🌐 chore: translate non-English comments to English in src-chat-and-helpers

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/features/WorkspaceSetting/hooks/useCategory.tsx

@@ -113,7 +113,7 @@ export const useWorkspaceSettingCategory = (): WorkspaceSettingCategoryGroup[] =
             key: WorkspaceSettingsTabs.Creds,
             label: t('tab.creds'),
           },
-          // Messenger (聊天平台) is intentionally omitted from workspace
+          // Messenger (chat platform) is intentionally omitted from workspace
           // settings: the System Bot binding is a per-user/personal identity
           // (the link is owned by `userId`, not the workspace), and reaching a
           // workspace's agents happens via the scope selector on the *personal*

src/helpers/executionTarget.ts

@@ -19,10 +19,10 @@ export interface ResolveExecutionTargetOptions {
  * Single source of truth for where an agent executes — one global
  * `agencyConfig.executionTarget` drives both desktop and web.
  *
- * - `none`    → 无设备 (no execution environment; plain chat)
- * - `local`   → 本机 (this machine, in-process; desktop only)
- * - `sandbox` → 云端沙箱 (server cloud sandbox)
- * - `device`  → 远程设备 (dispatched to `boundDeviceId`)
+ * - `none`    → no execution environment (plain chat)
+ * - `local`   → local machine (this machine, in-process; desktop only)
+ * - `sandbox` → cloud sandbox (server cloud sandbox)
+ * - `device`  → remote device (dispatched to `boundDeviceId`)
  *
  * `local` and `device` stay DISTINCT even when the bound device is this very
  * machine: `device` dispatches through the server gateway, so progress streams
@@ -102,7 +102,7 @@ export type ExecutionPlanUnroutedReason =
  * read it instead of re-resolving `agencyConfig.executionTarget`.
  */
 export type ExecutionPlan = { target: DeviceExecutionTarget } &
-  /** route execution / device tools to this device (includes 本机 — the local machine is a registered device) */
+  /** route execution / device tools to this device (includes local machine — the local machine is a registered device) */
   (| { deviceId: string; kind: 'device' }
     /**
      * Device-targeted but no routable device right now. The run proceeds without

src/store/chat/slices/aiChat/actions/gateway.ts

@@ -22,7 +22,7 @@ import { useUserStore } from '@/store/user';
 import { createGatewayEventHandler } from './gatewayEventHandler';
 
 /**
- * When the agent runs against the local machine ("本机"), resolve this desktop's
+ * When the agent runs against the local machine, resolve this desktop's
  * own gateway deviceId so it can be passed as the run's `deviceId`. The server
  * then presets `activeDeviceId` and injects `lobe-local-system` into the very
  * first LLM payload — skipping the extra `activateDevice` round-trip the model

src/store/chat/slices/aiChat/actions/heterogeneousAgentExecutor.ts

@@ -627,7 +627,7 @@ export const executeHeterogeneousAgent = async (
    * `toolState.payloads` reset that happens on every new step.
    *
    * Required for the **toolless middle step** case (): when a step
-   * produces only text (e.g. Monitor stdout drives Claude to reply "等一下…"
+   * produces only text (e.g. Monitor stdout drives Claude to reply "wait a moment…"
    * without invoking a tool), `toolState.payloads` is empty at the next step
    * boundary. Without this tracker, `stepParentId` would fall back to
    * `currentAssistantMessageId` (= the toolless assistant), forming an
@@ -1435,7 +1435,7 @@ export const executeHeterogeneousAgent = async (
       // — otherwise the handler reads `assistant.tools[]` while a parallel
       // `persistToolBatch` is still mid-flight and `replaceMessages` clobbers
       // the in-memory cumulative tools[] with a shorter snapshot. That's the
-      // "7 → 6 次技能调用" rollback users see on parallel CC tool batches.
+      // "7 → 6 skill calls" rollback users see on parallel CC tool batches.
       //
       // Other forwards (text / reasoning / tools_calling dispatches) stay
       // synchronous so live streaming UX isn't gated on DB round-trips.

src/store/chat/slices/message/utils/reconcileTools.ts

@@ -16,7 +16,7 @@ import type { ChatToolPayload, UIChatMessage } from '@lobechat/types';
  * `replaceMessages` (stale / out-of-order DB snapshot) or an optimistic
  * `internal_dispatchMessage updateMessage{tools}` can momentarily drop an
  * assistant's `tools[]` while the tool row + parentId survive (the extreme of
- * the "7→6 次技能调用" tools[] regression).
+ * the "7→6 skill-calls" tools[] regression).
  *
  * Fixing this at the RAW bucket write boundary — not only inside `parse` — keeps
  * `dbMessagesMap` (the Source of Truth that optimistic updates read & mutate)