEqual([])
     await expect(ctx.llm.listModels('missing')).rejects.toMatchObject({ code: 'NO_ADAPTER' })
-    await expect(ctx.llm.resolveModelContext('plain', 'unlisted')).resolves.toBeUndefined()
-    await expect(ctx.llm.resolveModelContext('missing', 'm')).rejects.toMatchObject({ code: 'NO_ADAPTER' })
+    await expect(ctx.llm.resolveModelInfo('plain', 'unlisted')).resolves.toEqual({
+      provider: 'plain', id: 'unlisted', name: 'unlisted',
+    })
+    await expect(ctx.llm.resolveModelInfo('missing', 'm')).rejects.toMatchObject({ code: 'NO_ADAPTER' })
+  })
+
+  it.each([
+    [{ provider: 1, id: 'model', name: 'Model' }, 'non-string provider'],
+    [{ provider: 'other', id: 'model', name: 'Model' }, 'mismatched provider'],
+    [{ provider: 'route', id: 1, name: 'Model' }, 'non-string id'],
+    [{ provider: 'route', id: 'other', name: 'Model' }, 'mismatched id'],
+    [{ provider: 'route', id: 'model', name: 1 }, 'non-string name'],
+    [{ provider: 'route', id: 'model', name: '' }, 'empty name'],
+    [{ provider: 'route', id: 'model', name: 'Model', description: 1 }, 'non-string description'],
+  ] as const)('rejects invalid exact model metadata (%s: %s)', async (metadata, _label) => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    const adapter = new class extends ScriptedAdapter {
+      override resolveModel(): Promise<LlmResolvedModelInfo> {
+        return Promise.resolve(metadata as unknown as LlmResolvedModelInfo)
+      }
+    }(SCRIPT)
+    ctx.llm.registerAdapter(['route'], adapter)
+
+    await expect(ctx.llm.resolveModelInfo('route', 'model'))
+      .rejects.toMatchObject({ code: 'INVALID_MODEL_INFO' })
   })
 
   it('resolves detached model context independently of advisory catalog membership', async () => {
@@ -681,11 +705,13 @@ describe('LlmService', () => {
       { unlisted: source },
     ))
 
-    const resolved = await ctx.llm.resolveModelContext('route', 'unlisted')
-    expect(resolved).toEqual({ contextWindow: 32_000 })
+    const resolved = await ctx.llm.resolveModelInfo('route', 'unlisted')
+    expect(resolved.context).toEqual({ contextWindow: 32_000 })
     source.contextWindow = 64_000
-    expect(resolved).toEqual({ contextWindow: 32_000 })
-    await expect(ctx.llm.resolveModelContext('route', 'other')).resolves.toBeUndefined()
+    expect(resolved.context).toEqual({ contextWindow: 32_000 })
+    await expect(ctx.llm.resolveModelInfo('route', 'other')).resolves.toEqual({
+      provider: 'route', id: 'other', name: 'other',
+    })
   })
 
   it('resolves detached adapter-owned reasoning metadata and materializes its default', async () => {
@@ -705,10 +731,10 @@ describe('LlmService', () => {
       { model: source },
     ))
 
-    const resolved = await ctx.llm.resolveModelReasoning('route', 'model')
-    expect(resolved).toEqual(source)
+    const resolved = await ctx.llm.resolveModelInfo('route', 'model')
+    expect(resolved.reasoning).toEqual(source)
     source.efforts[0]!.name = 'mutated'
-    expect(resolved?.efforts[0]?.name).toBe('Standard')
+    expect(resolved.reasoning?.efforts[0]?.name).toBe('Standard')
     await expect(ctx.llm.resolveCallConfig({ provider: 'route', model: 'model' })).resolves.toEqual({
       provider: 'route',
       model: 'model',
@@ -734,7 +760,7 @@ describe('LlmService', () => {
       {},
       { model: metadata as unknown as LlmModelReasoningInfo },
     ))
-    await expect(ctx.llm.resolveModelReasoning('route', 'model'))
+    await expect(ctx.llm.resolveModelInfo('route', 'model'))
       .rejects.toMatchObject({ code: 'INVALID_MODEL_REASONING' })
   })
 
@@ -764,13 +790,16 @@ describe('LlmService', () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
     const adapter = new class extends RecordingAdapter {
-      override resolveModelReasoning(
-        _provider: string,
-        _model: string,
-      ): Promise<LlmModelReasoningInfo> {
-        return Promise.resolve({
+      override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
+        const reasoning: LlmModelReasoningInfo = {
           efforts: [{ id: ReasoningEffortId('standard'), name: 'Standard' }],
           defaultEffort: ReasoningEffortId('standard'),
+        }
+        return Promise.resolve({
+          provider,
+          id: model,
+          name: model,
+          reasoning,
         })
       }
     }(SCRIPT)
@@ -799,19 +828,24 @@ describe('LlmService', () => {
     expect(Object.isFrozen(adapter.lastOptions)).toBe(true)
   })
 
-  it('pins one adapter registration across asynchronous reasoning resolution and dispatch', async () => {
+  it('pins one adapter registration across asynchronous exact-model resolution and dispatch', async () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
     const started = Promise.withResolvers<undefined>()
     const reasoning = Promise.withResolvers<LlmModelReasoningInfo>()
     const first = new class extends RecordingAdapter {
-      override resolveModelReasoning(
-        _provider: string,
-        _model: string,
+      override async resolveModel(
+        provider: string,
+        model: string,
         _signal?: AbortSignal,
-      ): Promise<LlmModelReasoningInfo> {
+      ): Promise<LlmResolvedModelInfo> {
         started.resolve(undefined)
-        return reasoning.promise
+        return {
+          provider,
+          id: model,
+          name: model,
+          reasoning: await reasoning.promise,
+        }
       }
     }(SCRIPT)
     const disposeFirst = ctx.llm.registerAdapter(['route'], first)
@@ -869,18 +903,18 @@ describe('LlmService', () => {
     })).toThrow(expect.objectContaining({ code: 'INVALID_PREPARED_CALL' }))
   })
 
-  it('passes cancellation through reasoning capability resolution', async () => {
+  it('passes cancellation through exact-model resolution', async () => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
     const started = Promise.withResolvers<undefined>()
     const adapter = new class extends ScriptedAdapter {
-      override resolveModelReasoning(
+      override resolveModel(
         _provider: string,
         _model: string,
         signal?: AbortSignal,
-      ): Promise<LlmModelReasoningInfo | undefined> {
+      ): Promise<LlmResolvedModelInfo> {
         started.resolve(undefined)
-        return new Promise((_resolve, reject) => {
+        return new Promise<LlmResolvedModelInfo>((_resolve, reject) => {
           if (signal === undefined) {
             reject(new Error('missing reasoning signal'))
             return
@@ -918,7 +952,7 @@ describe('LlmService', () => {
         [],
         { model: { contextWindow } },
       ))
-      await expect(ctx.llm.resolveModelContext('route', 'model'))
+      await expect(ctx.llm.resolveModelInfo('route', 'model'))
         .rejects.toMatchObject({ code: 'INVALID_MODEL_CONTEXT' })
     },
   )
diff --git a/packages/llm/token-meter/README.md b/packages/llm/token-meter/README.md
index 18f828ddd4..ef7579b2eb 100644
--- a/packages/llm/token-meter/README.md
+++ b/packages/llm/token-meter/README.md
@@ -4,7 +4,7 @@ Replay-aware token measurement through the singleton `ctx.tokenMeter` service. I
 
 ## Configuration
 
-The estimator has no settings. It intentionally uses one fixed heuristic: four characters per token plus structural overhead for roles, blocks, and request-envelope fields. Any key is rejected, including the obsolete global `contextWindow`; model capacity belongs to the adapter that owns an exact provider/model route and is available through `ctx.llm.resolveModelContext()`.
+The estimator has no settings. It intentionally uses one fixed heuristic: four characters per token plus structural overhead for roles, blocks, and request-envelope fields. Any key is rejected, including the obsolete global `contextWindow`; model capacity belongs to the adapter that owns an exact provider/model route and is available through `ctx.llm.resolveModelInfo().context`.
 
 ## Measurement contract
 
diff --git a/packages/support/llm-replay/src/index.ts b/packages/support/llm-replay/src/index.ts
index cd1993b1e2..3cac9d80b0 100644
--- a/packages/support/llm-replay/src/index.ts
+++ b/packages/support/llm-replay/src/index.ts
@@ -11,7 +11,7 @@ import { delimiter as pathDelimiter } from 'node:path'
 import type { Context } from 'cordis'
 import { decodeStorageRecord } from '@deepseek-ai/dsh-session'
 import type { SessionEvent } from '@deepseek-ai/dsh-session'
-import type { GenerateOptions, LlmModelContext, LlmModelInfo, LlmProviderInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
+import type { GenerateOptions, LlmModelInfo, LlmProviderInfo, LlmResolvedModelInfo, StreamChunk } from '@deepseek-ai/dsh-llm'
 import { LlmAdapter, LlmError, assertNever } from '@deepseek-ai/dsh-llm'
 
 /**
@@ -268,12 +268,20 @@ class ReplayAdapter extends LlmAdapter {
     })))
   }
 
-  override resolveModelContext(provider: string, model: string): Promise<LlmModelContext | undefined> {
+  override resolveModel(provider: string, model: string): Promise<LlmResolvedModelInfo> {
     const configured = this.providers.get(provider)
     /* v8 ignore next -- LlmService only asks about routes registered from this same map. */
-    if (configured === undefined) return Promise.resolve(undefined)
-    const contextWindow = configured.models?.find(candidate => candidate.id === model)?.contextWindow
-    return Promise.resolve(contextWindow === undefined ? undefined : { contextWindow })
+    if (configured === undefined) return Promise.resolve({ provider, id: model, name: model })
+    const configuredModel = configured.models?.find(candidate => candidate.id === model)
+    return Promise.resolve({
+      provider,
+      id: model,
+      name: configuredModel?.name ?? model,
+      ...configuredModel?.description === undefined ? {} : { description: configuredModel.description },
+      ...configuredModel?.contextWindow === undefined
+        ? {}
+        : { context: { contextWindow: configuredModel.contextWindow } },
+    })
   }
 
   override stream(options: GenerateOptions): AsyncIterable<StreamChunk> {
diff --git a/packages/support/llm-replay/tests/llm-replay.spec.ts b/packages/support/llm-replay/tests/llm-replay.spec.ts
index 645b655445..7ed638dc51 100644
--- a/packages/support/llm-replay/tests/llm-replay.spec.ts
+++ b/packages/support/llm-replay/tests/llm-replay.spec.ts
@@ -258,10 +258,12 @@ describe('installLlmReplay (through the real LlmService)', () => {
       { provider: 'deepseek', id: 'pro', name: 'Pro', description: 'Larger model' },
     ])
     await expect(ctx.llm.listModels('empty')).resolves.toEqual([])
-    await expect(ctx.llm.resolveModelContext('deepseek', 'flash')).resolves.toEqual({ contextWindow: 128_000 })
-    await expect(ctx.llm.resolveModelContext('deepseek', 'pro')).resolves.toBeUndefined()
-    await expect(ctx.llm.resolveModelContext('deepseek', 'unlisted')).resolves.toBeUndefined()
-    await expect(ctx.llm.resolveModelContext('empty', 'unlisted')).resolves.toBeUndefined()
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'flash')).resolves.toMatchObject({
+      context: { contextWindow: 128_000 },
+    })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'pro')).resolves.not.toHaveProperty('context')
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'unlisted')).resolves.not.toHaveProperty('context')
+    await expect(ctx.llm.resolveModelInfo('empty', 'unlisted')).resolves.not.toHaveProperty('context')
     expect(await drain(ctx.llm.stream({ provider: 'deepseek', model: 'pro', messages: [] }))).toEqual(TEXT_CHUNKS)
 
     dispose()
diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md
index 3326c548eb..b008a3f821 100644
--- a/packages/ui/tui/README.md
+++ b/packages/ui/tui/README.md
@@ -10,7 +10,7 @@ This package owns interactive terminal presentation and input only. It injects `
 
 After terminal startup succeeds, the package provides the terminal-local `ctx.tui` extension service. A plugin that injects it can call `openOverlay()` with a component factory and constrained layout options; the host exposes the viewport, semantic theme, display-text escaping, redraw, close, and a lifetime signal, but not the pi-tui tree, terminal, focus controller, or overlay handle. Plugin overlays, the model selector, and user questions share one FIFO modal queue. Each request is an effect of the calling plugin fiber, so unload removes queued work or closes visible work before cleanup settles; terminal shutdown unloads dependents before stopping pi-tui. Overlay state is not logged or replayed. Component code is trusted and may render ANSI styling, but must pass untrusted text through `host.display()`. The [interactive-extension Agent Note](../../../.agents/notes/implemented/architecture/2026-07-22-tui-interactive-extension-service.md) owns the boundary and rejected alternatives.
 
-The TUI rebuilds resumed history from the active session surface, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the latest `todo/write` plan above the editor, and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes `<session title> — <configured title>`. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelContext()` for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode plus the current model and any explicitly selected reasoning effort; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. Surface replacement events rebuild the transcript so compacted history does not reappear.
+The TUI rebuilds resumed history from the active session surface, renders Markdown responses and reasoning, applies each tool's `presentCall` / `presentResult` intent to terminal, diff, or generic cards, keeps the latest `todo/write` plan above the editor, and presents `ctx.userInteraction` questions in a wide bottom-left keyboard panel with progress, numbered options, and aligned descriptions. The latest logged session title becomes the header subtitle, with `welcome` before a title exists, and the terminal window title becomes `<session title> — <configured title>`. A durable `llm/retry` event retracts the failed step's live chunks and renders the scheduled retry count, delay, and failure in the transcript; success, exhaustion, and cancellation then settle through ordinary session events. The footer totals each logged model step's usage once, including failed attempts, while treating committed-message usage as a fallback for logs without a usage chunk. Its idle view compares token-meter pressure with `ctx.llm.resolveModelInfo()` context for the current route, displays `context unknown` when the adapter has no capacity metadata, and also shows tool-card mode plus the current model and any explicitly selected reasoning effort; while the agent runs, an elapsed working indicator and `esc interrupt` replace that summary. Surface replacement events rebuild the transcript so compacted history does not reappear.
 
 An embedding may provide `TuiRuntime.formatCwd` when its logical workspace label differs from the session's host directory. The override changes only the footer label; tools continue to use the session `cwd`.
 
@@ -22,11 +22,11 @@ When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers
 
 While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/reasoning`, `/tools`, `/redraw`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool cards collapse long bodies into a configurable head/tail preview; Ctrl+O toggles every card between its preview and full output. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle.
 
-`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape closes it. Models without selectable effort metadata ignore Shift+Tab; the selector does not synthesize `off`, clamp a value, or transfer an effort between models. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
+`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape closes it. Models without selectable effort metadata ignore Shift+Tab; the selector renders the exact advertised list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
 
 `/reload` (EXPERIMENTAL, dev-only) re-reads every file-backed loader config tree and applies the diff to the running app — the HMR watcher's config path, invoked manually; it needs the cordis Loader in the context and degrades to a warning without one, runs only while the agent is idle, and refuses re-entry while a reload is in flight. Module-source hot reload remains watcher-owned. When a `skills` service is mounted, `/skill:<name> [instructions]` loads that skill's instructions into the conversation as a user turn; autocomplete lists the model-invocable skills, and any skill (including a model-disabled one) is loadable by its exact name.
 
-The footer sums the session's reported usage as `↑<uncached input> ↓<output>`, followed by `cache <rate>%` once any input has been billed — the share of billed prompt tokens (uncached input plus cache reads and writes) served from the provider cache, rounded to a percent. It also compares token-meter pressure with `ctx.llm.resolveModelContext()` for the current route (omitting the context share when the adapter has no capacity metadata) and shows the current model and tool-card mode; the right side clips first when the footer is narrow.
+The footer sums the session's reported usage as `↑<uncached input> ↓<output>`, followed by `cache <rate>%` once any input has been billed — the share of billed prompt tokens (uncached input plus cache reads and writes) served from the provider cache, rounded to a percent. It also compares token-meter pressure with `ctx.llm.resolveModelInfo()` context for the current route (omitting the context share when the adapter has no capacity metadata) and shows the current model and tool-card mode; the right side clips first when the footer is narrow.
 
 `/status` adds a point-in-time diagnostics card to the transcript and remains available while the agent runs. It reports the session id, title, working directory, selected provider/model, selected reasoning effort or default behavior, reasoning-block visibility, agent state, event/turn/step/tool-call counts, exact input/output/cache token buckets, KV-cache hit rate, token-meter context use and capacity, creation time, and latest event time. Missing titles, models, cache input, or context capacity are labeled instead of inferred. The card is terminal-only and does not duplicate the compact footer.
 
diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts
index 21db41789c..e6f19cde47 100644
--- a/packages/ui/tui/src/index.ts
+++ b/packages/ui/tui/src/index.ts
@@ -630,7 +630,7 @@ async function readModelChoices(
       models.push({ provider: provider.id, id: current.model, name: current.model })
     }
     return Promise.all(models.map(async (model): Promise<ModelChoice> => {
-      const reasoning = await ctx.llm.resolveModelReasoning(provider.id, model.id)
+      const reasoning = (await ctx.llm.resolveModelInfo(provider.id, model.id)).reasoning
       return {
         provider: provider.id,
         model: model.id,
@@ -2101,8 +2101,8 @@ export function createTuiChat(
     contextWindow = undefined
     const resolution = selected === undefined
       ? Promise.resolve({ kind: 'resolved', contextWindow: undefined } as const)
-      : ctx.llm.resolveModelContext(selected.provider, selected.model).then(
-        context => ({ kind: 'resolved', contextWindow: context?.contextWindow } as const),
+      : ctx.llm.resolveModelInfo(selected.provider, selected.model).then(
+        info => ({ kind: 'resolved', contextWindow: info.context?.contextWindow } as const),
         (error: unknown) => ({ kind: 'error', error } as const),
       )
     contextResolution = resolution
diff --git a/packages/ui/tui/tests/harness.ts b/packages/ui/tui/tests/harness.ts
index e42610644c..c74c469133 100644
--- a/packages/ui/tui/tests/harness.ts
+++ b/packages/ui/tui/tests/harness.ts
@@ -10,10 +10,9 @@ import AgentRegistry, {
 } from '@deepseek-ai/dsh-agent'
 import type {
   ContentBlock,
-  LlmModelContext,
   LlmModelInfo,
-  LlmModelReasoningInfo,
   LlmProviderInfo,
+  LlmResolvedModelInfo,
 } from '@deepseek-ai/dsh-llm'
 import CommandService from '@deepseek-ai/dsh-commands'
 import SessionStore, { SessionId, type Session, type SessionHeader } from '@deepseek-ai/dsh-session'
@@ -53,11 +52,10 @@ export interface TuiHarnessOptions {
     providers: LlmProviderInfo[]
     models: LlmModelInfo[]
     listModels?: (provider: string) => Promise<LlmModelInfo[]>
-    resolveModelContext?: (provider: string, model: string) => Promise<LlmModelContext | undefined>
-    resolveModelReasoning?: (
+    resolveModelInfo?: (
       provider: string,
       model: string,
-    ) => Promise<LlmModelReasoningInfo | undefined>
+    ) => Promise<Pick<LlmResolvedModelInfo, 'context' | 'reasoning'>>
   }
   /** Provide a fake `sessionPersistence` service so resume surfaces can list sessions. */
   sessionPersistence?: {
@@ -128,12 +126,20 @@ export async function createTuiTestHarness<TerminalType extends Terminal, Exit e
         return catalog.listModels?.(provider)
           ?? Promise.resolve(catalog.models.filter(model => model.provider === provider).map(model => ({ ...model })))
       },
-      resolveModelContext(provider: string, model: string) {
-        return catalog.resolveModelContext?.(provider, model)
-          ?? Promise.resolve({ contextWindow: options.contextWindow ?? 128_000 })
-      },
-      resolveModelReasoning(provider: string, model: string) {
-        return catalog.resolveModelReasoning?.(provider, model) ?? Promise.resolve(undefined)
+      async resolveModelInfo(provider: string, model: string) {
+        const advertised = catalog.models.find(candidate =>
+          candidate.provider === provider && candidate.id === model)
+        const capabilities = await (catalog.resolveModelInfo?.(provider, model)
+          ?? Promise.resolve({
+            context: { contextWindow: options.contextWindow ?? 128_000 },
+          }))
+        return {
+          provider,
+          id: model,
+          name: advertised?.name ?? model,
+          ...advertised?.description === undefined ? {} : { description: advertised.description },
+          ...capabilities,
+        }
       },
     } as never)
   }
diff --git a/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
index a7db8ccba0..74ca319f8c 100644
--- a/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
@@ -26,7 +26,7 @@ buffer
   style 8-8 fg=bright-blue
   style 38-77 fg=bright-black
   style 83-83 fg=bright-blue
-15| "        │ → deepseek/deepseek-v4-pro        DeepSeek V4 Pro — Max                  │        "
+15| "        │ → deepseek/deepseek-v4-pro        DeepSeek V4 Pro — Off                  │        "
   style 8-8 fg=bright-blue
   style 10-64 fg=bright-blue inverse
   style 83-83 fg=bright-blue
diff --git a/packages/ui/tui/tests/snapshots/model-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
index 70b61f1022..67faad5c40 100644
--- a/packages/ui/tui/tests/snapshots/model-switching.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
@@ -8,10 +8,10 @@ buffer
   style 10-16 bold
 1| " Snapshot agent ready."
   style 1-21 fg=bright-black
-2| " deepseek-v4-pro max  •  main-session"
+2| " deepseek-v4-pro off  •  main-session"
   style 1-36 dim
 3| <blank>
-4| " Model selected: deepseek/deepseek-v4-pro. Reasoning effort: Max. New steps will use it.    "
+4| " Model selected: deepseek/deepseek-v4-pro. Reasoning effort: Off. New steps will use it.    "
   style 1-87 fg=bright-black
 5| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
@@ -19,7 +19,7 @@ buffer
   style 1-1 inverse
 7| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
-8| "deepseek-v4-pro max  /workspace/project  ↑0 ↓0                   0% context  tools:collapsed"
+8| "deepseek-v4-pro off  /workspace/project  ↑0 ↓0                   0% context  tools:collapsed"
   style 0-45 dim
   style 65-91 dim
 9-31| <blank>
diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts
index 9e30d3525e..539ffc57e5 100644
--- a/packages/ui/tui/tests/tui.snapshot.ts
+++ b/packages/ui/tui/tests/tui.snapshot.ts
@@ -640,12 +640,16 @@ describe('TUI terminal-state snapshots', () => {
           { provider: 'deepseek', id: 'deepseek-v4-flash', name: 'DeepSeek V4 Flash' },
           { provider: 'deepseek', id: 'deepseek-v4-pro', name: 'DeepSeek V4 Pro' },
         ],
-        resolveModelReasoning: () => Promise.resolve({
-          efforts: [
-            { id: ReasoningEffortId('high'), name: 'High' },
-            { id: ReasoningEffortId('max'), name: 'Max' },
-          ],
-          defaultEffort: ReasoningEffortId('high'),
+        resolveModelInfo: () => Promise.resolve({
+          context: { contextWindow: 128_000 },
+          reasoning: {
+            efforts: [
+              { id: ReasoningEffortId('off'), name: 'Off' },
+              { id: ReasoningEffortId('high'), name: 'High' },
+              { id: ReasoningEffortId('max'), name: 'Max' },
+            ],
+            defaultEffort: ReasoningEffortId('high'),
+          },
         }),
       },
     }, { columns: 92, rows: 32 })
@@ -657,6 +661,7 @@ describe('TUI terminal-state snapshots', () => {
     await renderAfter(harness, () => {
       harness.terminal.send('\x1b[B')
       harness.terminal.send('\x1b[Z')
+      harness.terminal.send('\x1b[Z')
     })
     await checkpoint('model-effort-switching', harness.terminal, { includeScrollback: true })
     await renderAfter(harness, () => {
diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts
index 8af6252554..02b37efa8c 100644
--- a/packages/ui/tui/tests/tui.spec.ts
+++ b/packages/ui/tui/tests/tui.spec.ts
@@ -5,7 +5,11 @@ import { describe, expect, it, vi } from 'vitest'
 import { Context } from 'cordis'
 import { CombinedAutocompleteProvider, type Terminal } from '@earendil-works/pi-tui'
 import AgentRegistry, { agentEvents, assembleContextFor, AgentMessageId, type Agent } from '@deepseek-ai/dsh-agent'
-import { ReasoningEffortId, type LlmCallConfig } from '@deepseek-ai/dsh-llm'
+import {
+  ReasoningEffortId,
+  type LlmCallConfig,
+  type LlmModelReasoningInfo,
+} from '@deepseek-ai/dsh-llm'
 import { GOAL_CHANGE_VERSION, GoalId, renderGoalChange, type GoalSnapshotChangeMeta } from '@deepseek-ai/dsh-goal'
 import CommandService, { type CommandInvocation } from '@deepseek-ai/dsh-commands'
 import SessionStore, { SessionId, type JsonValue, type SessionEvent, type SessionHeader, type TurnEndReason } from '@deepseek-ai/dsh-session'
@@ -143,7 +147,11 @@ function provideLlmCatalog(ctx: Context): void {
   ctx.provide('llm', {
     listProviders: () => [],
     listModels: () => Promise.resolve([]),
-    resolveModelContext: () => Promise.resolve(undefined),
+    resolveModelInfo: (provider: string, model: string) => Promise.resolve({
+      provider,
+      id: model,
+      name: model,
+    }),
   } as never)
 }
 
@@ -1155,7 +1163,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
     expect(result.terminal.output).toContain('restored answer')
     expect(result.terminal.output).toContain('write tests')
     expect(result.terminal.output).toContain('↑1.3k ↓42')
-    // Context resolution is async (resolveModelContext); settle before reading.
+    // Exact model resolution is async; settle before reading.
     await tick()
     expect(result.terminal.output).toContain('42% context  tools:collapsed')
     // Narrow terminals clip the right-hand context/tools segment first; the
@@ -1741,7 +1749,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
       catalog: {
         providers: [],
         models: [],
-        resolveModelContext: () => Promise.resolve(undefined),
+        resolveModelInfo: () => Promise.resolve({}),
       },
     })
     result.terminal.send('/status')
@@ -2288,6 +2296,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
 
   it('opens a keyboard selector and switches the session model without sending slash text to the agent', async () => {
     const initialContext = Promise.withResolvers<{ contextWindow: number }>()
+    let deferInitialContext = true
     const result = await setup({
       agentOptions: { provider: 'alpha', model: 'a1' },
       contextTokens: 50,
@@ -2299,37 +2308,41 @@ describe('pi-tui chat lifecycle and transcript', () => {
           { provider: 'beta', id: 'b1', name: 'Beta One' },
           { provider: 'beta', id: 'shared', name: 'Beta Shared' },
         ],
-        resolveModelContext: (provider, model) => provider === 'alpha' && model === 'a1'
-          ? initialContext.promise
-          : Promise.resolve({ contextWindow: 200 }),
-        resolveModelReasoning: (provider, model) => {
+        async resolveModelInfo(provider, model) {
+          const shouldDeferContext = provider === 'alpha' && model === 'a1' && deferInitialContext
+          if (shouldDeferContext) deferInitialContext = false
+          const context = shouldDeferContext
+            ? await initialContext.promise
+            : { contextWindow: 200 }
+          let reasoning: LlmModelReasoningInfo | undefined
           if (model === 'a1') {
-            return Promise.resolve({
+            reasoning = {
               efforts: [
                 { id: ReasoningEffortId('low'), name: 'Low' },
                 { id: ReasoningEffortId('high'), name: 'High' },
               ],
               defaultEffort: ReasoningEffortId('low'),
-            })
-          }
-          if (model === 'b1') {
-            return Promise.resolve({
+            }
+          } else if (model === 'b1') {
+            reasoning = {
               efforts: [
                 { id: ReasoningEffortId('high'), name: 'High' },
                 { id: ReasoningEffortId('max'), name: 'Max' },
               ],
               defaultEffort: ReasoningEffortId('high'),
-            })
-          }
-          if (provider === 'alpha' && model === 'shared') {
-            return Promise.resolve({
+            }
+          } else if (provider === 'alpha' && model === 'shared') {
+            reasoning = {
               efforts: [
                 { id: ReasoningEffortId('standard'), name: 'Standard' },
                 { id: ReasoningEffortId('ultra'), name: 'Ultra' },
               ],
-            })
+            }
+          }
+          return {
+            context,
+            ...reasoning === undefined ? {} : { reasoning },
           }
-          return Promise.resolve(undefined)
         },
       },
     })
@@ -2522,7 +2535,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
       catalog: {
         providers: [{ id: 'alpha', name: 'Alpha' }],
         models: [{ provider: 'alpha', id: 'a1', name: 'Alpha One' }],
-        resolveModelContext: () => Promise.resolve(undefined),
+        resolveModelInfo: () => Promise.resolve({}),
       },
     })
     unset.terminal.send('/model')
@@ -2554,7 +2567,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
         providers: [{ id: 'deepseek', name: 'DeepSeek' }],
         models: [],
         listModels: () => Promise.reject(new Error('catalog offline')),
-        resolveModelContext: () => Promise.reject(new Error('capacity offline')),
+        resolveModelInfo: () => Promise.reject(new Error('capacity offline')),
       },
     })
     failed.terminal.send('/model')
@@ -2569,7 +2582,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
       catalog: {
         providers: [{ id: 'deepseek', name: 'DeepSeek' }],
         models: [{ provider: 'deepseek', id: 'model-1', name: 'Model One' }],
-        resolveModelReasoning: () => Promise.reject(new Error('reasoning metadata offline')),
+        resolveModelInfo: () => Promise.reject(new Error('reasoning metadata offline')),
       },
     })
     reasoningFailed.terminal.send('/model')
@@ -2619,7 +2632,7 @@ describe('pi-tui chat lifecycle and transcript', () => {
       catalog: {
         providers: [{ id: 'deepseek', name: 'DeepSeek' }],
         models: [],
-        resolveModelContext: () => context.promise,
+        resolveModelInfo: () => context.promise.then(value => ({ context: value })),
       },
     })
     await contextResult.controller.dispose()
diff --git a/scripts/gen-cordis-catalog.ts b/scripts/gen-cordis-catalog.ts
index 30b3ff7518..70d6e38968 100644
--- a/scripts/gen-cordis-catalog.ts
+++ b/scripts/gen-cordis-catalog.ts
@@ -41,6 +41,7 @@ export const LINK_MAP: Record<string, string> = {
   LlmCallConfig: 'core.md',
   LlmModelContext: 'core.md',
   LlmModelReasoningInfo: 'core.md',
+  LlmResolvedModelInfo: 'core.md',
   LlmFailure: 'llm-streaming.md',
   LlmModelInfo: 'core.md',
   LlmProviderInfo: 'core.md',
diff --git a/scripts/type-equiv.manifest.json b/scripts/type-equiv.manifest.json
index c003b4b199..a01e7f01f4 100644
--- a/scripts/type-equiv.manifest.json
+++ b/scripts/type-equiv.manifest.json
@@ -61,6 +61,11 @@
       "symbol": "LlmModelReasoningInfo",
       "source": "packages/llm/llm/src/types.ts"
     },
+    {
+      "doc": "docs/core-data-structures/core.md",
+      "symbol": "LlmResolvedModelInfo",
+      "source": "packages/llm/llm/src/types.ts"
+    },
     {
       "doc": "docs/core-data-structures/core.md",
       "symbol": "GenerateOptions",
@@ -1309,6 +1314,11 @@
       "symbol": "LlmModelReasoningInfo",
       "source": "packages/llm/llm/src/types.ts"
     },
+    {
+      "doc": "docs/core-data-structures/core.zh.md",
+      "symbol": "LlmResolvedModelInfo",
+      "source": "packages/llm/llm/src/types.ts"
+    },
     {
       "doc": "docs/core-data-structures/core.zh.md",
       "symbol": "GenerateOptions",

From 970f893d6b0f2fbcf7fa4a0cab5e9f9d8cbff039 Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sun, 26 Jul 2026 13:30:01 +0800
Subject: [PATCH 24/36] feat(tui): add pi-ai Anthropic example

---
 ...-14-provider-routed-llm-adapters.i18n.yaml |  4 +-
 ...2026-07-14-provider-routed-llm-adapters.md |  1 +
 ...6-07-14-provider-routed-llm-adapters.zh.md |  1 +
 examples/tui-agent/README.md                  | 10 ++++-
 examples/tui-agent/pi-ai-anthropic.cordis.yml | 39 +++++++++++++++++++
 .../tui-agent/tests/tui-keyless-smoke.e2e.ts  | 21 ++++++++++
 6 files changed, 73 insertions(+), 3 deletions(-)
 create mode 100644 examples/tui-agent/pi-ai-anthropic.cordis.yml

diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
index 3a6b18cecf..16a2f7752a 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-14-provider-routed-llm-adapters.md: 98205d18d07752e0cdba86d7cba80368d45fd816
-2026-07-14-provider-routed-llm-adapters.zh.md: c35225a86baf4c2d09732b5940abbc8046d365fb
+2026-07-14-provider-routed-llm-adapters.md: a0f43751a17134f42e3f1a73879c79521fa7921b
+2026-07-14-provider-routed-llm-adapters.zh.md: a8712166e74e323e3091251766fd26fefd0690d0
diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
index 98205d18d0..a0f43751a1 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
@@ -84,6 +84,7 @@ The on-disk session format remains the pre-release pinned version `0`, with no c
 
 - Unit coverage exercises registry conflicts, request reconstruction, session validation, profile resolution, single-attempt option forwarding, native API selection including OpenAI Responses, conversion, replay validation, error mapping, caller cancellation, idle-timeout transport termination, content rewrites, and same-instance versus different-instance replay dispatch.
 - Keyless loop/session tests and ACP snapshots exercise durable provider/model metadata, resume and fork propagation, workflow/subagent overrides, and unchanged user-visible transcripts; the key-gated DeepSeek e2e retains real provider streaming and tool follow-up coverage.
+- The TUI's `pi-ai-anthropic.cordis.yml` overlay disables the native provider owner, mounts an explicit Anthropic profile over deployment-supplied credentials and endpoint, and selects a catalog model. Its keyless PTY smoke opens `/model`, observes the profile's default reasoning effort, and selects the next adapter-advertised effort without provider I/O.
 - Public JSDoc, package READMEs, architecture and core-data-structure docs, generated catalogs, examples, session fixtures, and Python SDK pairs use provider/model targets consistently and are checked by the repository documentation and type-equivalence gates.
 
 ## Risks
diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
index c35225a86b..a8712166e7 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
@@ -84,6 +84,7 @@ JSON-RPC 运行时显式接收 provider 与 model。仅当 `deepseek` 提供方
 
 - 单元测试覆盖注册表冲突、请求重建、会话验证、配置解析、单次请求的选项转发、包括 OpenAI Responses 在内的原生 API 选择、转换、回放验证、错误映射、调用方取消、空闲超时导致的传输终止、内容重写，以及同一实例与不同实例间的回放分发。
 - 无密钥的 agent loop/会话测试和 ACP 快照覆盖持久化 provider/model 元数据、恢复与 fork 传播、工作流/subagent 覆盖，以及不变的用户可见 transcript（文本记录）；密钥门控的 DeepSeek e2e 测试保留真实提供方的流式输出与工具后续调用覆盖率。
+- TUI 的 `pi-ai-anthropic.cordis.yml` 覆盖配置会禁用原生提供方所有者，在部署提供的凭据和端点之上挂载显式 Anthropic 配置，并选择一个目录模型。其无密钥 PTY 冒烟测试会打开 `/model`，观察该配置的默认推理级别，并在不执行提供方 I/O 的情况下选择适配器公布的下一个推理级别。
 - 公共 JSDoc、package README、架构与核心数据结构文档、生成目录、示例、会话 fixture（测试前置数据）和 Python SDK 配对文档统一使用 provider/model 目标，并由仓库文档与类型等价门禁校验。
 
 ## 风险
diff --git a/examples/tui-agent/README.md b/examples/tui-agent/README.md
index 47ba6e45a2..fef7f99a91 100644
--- a/examples/tui-agent/README.md
+++ b/examples/tui-agent/README.md
@@ -13,6 +13,14 @@ pnpm run demo:tui
 
 Both the demo script and the installable `dsh` CLI ([`apps/cli`](../../apps/cli/README.md)) boot this example's `cordis.yml` as the shipped default config; `dsh` additionally applies the personal overlay from `~/.dsh` and uses the invoking directory as the workspace.
 
+### Anthropic-compatible pi-ai route
+
+The checked-in `pi-ai-anthropic.cordis.yml` overlay disables the native DeepSeek adapter, mounts one explicit `anthropic` profile through `dsh-llm-pi-ai`, and selects `claude-sonnet-4-6` with `high` reasoning by default. It deliberately reuses `DEEPSEEK_API_KEY` and `DEEPSEEK_BASE_URL` for a deployment whose private endpoint accepts the Anthropic Messages protocol; a DeepSeek Chat Completions endpoint is not interchangeable. Run it from the repository root:
+
+```sh
+pnpm run demo:tui --config examples/tui-agent/pi-ai-anthropic.cordis.yml
+```
+
 Type a coding task. The agent works through the `read`/`write`/`edit` filesystem tools for ordinary file operations and `bash` (+ the generic `task_output` / `task_list` / `task_kill` for background tasks) for shell commands, searches, and test runs, each in a fresh `bash -c` (the system prompt tells the model to pass `workdir` instead of `cd`). Both the fs tools and bash resolve relative paths against the session workspace. It can also delegate with `subagent`/`subagent_fork`.
 
 The `todo_write` task tracker is opt-in and not in the shipped config: add `@deepseek-ai/dsh-tool-todo` to `cordis.yml` (or a personal-config overlay under `~/.dsh`) to expose it. Once loaded, the model records a whole-list plan to the session log and the TUI renders it.
@@ -51,7 +59,7 @@ This example is a thin leaf `cordis.yml`: it picks the swappable backends, loads
 | Entry | Demonstrates |
 |---|---|
 | `hmr` (`@cordisjs/plugin-hmr`) | the dev/demo edit-reload loop — a **leaf** entry (not baked into the app) because it depends on the Loader's internal module access |
-| `llm-deepseek` | real `LlmAdapter` via config (`!!js process.env.…` secrets); swap one line to `@deepseek-ai/dsh-llm-pi-ai` for the library-backed twin |
+| `llm-deepseek` | the default native adapter; `pi-ai-anthropic.cordis.yml` disables that owner and mounts an explicit library-backed Anthropic profile |
 | `bash` (`dsh-bash-local`) | the executor implementation — the swappable half of the bash seam. The model-facing `bash` schema (`tool-bash`) and generic `task_*` controls (`tool-tasks`) come from `dsh-agent-spine-demo`, so only the executor is a leaf choice |
 | `tui-agent` (`@deepseek-ai/dsh-tui-demo`) | the app bundle: the agent-spine demo + JSONL persistence + the pi-tui channel + a pre-created `main` agent |
 | `subagent`, `subagent-spawn`, `subagent-fork` | the subagent provider registry plus the two in-process backends: a fresh child and a child seeded with the parent's completed-turn prefix |
diff --git a/examples/tui-agent/pi-ai-anthropic.cordis.yml b/examples/tui-agent/pi-ai-anthropic.cordis.yml
new file mode 100644
index 0000000000..c1ec3b2968
--- /dev/null
+++ b/examples/tui-agent/pi-ai-anthropic.cordis.yml
@@ -0,0 +1,39 @@
+# Replace the native DeepSeek adapter with pi-ai's Anthropic route. The
+# configured endpoint must implement the Anthropic Messages protocol even
+# though this deployment reuses the repository's DeepSeek-named environment
+# variables.
+- id: base
+  name: '@cordisjs/plugin-include'
+  config:
+    path: ./cordis.yml
+    patches:
+      - id: llm-deepseek
+        name: '@deepseek-ai/dsh-llm-deepseek'
+        disabled: true
+      - id: tui-agent
+        name: '@deepseek-ai/dsh-tui-demo'
+        config:
+          provider: anthropic
+          model: claude-sonnet-4-6
+          resumeSessionId: !!js "typeof resumeSessionId === 'string' ? resumeSessionId : undefined"
+          persistenceRoot: './.sessions'
+          resumeCommand: 'dsh --config examples/tui-agent/pi-ai-anthropic.cordis.yml --resume {session}'
+          workspaceContext:
+            maxBytes: 65536
+          ui:
+            showReasoning: true
+            maxToolOutputLines: 6
+          persona: |
+            You are a coding agent powered by the {{model}} model.
+
+            Verify your work by running the code or tests. Keep answers brief and
+            factual.
+      - insert:
+          - id: llm-pi-ai
+            name: '@deepseek-ai/dsh-llm-pi-ai'
+            config:
+              providers:
+                - provider: anthropic
+                  apiKey: !!js process.env.DEEPSEEK_API_KEY
+                  baseURL: !!js process.env.DEEPSEEK_BASE_URL
+                  reasoning: high
diff --git a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
index 169b582492..79f0155cbd 100644
--- a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
+++ b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
@@ -11,6 +11,7 @@ import { runTuiPtySmoke, type TuiPtySmokeOptions } from './pty-harness.ts'
 const dshBinScript = fileURLToPath(new URL('../../../apps/cli/src/bin.ts', import.meta.url))
 const configPath = fileURLToPath(new URL('../cordis.yml', import.meta.url))
 const codeModeConfigPath = fileURLToPath(new URL('../code-mode.cordis.yml', import.meta.url))
+const piAiAnthropicConfigPath = fileURLToPath(new URL('../pi-ai-anthropic.cordis.yml', import.meta.url))
 const scriptedConfigPath = fileURLToPath(new URL('./fixtures/tui-scripted.cordis.yml', import.meta.url))
 const tsconfigPath = fileURLToPath(new URL('../../../tsconfig.json', import.meta.url))
 
@@ -246,6 +247,26 @@ describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => {
     expect(output).toContain('\u001B[?2004l')
   }, LOADER_SMOKE_TEST_TIMEOUT_MS)
 
+  it('boots the pi-ai Anthropic overlay and selects an advertised reasoning effort', async () => {
+    const output = await smoke({
+      label: 'tui-agent pi-ai anthropic',
+      tempDirPrefix: 'tui-agent-pi-ai-anthropic-',
+      configPath: piAiAnthropicConfigPath,
+      env: {
+        DEEPSEEK_API_KEY: 'keyless-tui-no-call',
+        DEEPSEEK_BASE_URL: 'http://127.0.0.1:1',
+      },
+      actions: [
+        { waitFor: 'main-session-', send: '/model\r' },
+        { waitFor: 'Claude Sonnet 4.6 — High — current', send: '\x1b[Z\r' },
+        { waitFor: 'Reasoning effort: Max.', send: '/exit\r' },
+      ],
+    })
+    expect(output).toContain('anthropic/claude-sonnet-4-6')
+    expect(output).toContain('Reasoning effort: Max.')
+    expect(output).toContain('\u001B[?2004l')
+  }, LOADER_SMOKE_TEST_TIMEOUT_MS)
+
 })
 
 describe('dsh CLI keyless smoke (apps/cli through the same PTY)', () => {

From 15d6f4c0852801c25ac23f40ecb0d932b4a5de60 Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sun, 26 Jul 2026 13:42:02 +0800
Subject: [PATCH 25/36] feat(llm-deepseek): expose off as a dynamic reasoning
 effort

---
 ...ed-reasoning-effort-capabilities.i18n.yaml |   4 +-
 ...ter-owned-reasoning-effort-capabilities.md |   4 +-
 ...-owned-reasoning-effort-capabilities.zh.md |   4 +-
 docs/config-catalog.md                        |   6 +-
 docs/cookbook/adding-an-llm-adapter.i18n.yaml |   4 +-
 docs/cookbook/adding-an-llm-adapter.md        |   2 +-
 docs/cookbook/adding-an-llm-adapter.zh.md     |   2 +-
 packages/llm/llm-deepseek/README.md           |   9 +-
 packages/llm/llm-deepseek/src/adapter.ts      |  26 +++-
 packages/llm/llm-deepseek/src/index.ts        |  14 +-
 packages/llm/llm-deepseek/src/serialize.ts    |  47 +++++--
 .../llm/llm-deepseek/tests/adapter.e2e.ts     |  30 ++--
 .../llm/llm-deepseek/tests/adapter.spec.ts    | 130 +++++++++++++++---
 .../llm/llm-deepseek/tests/serialize.spec.ts  |  31 +++++
 14 files changed, 240 insertions(+), 73 deletions(-)

diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
index c3d5b45684..e9adb2cd8b 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-24-adapter-owned-reasoning-effort-capabilities.md: ffd633a3876c96f312fb274159f3f8422f8811ab
-2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: 05c9941ad9c0906ab5400b5a215b92381fba3119
+2026-07-24-adapter-owned-reasoning-effort-capabilities.md: cc66e4ec151fcc04445a91f4a3527cbddd130c33
+2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md: e0d28e1aca370068478e8fb1704defeaac3ab351
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
index ffd633a387..cc66e4ec15 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.md
@@ -14,7 +14,7 @@ Reasoning strength was adapter configuration only, so a conversation could not d
 
 `LlmCallConfig` and `GenerateOptions` carry the optional effort. The agent loop prepares the post-`agent/request` config under the active turn signal before writing `request/header`, so defaults and dynamic changes are model-visible only after becoming durable facts. The prepared call retains the exact adapter registration across asynchronous exact-model resolution, durable header logging, and dispatch; direct `LlmService.stream()` calls likewise capture their final registration before awaiting resolution. A route with no registered adapter retains its proposed config so an `llm/stream` middleware can own and short-circuit it; terminal dispatch still rejects an unhandled route. A resumed loop retains the logged effort only when its initial provider/model route is unchanged; a route change discards the previous model's opaque id.
 
-The native DeepSeek adapter advertises `high` and `max`, defaults to configured effort or `high`, and exposes no effort capability while thinking is disabled. The pi-ai adapter publishes each exact model's `getSupportedThinkingLevels()` result unchanged, including `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai. Its common stream options represent `off` by omitting `reasoning`, as required by pi-ai's own API.
+The native DeepSeek adapter advertises `off`, `high`, and `max` when deployment policy permits thinking, and defaults to the configured effort or `high`. Its adapter-owned `off` maps to `thinking.type: disabled` with no `reasoning_effort`; `high` and `max` enable thinking and carry their official wire effort. A `thinking: disabled` deployment publishes only `off` and rejects attempts to enable thinking before provider I/O. The pi-ai adapter publishes each exact model's `getSupportedThinkingLevels()` result unchanged, including `off`, preserves an absent profile default as a provider default, and leaves provider wire-value mapping inside pi-ai. Its common stream options represent `off` by omitting `reasoning`, as required by pi-ai's own API.
 
 ## Alternatives considered
 
@@ -28,6 +28,6 @@ The native DeepSeek adapter advertises `high` and `max`, defaults to configured
 
 ## Consequences
 
-Clients can query one exact route once and render its identity, context capacity, and adapter-owned reasoning choices without knowing a global enum. Adapter configuration remains the deployment-default owner, while `agent/request` can replace the effective effort on each step. Invalid exact identity, context, or reasoning metadata fails with `INVALID_MODEL_INFO`, `INVALID_MODEL_CONTEXT`, or `INVALID_MODEL_REASONING`; unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
+Clients can query one exact route once and render its identity, context capacity, and adapter-owned reasoning choices without knowing a global enum or synthesizing `off`. Adapter configuration remains the deployment-default and policy owner, while `agent/request` can replace the effective effort on each step within that policy. Invalid exact identity, context, or reasoning metadata fails with `INVALID_MODEL_INFO`, `INVALID_MODEL_CONTEXT`, or `INVALID_MODEL_REASONING`; unsupported explicit or configured values fail with `UNSUPPORTED_REASONING_EFFORT` before provider I/O.
 
 The aggregate exact-model query is asynchronous and may fail for adapters backed by authoritative catalogs. Its optional signal is the caller's cancellation boundary; an asynchronous adapter must settle promptly after abort so loop disposal can reach quiescence. Keyless service, adapter, loop, session, and request-header tests pin validation, defaulting, dynamic changes, logging, resume behavior, HMR registration ownership, and cancellation; runnable snapshots pin the resolved effort in real assembled request headers, while key-gated adapter tests exercise provider serialization.
diff --git a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
index 05c9941ad9..e0d28e1aca 100644
--- a/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-24-adapter-owned-reasoning-effort-capabilities.zh.md
@@ -14,7 +14,7 @@ Status: implemented
 
 `LlmCallConfig` 和 `GenerateOptions` 携带可选的推理强度。agent loop（智能体循环）在活跃轮次信号的控制下准备 `agent/request` 处理完成后的配置，再写入 `request/header`，因此默认值和动态变更只有成为持久化事实后才对模型可见。准备完成的调用在异步确切模型解析、请求头持久记录和分派全程保留同一项确切的适配器注册；直接调用 `LlmService.stream()` 时，也会在等待解析前捕获最终的适配器注册。没有已注册适配器的路由会保留原定配置，使 `llm/stream` 中间件可以接管并短路该请求；若仍未得到处理，最终分发会拒绝该路由。恢复后的主循环仅在初始提供方/模型路由未变时保留日志中记录的推理强度；如果路由发生变化，则丢弃上一模型的不透明 ID。
 
-原生 DeepSeek 适配器声明 `high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`；禁用思考时不暴露推理强度能力。pi-ai 适配器原样发布每个确切模型的 `getSupportedThinkingLevels()` 结果，其中包括 `off`；profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。按照 pi-ai 自身 API 的要求，其通用流选项通过省略 `reasoning` 来表示 `off`。
+当部署策略允许思考时，原生 DeepSeek 适配器声明 `off`、`high` 和 `max`，默认使用配置指定的推理强度，若未配置则使用 `high`。由适配器持有的 `off` 映射为 `thinking.type: disabled`，且不带 `reasoning_effort`；`high` 和 `max` 会启用思考并携带各自的官方协议强度值。配置为 `thinking: disabled` 的部署仅声明 `off`，并会在提供方 I/O 前拒绝启用思考的尝试。pi-ai 适配器原样发布每个确切模型的 `getSupportedThinkingLevels()` 结果，其中包括 `off`；profile 未指定默认值时保留提供方默认行为，并将提供方协议值的映射留在 pi-ai 内部。按照 pi-ai 自身 API 的要求，其通用流选项通过省略 `reasoning` 来表示 `off`。
 
 ## 备选方案
 
@@ -28,6 +28,6 @@ Status: implemented
 
 ## 影响
 
-客户端只需查询一次确切路由，即可渲染其身份、上下文容量和由适配器持有的推理选项，而无需了解全局枚举。适配器配置仍负责提供部署默认值，`agent/request` 则可以在每个步骤替换实际生效的推理强度。确切身份、上下文或推理元数据无效时，分别抛出 `INVALID_MODEL_INFO`、`INVALID_MODEL_CONTEXT` 或 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
+客户端只需查询一次确切路由，即可渲染其身份、上下文容量和由适配器持有的推理选项，而无需了解全局枚举或自行合成 `off`。适配器配置仍是部署默认值和策略的归属方，`agent/request` 则可以在该策略范围内为每个步骤替换实际生效的推理强度。确切身份、上下文或推理元数据无效时，分别抛出 `INVALID_MODEL_INFO`、`INVALID_MODEL_CONTEXT` 或 `INVALID_MODEL_REASONING`；显式指定或配置指定的值不受支持时，会在提供方 I/O 前抛出 `UNSUPPORTED_REASONING_EFFORT`。
 
 确切模型元数据的聚合查询采用异步方式，并且对于由权威目录支持的适配器可能失败。可选信号构成调用方的取消边界；异步适配器必须在信号中止后迅速完成结算，使主循环的资源释放达到完全停稳。无密钥的服务、适配器、主循环、会话和请求头测试为校验、默认值解析、动态变更、日志记录、恢复行为、HMR（热模块替换）期间的注册所有权和取消提供回归保障；可运行快照锁定实际组装请求头中的已解析推理强度，仅在有密钥时运行的适配器测试则覆盖提供方序列化。
diff --git a/docs/config-catalog.md b/docs/config-catalog.md
index a96858c3e6..cc269754bf 100644
--- a/docs/config-catalog.md
+++ b/docs/config-catalog.md
@@ -574,10 +574,10 @@ export interface Config {
   apiKey?: string
   /** Endpoint base; falls back to $DEEPSEEK_BASE_URL, then the public API. */
   baseURL?: string
-  /** Thinking-mode default for every request (provider default: enabled). */
+  /** Deployment thinking policy; `disabled` limits every conversation request to `off`. */
   thinking?: 'enabled' | 'disabled'
-  /** Default thinking effort when thinking is enabled (default `high`). */
-  reasoningEffort?: 'high' | 'max'
+  /** Default thinking effort (default `high`); `off` disables thinking per request. */
+  reasoningEffort?: 'off' | 'high' | 'max'
   /** Positive context capacity used when the selected model has no exact value. */
   defaultContextWindow?: number
   /** Advisory models shown by discovery consumers; defaults to V4 Flash and V4 Pro. */
diff --git a/docs/cookbook/adding-an-llm-adapter.i18n.yaml b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
index ca8c27ddae..ebd3cf07a8 100644
--- a/docs/cookbook/adding-an-llm-adapter.i18n.yaml
+++ b/docs/cookbook/adding-an-llm-adapter.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-adding-an-llm-adapter.md: 255dce40e5c10a4c94e34b358ad71f0283c1227f
-adding-an-llm-adapter.zh.md: 67d85f96326474f6acb1b604b2ed31eba18eab4f
+adding-an-llm-adapter.md: 659fdab30e753c9a7acf9e06f85ebfc83480ec50
+adding-an-llm-adapter.zh.md: 5e2a84c82f9a8025c951b1bb2b44ba8b8950f5e3
diff --git a/docs/cookbook/adding-an-llm-adapter.md b/docs/cookbook/adding-an-llm-adapter.md
index 255dce40e5..659fdab30e 100644
--- a/docs/cookbook/adding-an-llm-adapter.md
+++ b/docs/cookbook/adding-an-llm-adapter.md
@@ -32,7 +32,7 @@ Registration is effect-based (HMR-safe); one adapter per provider route — dupl
 - A `GenerateOptions` field your provider cannot honor (e.g. a `stop` list on a provider without stop sequences): throw `LlmError(..., 'UNSUPPORTED')` rather than silently dropping it.
 - If the provider requires response ids, signatures, or other native metadata on follow-up calls, emit the minimal lossless-JSON projection as `finish.replayState`. Validate it when rebuilding history. `LlmService` passes it only when the historical provider route and target provider route are currently owned by the exact same adapter instance; your adapter decides whether same-model, cross-model, or cross-provider restoration is legal. Never infer native replay from provider/model names alone when state is absent.
 
-Provider-specific thinking-mode toggles remain in the adapter's Config. Exact model metadata uses one provider-neutral capability seam: implement `resolveModel()` with provider/model identity and optional `context` and `reasoning` fields, declare a configured `defaultEffort` only when one exists, and honor the resolver's optional `AbortSignal`. Reasoning efforts are ordered opaque ids mapped to provider requests by the adapter. Preserve the adapter's authoritative selectable list—including `off` when its upstream capability API returns it—without exposing final wire spellings or clamping unsupported values.
+Provider-specific thinking-mode toggles remain in the adapter's Config. Exact model metadata uses one provider-neutral capability seam: implement `resolveModel()` with provider/model identity and optional `context` and `reasoning` fields, declare a configured `defaultEffort` only when one exists, and honor the resolver's optional `AbortSignal`. Reasoning efforts are ordered opaque ids mapped to provider requests by the adapter. Preserve the adapter's authoritative selectable list, including an adapter-defined `off` when supported, without exposing final wire spellings or clamping unsupported values; an id need not equal its wire representation.
 
 ## Structure that worked
 
diff --git a/docs/cookbook/adding-an-llm-adapter.zh.md b/docs/cookbook/adding-an-llm-adapter.zh.md
index 67d85f9632..5e2a84c82f 100644
--- a/docs/cookbook/adding-an-llm-adapter.zh.md
+++ b/docs/cookbook/adding-an-llm-adapter.zh.md
@@ -32,7 +32,7 @@ export function apply(ctx: Context, config: Config) {
 - 如果 `GenerateOptions` 中某个字段你的提供方无法支持（例如提供方不支持 stop sequences 时收到 `stop` 列表）：抛出 `LlmError(..., 'UNSUPPORTED')`，而非静默丢弃。
 - 如果提供方在后续调用中需要响应 ID、签名或其他原生元数据，请将其最小无损 JSON 投影作为 `finish.replayState` 发出。重建历史时验证该状态。只有历史提供方路由和目标提供方路由当前由完全相同的适配器实例拥有时，`LlmService` 才会传递该状态；由适配器决定同模型、跨模型或跨提供方恢复是否合法。状态缺失时，切勿仅根据提供方/模型名称推断原生回放。
 
-提供方特有的 thinking 模式开关仍放在适配器的 Config 中。确切模型元数据使用一处提供方无关的能力 seam：实现 `resolveModel()`，返回提供方/模型身份以及可选的 `context` 和 `reasoning` 字段；仅当存在配置指定的默认值时才声明 `defaultEffort`；响应传给解析器的可选 `AbortSignal`。推理强度是由适配器映射到提供方请求的有序不透明 ID。请保留适配器给出的权威可选列表，包括其上游能力 API 返回的 `off`，但不要暴露最终协议值的具体拼写，也不要自动调整不支持的值。
+提供方特有的 thinking 模式开关仍放在适配器的 Config 中。确切模型元数据使用一处提供方无关的能力 seam：实现 `resolveModel()`，返回提供方/模型身份以及可选的 `context` 和 `reasoning` 字段；仅当存在配置指定的默认值时才声明 `defaultEffort`；响应传给解析器的可选 `AbortSignal`。推理强度是由适配器映射到提供方请求的有序不透明 ID。请保留适配器给出的权威可选列表，包括适配器在支持时定义的 `off`；不得暴露最终协议值的具体拼写，也不得自动调整不支持的值。ID 无需与其协议表示相同。
 
 ## 经验证有效的结构
 
diff --git a/packages/llm/llm-deepseek/README.md b/packages/llm/llm-deepseek/README.md
index a92c48463d..9cfc69ee0b 100644
--- a/packages/llm/llm-deepseek/README.md
+++ b/packages/llm/llm-deepseek/README.md
@@ -15,7 +15,7 @@ The package root exposes the Cordis plugin contract and `DeepSeekAdapter`; wire
     apiKey: !!js process.env.DEEPSEEK_API_KEY    # or rely on the env fallback
     baseURL: !!js process.env.DEEPSEEK_BASE_URL  # default: https://api.deepseek.com
     thinking: enabled        # optional; provider default is enabled
-    reasoningEffort: high    # optional; high | max — omitted ⇒ high
+    reasoningEffort: high    # optional; off | high | max — omitted ⇒ high
     streamIdleTimeoutMs: 300000 # optional; positive finite Node timer delay; five-minute default
     defaultContextWindow: 256000 # optional positive-integer fallback for models without an exact value
     models:                  # optional; defaults to V4 Flash and V4 Pro
@@ -30,9 +30,9 @@ The plugin registers the single provider route `deepseek`. A request selects it
 
 `contextWindow` is optional per configured model and is not exposed through the advisory catalog. `ctx.llm.resolveModelInfo('deepseek', model).context` returns an exact model value first, then `defaultContextWindow` for an entry without capacity or an unlisted pass-through id. When neither value exists, `context` is absent without invalidating routing. Pressure-sensitive plugins therefore get deployment-owned capacity without treating the model selector as authoritative. Registering another adapter for `deepseek` throws `LlmError('DUPLICATE_ADAPTER')`.
 
-The same exact-model result exposes ordered `high` and `max` efforts under `reasoning` for every pass-through model while thinking is enabled. `reasoningEffort` selects the deployment default and falls back to `high` when omitted. `agent/request` can replace it on each conversation step; the resolved value is logged in `request/header` and serialized as the official top-level `reasoning_effort` field. An unsupported value fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O.
+The same exact-model result exposes ordered `off`, `high`, and `max` efforts under `reasoning` for every pass-through model when deployment policy permits thinking. `reasoningEffort` selects the deployment default and falls back to `high` when omitted. `agent/request` can replace it on each conversation step; the resolved value is logged in `request/header`. `high` and `max` enable thinking and serialize as the official top-level `reasoning_effort`; adapter-owned `off` instead serializes `thinking.type: disabled` and omits `reasoning_effort`. An unsupported value fails with `UNSUPPORTED_REASONING_EFFORT` before network I/O.
 
-`thinking: disabled` removes the reasoning capability and omits `reasoning_effort`; combining it with a configured default fails plugin loading, and a per-request effort fails as unsupported. A request with `GenerateOptions.purpose: 'session-title'` also forces thinking disabled and omits the already-resolved effort, reserving its bounded output for visible title text without changing conversation or compaction defaults.
+`thinking: disabled` is a deployment lock that publishes only `off` with `off` as its default. Omitting `reasoningEffort` or configuring it as `off` is valid; configuring `high` or `max` fails plugin loading, and a direct per-request attempt to enable thinking fails before network I/O. A request with `GenerateOptions.purpose: 'session-title'` also forces thinking disabled and omits the already-resolved effort, reserving its bounded output for visible title text without changing conversation or compaction defaults.
 
 `streamIdleTimeoutMs` bounds each outstanding provider read, including the initial `fetch`, without counting time the consumer spends between chunks. One stable abort signal reaches the request and body reader for the whole call; expiry stops the transport and throws `LlmError('TIMEOUT')`, while an earlier caller abort throws `LlmError('ABORTED')`. The adapter makes exactly one provider request per `stream()` call; agent-level retry is a separate plugin policy.
 
@@ -43,6 +43,7 @@ Every request carries the shared attribution header from dsh-llm's `attributionH
 ## Wire-format notes (verified live + against the official docs)
 
 - Streaming only (`stream_options.include_usage` always on). `usage` may arrive attached to the finish chunk or as a trailing usage-only chunk — the translator defers both to `[DONE]`, so `usage` always precedes `finish` and nothing follows `finish`.
+- The adapter-owned `off` effort maps to `thinking: {type: 'disabled'}` and never crosses the wire as `reasoning_effort: 'off'`.
 - The first thinking-mode chunk carries `reasoning_content: ""` — handled (no spurious reasoning block).
 - **Reasoning passback rule**: on assistant turns that carried tool calls, `reasoning_content` is serialized back in history (required by the API in thinking mode); on tool-call-free turns it is dropped (ignored anyway — saves tokens).
 - Cache accounting: `cacheReadTokens` ← `prompt_cache_hit_tokens` / `prompt_tokens_details.cached_tokens`; DeepSeek reports no cache-write metric.
@@ -53,7 +54,7 @@ Non-2xx responses throw `LlmError` with stable codes: `AUTH` (401/403), `QUOTA`
 
 ## Testing
 
-Unit suites run against a local `node:http` mock SSE server (no network), including structured HTTP facts, malformed/truncated streams, caller abort, connection failure, and proof that idle timeout aborts the actual body. Real-API coverage lives in `tests/adapter.e2e.ts` (`pnpm run test:e2e`, key-gated): V4 Flash + V4 Pro across thinking enabled/disabled and both official effort levels, including the thinking+tools round trip with reasoning passback.
+Unit suites run against a local `node:http` mock SSE server (no network), including dynamic `high`/`off`/`max` selection, structured HTTP facts, malformed/truncated streams, caller abort, connection failure, and proof that idle timeout aborts the actual body. Real-API coverage lives in `tests/adapter.e2e.ts` (`pnpm run test:e2e`, key-gated): V4 Flash + V4 Pro across thinking enabled/disabled and both official effort levels, including the thinking+tools round trip with reasoning passback.
 
 ## Model Experience
 
diff --git a/packages/llm/llm-deepseek/src/adapter.ts b/packages/llm/llm-deepseek/src/adapter.ts
index 3f6d50dbb8..ca031b8c1e 100644
--- a/packages/llm/llm-deepseek/src/adapter.ts
+++ b/packages/llm/llm-deepseek/src/adapter.ts
@@ -51,12 +51,17 @@ export interface DeepSeekAdapterOptions {
 /** Default maximum idle interval while an adapter stream read is outstanding. */
 export const DEFAULT_STREAM_IDLE_TIMEOUT_MS = 300_000
 const STREAM_IDLE_TIMEOUT_CODE = 'LLM_STREAM_IDLE_TIMEOUT'
+const OFF_REASONING_EFFORT = ReasoningEffortId('off')
 const HIGH_REASONING_EFFORT = ReasoningEffortId('high')
 const MAX_REASONING_EFFORT = ReasoningEffortId('max')
 const REASONING_EFFORTS = [
+  { id: OFF_REASONING_EFFORT, name: 'Off' },
   { id: HIGH_REASONING_EFFORT, name: 'High' },
   { id: MAX_REASONING_EFFORT, name: 'Max' },
 ] as const
+const OFF_ONLY_REASONING_EFFORTS = [
+  { id: OFF_REASONING_EFFORT, name: 'Off' },
+] as const
 
 function modelInfo(provider: string, model: DeepSeekCatalogModel): LlmModelInfo {
   return {
@@ -113,8 +118,10 @@ export class DeepSeekAdapter extends LlmAdapter {
 
   constructor(private readonly options: DeepSeekAdapterOptions) {
     super()
-    if (options.defaults?.thinking === 'disabled' && options.defaults.reasoningEffort !== undefined) {
-      throw new Error('llm-deepseek: reasoningEffort cannot be configured when thinking is disabled')
+    if (options.defaults?.thinking === 'disabled'
+      && options.defaults.reasoningEffort !== undefined
+      && options.defaults.reasoningEffort !== 'off') {
+      throw new Error('llm-deepseek: only reasoningEffort "off" can be configured when thinking is disabled')
     }
     if (options.defaultContextWindow !== undefined
       && (!Number.isInteger(options.defaultContextWindow) || options.defaultContextWindow <= 0)) {
@@ -152,13 +159,20 @@ export class DeepSeekAdapter extends LlmAdapter {
         : modelInfo(provider, configured),
       ...contextWindow === undefined ? {} : { context: { contextWindow } },
       ...this.options.defaults?.thinking === 'disabled'
-        ? {}
+        ? {
+          reasoning: {
+            efforts: OFF_ONLY_REASONING_EFFORTS,
+            defaultEffort: OFF_REASONING_EFFORT,
+          },
+        }
         : {
           reasoning: {
             efforts: REASONING_EFFORTS,
-            defaultEffort: this.options.defaults?.reasoningEffort === 'max'
-              ? MAX_REASONING_EFFORT
-              : HIGH_REASONING_EFFORT,
+            defaultEffort: this.options.defaults?.reasoningEffort === 'off'
+              ? OFF_REASONING_EFFORT
+              : this.options.defaults?.reasoningEffort === 'max'
+                ? MAX_REASONING_EFFORT
+                : HIGH_REASONING_EFFORT,
           },
         },
     })
diff --git a/packages/llm/llm-deepseek/src/index.ts b/packages/llm/llm-deepseek/src/index.ts
index 1dad768440..b5468f4bdb 100644
--- a/packages/llm/llm-deepseek/src/index.ts
+++ b/packages/llm/llm-deepseek/src/index.ts
@@ -37,10 +37,10 @@ export interface Config {
   apiKey?: string
   /** Endpoint base; falls back to $DEEPSEEK_BASE_URL, then the public API. */
   baseURL?: string
-  /** Thinking-mode default for every request (provider default: enabled). */
+  /** Deployment thinking policy; `disabled` limits every conversation request to `off`. */
   thinking?: 'enabled' | 'disabled'
-  /** Default thinking effort when thinking is enabled (default `high`). */
-  reasoningEffort?: 'high' | 'max'
+  /** Default thinking effort (default `high`); `off` disables thinking per request. */
+  reasoningEffort?: 'off' | 'high' | 'max'
   /** Positive context capacity used when the selected model has no exact value. */
   defaultContextWindow?: number
   /** Advisory models shown by discovery consumers; defaults to V4 Flash and V4 Pro. */
@@ -60,7 +60,7 @@ export const Config: z<Config> = z.object({
   apiKey: z.string(),
   baseURL: z.string(),
   thinking: z.union(['enabled', 'disabled']),
-  reasoningEffort: z.union(['high', 'max']),
+  reasoningEffort: z.union(['off', 'high', 'max']),
   defaultContextWindow: z.number().step(1).min(1),
   models: z.array(catalogModel).default(DEFAULT_MODELS),
   streamIdleTimeoutMs: z.number().min(Number.MIN_VALUE).max(MAX_TIMER_DELAY_MS).default(DEFAULT_STREAM_IDLE_TIMEOUT_MS),
@@ -95,8 +95,10 @@ function resolveModels(models: readonly DeepSeekCatalogModel[] | undefined): Dee
 }
 
 export function apply(ctx: Context, config: Config): void {
-  if (config.thinking === 'disabled' && config.reasoningEffort !== undefined) {
-    throw new Error('llm-deepseek: reasoningEffort cannot be configured when thinking is disabled')
+  if (config.thinking === 'disabled'
+    && config.reasoningEffort !== undefined
+    && config.reasoningEffort !== 'off') {
+    throw new Error('llm-deepseek: only reasoningEffort "off" can be configured when thinking is disabled')
   }
   const apiKey = config.apiKey ?? process.env.DEEPSEEK_API_KEY
   if (apiKey === undefined || apiKey.length === 0) {
diff --git a/packages/llm/llm-deepseek/src/serialize.ts b/packages/llm/llm-deepseek/src/serialize.ts
index 70d51c8549..fb6b8d9117 100644
--- a/packages/llm/llm-deepseek/src/serialize.ts
+++ b/packages/llm/llm-deepseek/src/serialize.ts
@@ -13,20 +13,44 @@ import type { WireMessage, WireRequest, WireTool } from './types.ts'
 /** Adapter-level request defaults (from plugin config). */
 export interface RequestDefaults {
   thinking?: 'enabled' | 'disabled' | undefined
-  reasoningEffort?: 'high' | 'max' | undefined
+  reasoningEffort?: 'off' | 'high' | 'max' | undefined
 }
 
-/** Validate the adapter-owned effort before assigning its narrower wire type. */
-function reasoningEffort(options: GenerateOptions): 'high' | 'max' | undefined {
-  const effort = options.reasoningEffort
-  if (effort === undefined) return undefined
-  if (effort === 'high' || effort === 'max') return effort as 'high' | 'max'
+interface ResolvedThinking {
+  thinking?: 'enabled' | 'disabled'
+  reasoningEffort?: 'high' | 'max'
+}
+
+/** Validate the adapter-owned effort before resolving its DeepSeek wire fields. */
+function reasoningEffort(effort: NonNullable<GenerateOptions['reasoningEffort']>): 'off' | 'high' | 'max' {
+  if (effort === 'off' || effort === 'high' || effort === 'max') {
+    return effort as 'off' | 'high' | 'max'
+  }
   throw new LlmError(
     `DeepSeek does not support reasoning effort "${effort}"`,
     'UNSUPPORTED_REASONING_EFFORT',
   )
 }
 
+/** Resolve one legal thinking/effort pair without exposing `off` as a wire effort. */
+function resolveThinking(options: GenerateOptions, defaults: RequestDefaults): ResolvedThinking {
+  if (options.purpose === 'session-title') return { thinking: 'disabled' }
+  const effort = options.reasoningEffort === undefined
+    ? defaults.reasoningEffort
+    : reasoningEffort(options.reasoningEffort)
+  if (defaults.thinking === 'disabled' && effort !== undefined && effort !== 'off') {
+    throw new LlmError(
+      `DeepSeek deployment does not support reasoning effort "${effort}"`,
+      'UNSUPPORTED_REASONING_EFFORT',
+    )
+  }
+  if (effort === 'off') return { thinking: 'disabled' }
+  if (effort === 'high' || effort === 'max') {
+    return { thinking: 'enabled', reasoningEffort: effort }
+  }
+  return defaults.thinking === undefined ? {} : { thinking: defaults.thinking }
+}
+
 /** Join the text blocks of a message (used for user/tool-result content). */
 function flattenText(blocks: ContentBlock[]): string {
   return blocks
@@ -132,18 +156,17 @@ export function serializeRequest(options: GenerateOptions, defaults: RequestDefa
   }))
   // A short title budget must produce visible text; conversation and
   // compaction calls continue to inherit the adapter's thinking defaults.
-  const thinking = options.purpose === 'session-title' ? 'disabled' : defaults.thinking
-  const resolvedReasoningEffort = options.purpose === 'session-title'
-    ? undefined
-    : reasoningEffort(options)
+  const resolvedThinking = resolveThinking(options, defaults)
 
   return {
     model: options.model,
     messages,
     stream: true,
     stream_options: { include_usage: true },
-    ...thinking !== undefined ? { thinking: { type: thinking } } : {},
-    ...resolvedReasoningEffort !== undefined ? { reasoning_effort: resolvedReasoningEffort } : {},
+    ...resolvedThinking.thinking !== undefined ? { thinking: { type: resolvedThinking.thinking } } : {},
+    ...resolvedThinking.reasoningEffort !== undefined
+      ? { reasoning_effort: resolvedThinking.reasoningEffort }
+      : {},
     ...tools !== undefined && tools.length > 0 ? { tools } : {},
     ...options.temperature !== undefined ? { temperature: options.temperature } : {},
     ...options.maxTokens !== undefined ? { max_tokens: options.maxTokens } : {},
diff --git a/packages/llm/llm-deepseek/tests/adapter.e2e.ts b/packages/llm/llm-deepseek/tests/adapter.e2e.ts
index 8897d500d5..742a69d20a 100644
--- a/packages/llm/llm-deepseek/tests/adapter.e2e.ts
+++ b/packages/llm/llm-deepseek/tests/adapter.e2e.ts
@@ -50,31 +50,29 @@ const weatherTool: ToolSchema = {
 }
 
 describe.skipIf(!process.env.DEEPSEEK_API_KEY)('llm-deepseek e2e (real API)', () => {
-  it('flash + thinking disabled: plain text generation', async () => {
-    const ctx = await harness(FLASH, { thinking: 'disabled' })
-    const result = await assemble(ctx,{
+  it('flash dynamically switches from off to high', async () => {
+    const ctx = await harness(FLASH, { reasoningEffort: 'off' })
+    const withoutThinking = await assemble(ctx,{
       model: FLASH,
       messages: ask('Reply with exactly the word: pong'),
       maxTokens: 50,
     })
-    expect(result.finish.kind).toBe('stop')
-    expect(textOf(result).toLowerCase()).toContain('pong')
-    expect(result.message.content.some(block => block.type === 'reasoning')).toBe(false)
-    expect(result.usage?.inputTokens).toBeGreaterThan(0)
-    expect(result.usage?.outputTokens).toBeGreaterThan(0)
-  })
+    expect(withoutThinking.finish.kind).toBe('stop')
+    expect(textOf(withoutThinking).toLowerCase()).toContain('pong')
+    expect(withoutThinking.message.content.some(block => block.type === 'reasoning')).toBe(false)
+    expect(withoutThinking.usage?.inputTokens).toBeGreaterThan(0)
+    expect(withoutThinking.usage?.outputTokens).toBeGreaterThan(0)
 
-  it('flash + thinking enabled (effort high): reasoning blocks + reasoning tokens', async () => {
-    const ctx = await harness(FLASH, { thinking: 'enabled', reasoningEffort: 'high' })
-    const result = await assemble(ctx,{
+    const withThinking = await assemble(ctx,{
       model: FLASH,
+      reasoningEffort: ReasoningEffortId('high'),
       messages: ask('Which is larger, 9.11 or 9.8? Answer with just the number.'),
       maxTokens: 2000,
     })
-    expect(result.finish.kind).toBe('stop')
-    expect(result.message.content.some(block => block.type === 'reasoning')).toBe(true)
-    expect(textOf(result)).toContain('9.8')
-    expect(result.usage?.reasoningTokens).toBeGreaterThan(0)
+    expect(withThinking.finish.kind).toBe('stop')
+    expect(withThinking.message.content.some(block => block.type === 'reasoning')).toBe(true)
+    expect(textOf(withThinking)).toContain('9.8')
+    expect(withThinking.usage?.reasoningTokens).toBeGreaterThan(0)
   })
 
   it.each(['high', 'max'] as const)(
diff --git a/packages/llm/llm-deepseek/tests/adapter.spec.ts b/packages/llm/llm-deepseek/tests/adapter.spec.ts
index 9ac66ce5d1..341d563b1b 100644
--- a/packages/llm/llm-deepseek/tests/adapter.spec.ts
+++ b/packages/llm/llm-deepseek/tests/adapter.spec.ts
@@ -175,12 +175,13 @@ describe('DeepSeekAdapter against a mock server', () => {
     expect(server.headers[0]?.['x-deepseek-harness-compact']).toBe('1')
   })
 
-  it('forwards the configured reasoning default and a dynamic request override', async () => {
+  it('switches dynamically from the configured high default through off to max', async () => {
     const server = await mockServer([
       { kind: 'sse', events: textEvents },
       { kind: 'sse', events: textEvents },
+      { kind: 'sse', events: textEvents },
     ])
-    const ctx = await harness(server.url, { thinking: 'enabled', reasoningEffort: 'max' })
+    const ctx = await harness(server.url, { thinking: 'enabled', reasoningEffort: 'high' })
 
     await assemble(ctx,{
       model: 'deepseek-v4-flash',
@@ -188,20 +189,29 @@ describe('DeepSeekAdapter against a mock server', () => {
     })
     await assemble(ctx,{
       model: 'deepseek-v4-flash',
-      reasoningEffort: ReasoningEffortId('high'),
+      reasoningEffort: ReasoningEffortId('off'),
       messages: [{ role: 'user', content: [{ type: 'text', text: 'hi again' }] }],
     })
+    await assemble(ctx,{
+      model: 'deepseek-v4-flash',
+      reasoningEffort: ReasoningEffortId('max'),
+      messages: [{ role: 'user', content: [{ type: 'text', text: 'one more time' }] }],
+    })
     expect(server.requests[0]).toMatchObject({
       thinking: { type: 'enabled' },
-      reasoning_effort: 'max',
+      reasoning_effort: 'high',
     })
     expect(server.requests[1]).toMatchObject({
+      thinking: { type: 'disabled' },
+    })
+    expect(server.requests[1]).not.toHaveProperty('reasoning_effort')
+    expect(server.requests[2]).toMatchObject({
       thinking: { type: 'enabled' },
-      reasoning_effort: 'high',
+      reasoning_effort: 'max',
     })
   })
 
-  it('omits reasoning capability and effort when thinking is disabled', async () => {
+  it('publishes only off and omits the wire effort when thinking is disabled', async () => {
     const server = await mockServer([{ kind: 'sse', events: textEvents }])
     const ctx = await harness(server.url, { thinking: 'disabled' })
 
@@ -214,7 +224,12 @@ describe('DeepSeekAdapter against a mock server', () => {
     })
     expect(server.requests[0]).not.toHaveProperty('reasoning_effort')
     await expect(ctx.llm.resolveModelInfo('deepseek', 'deepseek-v4-flash'))
-      .resolves.not.toHaveProperty('reasoning')
+      .resolves.toMatchObject({
+        reasoning: {
+          efforts: [{ id: ReasoningEffortId('off'), name: 'Off' }],
+          defaultEffort: ReasoningEffortId('off'),
+        },
+      })
   })
 
   it('rejects a per-request effort before I/O when thinking is disabled', async () => {
@@ -229,6 +244,29 @@ describe('DeepSeekAdapter against a mock server', () => {
     expect(server.requests).toHaveLength(0)
   })
 
+  it.each(['high', 'max'])(
+    'rejects direct adapter effort %s before I/O when thinking is disabled',
+    async (effort) => {
+      const server = await mockServer([])
+      const adapter = new DeepSeekAdapter({
+        apiKey: 'test-key',
+        baseURL: server.url,
+        defaults: { thinking: 'disabled' },
+      })
+
+      const stream = adapter.stream({
+        provider: 'deepseek',
+        model: 'deepseek-v4-flash',
+        reasoningEffort: ReasoningEffortId(effort),
+        messages: [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }],
+      })
+      await expect(async () => {
+        for await (const _chunk of stream) { /* drain */ }
+      }).rejects.toMatchObject({ code: 'UNSUPPORTED_REASONING_EFFORT' })
+      expect(server.requests).toHaveLength(0)
+    },
+  )
+
   it.each([
     [401, 'AUTH'],
     [403, 'AUTH'],
@@ -581,6 +619,7 @@ describe('plugin registration and config', () => {
         context: { contextWindow: 128_000 },
         reasoning: {
           efforts: [
+            { id: ReasoningEffortId('off'), name: 'Off' },
             { id: ReasoningEffortId('high'), name: 'High' },
             { id: ReasoningEffortId('max'), name: 'Max' },
           ],
@@ -589,24 +628,83 @@ describe('plugin registration and config', () => {
       })
   })
 
-  it('rejects a configured reasoning effort when thinking is disabled', async () => {
+  it.each(['off', 'max'] as const)('uses the configured %s reasoning default', async (effort) => {
     const ctx = new Context()
     await ctx.plugin(LlmService)
-    await expect(ctx.plugin(LlmDeepSeek, {
+    await ctx.plugin(LlmDeepSeek, {
+      apiKey: 'k',
+      baseURL: 'http://127.0.0.1:1',
+      reasoningEffort: effort,
+    })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'unlisted-pass-through'))
+      .resolves.toMatchObject({
+        reasoning: {
+          efforts: [
+            { id: ReasoningEffortId('off'), name: 'Off' },
+            { id: ReasoningEffortId('high'), name: 'High' },
+            { id: ReasoningEffortId('max'), name: 'Max' },
+          ],
+          defaultEffort: ReasoningEffortId(effort),
+        },
+      })
+  })
+
+  it('accepts off as the default when thinking is deployment-disabled', async () => {
+    const ctx = new Context()
+    await ctx.plugin(LlmService)
+    await ctx.plugin(LlmDeepSeek, {
       apiKey: 'k',
       baseURL: 'http://127.0.0.1:1',
       thinking: 'disabled',
-      reasoningEffort: 'high',
-    })).rejects.toThrow(/reasoningEffort cannot be configured/)
-    expect(ctx.llm.listProviders()).toEqual([])
+      reasoningEffort: 'off',
+    })
+    await expect(ctx.llm.resolveModelInfo('deepseek', 'unlisted-pass-through'))
+      .resolves.toMatchObject({
+        reasoning: {
+          efforts: [{ id: ReasoningEffortId('off'), name: 'Off' }],
+          defaultEffort: ReasoningEffortId('off'),
+        },
+      })
   })
 
-  it('rejects a disabled-thinking effort at the direct constructor boundary', () => {
-    expect(() => new DeepSeekAdapter({
+  it.each(['high', 'max'] as const)(
+    'rejects configured reasoning effort %s when thinking is disabled',
+    async (reasoningEffort) => {
+      const ctx = new Context()
+      await ctx.plugin(LlmService)
+      await expect(ctx.plugin(LlmDeepSeek, {
+        apiKey: 'k',
+        baseURL: 'http://127.0.0.1:1',
+        thinking: 'disabled',
+        reasoningEffort,
+      })).rejects.toThrow(/only reasoningEffort "off"/)
+      expect(ctx.llm.listProviders()).toEqual([])
+    },
+  )
+
+  it.each(['high', 'max'] as const)(
+    'rejects disabled-thinking effort %s at the direct constructor boundary',
+    (reasoningEffort) => {
+      expect(() => new DeepSeekAdapter({
+        apiKey: 'k',
+        baseURL: 'http://127.0.0.1:1',
+        defaults: { thinking: 'disabled', reasoningEffort },
+      })).toThrow(/only reasoningEffort "off"/)
+    },
+  )
+
+  it('accepts disabled thinking with off at the direct constructor boundary', async () => {
+    const adapter = new DeepSeekAdapter({
       apiKey: 'k',
       baseURL: 'http://127.0.0.1:1',
-      defaults: { thinking: 'disabled', reasoningEffort: 'high' },
-    })).toThrow(/reasoningEffort cannot be configured/)
+      defaults: { thinking: 'disabled', reasoningEffort: 'off' },
+    })
+    await expect(adapter.resolveModel('deepseek', 'pass-through')).resolves.toMatchObject({
+      reasoning: {
+        efforts: [{ id: ReasoningEffortId('off'), name: 'Off' }],
+        defaultEffort: ReasoningEffortId('off'),
+      },
+    })
   })
 
   it('uses the default model catalog when apply is called directly', async () => {
diff --git a/packages/llm/llm-deepseek/tests/serialize.spec.ts b/packages/llm/llm-deepseek/tests/serialize.spec.ts
index 0f294a5e3a..213e910c3f 100644
--- a/packages/llm/llm-deepseek/tests/serialize.spec.ts
+++ b/packages/llm/llm-deepseek/tests/serialize.spec.ts
@@ -183,6 +183,31 @@ describe('serializeRequest', () => {
     expect(wire.reasoning_effort).toBe('max')
   })
 
+  it('maps off to disabled thinking without a wire reasoning effort', () => {
+    const wire = serializeRequest(
+      request({ messages: history, reasoningEffort: ReasoningEffortId('off') }),
+      { thinking: 'enabled', reasoningEffort: 'max' },
+    )
+    expect(wire.thinking).toEqual({ type: 'disabled' })
+    expect(wire.reasoning_effort).toBeUndefined()
+  })
+
+  it('re-enables thinking when max overrides an off default', () => {
+    const wire = serializeRequest(
+      request({ messages: history, reasoningEffort: ReasoningEffortId('max') }),
+      { reasoningEffort: 'off' },
+    )
+    expect(wire.thinking).toEqual({ type: 'enabled' })
+    expect(wire.reasoning_effort).toBe('max')
+  })
+
+  it('rejects enabling thinking when the deployment is locked to disabled', () => {
+    expect(() => serializeRequest(
+      request({ messages: history, reasoningEffort: ReasoningEffortId('high') }),
+      { thinking: 'disabled' },
+    )).toThrow(expect.objectContaining({ code: 'UNSUPPORTED_REASONING_EFFORT' }))
+  })
+
   it('disables thinking for session-title requests without changing adapter defaults', () => {
     const wire = serializeRequest(
       request({
@@ -202,6 +227,12 @@ describe('serializeRequest', () => {
     expect(wire.reasoning_effort).toBeUndefined()
   })
 
+  it('preserves an explicit enabled default without inventing a wire effort', () => {
+    const wire = serializeRequest(request({ messages: history }), { thinking: 'enabled' })
+    expect(wire.thinking).toEqual({ type: 'enabled' })
+    expect(wire.reasoning_effort).toBeUndefined()
+  })
+
   it('rejects an effort outside the DeepSeek capability', () => {
     expect(() => serializeRequest(request({
       messages: history,

From 8b34ccd9943392b4d1bc90d9584bd83b451388ef Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sun, 26 Jul 2026 14:00:37 +0800
Subject: [PATCH 26/36] chore(tui): remove temporary Anthropic test overlay

---
 ...-14-provider-routed-llm-adapters.i18n.yaml |  4 +-
 ...2026-07-14-provider-routed-llm-adapters.md |  1 -
 ...6-07-14-provider-routed-llm-adapters.zh.md |  1 -
 examples/tui-agent/README.md                  | 10 +----
 examples/tui-agent/pi-ai-anthropic.cordis.yml | 39 -------------------
 .../tui-agent/tests/tui-keyless-smoke.e2e.ts  | 21 ----------
 6 files changed, 3 insertions(+), 73 deletions(-)
 delete mode 100644 examples/tui-agent/pi-ai-anthropic.cordis.yml

diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
index 16a2f7752a..3a6b18cecf 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-14-provider-routed-llm-adapters.md: a0f43751a17134f42e3f1a73879c79521fa7921b
-2026-07-14-provider-routed-llm-adapters.zh.md: a8712166e74e323e3091251766fd26fefd0690d0
+2026-07-14-provider-routed-llm-adapters.md: 98205d18d07752e0cdba86d7cba80368d45fd816
+2026-07-14-provider-routed-llm-adapters.zh.md: c35225a86baf4c2d09732b5940abbc8046d365fb
diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
index a0f43751a1..98205d18d0 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.md
@@ -84,7 +84,6 @@ The on-disk session format remains the pre-release pinned version `0`, with no c
 
 - Unit coverage exercises registry conflicts, request reconstruction, session validation, profile resolution, single-attempt option forwarding, native API selection including OpenAI Responses, conversion, replay validation, error mapping, caller cancellation, idle-timeout transport termination, content rewrites, and same-instance versus different-instance replay dispatch.
 - Keyless loop/session tests and ACP snapshots exercise durable provider/model metadata, resume and fork propagation, workflow/subagent overrides, and unchanged user-visible transcripts; the key-gated DeepSeek e2e retains real provider streaming and tool follow-up coverage.
-- The TUI's `pi-ai-anthropic.cordis.yml` overlay disables the native provider owner, mounts an explicit Anthropic profile over deployment-supplied credentials and endpoint, and selects a catalog model. Its keyless PTY smoke opens `/model`, observes the profile's default reasoning effort, and selects the next adapter-advertised effort without provider I/O.
 - Public JSDoc, package READMEs, architecture and core-data-structure docs, generated catalogs, examples, session fixtures, and Python SDK pairs use provider/model targets consistently and are checked by the repository documentation and type-equivalence gates.
 
 ## Risks
diff --git a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
index a8712166e7..c35225a86b 100644
--- a/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
+++ b/.agents/notes/implemented/architecture/2026-07-14-provider-routed-llm-adapters.zh.md
@@ -84,7 +84,6 @@ JSON-RPC 运行时显式接收 provider 与 model。仅当 `deepseek` 提供方
 
 - 单元测试覆盖注册表冲突、请求重建、会话验证、配置解析、单次请求的选项转发、包括 OpenAI Responses 在内的原生 API 选择、转换、回放验证、错误映射、调用方取消、空闲超时导致的传输终止、内容重写，以及同一实例与不同实例间的回放分发。
 - 无密钥的 agent loop/会话测试和 ACP 快照覆盖持久化 provider/model 元数据、恢复与 fork 传播、工作流/subagent 覆盖，以及不变的用户可见 transcript（文本记录）；密钥门控的 DeepSeek e2e 测试保留真实提供方的流式输出与工具后续调用覆盖率。
-- TUI 的 `pi-ai-anthropic.cordis.yml` 覆盖配置会禁用原生提供方所有者，在部署提供的凭据和端点之上挂载显式 Anthropic 配置，并选择一个目录模型。其无密钥 PTY 冒烟测试会打开 `/model`，观察该配置的默认推理级别，并在不执行提供方 I/O 的情况下选择适配器公布的下一个推理级别。
 - 公共 JSDoc、package README、架构与核心数据结构文档、生成目录、示例、会话 fixture（测试前置数据）和 Python SDK 配对文档统一使用 provider/model 目标，并由仓库文档与类型等价门禁校验。
 
 ## 风险
diff --git a/examples/tui-agent/README.md b/examples/tui-agent/README.md
index fef7f99a91..a71eb848ab 100644
--- a/examples/tui-agent/README.md
+++ b/examples/tui-agent/README.md
@@ -13,14 +13,6 @@ pnpm run demo:tui
 
 Both the demo script and the installable `dsh` CLI ([`apps/cli`](../../apps/cli/README.md)) boot this example's `cordis.yml` as the shipped default config; `dsh` additionally applies the personal overlay from `~/.dsh` and uses the invoking directory as the workspace.
 
-### Anthropic-compatible pi-ai route
-
-The checked-in `pi-ai-anthropic.cordis.yml` overlay disables the native DeepSeek adapter, mounts one explicit `anthropic` profile through `dsh-llm-pi-ai`, and selects `claude-sonnet-4-6` with `high` reasoning by default. It deliberately reuses `DEEPSEEK_API_KEY` and `DEEPSEEK_BASE_URL` for a deployment whose private endpoint accepts the Anthropic Messages protocol; a DeepSeek Chat Completions endpoint is not interchangeable. Run it from the repository root:
-
-```sh
-pnpm run demo:tui --config examples/tui-agent/pi-ai-anthropic.cordis.yml
-```
-
 Type a coding task. The agent works through the `read`/`write`/`edit` filesystem tools for ordinary file operations and `bash` (+ the generic `task_output` / `task_list` / `task_kill` for background tasks) for shell commands, searches, and test runs, each in a fresh `bash -c` (the system prompt tells the model to pass `workdir` instead of `cd`). Both the fs tools and bash resolve relative paths against the session workspace. It can also delegate with `subagent`/`subagent_fork`.
 
 The `todo_write` task tracker is opt-in and not in the shipped config: add `@deepseek-ai/dsh-tool-todo` to `cordis.yml` (or a personal-config overlay under `~/.dsh`) to expose it. Once loaded, the model records a whole-list plan to the session log and the TUI renders it.
@@ -59,7 +51,7 @@ This example is a thin leaf `cordis.yml`: it picks the swappable backends, loads
 | Entry | Demonstrates |
 |---|---|
 | `hmr` (`@cordisjs/plugin-hmr`) | the dev/demo edit-reload loop — a **leaf** entry (not baked into the app) because it depends on the Loader's internal module access |
-| `llm-deepseek` | the default native adapter; `pi-ai-anthropic.cordis.yml` disables that owner and mounts an explicit library-backed Anthropic profile |
+| `llm-deepseek` | the default native adapter |
 | `bash` (`dsh-bash-local`) | the executor implementation — the swappable half of the bash seam. The model-facing `bash` schema (`tool-bash`) and generic `task_*` controls (`tool-tasks`) come from `dsh-agent-spine-demo`, so only the executor is a leaf choice |
 | `tui-agent` (`@deepseek-ai/dsh-tui-demo`) | the app bundle: the agent-spine demo + JSONL persistence + the pi-tui channel + a pre-created `main` agent |
 | `subagent`, `subagent-spawn`, `subagent-fork` | the subagent provider registry plus the two in-process backends: a fresh child and a child seeded with the parent's completed-turn prefix |
diff --git a/examples/tui-agent/pi-ai-anthropic.cordis.yml b/examples/tui-agent/pi-ai-anthropic.cordis.yml
deleted file mode 100644
index c1ec3b2968..0000000000
--- a/examples/tui-agent/pi-ai-anthropic.cordis.yml
+++ /dev/null
@@ -1,39 +0,0 @@
-# Replace the native DeepSeek adapter with pi-ai's Anthropic route. The
-# configured endpoint must implement the Anthropic Messages protocol even
-# though this deployment reuses the repository's DeepSeek-named environment
-# variables.
-- id: base
-  name: '@cordisjs/plugin-include'
-  config:
-    path: ./cordis.yml
-    patches:
-      - id: llm-deepseek
-        name: '@deepseek-ai/dsh-llm-deepseek'
-        disabled: true
-      - id: tui-agent
-        name: '@deepseek-ai/dsh-tui-demo'
-        config:
-          provider: anthropic
-          model: claude-sonnet-4-6
-          resumeSessionId: !!js "typeof resumeSessionId === 'string' ? resumeSessionId : undefined"
-          persistenceRoot: './.sessions'
-          resumeCommand: 'dsh --config examples/tui-agent/pi-ai-anthropic.cordis.yml --resume {session}'
-          workspaceContext:
-            maxBytes: 65536
-          ui:
-            showReasoning: true
-            maxToolOutputLines: 6
-          persona: |
-            You are a coding agent powered by the {{model}} model.
-
-            Verify your work by running the code or tests. Keep answers brief and
-            factual.
-      - insert:
-          - id: llm-pi-ai
-            name: '@deepseek-ai/dsh-llm-pi-ai'
-            config:
-              providers:
-                - provider: anthropic
-                  apiKey: !!js process.env.DEEPSEEK_API_KEY
-                  baseURL: !!js process.env.DEEPSEEK_BASE_URL
-                  reasoning: high
diff --git a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
index 79f0155cbd..169b582492 100644
--- a/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
+++ b/examples/tui-agent/tests/tui-keyless-smoke.e2e.ts
@@ -11,7 +11,6 @@ import { runTuiPtySmoke, type TuiPtySmokeOptions } from './pty-harness.ts'
 const dshBinScript = fileURLToPath(new URL('../../../apps/cli/src/bin.ts', import.meta.url))
 const configPath = fileURLToPath(new URL('../cordis.yml', import.meta.url))
 const codeModeConfigPath = fileURLToPath(new URL('../code-mode.cordis.yml', import.meta.url))
-const piAiAnthropicConfigPath = fileURLToPath(new URL('../pi-ai-anthropic.cordis.yml', import.meta.url))
 const scriptedConfigPath = fileURLToPath(new URL('./fixtures/tui-scripted.cordis.yml', import.meta.url))
 const tsconfigPath = fileURLToPath(new URL('../../../tsconfig.json', import.meta.url))
 
@@ -247,26 +246,6 @@ describe('tui-agent keyless smoke (real Loader tree in a PTY)', () => {
     expect(output).toContain('\u001B[?2004l')
   }, LOADER_SMOKE_TEST_TIMEOUT_MS)
 
-  it('boots the pi-ai Anthropic overlay and selects an advertised reasoning effort', async () => {
-    const output = await smoke({
-      label: 'tui-agent pi-ai anthropic',
-      tempDirPrefix: 'tui-agent-pi-ai-anthropic-',
-      configPath: piAiAnthropicConfigPath,
-      env: {
-        DEEPSEEK_API_KEY: 'keyless-tui-no-call',
-        DEEPSEEK_BASE_URL: 'http://127.0.0.1:1',
-      },
-      actions: [
-        { waitFor: 'main-session-', send: '/model\r' },
-        { waitFor: 'Claude Sonnet 4.6 — High — current', send: '\x1b[Z\r' },
-        { waitFor: 'Reasoning effort: Max.', send: '/exit\r' },
-      ],
-    })
-    expect(output).toContain('anthropic/claude-sonnet-4-6')
-    expect(output).toContain('Reasoning effort: Max.')
-    expect(output).toContain('\u001B[?2004l')
-  }, LOADER_SMOKE_TEST_TIMEOUT_MS)
-
 })
 
 describe('dsh CLI keyless smoke (apps/cli through the same PTY)', () => {

From ac792bfd78a62621c7624270e412edb05203d5f7 Mon Sep 17 00:00:00 2001
From: Yichen Jiang <jyc132780208@sina.com>
Date: Sun, 26 Jul 2026 14:23:06 +0800
Subject: [PATCH 27/36] fix(llm): honor model metadata cancellation and
 defaults

---
 ...cated-full-screen-tui-front-door.i18n.yaml |  4 +-
 ...17-dedicated-full-screen-tui-front-door.md |  2 +-
 ...dedicated-full-screen-tui-front-door.zh.md |  2 +-
 packages/compact/compact-basic/src/index.ts   |  2 +-
 .../compact-basic/tests/compact-basic.spec.ts | 12 ++++++
 packages/ui/tui/README.md                     |  2 +-
 packages/ui/tui/src/index.ts                  | 38 ++++++++++++-------
 .../model-effort-switching.expected.txt       |  4 +-
 .../snapshots/model-selector.expected.txt     |  4 +-
 .../snapshots/model-switching.expected.txt    | 24 ++++++------
 packages/ui/tui/tests/tui.snapshot.ts         | 10 +++--
 packages/ui/tui/tests/tui.spec.ts             | 30 +++++++++++++++
 12 files changed, 96 insertions(+), 38 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
index 0767603128..6d44658ecc 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-17-dedicated-full-screen-tui-front-door.md: 3b018de9d2360e68ad00cdb7e4d5178c9e3bc6ea
-2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 46165ce41bb0704b64d4447e683b70839e188562
+2026-07-17-dedicated-full-screen-tui-front-door.md: 8d7c7b00c8d9b15ea3f2419ed44ca88209e60dad
+2026-07-17-dedicated-full-screen-tui-front-door.zh.md: 49e9541c9f98c9f3beba11945ff452fc38bd9ede
diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
index 3b018de9d2..8d7c7b00c8 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.md
@@ -24,7 +24,7 @@ The TUI rebuilds the transcript from the active `session.surface` and reprojects
 
 Editor input calls `agent.send()` while idle and `agent.steer()` while a turn is running. Cancellation, reasoning visibility, tool-card expansion, redraw, transcript clearing, and exit are terminal-only controls. The idle footer derives context occupancy from `tokenMeter` and shows the selected model and explicit reasoning effort; during a run, elapsed activity and the Escape interrupt hint replace that summary. `/status` remains available in either state and appends a detailed terminal-only snapshot: session identity and timestamps, selected model, reasoning effort/default state and reasoning visibility, lifecycle counts folded from the event log, the same deduplicated usage buckets and KV-cache rate as the footer, and context use from `tokenMeter` plus the selected model's advertised capacity. The plugin registers the shared `userInteraction` provider and presents queued questions in a wide bottom-left keyboard panel with batch progress, numbered options, and aligned descriptions; agent behavior and answer logging remain owned by their existing services.
 
-The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. Each model row owns the adapter-advertised reasoning-effort order and default: Shift+Tab cycles that row's efforts, while a model without selectable metadata keeps default behavior. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model/reasoning-effort target per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
+The `/model` command presents the advisory `ctx.llm` catalog as a keyboard selector and changes only this TUI session's target; argument forms remain available for direct selection. Each model row owns the adapter-advertised reasoning-effort order and default: Shift+Tab cycles that row's efforts, includes provider-default behavior when the adapter advertises no default, and leaves models without selectable metadata unchanged. Agent-scoped prompt-assembly and request waterfalls snapshot one provider/model/reasoning-effort target per step, so `{{provider}}` / `{{model}}` interpolation and request routing cannot split when a command arrives during assembly. The latest logged request header restores a used target; a selection that never reaches a request remains process-local.
 
 ### Terminal ownership
 
diff --git a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
index 46165ce41b..49e9541c9f 100644
--- a/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-17-dedicated-full-screen-tui-front-door.zh.md
@@ -24,7 +24,7 @@ TUI 从活跃的 `session.surface` 重建 transcript（文本记录），并在
 
 agent 空闲时，编辑器输入调用 `agent.send()`；轮次运行中则调用 `agent.steer()`。取消、推理显隐、工具卡片展开、重绘、清空 transcript 和退出都只是终端控制。空闲态页脚根据 `tokenMeter` 得出上下文占用率，并显示所选模型和显式选定的推理强度；agent 运行期间，该摘要会替换为带已用时长的活动指示和 Escape 中断提示。`/status` 在这两种状态下均可用，并会追加一份仅在终端显示的详细快照，其中包括会话标识与时间戳、所选模型、推理强度（或默认状态）及推理显隐状态、从事件日志归并得出的生命周期计数、与页脚一致的去重用量分项和 KV 缓存命中率，以及 `tokenMeter` 给出的上下文用量和所选模型公布的容量。插件注册共享的 `userInteraction` 提供方，在左下角宽幅键盘操作面板中呈现排队的问题，面板显示批次进度、带编号的选项和对齐的描述；agent 行为和答案日志仍由既有服务负责。
 
-`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。每个模型行都持有适配器公布的推理强度顺序和默认值：按 Shift+Tab 可循环切换该行的推理强度；没有可选元数据的模型保持默认行为。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个步骤快照一次同一个提供方/模型/推理强度目标，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
+`/model` 命令将建议性的 `ctx.llm` 目录呈现为键盘选择器，并且只更改当前 TUI 会话的目标；带参数的形式仍可直接选择目标。每个模型行都持有适配器公布的推理强度顺序和默认值：按 Shift+Tab 可循环切换该行的推理强度；如果适配器没有公布默认值，循环中还会包含提供方默认行为；没有可选元数据的模型则保持不变。agent 作用域内的 prompt 组装和请求两条 waterfall（瀑布式事件）会为每个步骤快照一次同一个提供方/模型/推理强度目标，因此即使命令在组装期间到达，`{{provider}}` / `{{model}}` 插值与请求路由也不会分裂。系统通过日志中最新的请求头恢复已经使用过的目标；未被请求使用的选择只保留在当前进程中。
 
 ### 终端所有权
 
diff --git a/packages/compact/compact-basic/src/index.ts b/packages/compact/compact-basic/src/index.ts
index f7ab37ccc4..7920197739 100644
--- a/packages/compact/compact-basic/src/index.ts
+++ b/packages/compact/compact-basic/src/index.ts
@@ -272,7 +272,7 @@ export class BasicCompactService extends CompactService {
       return this.compactRegion(range.start, range.end, agent, signal)
     }
 
-    const context = (await this.ctx.llm.resolveModelInfo(target.provider, target.model)).context
+    const context = (await this.ctx.llm.resolveModelInfo(target.provider, target.model, signal)).context
     const targetKey = `${target.provider}/${target.model}`
     if (context === undefined) {
       throw new TargetPressureConfigError(
diff --git a/packages/compact/compact-basic/tests/compact-basic.spec.ts b/packages/compact/compact-basic/tests/compact-basic.spec.ts
index 9433e70f32..b89a9d1950 100644
--- a/packages/compact/compact-basic/tests/compact-basic.spec.ts
+++ b/packages/compact/compact-basic/tests/compact-basic.spec.ts
@@ -463,6 +463,18 @@ describe('pressure measurement and retention', () => {
       .resolves.not.toBeNull()
   })
 
+  it('forwards turn cancellation to proactive model metadata resolution', async () => {
+    const ctx = createContext()
+    const resolveModelInfo = vi.spyOn(ctx.llm, 'resolveModelInfo')
+    const compact = service(compactConfig, ctx)
+    const session = conversation()
+    const signal = new AbortController().signal
+
+    await expect(compact.compactIfNeeded(agent(session, MODEL), 'pressure', signal))
+      .resolves.not.toBeNull()
+    expect(resolveModelInfo).toHaveBeenCalledWith(MODEL, MODEL, signal)
+  })
+
   it('re-resolves capacity after a same-model-id provider switch in one session', async () => {
     const ctx = new Context()
     void new LlmService(ctx)
diff --git a/packages/ui/tui/README.md b/packages/ui/tui/README.md
index b008a3f821..17c0a37cb5 100644
--- a/packages/ui/tui/README.md
+++ b/packages/ui/tui/README.md
@@ -22,7 +22,7 @@ When optional `ctx.sessionReferences` is mounted, the same `@` menu also offers
 
 While the agent is running, ordinary editor submissions call `agent.steer()`; otherwise they call `agent.followup()`. A slash at the start of the submitted line enters `ctx.commands` instead: known commands execute directly, unknown commands produce a warning, and neither path automatically reaches the model. A command producer may explicitly schedule agent work; [`dsh-plan-mode`](../../plan/plan-mode/README.md#model-and-human-surfaces) uses that contract for `/plan [message]`. The TUI registers `/help`, `/model`, `/clear`, `/reasoning`, `/tools`, `/redraw`, `/reload`, `/resume`, `/status`, and `/exit` as agent-scoped definitions; every other effective command joins autocomplete and `/help` dynamically, as do `/skill:` completions. A status line above the editor reports the turn phase the TUI derives from session events — waiting for the first token, thinking, responding, or executing tools — with the elapsed time in that phase and the running step total, refreshed each second, and ends with the `Enter sends steering, Esc cancels` hint; while steering messages wait to reach the model it inserts a `N queued ·` badge before the hint that clears as each drains. Ctrl+C or Escape cancels a running turn. Tool cards collapse long bodies into a configurable head/tail preview; Ctrl+O toggles every card between its preview and full output. Ctrl+R toggles reasoning, Ctrl+L redraws, and Ctrl+D exits while idle.
 
-`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape closes it. Models without selectable effort metadata ignore Shift+Tab; the selector renders the exact advertised list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
+`/model` opens the advisory `ctx.llm` catalog as a keyboard selector: Up/Down moves, Shift+Tab cycles the focused model's adapter-advertised reasoning efforts in display order, Enter selects the model and effort, and Escape closes it. When an adapter does not advertise a default effort, the cycle also includes `provider default`, which clears an explicit selection; models without selectable effort metadata ignore Shift+Tab. The selector renders the exact advertised effort list—including `off` when present—and does not synthesize, clamp, or transfer an effort between models. `/model <model>` still selects an unambiguous model id directly, while `/model <provider>/<model>` selects an exact target and uses its adapter default when one exists. The configured target or latest logged request header initializes the selector, and an unlisted current model remains visible because catalogs are advisory. Selection is local to this TUI session. Prompt assembly snapshots the target for one step, replaces `{{provider}}` and `{{model}}`, and applies the same provider/model/reasoning-effort target through `agent/request`; a switch during assembly therefore starts with a later step. The request header durably records targets that reach the model, while an unused selection remains process-local.
 
 `/reload` (EXPERIMENTAL, dev-only) re-reads every file-backed loader config tree and applies the diff to the running app — the HMR watcher's config path, invoked manually; it needs the cordis Loader in the context and degrades to a warning without one, runs only while the agent is idle, and refuses re-entry while a reload is in flight. Module-source hot reload remains watcher-owned. When a `skills` service is mounted, `/skill:<name> [instructions]` loads that skill's instructions into the conversation as a user turn; autocomplete lists the model-invocable skills, and any skill (including a model-disabled one) is loadable by its exact name.
 
diff --git a/packages/ui/tui/src/index.ts b/packages/ui/tui/src/index.ts
index e6f19cde47..a88abb5586 100644
--- a/packages/ui/tui/src/index.ts
+++ b/packages/ui/tui/src/index.ts
@@ -586,6 +586,11 @@ interface ModelChoice extends AgentLlmTarget {
   reasoning?: LlmModelReasoningInfo
 }
 
+interface ModelDialogSelection {
+  choice: ModelChoice
+  reasoningEffort: ReasoningEffortId | undefined
+}
+
 function targetLabel(target: AgentLlmTarget): string {
   return `${target.provider}/${target.model}`
 }
@@ -1262,7 +1267,7 @@ class ModelDialog implements Component {
     current: AgentLlmTarget | undefined,
     maxVisible: number,
     private readonly palette: Palette,
-    done: (choice: ModelChoice) => void,
+    done: (selection: ModelDialogSelection) => void,
     cancel: () => void,
   ) {
     this.items = new Map()
@@ -1294,10 +1299,9 @@ class ModelDialog implements Component {
       const selected = choices.find(choice => targetLabel(choice) === item.value)
       /* v8 ignore next -- SelectList only returns values built from `choices`. */
       if (selected === undefined) return
-      const effort = this.efforts.get(item.value)
       done({
-        ...selected,
-        ...effort === undefined ? {} : { reasoningEffort: effort },
+        choice: selected,
+        reasoningEffort: this.efforts.get(item.value),
       })
     }
     this.list.onCancel = cancel
@@ -1324,11 +1328,13 @@ class ModelDialog implements Component {
     const choice = this.choices.get(selectedItem.value)
     if (choice?.reasoning === undefined) return
     const current = this.efforts.get(selectedItem.value)
-    const currentIndex = choice.reasoning.efforts.findIndex(effort => effort.id === current)
-    const next = choice.reasoning.efforts[(currentIndex + 1) % choice.reasoning.efforts.length]
-    /* v8 ignore next -- validated reasoning metadata always carries at least one effort. */
-    if (next === undefined) return
-    this.efforts.set(selectedItem.value, next.id)
+    const efforts: Array<ReasoningEffortId | undefined> = [
+      ...choice.reasoning.defaultEffort === undefined ? [undefined] : [],
+      ...choice.reasoning.efforts.map(effort => effort.id),
+    ]
+    const currentIndex = efforts.indexOf(current)
+    const next = efforts[(currentIndex + 1) % efforts.length]
+    this.efforts.set(selectedItem.value, next)
     const item = this.items.get(selectedItem.value)
     /* v8 ignore next -- items and choices are constructed from the same values. */
     if (item === undefined) return
@@ -2118,10 +2124,14 @@ export function createTuiChat(
   }
   resolveContextWindow(target.current)
 
-  const selectModel = (selected: ModelChoice): void => {
+  const selectModel = (
+    selected: ModelChoice,
+    explicitReasoning?: { effort: ReasoningEffortId | undefined },
+  ): void => {
     const sameRoute = target.current?.provider === selected.provider && target.current.model === selected.model
-    const reasoningEffort = selected.reasoningEffort
-      ?? (sameRoute ? target.current?.reasoningEffort ?? selected.reasoning?.defaultEffort : selected.reasoning?.defaultEffort)
+    const reasoningEffort = explicitReasoning === undefined
+      ? (sameRoute ? target.current?.reasoningEffort ?? selected.reasoning?.defaultEffort : selected.reasoning?.defaultEffort)
+      : explicitReasoning.effort
     if (sameRoute && target.current?.reasoningEffort === reasoningEffort) {
       const reasoning = targetReasoningLabel(selected, reasoningEffort)
       appendNotice(`Model is already ${targetLabel(selected)}${reasoning === undefined ? '' : ` with reasoning effort ${displayText(reasoning)}`}.`)
@@ -2154,9 +2164,9 @@ export function createTuiChat(
         target.current,
         resolved.maxModelOptions,
         palette,
-        (selected) => {
+        (selection) => {
           void session.close()
-          selectModel(selected)
+          selectModel(selection.choice, { effort: selection.reasoningEffort })
         },
         () => { void session.close() },
       ),
diff --git a/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
index 74ca319f8c..cef660ff49 100644
--- a/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-effort-switching.expected.txt
@@ -26,9 +26,9 @@ buffer
   style 8-8 fg=bright-blue
   style 38-77 fg=bright-black
   style 83-83 fg=bright-blue
-15| "        │ → deepseek/deepseek-v4-pro        DeepSeek V4 Pro — Off                  │        "
+15| "        │ → deepseek/deepseek-v4-pro        DeepSeek V4 Pro — provider default     │        "
   style 8-8 fg=bright-blue
-  style 10-64 fg=bright-blue inverse
+  style 10-77 fg=bright-blue inverse
   style 83-83 fg=bright-blue
 16| "        │                                                                          │        "
   style 8-8 fg=bright-blue
diff --git a/packages/ui/tui/tests/snapshots/model-selector.expected.txt b/packages/ui/tui/tests/snapshots/model-selector.expected.txt
index 1caba208d6..c86b1f1f3d 100644
--- a/packages/ui/tui/tests/snapshots/model-selector.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-selector.expected.txt
@@ -26,9 +26,9 @@ buffer
   style 8-8 fg=bright-blue
   style 10-77 fg=bright-blue inverse
   style 83-83 fg=bright-blue
-15| "        │   deepseek/deepseek-v4-pro        DeepSeek V4 Pro — High                 │        "
+15| "        │   deepseek/deepseek-v4-pro        DeepSeek V4 Pro — provider default     │        "
   style 8-8 fg=bright-blue
-  style 36-65 fg=bright-black
+  style 36-77 fg=bright-black
   style 83-83 fg=bright-blue
 16| "        │                                                                          │        "
   style 8-8 fg=bright-blue
diff --git a/packages/ui/tui/tests/snapshots/model-switching.expected.txt b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
index 67faad5c40..4e711e37c9 100644
--- a/packages/ui/tui/tests/snapshots/model-switching.expected.txt
+++ b/packages/ui/tui/tests/snapshots/model-switching.expected.txt
@@ -1,25 +1,27 @@
 terminal 92x32 buffer=normal length=32 base=0 viewport=0
 lifecycle started=1 stopped=0 progress=inactive
 title "DSH snapshot"
-cursor hidden column=1 viewportRow=6 bufferRow=6
+cursor hidden column=1 viewportRow=7 bufferRow=7
 buffer
 0| " DEEPSEEK HARNESS"
   style 1-8 fg=bright-blue bold
   style 10-16 bold
 1| " Snapshot agent ready."
   style 1-21 fg=bright-black
-2| " deepseek-v4-pro off  •  main-session"
-  style 1-36 dim
+2| " deepseek-v4-pro  •  main-session"
+  style 1-32 dim
 3| <blank>
-4| " Model selected: deepseek/deepseek-v4-pro. Reasoning effort: Off. New steps will use it.    "
-  style 1-87 fg=bright-black
-5| "────────────────────────────────────────────────────────────────────────────────────────────"
+4| " Model selected: deepseek/deepseek-v4-pro. Reasoning effort: provider default. New steps    "
+  style 1-91 fg=bright-black
+5| " will use it.                                                                               "
+  style 1-12 fg=bright-black
+6| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
-6| "                                                                                            "
+7| "                                                                                            "
   style 1-1 inverse
-7| "────────────────────────────────────────────────────────────────────────────────────────────"
+8| "────────────────────────────────────────────────────────────────────────────────────────────"
   style 0-91 dim
-8| "deepseek-v4-pro off  /workspace/project  ↑0 ↓0                   0% context  tools:collapsed"
-  style 0-45 dim
+9| "deepseek-v4-pro  /workspace/project  ↑0 ↓0                       0% context  tools:collapsed"
+  style 0-41 dim
   style 65-91 dim
-9-31| <blank>
+10-31| <blank>
diff --git a/packages/ui/tui/tests/tui.snapshot.ts b/packages/ui/tui/tests/tui.snapshot.ts
index 333b85c4f2..da3ad6e1aa 100644
--- a/packages/ui/tui/tests/tui.snapshot.ts
+++ b/packages/ui/tui/tests/tui.snapshot.ts
@@ -632,7 +632,7 @@ describe('TUI terminal-state snapshots', () => {
     await harness.terminal.dispose()
   })
 
-  it('pins the model selector and selection notice', async () => {
+  it('pins the model selector, effort cycling, and provider-default selection', async () => {
     const harness = await setupSnapshot({
       catalog: {
         providers: [{ id: 'deepseek', name: 'DeepSeek' }],
@@ -640,7 +640,7 @@ describe('TUI terminal-state snapshots', () => {
           { provider: 'deepseek', id: 'deepseek-v4-flash', name: 'DeepSeek V4 Flash' },
           { provider: 'deepseek', id: 'deepseek-v4-pro', name: 'DeepSeek V4 Pro' },
         ],
-        resolveModelInfo: () => Promise.resolve({
+        resolveModelInfo: (_provider, model) => Promise.resolve({
           context: { contextWindow: 128_000 },
           reasoning: {
             efforts: [
@@ -648,7 +648,9 @@ describe('TUI terminal-state snapshots', () => {
               { id: ReasoningEffortId('high'), name: 'High' },
               { id: ReasoningEffortId('max'), name: 'Max' },
             ],
-            defaultEffort: ReasoningEffortId('high'),
+            ...model === 'deepseek-v4-flash'
+              ? { defaultEffort: ReasoningEffortId('high') }
+              : {},
           },
         }),
       },
@@ -662,6 +664,8 @@ describe('TUI terminal-state snapshots', () => {
       harness.terminal.send('\x1b[B')
       harness.terminal.send('\x1b[Z')
       harness.terminal.send('\x1b[Z')
+      harness.terminal.send('\x1b[Z')
+      harness.terminal.send('\x1b[Z')
     })
     await checkpoint('model-effort-switching', harness.terminal, { includeScrollback: true })
     await renderAfter(harness, () => {
diff --git a/packages/ui/tui/tests/tui.spec.ts b/packages/ui/tui/tests/tui.spec.ts
index 34f5241004..d1ae8514ca 100644
--- a/packages/ui/tui/tests/tui.spec.ts
+++ b/packages/ui/tui/tests/tui.spec.ts
@@ -2387,6 +2387,36 @@ describe('pi-tui chat lifecycle and transcript', () => {
     await tick()
     expect(result.terminal.output.slice(providerDefaultOutput)).toContain('Reasoning effort: Standard.')
 
+    const resetDefaultOutput = result.terminal.output.length
+    result.terminal.send('/model')
+    result.terminal.send('\r')
+    await vi.waitFor(() => {
+      expect(result.terminal.output.slice(resetDefaultOutput)).toContain('Alpha Shared — Standard — current')
+    })
+    result.terminal.send('\x1b[Z')
+    await tick()
+    expect(result.terminal.output.slice(resetDefaultOutput)).toContain('Alpha Shared — Ultra — current')
+    result.terminal.send('\x1b[Z')
+    await tick()
+    expect(result.terminal.output.slice(resetDefaultOutput)).toContain('Alpha Shared — provider default')
+    result.terminal.send('\r')
+    await tick()
+    expect(result.terminal.output.slice(resetDefaultOutput)).toContain('Reasoning effort: provider default.')
+    const explicitResetSeed: LlmCallConfig = {
+      provider: 'beta',
+      model: 'b1',
+      reasoningEffort: ReasoningEffortId('max'),
+    }
+    await result.ctx.systemPrompt.assemble(assembleContextFor(result.agent))
+    await expect(agentEvents(result.ctx, result.agent).waterfall(
+      'agent/request',
+      0,
+      0,
+      explicitResetSeed,
+      new AbortController().signal,
+      () => Promise.resolve(explicitResetSeed),
+    )).resolves.toEqual({ provider: 'alpha', model: 'shared' })
+
     const nonReasoningOutput = result.terminal.output.length
     result.terminal.send('/model')
     result.terminal.send('\r')

From 68d6ed2c9cc1d2f9a4cb14c3642803adc4566d7a Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Sun, 26 Jul 2026 17:11:18 +0800
Subject: [PATCH 28/36] docs(i18n): bring the tool-todo and runtime README
 Chinese sides along after the master merge

---
 packages/todo/tool-todo/README.i18n.yaml | 2 +-
 packages/todo/tool-todo/README.zh.md     | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/packages/todo/tool-todo/README.i18n.yaml b/packages/todo/tool-todo/README.i18n.yaml
index cd40049580..dff7032381 100644
--- a/packages/todo/tool-todo/README.i18n.yaml
+++ b/packages/todo/tool-todo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
 README.md: 3615f68953cfe6cdc0e6fc41bf75b8dfdf90a310
-README.zh.md: 6a9b817cb5af2c43b8e66100333e46665359eba8
+README.zh.md: ee0b99d1b4a444dc0b92217686140f691abf1e2b
diff --git a/packages/todo/tool-todo/README.zh.md b/packages/todo/tool-todo/README.zh.md
index 6a9b817cb5..ee0b99d1b4 100644
--- a/packages/todo/tool-todo/README.zh.md
+++ b/packages/todo/tool-todo/README.zh.md
@@ -20,7 +20,7 @@
 
 ## 渲染
 
-规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久列表；[TUI 应用](../../examples/tui-demo)将其显示为持久计划。
+规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久列表：[TUI 应用](../../examples/tui-demo)将其显示为持久计划，[web 客户端](../../client/ui-conversation)则基于 `ConversationSnapshot.todos` 渲染计划条和专属工具行（[Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)）。
 
 ## 导出形状
 

From dadb92302f957b7949ae65835f27467a544e8b27 Mon Sep 17 00:00:00 2001
From: imccyu <276526105+imccyu@users.noreply.github.com>
Date: Mon, 27 Jul 2026 10:09:42 +0800
Subject: [PATCH 29/36] adapt todo display to the slash/input/session
 architecture

- TodoPanel mounts through a 'conversation.input.dock' list entry
  (todoDockEntry, QueueDock posture, order -1 above the queue rows) instead
  of a ConversationRoot hardcode; the inner component is unchanged and takes
  useSession from the dock entry's standard kit.
- The verify-todo-display.mjs chromium probe is replaced by an assembled
  keyless snapshot (apps/web/tests/todo-display.snapshot.ts, the
  code-mode-fixture idiom) pinning the TodoRow summary/state, the dock
  panel content, and the collapse round-trip over built bundles.
- Fake snapshots across specs gain the todos field; bilingual note/READMEs
  updated for the dock mount and the snapshot.
---
 .../2026-07-23-web-todo-display.i18n.yaml     |   4 +-
 .../feature/2026-07-23-web-todo-display.md    |   7 +-
 .../feature/2026-07-23-web-todo-display.zh.md |   7 +-
 apps/web/tests/todo-display.snapshot.ts       | 186 ++++++++++++++++++
 .../ui-conversation/src/client/apply.ts       |   4 +
 .../src/client/skeleton/TodoPanel.tsx         |  39 +++-
 .../tests/chat-code-subcalls.spec.tsx         |   2 +-
 .../ui-conversation/tests/input-bar.spec.tsx  |   2 +-
 .../tests/input-matrix.spec.tsx               |   2 +-
 .../tests/input-scenarios.spec.tsx            |   2 +-
 .../ui-conversation/tests/queue-dock.spec.tsx |   2 +-
 packages/todo/tool-todo/README.i18n.yaml      |   2 +-
 packages/todo/tool-todo/README.zh.md          |   4 +-
 scripts/verify-todo-display.mjs               | 107 ----------
 14 files changed, 242 insertions(+), 128 deletions(-)
 create mode 100644 apps/web/tests/todo-display.snapshot.ts
 delete mode 100644 scripts/verify-todo-display.mjs

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index 86de69efbf..292c9b6200 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-2026-07-23-web-todo-display.md: 15368ee44c3a3aee1f1854964b9f3575d1b2fa92
-2026-07-23-web-todo-display.zh.md: c9ac6cda23721296a9cae5431adac4182feab72c
+2026-07-23-web-todo-display.md: b08c50b2f1a0aea03382b9e012ac364d697a21ee
+2026-07-23-web-todo-display.zh.md: daff6d4a18a621f3fd336ff6feb4433f5cbf9b16
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index 15368ee44c..b08c50b2f1 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -18,7 +18,7 @@ Consume `todo/write` as a Session side effect, not a surface node, and render it
 
 ### TodoPanel: the durable list as a persistent strip
 
-The skeleton pins the panel between the view area and the composer (the composer-card axis), hidden while empty, collapsible with the in-progress item as the collapsed one-line hint; ✓/●/○ glyphs mirror the TUI plan panel. It reads `snapshot.todos` via the framework `useSession` hook — no store, no service, no ctx. It lives inside `ConversationRoot` rather than the details column or its own slot: the details slot is single-occupant and selection-driven (a different lifetime than an always-on strip), and the slot table reserves no plan seat. The component is props-complete and framework-free, so a later relocation to a dedicated slot touches nothing inside it.
+The panel mounts through the `conversation.input.dock` slot (a plain registrant plugin, `todoDockEntry`, the QueueDock posture: `inject: ['slots', 'conversation']` as the load-order seam, `order: -1` above the queue rows), hidden while empty, collapsible with the in-progress item as the collapsed one-line hint; ✓/●/○ glyphs mirror the TUI plan panel. It reads `snapshot.todos` via the standard-kit `useSession` hook the dock entry receives — no store, no service, no ctx. The inner component stays props-complete and framework-free; the dock adapter is a one-line wrapper.
 
 ### TodoRow: the per-call row through the keyed toolview slot
 
@@ -27,9 +27,10 @@ The dedicated `todo_write` chat row is a plain registrant plugin (`todoToolview`
 ## Alternatives considered
 
 - **Fold todo writes into `nodes` as surface entries** — replayed windows would render every superseded list; the event is deliberately not a surface type.
-- **Details column or a dedicated slot for the panel** — the details slot is single-occupant and selection-driven; a new slot key needs a slot-table seat that design has not assigned. The panel is framework-free, so the relocation stays cheap if one lands.
+- **Hardcoding the panel inside `ConversationRoot`** — the original landing spot before the input-dock slot existed; the dock is the architecture's home for always-on strips above the composer, and a hardcode bypasses the slot registry's disposal and ordering.
+- **Details column for the panel** — the details slot is single-occupant and selection-driven, a different lifetime than an always-on strip.
 - **Host-computed view (a todo `ToolEventView`)** — presentation belongs to the client; the wire already carries the whole snapshot in the event payload.
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 63) plus `scripts/verify-todo-display.mjs` pin the full chain (panel visibility, row summary, details linkage, collapse, dark theme) in a real chromium. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index c9ac6cda23..daff6d4a18 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -18,7 +18,7 @@ Status: implemented
 
 ### TodoPanel：长驻列表作为一条常驻横条
 
-骨架把面板钉在视图区与 composer 之间（composer-card 轴），空列表时隐藏，可折叠——折叠态以进行中项作为单行提示；✓/●/○ 字形与 TUI plan 面板一致。它经框架 `useSession` hook 读取 `snapshot.todos`——无 store、无 service、无 ctx。它住在 `ConversationRoot` 之内，而非 details 列或自建 slot：details slot 单占用且由选中驱动（生命周期不同于一条常开横条），且 slot 表没有为 plan 预留席位。组件 props 完备且框架无关，因此日后迁往专用 slot 不触及其内部任何东西。
+面板经 `conversation.input.dock` slot 挂载（普通注册者插件 `todoDockEntry`，QueueDock 同款姿势：`inject: ['slots', 'conversation']` 载序 seam，`order: -1` 排在队列条上方），空列表时隐藏，可折叠——折叠态以进行中项作为单行提示；✓/●/○ 字形与 TUI plan 面板一致。它经 dock entry 收到的标准件 `useSession` hook 读取 `snapshot.todos`——无 store、无 service、无 ctx。内部组件保持 props 完备且框架无关；dock 适配件只是一行包装。
 
 ### TodoRow：经 keyed toolview slot 的逐调用行
 
@@ -27,9 +27,10 @@ Status: implemented
 ## Alternatives considered
 
 - **把 todo 写入作为 surface 条目折叠进 `nodes`**——回放的窗口会渲染每一份已被取代的列表；该事件被刻意设计成非 surface 类型。
-- **面板放进 details 列或专用 slot**——details slot 单占用且由选中驱动；新增一个 slot 键需要一个 slot 表席位，而设计尚未分配。面板框架无关，所以真要迁移，代价依然很低。
+- **面板硬编码进 `ConversationRoot`**——input-dock slot 出现之前的原始落点；dock 是本架构给"composer 上方常开横条"安排的家，硬编码绕开了 slot 注册表的 disposal 与定序。
+- **面板放进 details 列**——details slot 单占用且由选中驱动，生命周期不同于一条常开横条。
 - **host 计算的视图（一个 todo `ToolEventView`）**——呈现属于客户端；协议已在事件载荷里携带整份快照。
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 63 轮的 fixture（测试前置数据）加 `scripts/verify-todo-display.mjs` 在真实 chromium 里钉住整条链（面板可见性、行摘要、详情联动、折叠、深色主题）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
diff --git a/apps/web/tests/todo-display.snapshot.ts b/apps/web/tests/todo-display.snapshot.ts
new file mode 100644
index 0000000000..8d0127cc54
--- /dev/null
+++ b/apps/web/tests/todo-display.snapshot.ts
@@ -0,0 +1,186 @@
+// @vitest-environment jsdom
+// Todo display snapshot over the BUILT client graph (the code-mode-fixture
+// idiom: real bundles via AppWebEntry, keyless FixtureApiClient transport).
+// Opens the fixture history session and pins the todo_write turn's two
+// surfaces: the dedicated TodoRow in the chat flow (keyed toolview, summary
+// derived from the call args) and the TodoPanel plan strip riding the
+// 'conversation.input.dock' slot (fed by ConversationSnapshot.todos, seeded
+// by the tail history page), including the collapse interaction.
+import { readFileSync } from 'node:fs'
+import { join } from 'node:path'
+import { act, cleanup, fireEvent, screen, waitFor, within } from '@testing-library/react'
+import { afterEach, beforeEach, expect, it, vi } from 'vitest'
+import type { WebBootEntry } from '@deepseek-ai/dsh-client-modules/client'
+import { AppWebEntry } from '@deepseek-ai/dsh-client-web'
+
+const PLUGINS: readonly (WebBootEntry & { dir: string })[] = [
+  { id: '@deepseek-ai/dsh-client-connection', dir: 'connection', url: '/plugins/connection.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-runtime', dir: 'runtime', url: '/plugins/runtime.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-connection'], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-theme', dir: 'ui-theme', url: '/plugins/ui-theme.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-locale', dir: 'locale', url: '/plugins/locale.js', rev: 'fx', inject: [], immediately: true },
+  { id: '@deepseek-ai/dsh-client-ui-layout', dir: 'ui-layout', url: '/plugins/ui-layout.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-runtime'] },
+  { id: '@deepseek-ai/dsh-client-ui-sidebar', dir: 'ui-sidebar', url: '/plugins/ui-sidebar.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  { id: '@deepseek-ai/dsh-client-ui-conversation', dir: 'ui-conversation', url: '/plugins/ui-conversation.js', rev: 'fx', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
+  {
+    id: '@deepseek-ai/dsh-client-ui-workspace',
+    dir: 'ui-workspace',
+    url: '/plugins/ui-workspace.js',
+    rev: 'fx',
+    inject: [
+      '@deepseek-ai/dsh-client-runtime',
+      '@deepseek-ai/dsh-client-ui-conversation',
+      '@deepseek-ai/dsh-client-ui-sidebar',
+    ],
+  },
+]
+
+const bundles = new Map(PLUGINS.map(plugin => [
+  plugin.url,
+  readFileSync(join(process.cwd(), 'packages/client', plugin.dir, 'lib/client.js'), 'utf8'),
+]))
+
+interface FixtureWindow extends Window {
+  __DSH_BOOT__?: { rev: string; entries: WebBootEntry[] }
+  __ModuleLoader__?: unknown
+}
+
+class ResizeObserverStub {
+  observe(): void {}
+  disconnect(): void {}
+  unobserve(): void {}
+}
+
+const win = window as FixtureWindow
+let unmount: (() => void) | undefined
+
+beforeEach(() => {
+  localStorage.clear()
+  document.title = 'DeepSeek Harness'
+  vi.stubGlobal('ResizeObserver', ResizeObserverStub)
+  vi.stubGlobal('requestAnimationFrame', (callback: FrameRequestCallback) =>
+    setTimeout(() => { callback(0) }, 0) as unknown as number)
+  vi.stubGlobal('cancelAnimationFrame', (id: number) => { clearTimeout(id) })
+})
+
+afterEach(() => {
+  act(() => { unmount?.() })
+  unmount = undefined
+  cleanup()
+  delete win.__DSH_BOOT__
+  delete win.__ModuleLoader__
+  delete (globalThis as Record<string, unknown>).__fxTiming
+  document.body.innerHTML = ''
+  document.head.querySelectorAll('style[data-plugin]').forEach((style) => { style.remove() })
+  document.title = ''
+  history.replaceState(null, '', '/')
+  vi.unstubAllGlobals()
+})
+
+/** Boot the complete built client graph against the populated fixture branch. */
+function boot(): void {
+  history.replaceState(null, '', '/?fixture')
+  const root = document.createElement('div')
+  root.id = 'root'
+  document.body.appendChild(root)
+  win.__DSH_BOOT__ = { rev: 'fx', entries: PLUGINS.map(({ dir: _dir, ...plugin }) => plugin) }
+  act(() => {
+    const entry = new AppWebEntry(root, {
+      fetchBundle: (url) => {
+        const code = bundles.get(url)
+        return code === undefined ? Promise.reject(new Error(`missing built bundle ${url}`)) : Promise.resolve(code)
+      },
+      executeBundle: (code) => { (0, eval)(code) },
+    })
+    void entry.run()
+    unmount = () => { entry.dispose() }
+  })
+}
+
+/** Collapse decorative whitespace while preserving the text a user sees. */
+function visibleText(element: Element): string {
+  return (element.textContent ?? '').replace(/\s+/g, ' ').trim()
+}
+
+/** Open the fixture history session (the alpha log carrying the todo_write turn) and wait for its tail. */
+async function openFixtureSession(): Promise<void> {
+  const tree = await screen.findByRole('tree', { name: 'Sessions' }, { timeout: 10_000 })
+  // Anchor on the Workspace title, not the session-count meta: the count
+  // shifts when a blank session joins the group.
+  const group = (await within(tree).findByText('fixture')).closest<HTMLElement>('[role="treeitem"]')
+  if (group === null) throw new Error('fixture Workspace group missing')
+  if (group.getAttribute('aria-expanded') === 'false') {
+    fireEvent.click(within(group).getByText('fixture'))
+    await waitFor(() => {
+      expect(group.getAttribute('aria-expanded')).toBe('true')
+    })
+  }
+  const session = await within(tree).findByText('Fixture 历史会话')
+  fireEvent.click(session)
+  await waitFor(() => {
+    expect(document.querySelector('[data-sample="todo-row"]')).not.toBeNull()
+  }, { timeout: 10_000 })
+}
+
+it('renders the todo_write turn: dedicated tool row + the dock plan strip', async () => {
+  boot()
+  await openFixtureSession()
+
+  const row = document.querySelector('[data-sample="todo-row"]')
+  if (row === null) throw new Error('todo row missing')
+  const panel = document.querySelector('[data-testid="todo-panel"]')
+  if (panel === null) throw new Error('todo panel missing from the input dock')
+
+  expect({
+    row: visibleText(row),
+    rowState: row.getAttribute('data-state'),
+    panelHeader: visibleText(panel.querySelector('button') ?? panel),
+    panelItems: [...panel.querySelectorAll('li')].map(item => ({
+      status: item.getAttribute('data-status'),
+      text: visibleText(item),
+    })),
+  }).toMatchInlineSnapshot(`
+    {
+      "panelHeader": "Plan1/3",
+      "panelItems": [
+        {
+          "status": "completed",
+          "text": "✓梳理需求",
+        },
+        {
+          "status": "in_progress",
+          "text": "●实现 fixture 样本",
+        },
+        {
+          "status": "pending",
+          "text": "○浏览器验收",
+        },
+      ],
+      "row": "☰更新任务清单1/3 已完成 · 实现 fixture 样本",
+      "rowState": "ok",
+    }
+  `)
+})
+
+it('collapses the plan strip to the in-progress hint and restores it', async () => {
+  boot()
+  await openFixtureSession()
+
+  const panel = document.querySelector('[data-testid="todo-panel"]')
+  if (panel === null) throw new Error('todo panel missing from the input dock')
+  const header = panel.querySelector('button')
+  if (header === null) throw new Error('todo panel header missing')
+
+  fireEvent.click(header)
+  expect({
+    collapsedHeader: visibleText(header),
+    listGone: panel.querySelector('ul') === null,
+  }).toMatchInlineSnapshot(`
+    {
+      "collapsedHeader": "Plan1/3实现 fixture 样本",
+      "listGone": true,
+    }
+  `)
+
+  fireEvent.click(header)
+  expect(panel.querySelectorAll('li')).toHaveLength(3)
+})
diff --git a/packages/client/ui-conversation/src/client/apply.ts b/packages/client/ui-conversation/src/client/apply.ts
index 3005c84aff..208d75c5d4 100644
--- a/packages/client/ui-conversation/src/client/apply.ts
+++ b/packages/client/ui-conversation/src/client/apply.ts
@@ -14,6 +14,7 @@ import { InputBar } from './skeleton/InputBar.tsx'
 import { ChatView } from './chat/ChatView.tsx'
 import { bashToolviewSample } from './toolviews/bash-sample.tsx'
 import { todoToolview } from './toolviews/todo-row.tsx'
+import { todoDockEntry } from './skeleton/TodoPanel.tsx'
 import { queueDockEntry } from './queue/QueueDock.tsx'
 import { ConversationRoot } from './skeleton/ConversationRoot.tsx'
 import { ConversationSession } from './skeleton/ConversationSession.tsx'
@@ -186,6 +187,9 @@ export function apply(ctx: Context): void {
   // The todo_write row rides the same seam (a product registration, not a sample).
   ctx.plugin(todoToolview)
 
+  // The plan strip rides the input dock above the queue rows (same posture).
+  ctx.plugin(todoDockEntry)
+
   // The read-only queue dock entry (T9 file territory) rides the same
   // registration seam into the input dock declared above.
   ctx.plugin(queueDockEntry)
diff --git a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
index efaa2599b8..b7d8e4172c 100644
--- a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
@@ -1,10 +1,13 @@
-// TodoPanel: persistent plan strip pinned above the composer (the web
-// counterpart of the TUI plan panel; ACP maps the same event to its native
-// plan). Renders the latest todo/write whole-list snapshot off the session
-// snapshot — no data of its own, hidden while the list is empty. Zero
-// framework imports: useSession arrives via props from ConversationRoot.
+// TodoPanel: persistent plan strip above the composer (the web counterpart
+// of the TUI plan panel; ACP maps the same event to its native plan). Renders
+// the latest todo/write whole-list snapshot off the session snapshot — no
+// data of its own, hidden while the list is empty. Mounted through the
+// 'conversation.input.dock' slot (QueueDock posture): the standard session
+// kit supplies useSession, so the inner component stays framework-free.
 
 import { useState } from 'react'
+import type { Context } from 'cordis'
+import type { PropsRuntime } from '@deepseek-ai/dsh-client-ui-slots'
 import type { TodoItem } from '@deepseek-ai/dsh-client-runtime/client'
 import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
 import { IconChevronDownOutline14, IconChevronUpOutline14 } from '@deepseek-ai/dsh-client-ui-primitives'
@@ -57,3 +60,29 @@ export function TodoPanel({ useSession }: TodoPanelProps) {
     </section>
   )
 }
+
+/** Full props of a dock entry: InputZone owner share + session standard kit + global seat. */
+export type TodoDockProps = PropsRuntime<'conversation.input.dock'>
+
+/** Dock adapter: the standard kit's useSession feeds the strip. */
+export function TodoDock({ useSession }: TodoDockProps) {
+  return <TodoPanel useSession={useSession} />
+}
+
+/**
+ * The plan strip as a plain registrant plugin (QueueDock posture).
+ * `inject: ['conversation']` is the ordering seam: the conversation service
+ * mounts after ui-conversation's slot registrations, so the
+ * 'conversation.input.dock' declaration is on the ledger by then.
+ */
+export const todoDockEntry = {
+  name: 'conversation-todo-dock',
+  inject: ['slots', 'conversation'],
+  /**
+   * Register the plan strip into the input dock (list entry, above the queue rows).
+   * @param ctx - registrant context (disposal rides ctx.effect inside slots.register).
+   */
+  apply(ctx: Context): void {
+    ctx.slots.register({ name: 'conversation.input.dock', id: 'todo', order: -1 }, TodoDock)
+  },
+}
diff --git a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
index 125e421772..f48f35695a 100644
--- a/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
+++ b/packages/client/ui-conversation/tests/chat-code-subcalls.spec.tsx
@@ -56,7 +56,7 @@ function snapshotWith(
 ): ConversationSnapshot {
   return {
     sessionId: SID, nodes, foldDegraded: false, partial: null, runningCalls, codeDispatches,
-    pending: [], queue: [], running: runningCalls.length > 0, composerPhase: 'active', removed: false,
+    pending: [], queue: [], todos: [], running: runningCalls.length > 0, composerPhase: 'active', removed: false,
     openState: 'open', openError: null,
     hasMore: false, loadingOlder: false, promptError: null, blank: false, lastAgentError: null,
   } as ConversationSnapshot
diff --git a/packages/client/ui-conversation/tests/input-bar.spec.tsx b/packages/client/ui-conversation/tests/input-bar.spec.tsx
index 385a5d417e..06a8d266a0 100644
--- a/packages/client/ui-conversation/tests/input-bar.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-bar.spec.tsx
@@ -21,7 +21,7 @@ const SID = 's1' as SessionId
 function snapshotOf(overrides: Partial<ConversationSnapshot> = {}): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [], codeDispatches: new Map(),
-    pending: [], queue: [], running: false, composerPhase: 'active', removed: false,
+    pending: [], queue: [], todos: [], running: false, composerPhase: 'active', removed: false,
     openState: 'open', openError: null, hasMore: false, loadingOlder: false,
     promptError: null, blank: false, lastAgentError: null,
     ...overrides,
diff --git a/packages/client/ui-conversation/tests/input-matrix.spec.tsx b/packages/client/ui-conversation/tests/input-matrix.spec.tsx
index 6b60f7ea17..caa9b85ad9 100644
--- a/packages/client/ui-conversation/tests/input-matrix.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-matrix.spec.tsx
@@ -24,7 +24,7 @@ const SID = 's1' as SessionId
 function mountBar(shell: SessionInputShell, over?: { running?: boolean; disabled?: boolean }) {
   const session = createSnapshotStore<ConversationSnapshot>({
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [], codeDispatches: new Map(),
-    pending: [], queue: [], running: over?.running ?? false, composerPhase: 'active',
+    pending: [], queue: [], todos: [], running: over?.running ?? false, composerPhase: 'active',
     removed: over?.disabled ?? false, openState: 'open', openError: null, hasMore: false,
     loadingOlder: false, promptError: null, blank: false, lastAgentError: null,
   })
diff --git a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
index 9a99eccbf0..ca71a2f0f8 100644
--- a/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
+++ b/packages/client/ui-conversation/tests/input-scenarios.spec.tsx
@@ -111,7 +111,7 @@ async function scopedBench(register?: (slash: SlashService) => void) {
   const wiring = shell
   const sessionStore = createSnapshotStore<ConversationSnapshot>({
     sessionId, nodes: [], foldDegraded: false, partial: null, runningCalls: [], codeDispatches: new Map(),
-    pending: [], queue: [], running: false, composerPhase: 'active', removed: false,
+    pending: [], queue: [], todos: [], running: false, composerPhase: 'active', removed: false,
     openState: 'open', openError: null, hasMore: false, loadingOlder: false,
     promptError: null, blank: false, lastAgentError: null,
   })
diff --git a/packages/client/ui-conversation/tests/queue-dock.spec.tsx b/packages/client/ui-conversation/tests/queue-dock.spec.tsx
index d9b9e951bf..fa0c871bdb 100644
--- a/packages/client/ui-conversation/tests/queue-dock.spec.tsx
+++ b/packages/client/ui-conversation/tests/queue-dock.spec.tsx
@@ -19,7 +19,7 @@ const SID = 's1' as SessionId
 function snapshotWith(queue: QueuedMessage[]): ConversationSnapshot {
   return {
     sessionId: SID, nodes: [], foldDegraded: false, partial: null, runningCalls: [], codeDispatches: new Map(),
-    pending: [], queue, running: true, composerPhase: 'active', removed: false, openState: 'open', openError: null,
+    pending: [], queue, todos: [], running: true, composerPhase: 'active', removed: false, openState: 'open', openError: null,
     hasMore: false, loadingOlder: false, promptError: null, blank: false, lastAgentError: null,
   }
 }
diff --git a/packages/todo/tool-todo/README.i18n.yaml b/packages/todo/tool-todo/README.i18n.yaml
index dff7032381..66d516740c 100644
--- a/packages/todo/tool-todo/README.i18n.yaml
+++ b/packages/todo/tool-todo/README.i18n.yaml
@@ -3,4 +3,4 @@
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
 README.md: 3615f68953cfe6cdc0e6fc41bf75b8dfdf90a310
-README.zh.md: ee0b99d1b4a444dc0b92217686140f691abf1e2b
+README.zh.md: c4a7d829cc1583b65d2c8afa1683d957f68722e9
diff --git a/packages/todo/tool-todo/README.zh.md b/packages/todo/tool-todo/README.zh.md
index ee0b99d1b4..c4a7d829cc 100644
--- a/packages/todo/tool-todo/README.zh.md
+++ b/packages/todo/tool-todo/README.zh.md
@@ -16,11 +16,11 @@
 
 ## 验证
 
-除 schema 的类型／必填／枚举检查外，`execute` 还会拒绝空或重复的 `content`，以及同时存在多个 `in_progress` 任务的情况（连贯计划最多只有一个活跃任务）。顺序与保持列表最新的纪律由模型根据工具描述负责。
+除 schema 的类型／必填／枚举检查外，`execute` 还会拒绝空或重复的 `content`、同时存在多个 `in_progress` 任务的情况（连贯计划最多只有一个活跃任务），以及 `content`/`status` 之外的任何条目键——扩展条目形状（id、嵌套）会响亮失败而不是被静默压平，保证落日志的快照与模型自认为写入的内容一致。顺序与保持列表最新的纪律由模型根据工具描述负责。
 
 ## 渲染
 
-规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久列表：[TUI 应用](../../examples/tui-demo)将其显示为持久计划，[web 客户端](../../client/ui-conversation)则基于 `ConversationSnapshot.todos` 渲染计划条和专属工具行（[Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)）。
+规范结果为 `{ todos, counts: { pending, inProgress, completed } }`；其 Native 渲染器返回精简的更新确认。工具还会写入完整 `todo/write` 会话事件。UI 订阅事件流，并自行渲染该持久列表：[TUI 应用](../../examples/tui-demo)将其显示为持久计划，[web 客户端](../../client/ui-conversation)则基于 `ConversationSnapshot.todos` 渲染计划横条与专属工具行（[Agent Note](../../../.agents/notes/implemented/feature/2026-07-23-web-todo-display.md)）。
 
 ## 导出形状
 
diff --git a/scripts/verify-todo-display.mjs b/scripts/verify-todo-display.mjs
deleted file mode 100644
index 3e8f42e5fe..0000000000
--- a/scripts/verify-todo-display.mjs
+++ /dev/null
@@ -1,107 +0,0 @@
-// Manual acceptance probe: boot the real shell + 8 bundles in ?fixture mode,
-// open fx-alpha, assert the TodoPanel strip and the todo_write row render.
-import { existsSync } from 'node:fs'
-import { fileURLToPath } from 'node:url'
-import { createRequire } from 'node:module'
-import { startWebServer } from '@deepseek-ai/dsh-host-webserver'
-
-// playwright is a dependency of apps/web (the browser test owner), not the root.
-const { chromium } = createRequire(new URL('../apps/web/package.json', import.meta.url)).call(undefined, 'playwright')
-
-const root = fileURLToPath(new URL('..', import.meta.url))
-const bundle = (dir) => `${root}packages/client/${dir}/lib/client.js`
-const PLUGINS = [
-  { id: '@deepseek-ai/dsh-client-connection', dir: 'connection', inject: [], immediately: true },
-  { id: '@deepseek-ai/dsh-client-runtime', dir: 'runtime', inject: ['@deepseek-ai/dsh-client-connection'], immediately: true },
-  { id: '@deepseek-ai/dsh-client-ui-theme', dir: 'ui-theme', inject: [], immediately: true },
-  { id: '@deepseek-ai/dsh-client-i18n', dir: 'i18n', inject: [], immediately: true },
-  { id: '@deepseek-ai/dsh-client-ui-layout', dir: 'ui-layout', inject: ['@deepseek-ai/dsh-client-runtime'] },
-  { id: '@deepseek-ai/dsh-client-ui-sidebar', dir: 'ui-sidebar', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
-  { id: '@deepseek-ai/dsh-client-ui-conversation', dir: 'ui-conversation', inject: ['@deepseek-ai/dsh-client-ui-layout'] },
-  { id: '@deepseek-ai/dsh-client-ui-trajectory', dir: 'ui-trajectory', inject: ['@deepseek-ai/dsh-client-ui-conversation'] },
-]
-for (const p of PLUGINS) if (!existsSync(bundle(p.dir))) throw new Error(`bundle missing: ${p.dir}`)
-
-const rows = PLUGINS.map(p => ({
-  id: p.id, url: `/plugins/${p.id}/client.js?rev=verify`, rev: 'verify',
-  ...(p.inject.length > 0 ? { inject: p.inject } : {}),
-  ...(p.immediately ? { immediately: true } : {}),
-}))
-const graph = { rev: 'verify', entries: rows }
-const byId = new Map(PLUGINS.map(p => [p.id, bundle(p.dir)]))
-const port = 34567
-const errors = []
-const server = await startWebServer({
-  host: '127.0.0.1',
-  port,
-  distIndex: `${root}apps/web/dist/index.html`,
-  apiHandler: { fetch: () => Promise.resolve(new Response('fixture mode must not call /api', { status: 500 })) },
-  webPlugins: { graph: () => graph, clientPath: (id) => byId.get(id), onRebuilt: () => () => undefined },
-}, (err) => errors.push(`server: ${String(err)}`))
-
-const browser = await chromium.launch()
-const page = await browser.newPage({ viewport: { width: 1680, height: 1000 } })
-page.on('pageerror', e => errors.push(String(e)))
-await page.goto(`http://127.0.0.1:${port}/?fixture`, { waitUntil: 'load' })
-try {
-  await page.waitForSelector('[class*="frame"]', { timeout: 15000 })
-} catch (e) {
-  console.error('BODY:', (await page.evaluate(() => document.body.innerText)).slice(0, 400))
-  console.error('STATUS:', await page.evaluate(() => JSON.stringify(globalThis
-
-.__DSH_LOADER_STATUS__ ?? 'n/a')))
-  console.error('BOOT:', await page.evaluate(() => JSON.stringify(window.__DSH_BOOT__)))
-  const reqs = await page.evaluate(() => performance.getEntriesByType('resource').map(r => `${r.name.split('/').slice(-2).join('/')}=${r.responseStatus ?? '?'}`))
-  console.error('RES:', reqs.join(' '))
-  console.error('ERRORS:', errors.join(' ;; '))
-  throw e
-}
-
-// Open fx-alpha: expand the workspace group, then click the newest session row.
-await page.locator('[role="treeitem"]').first().click()
-// fx-alpha is the newest (running) session — the first option row.
-const sessionRow = page.locator('[role="treeitem"][aria-selected]').first()
-await sessionRow.waitFor({ timeout: 5000 })
-await sessionRow.click()
-await page.waitForSelector('[data-testid="todo-panel"]', { timeout: 15000 })
-console.log('✓ TodoPanel visible')
-
-const panelText = await page.locator('[data-testid="todo-panel"]').innerText()
-for (const expected of ['Plan', '1/3', '梳理需求', '实现 fixture 样本', '浏览器验收']) {
-  if (!panelText.includes(expected)) throw new Error(`TodoPanel missing "${expected}"; got: ${panelText}`)
-}
-console.log('✓ TodoPanel content: counts + all three items')
-
-await page.screenshot({ path: `${root}.artifacts/todo-01-panel.png` })
-
-// The todo_write row in the flow (turn 63 sample, already at the bottom).
-const row = page.locator('[data-sample="todo-row"]')
-await row.waitFor({ timeout: 10000 })
-const rowText = await row.innerText()
-if (!rowText.includes('更新任务清单') || !rowText.includes('1/3 已完成')) throw new Error(`TodoRow wrong: ${rowText}`)
-console.log('✓ TodoRow renders plan summary:', rowText.replace(/\n/g, ' '))
-await page.screenshot({ path: `${root}.artifacts/todo-02-row.png` })
-
-// Row click opens details with the raw args.
-await row.click()
-await page.waitForSelector('text=Input', { timeout: 5000 })
-console.log('✓ TodoRow click opens details')
-await page.screenshot({ path: `${root}.artifacts/todo-03-details.png` })
-
-// Collapse: list hides, active item hint appears in the header.
-await page.locator('[data-testid="todo-panel"] button').first().click()
-const collapsed = await page.locator('[data-testid="todo-panel"]').innerText()
-if (collapsed.includes('梳理需求')) throw new Error('collapse failed: list still visible')
-if (!collapsed.includes('实现 fixture 样本')) throw new Error('collapsed hint missing the active item')
-console.log('✓ Collapse hides list, shows active hint')
-await page.screenshot({ path: `${root}.artifacts/todo-04-collapsed.png` })
-
-// Dark theme spot check.
-await page.evaluate(() => document.body.setAttribute('data-ds-dark-theme', ''))
-await page.screenshot({ path: `${root}.artifacts/todo-05-dark.png` })
-console.log('✓ Dark screenshot taken')
-
-if (errors.length > 0) throw new Error(`page errors: ${errors.join('; ')}`)
-console.log('✓ No page errors — todo display acceptance PASSED')
-await browser.close()
-await server.close()

From 83d3179d34063a0d3f46dbab1775ae8a9014c133 Mon Sep 17 00:00:00 2001
From: imccyu <276526105+imccyu@users.noreply.github.com>
Date: Mon, 27 Jul 2026 10:14:37 +0800
Subject: [PATCH 30/36] disambiguate the snapshot's workspace-group lookup

Latest master seeds a blank session also titled "fixture"; anchor the
session-tree lookup on the expandable group row instead of the first text
match.
---
 apps/web/tests/todo-display.snapshot.ts | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/apps/web/tests/todo-display.snapshot.ts b/apps/web/tests/todo-display.snapshot.ts
index 8d0127cc54..3116bf4242 100644
--- a/apps/web/tests/todo-display.snapshot.ts
+++ b/apps/web/tests/todo-display.snapshot.ts
@@ -104,10 +104,13 @@ function visibleText(element: Element): string {
 /** Open the fixture history session (the alpha log carrying the todo_write turn) and wait for its tail. */
 async function openFixtureSession(): Promise<void> {
   const tree = await screen.findByRole('tree', { name: 'Sessions' }, { timeout: 10_000 })
-  // Anchor on the Workspace title, not the session-count meta: the count
-  // shifts when a blank session joins the group.
-  const group = (await within(tree).findByText('fixture')).closest<HTMLElement>('[role="treeitem"]')
-  if (group === null) throw new Error('fixture Workspace group missing')
+  // Anchor on the expandable Workspace group row: the title and the blank
+  // session row can both read "fixture", and the session-count meta shifts
+  // when a blank session joins the group.
+  const group = (await within(tree).findAllByText('fixture'))
+    .map(el => el.closest<HTMLElement>('[role="treeitem"]'))
+    .find(el => el?.getAttribute('aria-expanded') !== null)
+  if (group === null || group === undefined) throw new Error('fixture Workspace group missing')
   if (group.getAttribute('aria-expanded') === 'false') {
     fireEvent.click(within(group).getByText('fixture'))
     await waitFor(() => {

From 8d1a3b89c7c9749af8fc25693cffc69d4af6b5af Mon Sep 17 00:00:00 2001
From: imccyu <276526105+imccyu@users.noreply.github.com>
Date: Mon, 27 Jul 2026 10:25:12 +0800
Subject: [PATCH 31/36] fix runtime README pairing record left with merge
 conflict markers

The master merge committed the i18n.yaml with unresolved conflict hunks
(carried over from the adapt branch's own master merge); re-recorded via
verify-translation-pairing --write. 518 pairs consistent.
---
 packages/client/runtime/README.i18n.yaml | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml
index fbedbd07d4..3e687c64dd 100644
--- a/packages/client/runtime/README.i18n.yaml
+++ b/packages/client/runtime/README.i18n.yaml
@@ -2,10 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write
-<<<<<<< HEAD
 README.md: d80c4a13e3d1b0f1713a6788446ae24fe3550381
 README.zh.md: 727721824790f33b99fc7c79d96420434cc3a1d0
-=======
-README.md: 4724ebc75d441252245a0e811a4ae34f8b529a98
-README.zh.md: 6a0076742efccaf946910c77c77a9b74194b9dc5
->>>>>>> origin/master

From f2a9f4a40ea15cc0be54ab0f7c3556c8cc24d374 Mon Sep 17 00:00:00 2001
From: imccyu <276526105+imccyu@users.noreply.github.com>
Date: Mon, 27 Jul 2026 10:43:49 +0800
Subject: [PATCH 32/36] retire the stale ACP plan-mapping claim in the todo
 note

The automation-only ACP bridge deliberately omits todo presentation (its
edge test asserts plan updates are omitted; the todo-write tool note records
the mapping's retirement). Chinese counterpart updated, pair re-recorded;
the TodoPanel header comment drops the same claim.
---
 .../feature/2026-07-23-web-todo-display.i18n.yaml           | 6 +++---
 .../implemented/feature/2026-07-23-web-todo-display.md      | 6 +++---
 .../implemented/feature/2026-07-23-web-todo-display.zh.md   | 6 +++---
 .../ui-conversation/src/client/skeleton/TodoPanel.tsx       | 2 +-
 4 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index 292c9b6200..44d16b0c63 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-2026-07-23-web-todo-display.md: b08c50b2f1a0aea03382b9e012ac364d697a21ee
-2026-07-23-web-todo-display.zh.md: daff6d4a18a621f3fd336ff6feb4433f5cbf9b16
+#   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+2026-07-23-web-todo-display.md: 8b7bc8cbe0c072613c9b4ad01983df74466e81da
+2026-07-23-web-todo-display.zh.md: 4d5565f805386cf33eaae81e7dcc48e6260e3313
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index b08c50b2f1..8b7bc8cbe0 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -6,11 +6,11 @@ English | [中文](2026-07-23-web-todo-display.zh.md)
 
 ## Problem
 
-`todo_write` appends `todo/write` whole-list snapshots to the session log; the TUI renders a persistent plan panel and the ACP bridge maps the event to native `plan` updates. The web client dropped the event entirely: the host mux stream already forwards every session event, but `todo/write` is not a surface type (it never folds into `ConversationSnapshot.nodes`), and no side-effect branch accumulated it — the browser had no consumption point and no display surface.
+`todo_write` appends `todo/write` whole-list snapshots to the session log; the TUI renders a persistent plan panel (the automation-only ACP bridge deliberately omits todo presentation). The web client dropped the event entirely: the host mux stream already forwards every session event, but `todo/write` is not a surface type (it never folds into `ConversationSnapshot.nodes`), and no side-effect branch accumulated it — the browser had no consumption point and no display surface.
 
 ## Decision
 
-Consume `todo/write` as a Session side effect, not a surface node, and render it on two surfaces matching the split the TUI and ACP already draw.
+Consume `todo/write` as a Session side effect, not a surface node, and render it on two surfaces matching the split the TUI already draws.
 
 ### Side-effect channel, converging with window replay
 
@@ -33,4 +33,4 @@ The dedicated `todo_write` chat row is a plain registrant plugin (`todoToolview`
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The ACP bridge's todo → `plan` mapping and the TUI panel are untouched; the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The TUI panel is untouched (the automation-only ACP bridge deliberately omits todo presentation); the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index daff6d4a18..4d5565f805 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -6,11 +6,11 @@ Status: implemented
 
 ## Problem
 
-`todo_write` 把 `todo/write` 的整份列表快照追加进会话日志；TUI 渲染一块常驻的 plan 面板，ACP 桥接把该事件映射为原生 `plan` 更新。Web 客户端把这个事件整个丢弃了：host mux 流本已转发每一个会话事件，但 `todo/write` 不是 surface 类型（它从不 fold 进 `ConversationSnapshot.nodes`），也没有任何副作用分支累积它——浏览器既无消费点，也无展示面。
+`todo_write` 把 `todo/write` 的整份列表快照追加进会话日志；TUI 渲染一块常驻的 plan 面板（自动化专用的 ACP 桥接刻意不做 todo 呈现）。Web 客户端把这个事件整个丢弃了：host mux 流本已转发每一个会话事件，但 `todo/write` 不是 surface 类型（它从不 fold 进 `ConversationSnapshot.nodes`），也没有任何副作用分支累积它——浏览器既无消费点，也无展示面。
 
 ## Decision
 
-把 `todo/write` 当作 Session 副作用消费，而非 surface 节点，并在两个面上渲染它，这两个面正对应 TUI 与 ACP 已经绘制的那套划分。
+把 `todo/write` 当作 Session 副作用消费，而非 surface 节点，并在两个面上渲染它，这两个面正对应 TUI 已经绘制的那套划分。
 
 ### 副作用通道，与窗口回放收敛
 
@@ -33,4 +33,4 @@ Status: implemented
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。ACP 桥接的 todo → `plan` 映射与 TUI 面板均未受改动；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。TUI 面板未受改动（自动化专用的 ACP 桥接刻意不做 todo 呈现）；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
diff --git a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
index b7d8e4172c..d4d9a56240 100644
--- a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
@@ -1,5 +1,5 @@
 // TodoPanel: persistent plan strip above the composer (the web counterpart
-// of the TUI plan panel; ACP maps the same event to its native plan). Renders
+// of the TUI plan panel). Renders
 // the latest todo/write whole-list snapshot off the session snapshot — no
 // data of its own, hidden while the list is empty. Mounted through the
 // 'conversation.input.dock' slot (QueueDock posture): the standard session

From 8dce7981c4b18db22eab11f2320b1ecdf755c513 Mon Sep 17 00:00:00 2001
From: imccyu <276526105+imccyu@users.noreply.github.com>
Date: Mon, 27 Jul 2026 11:38:02 +0800
Subject: [PATCH 33/36] refresh code-mode trajectory ordinals for the todo
 fixture turn

Turn 65 (todo_write) at the fx-alpha tail slides the 50-message history
window: two head-of-window messages drop out, so every trajectory cell
ordinal shifts down by two. Timing, labels, and cell content are unchanged.
---
 apps/web/tests/code-mode-fixture.snapshot.ts | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/apps/web/tests/code-mode-fixture.snapshot.ts b/apps/web/tests/code-mode-fixture.snapshot.ts
index 6f4085da44..a727844b1b 100644
--- a/apps/web/tests/code-mode-fixture.snapshot.ts
+++ b/apps/web/tests/code-mode-fixture.snapshot.ts
@@ -213,9 +213,9 @@ it('trajectory and waterfall surface the run_code sub-calls with real timing', a
   }).toMatchInlineSnapshot(`
     {
       "subCells": [
-        "#53Subbash · {"command":"ls notes","description":"List notes"}+0.8s",
-        "#54Subread · {"path":"notes/demo.txt"}+0.8s",
-        "#55Subread · {"path":"notes/missing.txt"}+0.8s",
+        "#51Subbash · {"command":"ls notes","description":"List notes"}+0.8s",
+        "#52Subread · {"path":"notes/demo.txt"}+0.8s",
+        "#53Subread · {"path":"notes/missing.txt"}+0.8s",
       ],
     }
   `)

From 8ebdad5076fe85335b169c175ad43226aced4a00 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 11:40:41 +0800
Subject: [PATCH 34/36] reset the plan when a tail history response omits the
 todo projection
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

An omitted `todos` on a tail request was treated as "no projection carried"
and preserved the prior value. Every installWindow caller is a tail request
(doOpen, its gap re-pull, repairGap; loadOlder prepends without it), which the
host answers with the full-log projection or omits only when the log holds no
todo/write — so the field's absence is the authoritative empty list. A live
write whose host crashed before persisting therefore left the rolled-back plan
on screen indefinitely; the assignment now clears it on the next open or
resync. Widened the parameter to an explicit `| undefined` so the two meanings
cannot be conflated again, and updated the JSDoc at both declaring seams plus
the bilingual README/note pair.
---
 .../2026-07-23-web-todo-display.i18n.yaml        |  4 ++--
 .../feature/2026-07-23-web-todo-display.md       |  4 ++--
 .../feature/2026-07-23-web-todo-display.zh.md    |  4 ++--
 packages/client/runtime/README.i18n.yaml         |  6 +++---
 packages/client/runtime/README.md                |  2 +-
 packages/client/runtime/README.zh.md             |  2 +-
 .../runtime/src/client/sessions/conversation.ts  |  3 ++-
 .../runtime/src/client/sessions/session.ts       | 13 +++++++++----
 packages/client/runtime/tests/session.spec.ts    | 16 ++++++++++++++++
 9 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index 44d16b0c63..0b209eced2 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-23-web-todo-display.md
-2026-07-23-web-todo-display.md: 8b7bc8cbe0c072613c9b4ad01983df74466e81da
-2026-07-23-web-todo-display.zh.md: 4d5565f805386cf33eaae81e7dcc48e6260e3313
+2026-07-23-web-todo-display.md: a8ee5a5d2799b6e2f9ff442e0bd6230b46e9374b
+2026-07-23-web-todo-display.zh.md: 0bf7157789a0be8f2c68d462bc815db8cf07d96e
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index 8b7bc8cbe0..a8ee5a5d27 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -14,7 +14,7 @@ Consume `todo/write` as a Session side effect, not a surface node, and render it
 
 ### Side-effect channel, converging with window replay
 
-`applyEventSideEffects` gains a `todo/write` case (whole list, last write wins). Unlike partial/openCalls, `rebuildDerivedFromWindow` deliberately does NOT reset it: the value is session-level — seeded by the tail history page's full-log projection — and an arbitrary window may not contain the latest write, so rebuilds (paging, reconnect stitch, resync) preserve it and only an in-window or live write overwrites it. `ConversationSnapshot.todos` is the read surface. This follows the event's own contract ("log-only UI state; never derived history"): surfacing each write as a conversation node would render superseded lists as if they were still standing.
+`applyEventSideEffects` gains a `todo/write` case (whole list, last write wins). Unlike partial/openCalls, `rebuildDerivedFromWindow` deliberately does NOT reset it: the value is session-level — taken from the tail history page's full-log projection — and an arbitrary window may not contain the latest write, so an older-page prepend keeps it and only an in-window or live write overwrites it. Every `installWindow` caller is a tail request (`doOpen`, its gap re-pull, `repairGap`; `loadOlder` prepends without it), which the host answers with the projection or omits it only when the full log holds no `todo/write` — so an absent field is the authoritative empty list and is assigned as such. That distinction matters on rollback: a live write whose host crashed before persisting leaves the log empty, and preserving the prior value instead would strand the rolled-back plan on screen indefinitely. `ConversationSnapshot.todos` is the read surface. This follows the event's own contract ("log-only UI state; never derived history"): surfacing each write as a conversation node would render superseded lists as if they were still standing.
 
 ### TodoPanel: the durable list as a persistent strip
 
@@ -33,4 +33,4 @@ The dedicated `todo_write` chat row is a plain registrant plugin (`todoToolview`
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The TUI panel is untouched (the automation-only ACP bridge deliberately omits todo presentation); the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; the seeded value is preserved across window rebuilds and overwritten by any later write.
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The TUI panel is untouched (the automation-only ACP bridge deliberately omits todo presentation); the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; that value survives an older-page prepend, is overwritten by any later write, and resets to empty when a tail response carries no projection.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index 4d5565f805..0bf7157789 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -14,7 +14,7 @@ Status: implemented
 
 ### 副作用通道，与窗口回放收敛
 
-`applyEventSideEffects` 新增一个 `todo/write` 分支（整份列表，后写覆盖先写）。与 partial/openCalls 不同，`rebuildDerivedFromWindow` 刻意不重置它：该值是会话级的——由尾页 history 携带的全量 log 投影播种——而任意窗口未必包含最近一次写入，因此窗口重建（分页、重连缝合、resync）保留它，只有窗口内或实时的写入才会覆盖。`ConversationSnapshot.todos` 是读取面。这遵循事件自身的契约（「仅日志 UI 状态，绝非派生历史」）：把每次写入作为对话节点呈现，会让已被取代的列表看起来仍然有效。
+`applyEventSideEffects` 新增一个 `todo/write` 分支（整份列表，后写覆盖先写）。与 partial/openCalls 不同，`rebuildDerivedFromWindow` 刻意不重置它：该值是会话级的——取自尾页 history 携带的全量 log 投影——而任意窗口未必包含最近一次写入，因此往前翻页保留它，只有窗口内或实时的写入才会覆盖。`installWindow` 的每个调用方都是尾页请求（`doOpen`、其补洞重拉、`repairGap`；`loadOlder` 只往前拼接、不走它），而 host 对尾页请求要么带上投影、要么仅在全量 log 没有任何 `todo/write` 时省略——因此字段缺失就是权威的空列表，直接照此赋值。这个区分在回滚场景上要紧：实时写入若在 host 持久化前崩溃，log 里就是空的，此时保留旧值会让已回滚的计划永远留在屏幕上。`ConversationSnapshot.todos` 是读取面。这遵循事件自身的契约（「仅日志 UI 状态，绝非派生历史」）：把每次写入作为对话节点呈现，会让已被取代的列表看起来仍然有效。
 
 ### TodoPanel：长驻列表作为一条常驻横条
 
@@ -33,4 +33,4 @@ Status: implemented
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。TUI 面板未受改动（自动化专用的 ACP 桥接刻意不做 todo 呈现）；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；播种值跨窗口重建保留，之后的任何写入照常覆盖。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。TUI 面板未受改动（自动化专用的 ACP 桥接刻意不做 todo 呈现）；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；该值跨往前翻页保留，之后的任何写入照常覆盖，而尾页响应不带投影时复位为空。
diff --git a/packages/client/runtime/README.i18n.yaml b/packages/client/runtime/README.i18n.yaml
index 3e687c64dd..055ce4eb86 100644
--- a/packages/client/runtime/README.i18n.yaml
+++ b/packages/client/runtime/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: d80c4a13e3d1b0f1713a6788446ae24fe3550381
-README.zh.md: 727721824790f33b99fc7c79d96420434cc3a1d0
+#   pnpm run verify-translation-pairing --write packages/client/runtime/README.md
+README.md: d7b7bbc4e4e05893689a8f2dcac82763b4c67ef8
+README.zh.md: d2054170cd6f31505793812fff46cc0f2356ad75
diff --git a/packages/client/runtime/README.md b/packages/client/runtime/README.md
index d80c4a13e3..d7b7bbc4e4 100644
--- a/packages/client/runtime/README.md
+++ b/packages/client/runtime/README.md
@@ -2,7 +2,7 @@
 
 English | [中文](README.zh.md)
 
-Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects, list/scope/history state; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into both managers. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. `ConversationSnapshot` carries `todos` — the session's current todo projection: seeded by the tail history page's full-log value (host-computed, independent of the page window), preserved across window rebuilds, and overwritten by each live `todo/write` (last write wins).
+Client cordis boot and React-free object services: SlotsService wraps SlotCore and supplies renderer data sources; SessionsService owns Session objects, list/scope/history state; WorkspacesService depends on SessionsService and owns Workspace objects, list/actions, default-target derivation, and the New Session blank-reuse entry (`connectWorkspace`). The runtime fans the shared Host stream into both managers. Client sessions are always Host-born (Session+Agent+cwd in one `session.create`); the client holds no pre-entity session state — a session's Agent scope (the client mirror of host dsh-scope, keyed by the shared agent/session id) is born when its row enters the list mirror and dies with the prune. Contract: api-contracts v3 §4. `ConversationSnapshot` carries `todos` — the session's current todo projection: taken from the tail history page's full-log value (host-computed, independent of the page window), preserved across an older-page prepend, and overwritten by each live `todo/write` (last write wins). A tail response that omits the field means the log holds no `todo/write`, so the list resets to empty — a plan the log never kept (a write lost to a host crash) disappears on the next open or resync.
 
 ## Workspace and Session lists
 
diff --git a/packages/client/runtime/README.zh.md b/packages/client/runtime/README.zh.md
index 7277218247..d2054170cd 100644
--- a/packages/client/runtime/README.zh.md
+++ b/packages/client/runtime/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象、列表／scope／history 状态；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给两个 manager。客户端 Session 一律由 Host 出生（一次 `session.create` 同瞬产出 Session+Agent+cwd）；客户端不持有任何实体化之前的会话状态——Agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时出生，随 prune 死亡。契约：api-contracts v3 §4。`ConversationSnapshot` 携带 `todos`——会话当前的 todo 投影：由尾页 history 携带的全量 log 值播种（host 计算，独立于分页窗口），跨窗口重建保留，并被每次实时 `todo/write` 覆盖（后写胜出）。
+客户端 cordis 启动与不依赖 React 的对象服务：SlotsService 包装 SlotCore 并提供 renderer 数据源；SessionsService 拥有 Session 对象、列表／scope／history 状态；WorkspacesService 依赖 SessionsService，拥有 Workspace 对象、列表／操作、默认目标派生，以及 New Session 空会话复用入口（`connectWorkspace`）。运行时把共享 Host 流分发给两个 manager。客户端 Session 一律由 Host 出生（一次 `session.create` 同瞬产出 Session+Agent+cwd）；客户端不持有任何实体化之前的会话状态——Agent scope（host dsh-scope 的客户端镜像，以 agent/session 共用 id 为键）在会话行进入列表镜像时出生，随 prune 死亡。契约：api-contracts v3 §4。`ConversationSnapshot` 携带 `todos`——会话当前的 todo 投影：取自尾页 history 携带的全量 log 值（host 计算，独立于分页窗口），跨往前翻页保留，并被每次实时 `todo/write` 覆盖（后写胜出）。尾页响应省略该字段即表示 log 中没有任何 `todo/write`，因此列表复位为空——log 从未留下的计划（写入因 host 崩溃丢失）会在下一次打开或 resync 时消失。
 
 ## Workspace 与 Session 列表
 
diff --git a/packages/client/runtime/src/client/sessions/conversation.ts b/packages/client/runtime/src/client/sessions/conversation.ts
index b3136640b7..e1b2c42962 100644
--- a/packages/client/runtime/src/client/sessions/conversation.ts
+++ b/packages/client/runtime/src/client/sessions/conversation.ts
@@ -244,6 +244,7 @@ export interface ConversationSnapshot {
    */
   blank: boolean
   lastAgentError: string | null
-  /** Latest `todo/write` whole-list snapshot in the window (last write wins); empty = no plan. */
+  /** Current whole-list `todo/write` projection — the tail page's full-log value, then each live
+   *  write (last write wins); empty = the log holds no plan. */
   todos: readonly TodoItem[]
 }
diff --git a/packages/client/runtime/src/client/sessions/session.ts b/packages/client/runtime/src/client/sessions/session.ts
index c5935bf03b..1dd0283429 100644
--- a/packages/client/runtime/src/client/sessions/session.ts
+++ b/packages/client/runtime/src/client/sessions/session.ts
@@ -99,7 +99,8 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
   private queueCache: { rev: number; value: QueuedMessage[] } | null = null
   private frozenRev = 0
   private nodesCache: { folded: readonly ConversationNode[]; frozenRev: number; value: readonly ConversationNode[] } | null = null
-  /** Latest todo/write whole-list snapshot in the window (last write wins on replay). */
+  /** Current whole-list todo/write projection: each tail history response replaces it (an omitted
+   *  field is the authoritative empty list) and every live write overwrites it. */
   private todos: readonly TodoItem[] = []
   /** `run_code` sub-dispatches by parent callId (window-derived, like openCalls). Appends
    *  copy-on-write the per-parent array so published snapshot references never mutate. */
@@ -505,15 +506,19 @@ export class Session implements ObservableSnapshot<ConversationSnapshot> {
    *  Stitching MUST NOT route through acceptLiveEvent: openState is still 'loading' here
    *  (doOpen flips it after install), so recursing would push every buffered event straight
    *  back into liveBuffer where nothing ever drains it — a silent drop loop (audit S1). */
-  private installWindow(entries: HistoryEntry[], hasMore: boolean, todos?: readonly TodoItem[]): void {
+  private installWindow(entries: HistoryEntry[], hasMore: boolean, todos: readonly TodoItem[] | undefined): void {
     this.events = entries.map(e => e.event)
     this.views = entries.map(e => e.view)
     this.baseSeq = this.events[0]?.seq ?? 0
     this.hasMore = hasMore
     // Session-level projection from the tail page (full-log latest todo/write,
     // independent of the window); an in-window write below re-derives the same
-    // value, and later live events keep overwriting it.
-    if (todos !== undefined) this.todos = todos
+    // value, and later live events keep overwriting it. Every caller here is a
+    // tail request (no beforeSeq), which the host answers with the projection
+    // or omits it only when the full log holds no todo/write — so an absent
+    // field is the authoritative empty list, not a missing carrier. Assigning
+    // it clears a plan the log never kept (a write lost to a host crash).
+    this.todos = todos ?? []
     this.foldAdapter.reset(this.events, this.baseSeq, this.views)
     this.rebuildDerivedFromWindow()
     const buffered = this.liveBuffer
diff --git a/packages/client/runtime/tests/session.spec.ts b/packages/client/runtime/tests/session.spec.ts
index ae2739991e..6c223ef58b 100644
--- a/packages/client/runtime/tests/session.spec.ts
+++ b/packages/client/runtime/tests/session.spec.ts
@@ -217,6 +217,22 @@ describe('live event path', () => {
     await Promise.resolve()
     expect(session.getSnapshot().todos).toEqual(current)
   })
+
+  it('clears the plan when a tail response omits the projection (a write the log never kept)', async () => {
+    // Live write lands, then the host crashes before persisting it: the
+    // authoritative log holds no todo/write, so the resync tail response
+    // carries no projection — an omitted field on a tail request is the empty
+    // list, not a missing carrier, and the rolled-back plan must disappear.
+    const { api, session } = await opened(plainTurn(0, 0, 'a', 'b'))
+    session.handleMuxEnvelope('r' as never, {
+      type: 'session/event', sessionId: SID,
+      event: ev.todoWrite(6, [{ content: '丢失的计划', status: 'in_progress' as const }]),
+    })
+    expect(session.getSnapshot().todos).toEqual([{ content: '丢失的计划', status: 'in_progress' }])
+    api.onHistory = () => histResponse(plainTurn(0, 0, 'a', 'b'))
+    await session.resync()
+    expect(session.getSnapshot().todos).toEqual([])
+  })
 })
 
 describe('paging', () => {

From d2bb2a809b8a29b7a45fcb9168f982f089ae6e4b Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 12:08:46 +0800
Subject: [PATCH 35/36] fix(web-todo): dock-owned selection, keyboard-openable
 row, documented tail todos field

Three ds-review-bot round-8 findings on the todo display surfaces.

TodoPanel took the whole `useSession` hook and cast the snapshot to reach
`todos`, which put slot plumbing and an unchecked cast inside the presentation
component. The panel now takes `todos: readonly TodoItem[]`; TodoDock does the
selecting, matching the QueueDock posture the dock slot already establishes.

The todo row carried `onClick` with no keyboard route, so its details panel was
mouse-only. It now takes ToolRow's route verbatim: `role="button"`,
`tabIndex={0}`, and an Enter/Space handler that claims the event. The row stays
a `<div>` because a `<button>` flattens its inline spans.

`session.history`'s tail-only `todos` field was documented at the TS signature
but not in the apiproxy README pair, and the Agent Note claimed the feature
added no new wire vocabulary. Both README sides now state the tail/omission
semantics (an omitted field on a tail response is the empty plan, not unchanged
state), and the note records the one added field instead of denying it.

Tests: TodoPanel specs render the plain list; new TodoDock specs cover selection,
live follow, rollback-to-empty, and the registration shape; a new row spec pins
Enter/Space activation and non-activating keys.
---
 .../2026-07-23-web-todo-display.i18n.yaml     |  4 +-
 .../feature/2026-07-23-web-todo-display.md    |  2 +-
 .../feature/2026-07-23-web-todo-display.zh.md |  2 +-
 .../src/client/skeleton/TodoPanel.tsx         | 21 +++---
 .../src/client/toolviews/todo-row.tsx         | 18 ++++-
 .../ui-conversation/tests/todo-panel.spec.tsx | 75 +++++++++++++------
 packages/host/apiproxy/README.i18n.yaml       |  6 +-
 packages/host/apiproxy/README.md              |  2 +
 packages/host/apiproxy/README.zh.md           |  2 +
 9 files changed, 92 insertions(+), 40 deletions(-)

diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
index 0b209eced2..4bccfa396e 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.i18n.yaml
@@ -2,5 +2,5 @@
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
 #   pnpm run verify-translation-pairing --write .agents/notes/implemented/feature/2026-07-23-web-todo-display.md
-2026-07-23-web-todo-display.md: a8ee5a5d2799b6e2f9ff442e0bd6230b46e9374b
-2026-07-23-web-todo-display.zh.md: 0bf7157789a0be8f2c68d462bc815db8cf07d96e
+2026-07-23-web-todo-display.md: 830f55c86c893c4942a1a9d3b8529395d5f5e38b
+2026-07-23-web-todo-display.zh.md: e68928d7eddaaa92ac831722a738ee2002342b38
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
index a8ee5a5d27..830f55c86c 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.md
@@ -33,4 +33,4 @@ The dedicated `todo_write` chat row is a plain registrant plugin (`todoToolview`
 
 ## Consequences
 
-Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The TUI panel is untouched (the automation-only ACP bridge deliberately omits todo presentation); the web surfaces render the same event with no new wire vocabulary. Cold-load reconstruction is host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; that value survives an older-page prepend, is overwritten by any later write, and resets to empty when a tail response carries no projection.
+Replay correctness is owned by one code path: any future change to window rebuild keeps todos consistent for free, and the fixture (fx-alpha turn 65) plus the assembled keyless snapshot (`apps/web/tests/todo-display.snapshot.ts`) pin the full chain (row summary and state, dock panel content, collapse round-trip) over the built client graph. `todos` is a required `ConversationSnapshot` field, so scripted fakes in specs must carry it. The TUI panel is untouched (the automation-only ACP bridge deliberately omits todo presentation); the web surfaces render the same event, adding one wire field and no new event type. That field is how cold-load reconstruction stays host-backed: the tail history page carries `todos` — the full-log latest `todo/write`, computed independently of the page window (the same backscan posture the view pairing uses) — so a reopened session restores the plan even when the last write precedes the window; that value survives an older-page prepend, is overwritten by any later write, and resets to empty when a tail response carries no projection.
diff --git a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
index 0bf7157789..e68928d7ed 100644
--- a/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
+++ b/.agents/notes/implemented/feature/2026-07-23-web-todo-display.zh.md
@@ -33,4 +33,4 @@ Status: implemented
 
 ## Consequences
 
-回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。TUI 面板未受改动（自动化专用的 ACP 桥接刻意不做 todo 呈现）；Web 各面渲染同一个事件，不引入任何新的协议词汇。冷加载重建由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；该值跨往前翻页保留，之后的任何写入照常覆盖，而尾页响应不带投影时复位为空。
+回放正确性由一条代码路径掌管：未来对窗口重建的任何改动都免费保持 todos 一致；fx-alpha 第 65 轮的 fixture（测试前置数据）加 assembled keyless snapshot（`apps/web/tests/todo-display.snapshot.ts`）在构建产物客户端全图上钉住整条链（行摘要与状态、dock 面板内容、折叠往返）。`todos` 是 `ConversationSnapshot` 的必填字段，所以 spec 里脚本化的 fake 必须带上它。TUI 面板未受改动（自动化专用的 ACP 桥接刻意不做 todo 呈现）；Web 各面渲染同一个事件，只新增一个协议字段，不新增事件类型。冷加载重建正是靠这个字段由 host 兜底：history 尾页附带 `todos`——全量 log 上最新一次 `todo/write` 的投影，独立于分页窗口计算（与 view 配对同一种 backscan 姿势）——因此重开会话时即使最后一次写入落在窗口之前，计划也照常恢复；该值跨往前翻页保留，之后的任何写入照常覆盖，而尾页响应不带投影时复位为空。
diff --git a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
index d4d9a56240..283eeb3e5e 100644
--- a/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
+++ b/packages/client/ui-conversation/src/client/skeleton/TodoPanel.tsx
@@ -1,20 +1,19 @@
 // TodoPanel: persistent plan strip above the composer (the web counterpart
-// of the TUI plan panel). Renders
-// the latest todo/write whole-list snapshot off the session snapshot — no
-// data of its own, hidden while the list is empty. Mounted through the
-// 'conversation.input.dock' slot (QueueDock posture): the standard session
-// kit supplies useSession, so the inner component stays framework-free.
+// of the TUI plan panel). Renders the latest todo/write whole-list snapshot —
+// no data of its own, hidden while the list is empty. Mounted through the
+// 'conversation.input.dock' slot (QueueDock posture): the dock adapter does
+// the selecting, so the panel takes the plain list and stays framework-free.
 
 import { useState } from 'react'
 import type { Context } from 'cordis'
 import type { PropsRuntime } from '@deepseek-ai/dsh-client-ui-slots'
 import type { TodoItem } from '@deepseek-ai/dsh-client-runtime/client'
-import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
 import { IconChevronDownOutline14, IconChevronUpOutline14 } from '@deepseek-ai/dsh-client-ui-primitives'
 import css from './TodoPanel.module.css'
 
 export interface TodoPanelProps {
-  useSession: UseSession
+  /** The session's current plan (empty renders nothing) — selected by the dock adapter. */
+  todos: readonly TodoItem[]
 }
 
 /** Status glyphs mirror the TUI plan panel (✓ done / ● active / ○ pending). */
@@ -22,8 +21,7 @@ const STATUS_GLYPHS: Record<TodoItem['status'], string> = {
   completed: '✓', in_progress: '●', pending: '○',
 }
 
-export function TodoPanel({ useSession }: TodoPanelProps) {
-  const todos = useSession(s => (s as { todos: readonly TodoItem[] }).todos)
+export function TodoPanel({ todos }: TodoPanelProps) {
   const [collapsed, setCollapsed] = useState(false)
   if (todos.length === 0) return null
 
@@ -64,9 +62,10 @@ export function TodoPanel({ useSession }: TodoPanelProps) {
 /** Full props of a dock entry: InputZone owner share + session standard kit + global seat. */
 export type TodoDockProps = PropsRuntime<'conversation.input.dock'>
 
-/** Dock adapter: the standard kit's useSession feeds the strip. */
+/** Dock adapter: selects the plan off the session snapshot and hands the strip a plain list. */
 export function TodoDock({ useSession }: TodoDockProps) {
-  return <TodoPanel useSession={useSession} />
+  const todos = useSession(s => s.todos)
+  return <TodoPanel todos={todos} />
 }
 
 /**
diff --git a/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
index 665fc4dfab..353e7a5441 100644
--- a/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
+++ b/packages/client/ui-conversation/src/client/toolviews/todo-row.tsx
@@ -5,6 +5,7 @@
 // durable list itself renders in the TodoPanel above the composer, so the
 // row stays one line.
 
+import type { KeyboardEvent } from 'react'
 import type { Context } from 'cordis'
 import { StateDot } from '@deepseek-ai/dsh-client-ui-primitives'
 import type { ToolRowProps } from '../contract/slots.ts'
@@ -46,8 +47,23 @@ export function TodoRow({ toolName, block, openDetails }: ToolRowProps) {
   const model = toolRowModel(toolName, block)
   const argsRaw = ('kind' in block ? block.call?.argsRaw : block.argsRaw) ?? ''
   const summary = summarize(argsRaw) ?? model.summary
+  // Button semantics, not a <button>: the row carries inline spans a button
+  // would flatten, and ToolRow takes the same role/tabIndex/Enter-Space route.
+  const openFromKeyboard = (event: KeyboardEvent<HTMLDivElement>) => {
+    if (event.key !== 'Enter' && event.key !== ' ') return
+    event.preventDefault()
+    openDetails()
+  }
   return (
-    <div className={css.row} data-sample="todo-row" data-state={model.state} onClick={openDetails}>
+    <div
+      className={css.row}
+      data-sample="todo-row"
+      data-state={model.state}
+      role="button"
+      tabIndex={0}
+      onClick={openDetails}
+      onKeyDown={openFromKeyboard}
+    >
       {model.state === 'ok'
         ? <span className={css.badge} aria-hidden>☰</span>
         : <StateDot state={model.state === 'running' ? 'ongoing' : model.state === 'stopped' ? 'warning' : 'error'} />}
diff --git a/packages/client/ui-conversation/tests/todo-panel.spec.tsx b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
index 5261e1d9e4..5bc3aa5c3d 100644
--- a/packages/client/ui-conversation/tests/todo-panel.spec.tsx
+++ b/packages/client/ui-conversation/tests/todo-panel.spec.tsx
@@ -1,27 +1,24 @@
 // @vitest-environment jsdom
 /**
  * Todo display acceptance: the TodoPanel plan strip (empty-hidden, status
- * rows, collapse with active hint) and the todo_write toolview row (progress
- * summary from args, generic fallback on malformed JSON, error badge).
+ * rows, collapse with active hint), its TodoDock adapter (selects the plan off
+ * the session snapshot and follows changes), and the todo_write toolview row
+ * (progress summary from args, generic fallback on malformed JSON, error badge,
+ * keyboard activation).
  */
 import { act, cleanup, fireEvent, render, screen } from '@testing-library/react'
 import { afterEach, describe, expect, it, vi } from 'vitest'
 import { bindSnapshotSelector } from '@deepseek-ai/dsh-client-web-react'
 import { createSnapshotStore } from '@deepseek-ai/dsh-client-runtime/client'
 import type { TodoItem, ToolResultNode } from '@deepseek-ai/dsh-client-runtime/client'
-import type { UseSession } from '@deepseek-ai/dsh-client-ui-slots'
 import type { ToolRowProps } from '@deepseek-ai/dsh-client-ui-conversation/client'
 // Export discipline: packages/client/AGENTS.md.
 import { TodoRow, todoToolview } from '../src/client/toolviews/todo-row.tsx'
-import { TodoPanel } from '../src/client/skeleton/TodoPanel.tsx'
+import type { TodoDockProps } from '../src/client/skeleton/TodoPanel.tsx'
+import { TodoDock, TodoPanel, todoDockEntry } from '../src/client/skeleton/TodoPanel.tsx'
 
 afterEach(cleanup)
 
-function sessionWith(todos: readonly TodoItem[]) {
-  const store = createSnapshotStore<{ todos: readonly TodoItem[] }>({ todos })
-  return { store, useSession: bindSnapshotSelector(store) as unknown as UseSession }
-}
-
 const LIST: TodoItem[] = [
   { content: '搭骨架', status: 'completed' },
   { content: '写组件', status: 'in_progress' },
@@ -29,17 +26,14 @@ const LIST: TodoItem[] = [
 ]
 
 describe('TodoPanel', () => {
-  it('renders nothing while the list is empty, appears when todos land', () => {
-    const { store, useSession } = sessionWith([])
-    render(<TodoPanel useSession={useSession} />)
-    expect(screen.queryByTestId('todo-panel')).toBeNull()
-    act(() => { store.set({ todos: LIST }) })
-    expect(screen.getByTestId('todo-panel')).toBeTruthy()
+  it('renders nothing while the list is empty', () => {
+    const { container } = render(<TodoPanel todos={[]} />)
+    expect(container.innerHTML).toBe('')
   })
 
   it('shows progress, one row per item with its status, and strikes done items', () => {
-    const { useSession } = sessionWith(LIST)
-    render(<TodoPanel useSession={useSession} />)
+    render(<TodoPanel todos={LIST} />)
+    expect(screen.getByTestId('todo-panel')).toBeTruthy()
     expect(screen.getByText('1/3')).toBeTruthy()
     const items = screen.getAllByRole('listitem')
     expect(items.map(li => li.getAttribute('data-status'))).toEqual(['completed', 'in_progress', 'pending'])
@@ -48,8 +42,7 @@ describe('TodoPanel', () => {
   })
 
   it('collapse hides the list and surfaces the active item in the header; expand restores', () => {
-    const { useSession } = sessionWith(LIST)
-    render(<TodoPanel useSession={useSession} />)
+    render(<TodoPanel todos={LIST} />)
     const header = screen.getByRole('button', { expanded: true })
     fireEvent.click(header)
     expect(screen.queryByRole('list')).toBeNull()
@@ -60,14 +53,39 @@ describe('TodoPanel', () => {
   })
 
   it('collapsed header omits the hint when nothing is in progress', () => {
-    const { useSession } = sessionWith([{ content: '都完了', status: 'completed' }])
-    render(<TodoPanel useSession={useSession} />)
+    render(<TodoPanel todos={[{ content: '都完了', status: 'completed' }]} />)
     fireEvent.click(screen.getByRole('button', { expanded: true }))
     expect(screen.queryByText('都完了')).toBeNull()
     expect(screen.getByText('1/1')).toBeTruthy()
   })
 })
 
+/** Dock props stub: the adapter reads useSession only; the rest of the owner share is unused. */
+function dockProps(store: ReturnType<typeof createSnapshotStore<{ todos: readonly TodoItem[] }>>): TodoDockProps {
+  return { useSession: bindSnapshotSelector(store) } as unknown as TodoDockProps
+}
+
+describe('TodoDock', () => {
+  it('selects the plan off the session snapshot and follows later writes', () => {
+    const store = createSnapshotStore<{ todos: readonly TodoItem[] }>({ todos: [] })
+    render(<TodoDock {...dockProps(store)} />)
+    expect(screen.queryByTestId('todo-panel')).toBeNull()
+    act(() => { store.set({ todos: LIST }) })
+    expect(screen.getByText('1/3')).toBeTruthy()
+    // A rollback to the empty list retires the strip (the panel owns no data).
+    act(() => { store.set({ todos: [] }) })
+    expect(screen.queryByTestId('todo-panel')).toBeNull()
+  })
+
+  it('ships the registrant plugin shape (list entry above the queue rows)', () => {
+    expect(todoDockEntry.name).toBe('conversation-todo-dock')
+    expect(todoDockEntry.inject).toEqual(['slots', 'conversation'])
+    const register = vi.fn()
+    todoDockEntry.apply({ slots: { register } } as never)
+    expect(register).toHaveBeenCalledWith({ name: 'conversation.input.dock', id: 'todo', order: -1 }, TodoDock)
+  })
+})
+
 const resultNode = (argsRaw: string, over?: Partial<ToolResultNode>): ToolResultNode => ({
   kind: 'tool-result', seq: 10, time: 2_000, callTime: 1_000, callId: 'c1',
   call: { name: 'todo_write', argsRaw },
@@ -126,6 +144,21 @@ describe('TodoRow', () => {
     expect(openDetails).toHaveBeenCalledTimes(1)
   })
 
+  it('opens details from the keyboard on Enter and Space, ignoring other keys', () => {
+    const openDetails = vi.fn()
+    render(<TodoRow {...rowProps(resultNode(ARGS), openDetails)} />)
+    const row = screen.getByRole('button')
+    expect(row.getAttribute('tabindex')).toBe('0')
+    fireEvent.keyDown(row, { key: 'Enter' })
+    fireEvent.keyDown(row, { key: ' ' })
+    expect(openDetails).toHaveBeenCalledTimes(2)
+    // Space must not also scroll the flow: the handler claims the event.
+    expect(fireEvent.keyDown(row, { key: ' ' })).toBe(false)
+    fireEvent.keyDown(row, { key: 'a' })
+    fireEvent.keyDown(row, { key: 'ArrowDown' })
+    expect(openDetails).toHaveBeenCalledTimes(3)
+  })
+
   it.each([
     { label: 'null root', argsRaw: 'null' },
     { label: 'non-object root', argsRaw: '42' },
diff --git a/packages/host/apiproxy/README.i18n.yaml b/packages/host/apiproxy/README.i18n.yaml
index eb06e14d2d..ecd0150f73 100644
--- a/packages/host/apiproxy/README.i18n.yaml
+++ b/packages/host/apiproxy/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: 43ad70fa8b865b0b80496bbb67013f24e9e3a33f
-README.zh.md: cc95a7512fb872add816bf0456a93dfcf7b84c10
+#   pnpm run verify-translation-pairing --write packages/host/apiproxy/README.md
+README.md: 775ceec5b2171b81977650b739b5b94662902dd5
+README.zh.md: d28d6cdad43b7c944f6bdf17e80e99494b39308f
diff --git a/packages/host/apiproxy/README.md b/packages/host/apiproxy/README.md
index 43ad70fa8b..775ceec5b2 100644
--- a/packages/host/apiproxy/README.md
+++ b/packages/host/apiproxy/README.md
@@ -14,6 +14,8 @@ The mux stream projects the latest log-backed title as a validated `session/titl
 
 Workspace and Session lists are separate reconnect baselines. `workspace.create` creates a unique name or adopts an existing directory, `session.create` accepts an optional preallocated Session id, and `host/workspace-changed` plus `host/session-added` carry committed increments in either arrival order. `SessionSummary.blank` and the `host/session-added` frame carry the derived zero-events bit: clients hide blank sessions and reuse them per workspace, flip blank on the first `host/session-status(running:true)`, and treat `session.list` as the reconnect authority; cold summaries are never blank because lazy persistence keeps never-appended sessions out of `list()`.
 
+`session.history` pages on message boundaries, and its tail page (no `beforeSeq`) carries two session-level extras the page window cannot supply: the in-flight partial's chunk events, and `todos` — the latest `todo/write` whole-list projection over the full log. Older pages omit `todos` because the projection is session-level, not per-page; a tail response that omits it means the whole log holds no `todo/write`, so clients read the absent field as the empty plan rather than as unchanged state.
+
 The `command.*` and `skill.*` domains expose the host command registry and skill catalog to clients. Every method addresses one session's agent by `sessionId` (a served session always has an Agent; `command.*` resumes cold sessions through the same path as `session.*`, while `skill.list` resolves the project root from the session header without touching the Agent registry). `command.execute` runs a slash-command line host-side and returns a detached result; the carrier's request signal cancels the running handler. `host/commands-changed` is the catalog invalidation frame: clients refetch `command.list` instead of diffing.
 
 ## Carrier layer (`/client` + root)
diff --git a/packages/host/apiproxy/README.zh.md b/packages/host/apiproxy/README.zh.md
index cc95a7512f..d28d6cdad4 100644
--- a/packages/host/apiproxy/README.zh.md
+++ b/packages/host/apiproxy/README.zh.md
@@ -14,6 +14,8 @@ mux 流会在每个已附加会话的订阅基线之后，以及对应的实时
 
 Workspace 列表与 Session 列表是相互独立的重连基线。`workspace.create` 会创建唯一名称或接纳现有目录，`session.create` 接受可选的预分配 Session id，`host/workspace-changed` 与 `host/session-added` 则以任意到达顺序携带已提交的增量。`SessionSummary.blank` 与 `host/session-added` 帧携带派生的零事件位：客户端隐藏空白会话并按 workspace 复用它们，在首个 `host/session-status(running:true)` 时翻转 blank，并以 `session.list` 作为重连权威；冷会话摘要永远不是空白——惰性持久化让从未追加过事件的会话根本不出现在 `list()` 中。
 
+`session.history` 按消息边界分页，其尾页（不带 `beforeSeq`）额外携带两项页窗口本身无法提供的会话级数据：进行中局部消息的 chunk 事件，以及 `todos`——整份日志上最后一次 `todo/write` 的整表投影。较早的页面不带 `todos`，因为该投影是会话级而非分页级的；尾页响应缺少该字段意味着整份日志中没有任何 `todo/write`，因此客户端要把缺失字段读作空计划，而不是读作「状态未变」。
+
 `command.*` 与 `skill.*` 领域向客户端暴露宿主命令注册表和技能目录。每个方法都通过 `sessionId` 寻址一个会话的 Agent（被服务的会话必有 Agent；`command.*` 经由与 `session.*` 相同的路径恢复冷会话，而 `skill.list` 从会话头解析项目根目录，不触碰 Agent 注册表）。`command.execute` 在宿主侧运行一条斜杠命令行并返回脱耦结果；载体的请求信号可取消正在运行的处理器。`host/commands-changed` 是目录失效帧：客户端重新拉取 `command.list` 而不是做差分。
 
 ## 载体层（`/client` + 根路径）

From 710a3b1686eac0e0d1bb1574771ef079847d45b4 Mon Sep 17 00:00:00 2001
From: Chinesezjc <jczhai@deepseek.com>
Date: Mon, 27 Jul 2026 12:20:17 +0800
Subject: [PATCH 36/36] docs(ui-conversation): document the two todo
 registrations in the owning README pair

The todo row and the todo plan strip were registered without a contract in this
package's README, which still described only the bash sample. Both sides now
carry the two registrations: TodoRow's `todo_write` toolview key with its
attempted-write summary and non-ok dot semantics, and TodoDock's
`conversation.input.dock` entry at `order: -1` with the dock-owned selection,
empty-hidden panel, collapse behavior, and the composer-takeover consequence
that hides the whole dock. README.i18n.yaml re-recorded.
---
 packages/client/ui-conversation/README.i18n.yaml | 6 +++---
 packages/client/ui-conversation/README.md        | 4 +++-
 packages/client/ui-conversation/README.zh.md     | 4 +++-
 3 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/packages/client/ui-conversation/README.i18n.yaml b/packages/client/ui-conversation/README.i18n.yaml
index 8117d42ea4..2ffa02d313 100644
--- a/packages/client/ui-conversation/README.i18n.yaml
+++ b/packages/client/ui-conversation/README.i18n.yaml
@@ -1,6 +1,6 @@
 # Bilingual-pair consistency record (docs/i18n/README.md): the git blob hash of each
 # side as of the last confirmed-consistent state. Both languages carry equal authority;
 # after editing either side, bring the other along and re-record with:
-#   pnpm run verify-translation-pairing --write
-README.md: b9ec555f158722ea1f41e01c4b3f7131d3fe3467
-README.zh.md: b1e3c1f4331148ebf1c58b4bcd4869270bb44311
+#   pnpm run verify-translation-pairing --write packages/client/ui-conversation/README.md
+README.md: b242812411d513931ecd2767622f9e23fb0aaa34
+README.zh.md: 77f68e02d8d9161c413ae7d224121bc53547ba12
diff --git a/packages/client/ui-conversation/README.md b/packages/client/ui-conversation/README.md
index b9ec555f15..b242812411 100644
--- a/packages/client/ui-conversation/README.md
+++ b/packages/client/ui-conversation/README.md
@@ -2,7 +2,7 @@
 
 English | [中文](README.zh.md)
 
-Conversation domain: skeleton (header/tabs/composer/empty state), chat view (grouped step-summary flow, streaming tail isolation, stats line, per-tool row slot with a bash sample registrant), minimal details panel, scope-addressed ConversationService. Contract: api-contracts v3 §7 plus the slot terminal design (store seat / props shares).
+Conversation domain: skeleton (header/tabs/composer/empty state), chat view (grouped step-summary flow, streaming tail isolation, stats line, per-tool row slot with a bash sample registrant and the todo row), input dock (queue rows plus the todo plan strip), minimal details panel, scope-addressed ConversationService. Contract: api-contracts v3 §7 plus the slot terminal design (store seat / props shares).
 
 The no-session hero renders the frontend Session Intent from the Session list projection, including its frontend Workspace Intent when no real Workspace exists. It declares `conversation.empty.workspace`, where ui-workspace registers the same picker used by the sidebar. WorkspacesService starts the cross-object flow; each Workspace or Session object owns its own materialization. The Session keeps its identity across publication and retains any prompt that still needs connection or delivery; ConversationRoot reads that `pendingPrompt` from `useSession` and edits or retries it through the scoped ConversationService.
 
@@ -12,6 +12,8 @@ Generic tool rows classify the built-in bash, read, search, write, edit, and run
 
 Tool rows are slots too — the standalone tool ring (`ToolViewRegistry`/`ctx.toolviews`/outlet) is retired. The chat entry declares the keyed `'conversation.chat.toolview'` hole (session scope; the key space is runtime-open); its render site dispatches per row via `entryKey: toolName` with `GenericToolCard` as the call-site `fallback`. The owner payload is the uniform `ToolRowOwnerProps` (`callId`/`toolName`/`block`/`openDetails`) and `ToolRowProps` pre-composes it with the session standard kit. A registrant is a plain plugin: `ctx.slots.register({ name: 'conversation.chat.toolview', key: '<tool>', inject? }, Row)` with `inject: ['slots', 'conversation']` as the load-order seam (apply mounts ConversationService after the chat registration, so the service being present guarantees the slot is declared); session differentiation happens inside the component (`useSessions` reading `parentId` — the bash sample is the third-party-posture exemplar). Trajectory/waterfall toolview slots share this shape and land with their own render sites (RendersCheck rejects a declaration nobody renders).
 
+The todo surfaces are two registrations over that shape, both plain registrant plugins with `inject: ['slots', 'conversation']`. `TodoRow` takes the `'conversation.chat.toolview'` key `todo_write` and summarizes what the call attempted (`<done>/<total> 已完成 · <active item>` parsed from its args, falling back to the generic summary on malformed or wrongly-shaped model JSON, and keeping the generic dot for non-ok execution states so a cancelled call never reads as a completed update). `TodoDock` takes the `'conversation.input.dock'` list slot at `order: -1` — above the queue rows — and is the durable plan strip: it selects `todos` off the session snapshot and renders `TodoPanel`, which takes the plain list, hides itself while the list is empty, and collapses to a one-line header carrying the in-progress item. The dock adapter owns the selection so the panel stays a pure function of its props; the persistent list lives here rather than in the row so the row stays one line. Anything the input-zone composer chain hides (a `conversation.composer` takeover such as ui-question's) hides the whole dock, this strip included.
+
 Per-session UI state (selection, ordinary composer draft, active view) lives in the declared chat store (`stores.ts` `createChatStore`): apply constructs one handle and passes it to the conversation, chat-view, and details registrations, so the session slots share one instance per session (selection written by the chat view, read by details) and the framework owns instance lifecycle and draft persistence. The frontend Session Intent comes from the Session list projection; after publication, any retained prompt comes from that Session's conversation snapshot. Components are pure — the framework standard kit (`useSession`/`sessionId` when session-scoped, plus global `useSessions`/`useWorkspaces`) and the store faces (`useStore`/`actions`) arrive automatically from the registration declaration; inject factories contribute plain data and callbacks for runtime Session actions, send/stop, tabs, details, and paging.
 
 `src/client/` is organized for the future package split: `contract/` is the sole inter-domain shared face (`slots.ts` slot declarations + composed slot props including the tool-row contract, `views.ts` shared primitives, `tool-call-model.ts`); the `skeleton/`, `chat/`, and `toolviews/` (sample registrants) domain directories import contract files and never each other; `apply.ts` is the only assembly point allowed to import all three domains. The `/client` export surface is the contract only — `apply`/`inject`, the two service classes, and the `contract/` type families; implementation components (skeleton, chat rows) and the store factory stay internal and reach the page exclusively through apply's slot registrations (tests take them via the `./src/*` subpath).
diff --git a/packages/client/ui-conversation/README.zh.md b/packages/client/ui-conversation/README.zh.md
index b1e3c1f433..77f68e02d8 100644
--- a/packages/client/ui-conversation/README.zh.md
+++ b/packages/client/ui-conversation/README.zh.md
@@ -2,7 +2,7 @@
 
 [English](README.md) | 中文
 
-会话领域：骨架（标题栏／标签页／编辑器／空状态）、聊天视图（分组步骤摘要流、流式尾部隔离、统计行、逐工具行 slot 及一个 bash 示例注册方）、最小详情面板、按 scope 寻址的 ConversationService。契约：api-contracts v3 §7 加 slot 终端设计（store seat／props share）。
+会话领域：骨架（标题栏／标签页／编辑器／空状态）、聊天视图（分组步骤摘要流、流式尾部隔离、统计行、逐工具行 slot 及一个 bash 示例注册方与 todo 行）、输入区 dock（队列行加 todo 计划条）、最小详情面板、按 scope 寻址的 ConversationService。契约：api-contracts v3 §7 加 slot 终端设计（store seat／props share）。
 
 无会话主视觉区会渲染来自 Session 列表投影的前端 Session Intent；没有真实 Workspace 时，还会包含其前端 Workspace Intent。它声明 `conversation.empty.workspace`，ui-workspace 会在此注册侧边栏所用的同一选择器。WorkspacesService 启动跨对象流程；每个 Workspace 或 Session 对象拥有自身的物化。Session 在发布期间保持身份，并保留任何仍需连接或交付的提示词；ConversationRoot 读取该 `pendingPrompt`，其来源是 `useSession`，再通过 scope 内的 ConversationService 编辑或重试。
 
@@ -12,6 +12,8 @@
 
 工具行同样是 slot：独立工具环（`ToolViewRegistry`／`ctx.toolviews`／outlet）已经退役。聊天配置项声明键控的 `'conversation.chat.toolview'` 空位（Session scope；key 空间在运行时开放）；其渲染点逐行通过 `entryKey: toolName` 分发，并以 `GenericToolCard` 作为调用点 `fallback`。owner 载荷是统一的 `ToolRowOwnerProps`（`callId`／`toolName`／`block`／`openDetails`），`ToolRowProps` 则预先将其与 Session 标准工具包组合。注册方只是普通插件：`ctx.slots.register({ name: 'conversation.chat.toolview', key: '<tool>', inject? }, Row)`，以 `inject: ['slots', 'conversation']` 作为加载顺序 seam（apply 在聊天注册后挂载 ConversationService，因此服务存在即可保证 slot 已声明）；Session 区分在组件内部完成（`useSessions` 读取 `parentId`，bash 示例是第三方姿态的范例）。Trajectory/waterfall 工具视图 slot 共享此形状，并随各自的渲染点落地（RendersCheck 会拒绝没有任何渲染方的声明）。
 
+todo 两个面就是在该形状上的两个注册项，都是普通注册方插件，`inject: ['slots', 'conversation']`。`TodoRow` 占用 `'conversation.chat.toolview'` 的 `todo_write` key，摘要该次调用「试图写入」的内容（从其 args 解析出 `<已完成>/<总数> 已完成 · <进行中条目>`；模型 JSON 残缺或形状不对时回落到通用摘要；非 ok 执行状态保留通用状态点，使被取消的调用绝不读成一次已完成的更新）。`TodoDock` 以 `order: -1` 占用 `'conversation.input.dock'` 列表 slot（位于队列行之上），是常驻的计划条：它从会话快照中选取 `todos` 并渲染 `TodoPanel`，后者接收纯列表，在列表为空时自我隐藏，折叠时收成携带进行中条目的单行表头。选取由 dock 适配器负责，因此面板保持为其 props 的纯函数；常驻列表放在此处而非行内，行才能保持单行。输入区 composer 链隐藏的一切（例如 ui-question 对 `conversation.composer` 的接管）也会隐藏整个 dock，包括这条计划条。
+
 逐 Session UI 状态（选择、普通编辑器草稿、活跃视图）位于已声明的聊天 store（`stores.ts` `createChatStore`）中：apply 构造一个 handle，并将其传给会话、聊天视图和详情注册，因此 Session slot 每个 Session 共享一个实例（选择由聊天视图写入、详情读取），框架拥有实例生命周期与草稿持久化。前端 Session Intent 来自 Session 列表投影；发布后，任何保留的提示词都来自该 Session 的会话快照。组件保持纯粹：框架标准工具包（Session scope 下的 `useSession`／`sessionId`，以及全局 `useSessions`／`useWorkspaces`）和 store 表层（`useStore`／`actions`）会从注册声明自动到达；inject factory 为运行时 Session 操作、发送／停止、标签页、详情和分页贡献普通数据与回调。
 
 `src/client/` 按未来的包拆分组织：`contract/` 是唯一的跨领域共享表层（`slots.ts` slot 声明 + 组合后的 slot props，包括工具行契约、`views.ts` 共享原语、`tool-call-model.ts`）；`skeleton/`、`chat/` 和 `toolviews/`（示例注册方）领域目录只导入 contract 文件，彼此绝不导入；`apply.ts` 是唯一允许导入全部三个领域的组装点。`/client` 导出表层只包含契约：`apply`／`inject`、两个服务类和 `contract/` 类型家族；实现组件（骨架、聊天行）与 store factory 保持内部状态，只能通过 apply 的 slot 注册到达页面（测试通过 `./src/*` 子路径获取它们）。