fix(client): add /l vanity route to SW denylist; name failed offline steps (F2, F3)

F2: navigateFallbackDenylist was missing the server's `l/:alias` vanity
short-link, so a top-nav to /l/<alias> after SW registration got the
index.html app shell (which has no /l route) and dead-ended on Error404
instead of the server's 302 redirect. Add /^\/l(\/|$)/ mirroring main.ts.

F3: the partial-failure branch of "make page available offline" showed a
bare generic toast; include result.failed step labels in the message per
AGENTS.md (errors must be specific), matching the catch-branch below it.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.env.example

@@ -92,6 +92,19 @@ IFRAME_EMBED_ALLOWED=false
 # Example: https://intranet.example.com,https://portal.example.com
 IFRAME_ALLOWED_ORIGINS=
 
+# Comma-separated list of additional origins allowed to call the API via CORS.
+# The APP_URL origin and native mobile (Capacitor) origins are always allowed.
+# Leave empty for a same-origin (web-only) deployment.
+CORS_ALLOWED_ORIGINS=
+
+# Expose OpenAPI/Swagger docs at /api/docs (development/debugging aid only).
+SWAGGER_ENABLED=false
+
+# Capacitor (mobile shell): hosted client URL loaded by the iOS shell so the
+# AGPL web client is NOT bundled into the .ipa (see docs/mobile-app-plan.md §9).
+# Leave empty for Android bundled mode / local development.
+CAP_SERVER_URL=
+
 # Enable debug logging in production (default: false)
 DEBUG_MODE=false
 
@@ -170,20 +183,6 @@ MCP_DOCMOST_PASSWORD=
 # Default 900000 (15 min).
 # AI_MCP_CALL_TIMEOUT_MS=900000
 
-# --- Autonomous / detached agent runs (settings.ai.autonomousRuns) ---
-# Opt-in per workspace (AI settings; off by default). When on, a chat turn becomes
-# a server-side RUN that survives a browser disconnect — only an explicit Stop ends
-# it, and a client reconnects/live-follows the run.
-#
-# DEPLOY CONSTRAINT — SINGLE-INSTANCE ONLY in phase 1: Stop and the in-process
-# AbortController that backs it are process-local, so a Stop only aborts a run
-# executing on the SAME replica that owns it (cross-instance pub/sub stop is phase
-# 2 and not yet reliable). Do NOT enable autonomousRuns on a horizontally-scaled
-# deployment (multiple replicas behind a load balancer, or Docmost cloud
-# CLOUD=true) — run a single instance instead. The server logs a startup WARNING
-# when it detects a multi-instance deployment (CLOUD=true) so the constraint is
-# visible, and a startup sweep settles any run left dangling by a restart.
-
 # --- Anonymous public-share AI assistant ---
 # Opt-in per workspace (AI settings -> "public share assistant"; off by default).
 # When enabled, anonymous visitors of a published share can ask an AI about that

.gitignore

@@ -49,3 +49,8 @@ lerna-debug.log*
 
 # Self-hosted VAD / onnxruntime-web assets (copied from node_modules at dev/build time)
 apps/client/public/vad/
+
+# Capacitor native platform projects (generated locally via 'npx cap add ios|android')
+/ios
+/android
+.capacitor

AGENTS.md

@@ -259,7 +259,6 @@ The API server is a Fastify app with a global `/api` prefix (`main.ts` excludes
    - `core/ai-chat/tools/` — the agent's ~40 read+write tools. Every tool runs under the **calling user's** CASL permissions via a per-user loopback access token (`docmost-client.loader.ts`), so the agent can never exceed what the user could do. Only **reversible** operations are exposed (page history + trash; no permanent delete). Agent edits get an "AI agent" provenance badge in page history (`20260616T130000-agent-provenance` migration).
    - `core/ai-chat/embedding/` — RAG indexer + a BullMQ consumer on `AI_QUEUE` that embeds pages into `page_embeddings` (vector search), complementing Postgres full-text search. Pages are (re)indexed on edit; `AI_EMBEDDING_TIMEOUT_MS` bounds a hung embeddings endpoint.
    - `core/ai-chat/external-mcp/` — admins can attach external MCP servers (e.g. Tavily) to give the agent web access. **`ssrf-guard.ts` validates outbound MCP URLs against SSRF** — keep that guard in the path when touching external-MCP connection logic.
-   - `core/ai-chat/ai-chat-run.service.ts` + `ai_chat_runs` — **detached/autonomous agent runs** (`#184`), behind the per-workspace `settings.ai.autonomousRuns` flag (off by default). When on, a turn becomes a server-side RUN that survives a browser disconnect; only an explicit `POST /ai-chat/stop` ends it, and a client reconnects/live-follows via `POST /ai-chat/run`. **DEPLOY CONSTRAINT — single-instance only in phase 1:** Stop and the AbortController that backs it are process-local, so a Stop only aborts a run executing on the **same** replica that owns it (cross-instance pub/sub stop is phase 2). Do **not** enable `autonomousRuns` on a horizontally-scaled deployment (multiple replicas behind a load balancer, or Docmost cloud `CLOUD=true`) — run a single instance instead. The server logs a startup WARNING when it detects a multi-instance deployment (`CLOUD=true`) so the constraint is visible. The startup sweep settles any run left dangling by a restart.
 
 ### Client structure
 Vite SPA. Code is organized by feature under `apps/client/src/features/*` (mirrors the server domains: `page`, `space`, `comment`, `ai-chat`, `editor`, …). Conventions:

CHANGELOG.md

@@ -58,19 +58,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   append/prepend fragments, nor to COMMENT bodies — a comment may legitimately
   contain a standalone footnote definition, which canonicalization would drop.
   (#228)
-- **Detached, autonomous agent runs that survive a browser disconnect.** When the
-  new `settings.ai.autonomousRuns` workspace flag is on (off by default), an
-  AI-chat turn becomes a first-class, server-side RUN tracked in a new
-  `ai_chat_runs` table instead of a socket-bound stream: closing the tab or
-  losing the connection no longer aborts the turn — it keeps executing and
-  persisting server-side, and only an explicit Stop ends it. A client can
-  reconnect and live-follow (or stop) an in-flight run via `POST /ai-chat/run`
-  (resolve the latest run + its assistant message for a chat) and
-  `POST /ai-chat/stop` (stop by `runId` or `chatId`). A partial unique index
-  enforces one active run per chat, and a startup sweep settles any run left
-  dangling by a restart. Phase 1 is single-instance-only (cross-instance Stop is
-  not yet reliable); the server warns at startup on a horizontally-scaled
-  deployment. (#184)
 
 ### Changed
 
@@ -80,6 +67,33 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   toggle. Previously the create call defaulted to including sub-pages, silently
   exposing every child of a freshly shared page. (#216)
 
+- **Offline reading support**: opened pages, their sidebar tree, breadcrumb
+  children, and comments are cached in IndexedDB (TanStack Query persister plus
+  `y-indexeddb` for the page's Yjs document), and a PWA service worker
+  (vite-plugin-pwa) serves an app shell so previously opened pages stay readable
+  offline. The two offline stores (the persisted query cache and the Yjs page
+  documents) are cleared on logout AND on sign-in so a previous user's private
+  data does not remain in the browser; the same purge also defensively drops any
+  legacy service-worker `api-get-cache` left by older clients (current builds
+  serve `/api` as NetworkOnly, so there is no active service-worker API cache).
+- **Mobile bootstrap**: a `returnToken` opt-in on login so native/mobile clients
+  can request the access JWT in the response body (`data.authToken`) in addition
+  to the httpOnly cookie (the web client stays cookie-only); an optional
+  OpenAPI/Swagger UI at `/api/docs` gated by `SWAGGER_ENABLED` (off by default);
+  and new env vars `CORS_ALLOWED_ORIGINS`, `SWAGGER_ENABLED`, `CAP_SERVER_URL`.
+
+### Changed
+
+- **CORS is now an explicit allowlist** (replaces the previous unconfigured
+  `app.enableCors()`). The same-origin web client is unaffected, but any
+  separately-hosted cross-domain client must now be listed in
+  `CORS_ALLOWED_ORIGINS` (native Capacitor/Ionic/localhost WebView origins are
+  allowed automatically). Requests with no `Origin` header (server-to-server)
+  are still allowed. **Upgrade note:** the old bare `app.enableCors()` reflected
+  *any* origin (with `credentials:false`), so any previously-working cross-domain
+  REST/browser client is now rejected until its origin is added to
+  `CORS_ALLOWED_ORIGINS` (see `.env.example`).
+
 ### Fixed
 
 - **Internal links in exported Markdown no longer lose their visible text.** A

apps/client/index.html

@@ -10,6 +10,7 @@
     <meta name="theme-color" content="#0E1117" media="(prefers-color-scheme: dark)" />
     <meta name="theme-color" content="#f6f7f9" media="(prefers-color-scheme: light)" />
     <link rel="manifest" href="/manifest.json" />
+    <link rel="apple-touch-icon" href="/icons/app-icon-192x192.png" />
     <meta name="mobile-web-app-capable" content="yes" />
     <meta name="apple-touch-fullscreen" content="yes" />
     <meta name="apple-mobile-web-app-title" content="Gitmost" />

apps/client/package.json

@@ -33,7 +33,9 @@
     "@slidoapp/emoji-mart-data": "1.2.4",
     "@slidoapp/emoji-mart-react": "1.1.5",
     "@tabler/icons-react": "3.40.0",
+    "@tanstack/query-async-storage-persister": "5.90.17",
     "@tanstack/react-query": "5.90.17",
+    "@tanstack/react-query-persist-client": "5.90.17",
     "@tanstack/react-virtual": "3.13.24",
     "ai": "6.0.207",
     "alfaaz": "1.1.0",
@@ -45,6 +47,7 @@
     "highlightjs-sap-abap": "0.3.0",
     "i18next": "25.10.1",
     "i18next-http-backend": "3.0.6",
+    "idb-keyval": "6.2.5",
     "jotai": "2.18.1",
     "jotai-optics": "0.4.0",
     "js-cookie": "3.0.7",
@@ -95,6 +98,7 @@
     "typescript": "5.9.3",
     "typescript-eslint": "8.57.1",
     "vite": "8.0.5",
+    "vite-plugin-pwa": "1.3.0",
     "vitest": "4.1.6"
   }
 }

apps/client/public/locales/en-US/translation.json

@@ -464,6 +464,18 @@
   "Move page": "Move page",
   "Move page to a different space.": "Move page to a different space.",
   "Real-time editor connection lost. Retrying...": "Real-time editor connection lost. Retrying...",
+  "Offline — changes are saved locally and will sync when you reconnect": "Offline — changes are saved locally and will sync when you reconnect",
+  "Syncing changes…": "Syncing changes…",
+  "All changes synced": "All changes synced",
+  "Update available": "Update available",
+  "Reload": "Reload",
+  "Make available offline": "Make available offline",
+  "Saving page for offline use...": "Saving page for offline use...",
+  "Page is now available offline": "Page is now available offline",
+  "Failed to make page available offline": "Failed to make page available offline",
+  "You're offline": "You're offline",
+  "This page hasn't been saved for offline use, so it can't be loaded right now. Reconnect to the internet and try again.": "This page hasn't been saved for offline use, so it can't be loaded right now. Reconnect to the internet and try again.",
+  "Retry": "Retry",
   "Table of contents": "Table of contents",
   "Add headings (H1, H2, H3) to generate a table of contents.": "Add headings (H1, H2, H3) to generate a table of contents.",
   "Share": "Share",

apps/client/public/locales/ru-RU/translation.json

@@ -474,6 +474,18 @@
   "Move page": "Переместить страницу",
   "Move page to a different space.": "Переместите страницу в другое пространство.",
   "Real-time editor connection lost. Retrying...": "Соединение с редактором в реальном времени потеряно. Повторная попытка...",
+  "Offline — changes are saved locally and will sync when you reconnect": "Нет сети — изменения сохраняются локально и синхронизируются при восстановлении соединения",
+  "Syncing changes…": "Синхронизация изменений…",
+  "All changes synced": "Все изменения синхронизированы",
+  "Update available": "Доступно обновление",
+  "Reload": "Перезагрузить",
+  "Make available offline": "Сделать доступным офлайн",
+  "Saving page for offline use...": "Сохраняем страницу для офлайн-доступа…",
+  "Page is now available offline": "Страница доступна офлайн",
+  "Failed to make page available offline": "Не удалось сделать страницу доступной офлайн",
+  "You're offline": "Вы офлайн",
+  "This page hasn't been saved for offline use, so it can't be loaded right now. Reconnect to the internet and try again.": "Эта страница не была сохранена для офлайн-доступа, поэтому её нельзя загрузить сейчас. Подключитесь к интернету и попробуйте снова.",
+  "Retry": "Повторить",
   "Table of contents": "Оглавление",
   "Add headings (H1, H2, H3) to generate a table of contents.": "Добавьте заголовки (H1, H2, H3), чтобы создать оглавление.",
   "Share": "Поделиться",

apps/client/public/manifest.json

@@ -1,30 +1,19 @@
 {
+  "id": "/",
   "name": "Gitmost",
   "short_name": "Gitmost",
+  "description": "Gitmost - open-source collaborative documentation and knowledge base.",
+  "lang": "en",
   "start_url": "/",
+  "scope": "/",
   "display": "standalone",
+  "orientation": "any",
   "background_color": "#0E1117",
   "theme_color": "#0E1117",
   "icons": [
-    {
-      "src": "icons/favicon-16x16.png",
-      "type": "image/png",
-      "sizes": "16x16"
-    },
-    {
-      "src": "icons/favicon-32x32.png",
-      "type": "image/png",
-      "sizes": "32x32"
-    },
-    {
-      "src": "icons/app-icon-192x192.png",
-      "type": "image/png",
-      "sizes": "180x180 192x192"
-    },
-    {
-      "src": "icons/app-icon-512x512.png",
-      "type": "image/png",
-      "sizes": "512x512"
-    }
+    { "src": "icons/favicon-16x16.png", "type": "image/png", "sizes": "16x16" },
+    { "src": "icons/favicon-32x32.png", "type": "image/png", "sizes": "32x32" },
+    { "src": "icons/app-icon-192x192.png", "type": "image/png", "sizes": "192x192", "purpose": "any" },
+    { "src": "icons/app-icon-512x512.png", "type": "image/png", "sizes": "512x512", "purpose": "any" }
   ]
 }

apps/client/src/features/ai-chat/components/ai-chat-window.tsx

@@ -17,7 +17,7 @@ import {
   IconPlus,
   IconX,
 } from "@tabler/icons-react";
-import { useAtom, useAtomValue, useSetAtom } from "jotai";
+import { useAtom, useSetAtom } from "jotai";
 import { useMatch } from "react-router-dom";
 import { useTranslation } from "react-i18next";
 import { useQueryClient } from "@tanstack/react-query";
@@ -34,12 +34,9 @@ import {
   AI_CHATS_RQ_KEY,
   AI_CHAT_MESSAGES_RQ_KEY,
   useAiChatMessagesQuery,
-  useAiChatRunQuery,
   useAiChatsQuery,
   useAiRolesQuery,
 } from "@/features/ai-chat/queries/ai-chat-query.ts";
-import { shouldObserveRun } from "@/features/ai-chat/utils/run-polling.ts";
-import { workspaceAtom } from "@/features/user/atoms/current-user-atom";
 import ConversationList from "@/features/ai-chat/components/conversation-list.tsx";
 import ChatThread from "@/features/ai-chat/components/chat-thread.tsx";
 import { exportAiChat } from "@/features/ai-chat/services/ai-chat-service.ts";
@@ -165,61 +162,6 @@ export default function AiChatWindow() {
   const { data: messageRows, isLoading: messagesLoading } =
     useAiChatMessagesQuery(activeChatId ?? undefined);
 
-  // #184 reconnect-and-live-follow. Whether detached agent runs are enabled for
-  // this workspace. The reconnect endpoint itself is NOT flag-gated server-side
-  // (it is only owner-gated and returns `{ run: null }` when the chat has no
-  // run); but when the feature is off no runs are ever created, so polling it
-  // would always come back empty — we gate it off here to avoid pointless polls.
-  const workspace = useAtomValue(workspaceAtom);
-  const autonomousRunsEnabled =
-    workspace?.settings?.ai?.autonomousRuns === true;
-
-  // Whether THIS tab is the one actively streaming the open chat's run locally
-  // (it started the run here and holds the SSE). Reported up from ChatThread. We
-  // are the STREAMER while true and a passive OBSERVER while false — the basis of
-  // the observer-vs-streamer detection. Reset to false by the fresh ChatThread's
-  // mount effect on every chat switch.
-  const [localStreaming, setLocalStreaming] = useState(false);
-  const onStreamingChange = useCallback((streaming: boolean) => {
-    setLocalStreaming(streaming);
-  }, []);
-
-  // Poll the latest run of the open chat ONLY when we are a passive observer:
-  // feature on, a chat is open, and we are NOT the local streamer (the streamer
-  // already has the live SSE — polling/merging too would double-render). The
-  // query's own status-keyed refetchInterval stops once the run is terminal.
-  const { data: runData } = useAiChatRunQuery(
-    activeChatId ?? undefined,
-    autonomousRunsEnabled && !localStreaming,
-  );
-  const run = runData?.run ?? null;
-  // The run's incrementally-persisted assistant message to merge into the thread,
-  // but only while we are an observer (never when we are the streamer — guards
-  // against a stale poll fighting the live stream). Includes a terminal run so the
-  // final persisted output is shown on reopen.
-  const observedRow = shouldObserveRun(run, localStreaming)
-    ? (runData?.message ?? null)
-    : null;
-
-  // When the observed run reaches a terminal status, do a final messages refetch
-  // so the persisted final state (token/context badge, export source) is shown,
-  // then the query's refetchInterval has already stopped polling. Deduped per run
-  // id so it fires exactly once per run, not on every subsequent poll-less render.
-  const finalizedRunIdRef = useRef<string | null>(null);
-  useEffect(() => {
-    if (!run || !activeChatId) return;
-    if (run.status === "pending" || run.status === "running") {
-      // Active again (a new run) — re-arm so its terminal transition fires once.
-      finalizedRunIdRef.current = null;
-      return;
-    }
-    if (finalizedRunIdRef.current === run.id) return;
-    finalizedRunIdRef.current = run.id;
-    queryClient.invalidateQueries({
-      queryKey: AI_CHAT_MESSAGES_RQ_KEY(activeChatId),
-    });
-  }, [run, activeChatId, queryClient]);
-
   // The page the user is currently viewing. AiChatWindow lives in a pathless
   // parent layout route, so useParams() can't see :pageSlug. Match the full
   // pathname against the authenticated page route instead so "the current page"
@@ -694,12 +636,6 @@ export default function AiChatWindow() {
               assistantName={currentRole?.name}
               onTurnFinished={onTurnFinished}
               onServerChatId={onServerChatId}
-              // #184: live-follow a still-running run when we reopened the chat as
-              // a passive observer; null when there is nothing to observe or this
-              // tab is the streamer. onStreamingChange lets the window stop polling
-              // while we are the streamer.
-              observedRow={observedRow}
-              onStreamingChange={onStreamingChange}
             />
           )}
         </div>

apps/client/src/features/ai-chat/components/chat-thread.test.tsx

@@ -11,7 +11,6 @@ const h = vi.hoisted(() => ({
     onFinish: null as null | ((arg: Record<string, unknown>) => void),
     sendMessage: vi.fn(),
     stop: vi.fn(),
-    setMessages: vi.fn(),
     transport: null as null | {
       prepareSendMessagesRequest: (arg: {
         messages: unknown[];
@@ -31,8 +30,6 @@ vi.mock("@ai-sdk/react", () => ({
       status: h.state.status,
       stop: h.state.stop,
       error: null,
-      // #184: ChatThread reads setMessages to merge a polled observer run.
-      setMessages: h.state.setMessages,
     };
   },
 }));
@@ -143,56 +140,3 @@ describe("ChatThread — send now (#198)", () => {
     expect(prep({ messages: [], body: {} }).body.interrupted).toBe(false);
   });
 });
-
-// #184 passive-observer merge: when reconnecting to a still-running run, the
-// parent feeds the polled run message via `observedRow`; ChatThread merges it via
-// setMessages — but ONLY when this tab is NOT itself streaming (the streamer's
-// SSE owns the view, so a stale observedRow must never overwrite it).
-describe("ChatThread — observer run merge (#184)", () => {
-  beforeEach(() => {
-    h.state.onFinish = null;
-    h.state.setMessages.mockReset();
-  });
-
-  const observedRow = {
-    id: "a-run",
-    role: "assistant",
-    content: "step 1\nstep 2",
-    metadata: {
-      parts: [{ type: "text", text: "step 1\nstep 2" }],
-    },
-    createdAt: "2026-01-01T00:00:00Z",
-  } as const;
-
-  function renderObserver(status: string) {
-    h.state.status = status;
-    render(
-      <MantineProvider>
-        <ChatThread
-          chatId="c1"
-          initialRows={[]}
-          onTurnFinished={vi.fn()}
-          observedRow={observedRow as never}
-        />
-      </MantineProvider>,
-    );
-  }
-
-  it("merges the polled run message when this tab is a passive observer", () => {
-    renderObserver("ready");
-    expect(h.state.setMessages).toHaveBeenCalledTimes(1);
-    // The updater replaces/append the observed assistant row by id.
-    const updater = h.state.setMessages.mock.calls[0][0] as (
-      prev: { id: string; parts: { text: string }[] }[],
-    ) => { id: string; parts: { text: string }[] }[];
-    const merged = updater([{ id: "u1", parts: [{ text: "hi" }] }]);
-    expect(merged).toHaveLength(2);
-    expect(merged[1].id).toBe("a-run");
-    expect(merged[1].parts[0].text).toBe("step 1\nstep 2");
-  });
-
-  it("does NOT merge while THIS tab is the streamer (no double-render)", () => {
-    renderObserver("streaming");
-    expect(h.state.setMessages).not.toHaveBeenCalled();
-  });
-});

apps/client/src/features/ai-chat/components/chat-thread.tsx

@@ -24,7 +24,6 @@ import {
 } from "@/features/ai-chat/utils/role-launch.ts";
 import { describeChatError } from "@/features/ai-chat/utils/error-message.ts";
 import { extractServerChatId } from "@/features/ai-chat/utils/adopt-chat-id.ts";
-import { mergeObservedMessage } from "@/features/ai-chat/utils/run-polling.ts";
 import {
   dequeue,
   enqueueMessage,
@@ -87,19 +86,6 @@ interface ChatThreadProps {
    *  Copy/export button available mid-stream). Distinct from onTurnFinished,
    *  which fires only at the terminal outcome. */
   onServerChatId?: (serverChatId?: string) => void;
-  /** #184 reconnect-and-live-follow. When THIS tab reopened a chat whose agent
-   *  run is still going (it is a PASSIVE OBSERVER — it did not start the run here),
-   *  the parent polls the reconnect endpoint and feeds the run's incrementally-
-   *  persisted assistant message here; we merge it into the live list so new
-   *  steps/tool-calls appear as they are persisted. Null when there is nothing to
-   *  observe (no run, feature off, or this tab IS the streamer). The merge is
-   *  ADDITIONALLY guarded by our own `isStreaming`, so a stale value can never
-   *  fight the local stream when we are the streamer. */
-  observedRow?: IAiChatMessageRow | null;
-  /** Report this tab's live streaming status up to the parent, so it can stop
-   *  polling the run while WE are the active streamer (the SSE owns the view) and
-   *  resume once we go idle. Called from an effect on every transition. */
-  onStreamingChange?: (streaming: boolean) => void;
 }
 
 /**
@@ -145,8 +131,6 @@ export default function ChatThread({
   assistantName,
   onTurnFinished,
   onServerChatId,
-  observedRow,
-  onStreamingChange,
 }: ChatThreadProps) {
   const { t } = useTranslation();
 
@@ -290,7 +274,7 @@ export default function ChatThread({
     [],
   );
 
-  const { messages, sendMessage, status, stop, error, setMessages } = useChat({
+  const { messages, sendMessage, status, stop, error } = useChat({
     // Stable per-mount key. Existing chats use their real id; new chats use a
     // generated client id (never `undefined`) so the store is NOT re-created on
     // every render mid-stream (see `chatStoreId` above).
@@ -394,27 +378,6 @@ export default function ChatThread({
 
   const isStreaming = status === "submitted" || status === "streaming";
 
-  // #184: report our live streaming status up so the parent stops polling the run
-  // while WE are the streamer (the SSE owns the view) and resumes once we go idle.
-  // Effect (not render) so it never updates parent state during our own render;
-  // fires on mount with `false`, which also re-syncs the parent after a chat
-  // switch remounts this thread (a fresh mount is idle until the user sends).
-  useEffect(() => {
-    onStreamingChange?.(isStreaming);
-  }, [isStreaming, onStreamingChange]);
-
-  // #184 passive-observer merge: when the parent feeds a polled run message (we
-  // reopened a chat whose run is still going and did NOT start it here), merge it
-  // into the live list so new steps/tool-calls appear as they are persisted. Hard-
-  // gated by `!isStreaming`: if THIS tab is actually the streamer, the local SSE
-  // owns the view and a stale observedRow must never overwrite it. `observedRow`
-  // is a stable per-poll object, so this runs once per poll, not per render.
-  useEffect(() => {
-    if (isStreaming || !observedRow) return;
-    const observed = rowToUiMessage(observedRow);
-    setMessages((prev) => mergeObservedMessage(prev, observed));
-  }, [observedRow, isStreaming, setMessages]);
-
   // "Send now" on a queued message: interrupt the current turn and immediately
   // send THIS message, keeping the agent's partial output. Other queued messages
   // stay queued and flush normally after the new turn. Reuses the existing

apps/client/src/features/ai-chat/queries/ai-chat-query.ts

@@ -12,7 +12,6 @@ import {
   deleteAiChat,
   deleteAiRole,
   getAiChatMessages,
-  getAiChatRun,
   getAiChats,
   getAiRoleCatalog,
   getAiRoleCatalogBundle,
@@ -25,7 +24,6 @@ import {
 import {
   IAiChat,
   IAiChatMessageRow,
-  IAiChatRunResponse,
   IAiRole,
   IAiRoleCatalog,
   IAiRoleCatalogBundle,
@@ -36,7 +34,6 @@ import {
   IAiRoleUpdateFromCatalogResult,
 } from "@/features/ai-chat/types/ai-chat.types.ts";
 import { IPagination } from "@/lib/types.ts";
-import { runPollInterval } from "@/features/ai-chat/utils/run-polling.ts";
 
 export const AI_CHATS_RQ_KEY = ["ai-chats"];
 export const AI_ROLES_RQ_KEY = ["ai-roles"];
@@ -54,18 +51,16 @@ export const AI_CHAT_MESSAGES_RQ_KEY = (chatId: string) => [
   "ai-chat-messages",
   chatId,
 ];
-export const AI_CHAT_RUN_RQ_KEY = (chatId: string) => ["ai-chat-run", chatId];
 
 /** Paginated list of the current user's chats (auto-loads further pages). */
 export function useAiChatsQuery() {
   const query = useInfiniteQuery({
     queryKey: AI_CHATS_RQ_KEY,
-    queryFn: ({ pageParam }) => getAiChats({ cursor: pageParam, limit: 50 }),
+    queryFn: ({ pageParam }) =>
+      getAiChats({ cursor: pageParam, limit: 50 }),
     initialPageParam: undefined as string | undefined,
     getNextPageParam: (lastPage) =>
-      lastPage.meta.hasNextPage
-        ? (lastPage.meta.nextCursor ?? undefined)
-        : undefined,
+      lastPage.meta.hasNextPage ? (lastPage.meta.nextCursor ?? undefined) : undefined,
   });
 
   const data = useMemo<IPagination<IAiChat> | undefined>(() => {
@@ -95,9 +90,7 @@ export function useAiChatMessagesQuery(chatId: string | undefined) {
       getAiChatMessages({ chatId: chatId as string, cursor: pageParam }),
     initialPageParam: undefined as string | undefined,
     getNextPageParam: (lastPage) =>
-      lastPage.meta.hasNextPage
-        ? (lastPage.meta.nextCursor ?? undefined)
-        : undefined,
+      lastPage.meta.hasNextPage ? (lastPage.meta.nextCursor ?? undefined) : undefined,
     enabled: !!chatId,
   });
 
@@ -138,34 +131,6 @@ export function useAiChatMessagesQuery(chatId: string | undefined) {
   };
 }
 
-/**
- * Reconnect to a chat's latest agent run and LIVE-FOLLOW it (#184). While the run
- * is active the query re-polls every {@link runPollInterval} ms (driven off the
- * fetched `run.status`, the same status-keyed refetchInterval pattern as the
- * embeddings reindex polling); once the run reaches a terminal status — or there
- * is no run — the interval returns `false` and polling stops on its own. Polling
- * is thus naturally bounded by the run terminating; no separate timeout cap.
- *
- * `enabled` gates the whole thing: callers pass `false` when the autonomous-runs
- * feature is off (the endpoint is NOT flag-gated server-side, but with the feature
- * off the chat has no runs, so polling would only ever return `{ run: null }`) OR
- * when THIS tab is the one actively streaming the run (the live SSE owns the view,
- * so we must not also poll/merge). The global `retry: false` means a failed fetch
- * leaves `data` undefined, so refetchInterval(undefined run) returns false — a
- * failed fetch can never spin a tight loop.
- */
-export function useAiChatRunQuery(
-  chatId: string | undefined,
-  enabled: boolean,
-) {
-  return useQuery<IAiChatRunResponse, Error>({
-    queryKey: AI_CHAT_RUN_RQ_KEY(chatId ?? ""),
-    queryFn: () => getAiChatRun(chatId as string),
-    enabled: !!chatId && enabled,
-    refetchInterval: (query) => runPollInterval(query.state.data?.run),
-  });
-}
-
 export function useRenameAiChatMutation() {
   const queryClient = useQueryClient();
   const { t } = useTranslation();
@@ -315,14 +280,11 @@ export function useImportAiRolesFromCatalogMutation() {
     mutationFn: (payload) => importAiRolesFromCatalog(payload),
     onSuccess: (result) => {
       notifications.show({
-        message: t(
-          "Imported {{created}}, renamed {{renamed}}, skipped {{skipped}}",
-          {
-            created: result.created,
-            renamed: result.renamed,
-            skipped: result.skipped,
-          },
-        ),
+        message: t("Imported {{created}}, renamed {{renamed}}, skipped {{skipped}}", {
+          created: result.created,
+          renamed: result.renamed,
+          skipped: result.skipped,
+        }),
       });
       // Surface partial failures (e.g. unique-name races) as a red warning.
       if (result.errors.length > 0) {

apps/client/src/features/ai-chat/queries/ai-chat-run-query.test.tsx

@@ -1,92 +0,0 @@
-import { describe, it, expect, vi, beforeEach } from "vitest";
-import React from "react";
-import { renderHook, waitFor } from "@testing-library/react";
-import { QueryClient, QueryClientProvider } from "@tanstack/react-query";
-import type { IAiChatRunResponse } from "@/features/ai-chat/types/ai-chat.types.ts";
-
-// react-i18next is pulled in transitively by ai-chat-query.ts (the mutation hooks
-// use it); stub it so the module imports cleanly in this hook test.
-vi.mock("react-i18next", () => ({
-  useTranslation: () => ({ t: (key: string) => key }),
-}));
-
-vi.mock("@mantine/notifications", () => ({
-  notifications: { show: vi.fn() },
-}));
-
-// Mock the whole service module; only getAiChatRun is exercised here, but the
-// other named exports must exist so ai-chat-query.ts imports resolve.
-vi.mock("@/features/ai-chat/services/ai-chat-service.ts", () => ({
-  getAiChatRun: vi.fn(),
-  getAiChatMessages: vi.fn(),
-  getAiChats: vi.fn(),
-  getAiRoleCatalog: vi.fn(),
-  getAiRoleCatalogBundle: vi.fn(),
-  getAiRoles: vi.fn(),
-  importAiRolesFromCatalog: vi.fn(),
-  createAiRole: vi.fn(),
-  deleteAiChat: vi.fn(),
-  deleteAiRole: vi.fn(),
-  renameAiChat: vi.fn(),
-  updateAiRole: vi.fn(),
-  updateAiRoleFromCatalog: vi.fn(),
-}));
-
-import { getAiChatRun } from "@/features/ai-chat/services/ai-chat-service.ts";
-import { useAiChatRunQuery } from "@/features/ai-chat/queries/ai-chat-query.ts";
-
-function createWrapper() {
-  const queryClient = new QueryClient({
-    defaultOptions: { queries: { retry: false } },
-  });
-  return function Wrapper({ children }: { children: React.ReactNode }) {
-    return (
-      <QueryClientProvider client={queryClient}>{children}</QueryClientProvider>
-    );
-  };
-}
-
-const runningResponse: IAiChatRunResponse = {
-  run: { id: "run-1", chatId: "c1", status: "running" },
-  message: {
-    id: "a1",
-    role: "assistant",
-    content: "working...",
-    createdAt: "2026-01-01T00:00:00Z",
-  },
-};
-
-describe("useAiChatRunQuery — enable gating", () => {
-  beforeEach(() => {
-    vi.clearAllMocks();
-  });
-
-  it("fetches the run when enabled (passive observer, feature on)", async () => {
-    vi.mocked(getAiChatRun).mockResolvedValue(runningResponse);
-    const { result } = renderHook(() => useAiChatRunQuery("c1", true), {
-      wrapper: createWrapper(),
-    });
-    await waitFor(() => expect(result.current.isSuccess).toBe(true));
-    expect(getAiChatRun).toHaveBeenCalledWith("c1");
-    expect(result.current.data?.run?.status).toBe("running");
-  });
-
-  it("does NOT fetch when disabled (this tab is the streamer / feature off)", async () => {
-    vi.mocked(getAiChatRun).mockResolvedValue(runningResponse);
-    renderHook(() => useAiChatRunQuery("c1", false), {
-      wrapper: createWrapper(),
-    });
-    // Give any errant fetch a chance to fire, then assert none did.
-    await new Promise((r) => setTimeout(r, 20));
-    expect(getAiChatRun).not.toHaveBeenCalled();
-  });
-
-  it("does NOT fetch when there is no chat id", async () => {
-    vi.mocked(getAiChatRun).mockResolvedValue(runningResponse);
-    renderHook(() => useAiChatRunQuery(undefined, true), {
-      wrapper: createWrapper(),
-    });
-    await new Promise((r) => setTimeout(r, 20));
-    expect(getAiChatRun).not.toHaveBeenCalled();
-  });
-});

apps/client/src/features/ai-chat/services/ai-chat-service.ts

@@ -5,7 +5,6 @@ import {
   IAiChatListParams,
   IAiChatMessageRow,
   IAiChatMessagesParams,
-  IAiChatRunResponse,
   IAiRole,
   IAiRoleCatalog,
   IAiRoleCatalogBundle,
@@ -43,23 +42,6 @@ export async function getAiChatMessages(
   return req.data;
 }
 
-/**
- * Reconnect to the latest agent run of a chat (#184). Returns the run's
- * persisted lifecycle state and the assistant message it materializes (the
- * partial output while the run is in-flight, the final output once it finished).
- * The DB is the source of truth, so this works for an in-flight run (the browser
- * dropped, the run kept going) and a finished one alike; `{ run: null }` when the
- * chat has never had a run. Owner-gated server-side (the requesting user must own
- * the chat); it is NOT flag-gated — when the feature is off the chat simply has no
- * runs, so the endpoint returns `{ run: null }`.
- */
-export async function getAiChatRun(
-  chatId: string,
-): Promise<IAiChatRunResponse> {
-  const req = await api.post<IAiChatRunResponse>("/ai-chat/run", { chatId });
-  return req.data;
-}
-
 /**
  * Resolve the chat bound to a document (the current user's most-recent chat
  * created on that page), or null when there is none. Drives auto-open-on-page.

apps/client/src/features/ai-chat/types/ai-chat.types.ts

@@ -200,38 +200,6 @@ export interface IAiChatMessageRow {
   createdAt: string;
 }
 
-/**
- * A persisted agent-run row (#184), mirroring the `ai_chat_runs` fields the
- * client reads from `POST /ai-chat/run`. Only `status` is load-bearing for the
- * reconnect-and-live-update UX (it drives the poll cadence); the rest are carried
- * for display/diagnostics. The DB is the source of truth, so this resolves for an
- * in-flight run (the browser dropped, the run kept going) and a finished one.
- */
-export interface IAiChatRun {
-  id: string;
-  chatId: string;
-  // 'pending' | 'running' | 'succeeded' | 'failed' | 'aborted'. The first two are
-  // ACTIVE (keep polling); the rest are TERMINAL (stop polling).
-  status: "pending" | "running" | "succeeded" | "failed" | "aborted" | string;
-  error?: string | null;
-  stepCount?: number;
-  assistantMessageId?: string | null;
-  startedAt?: string | null;
-  finishedAt?: string | null;
-  createdAt?: string;
-  updatedAt?: string;
-}
-
-/**
- * Response of `POST /ai-chat/run` (#184): the latest run of a chat and the
- * assistant message it materializes (the partial/final output, projected from the
- * persisted rows). Both are `null` when the chat has never had a run.
- */
-export interface IAiChatRunResponse {
-  run: IAiChatRun | null;
-  message: IAiChatMessageRow | null;
-}
-
 export interface IAiChatListParams extends QueryParams {}
 
 export interface IAiChatMessagesParams {

apps/client/src/features/ai-chat/utils/run-polling.test.ts

@@ -1,104 +0,0 @@
-import { describe, it, expect } from "vitest";
-import type { UIMessage } from "@ai-sdk/react";
-import type { IAiChatRun } from "@/features/ai-chat/types/ai-chat.types.ts";
-import {
-  RUN_POLL_INTERVAL_MS,
-  isRunActive,
-  runPollInterval,
-  shouldObserveRun,
-  mergeObservedMessage,
-} from "./run-polling.ts";
-
-function makeRun(status: string): IAiChatRun {
-  return { id: "run-1", chatId: "c1", status };
-}
-
-function makeMsg(id: string, text: string): UIMessage {
-  return {
-    id,
-    role: "assistant",
-    parts: [{ type: "text", text }],
-  } as UIMessage;
-}
-
-describe("isRunActive", () => {
-  it("treats pending and running as active", () => {
-    expect(isRunActive(makeRun("pending"))).toBe(true);
-    expect(isRunActive(makeRun("running"))).toBe(true);
-  });
-
-  it("treats terminal / unknown / nullish as not active", () => {
-    expect(isRunActive(makeRun("succeeded"))).toBe(false);
-    expect(isRunActive(makeRun("failed"))).toBe(false);
-    expect(isRunActive(makeRun("aborted"))).toBe(false);
-    expect(isRunActive(makeRun("weird-future-status"))).toBe(false);
-    expect(isRunActive(null)).toBe(false);
-    expect(isRunActive(undefined)).toBe(false);
-  });
-});
-
-describe("runPollInterval (the refetchInterval helper)", () => {
-  it("returns 2000ms while the run is pending/running", () => {
-    expect(runPollInterval(makeRun("pending"))).toBe(RUN_POLL_INTERVAL_MS);
-    expect(runPollInterval(makeRun("running"))).toBe(RUN_POLL_INTERVAL_MS);
-    expect(RUN_POLL_INTERVAL_MS).toBe(2000);
-  });
-
-  it("returns false (stop polling) once the run is terminal", () => {
-    expect(runPollInterval(makeRun("succeeded"))).toBe(false);
-    expect(runPollInterval(makeRun("failed"))).toBe(false);
-    expect(runPollInterval(makeRun("aborted"))).toBe(false);
-  });
-
-  it("returns false (no polling) when there is no run", () => {
-    expect(runPollInterval(null)).toBe(false);
-    expect(runPollInterval(undefined)).toBe(false);
-  });
-});
-
-describe("shouldObserveRun (observer-vs-streamer decision)", () => {
-  it("observes an active run when this tab is NOT the local streamer", () => {
-    expect(shouldObserveRun(makeRun("running"), false)).toBe(true);
-    expect(shouldObserveRun(makeRun("pending"), false)).toBe(true);
-  });
-
-  it("observes a terminal run too (so the final output shows on reopen)", () => {
-    expect(shouldObserveRun(makeRun("succeeded"), false)).toBe(true);
-  });
-
-  it("does NOT observe when this tab IS the streamer (no double-render)", () => {
-    expect(shouldObserveRun(makeRun("running"), true)).toBe(false);
-    expect(shouldObserveRun(makeRun("succeeded"), true)).toBe(false);
-  });
-
-  it("does NOT observe when there is no run", () => {
-    expect(shouldObserveRun(null, false)).toBe(false);
-    expect(shouldObserveRun(undefined, false)).toBe(false);
-  });
-});
-
-describe("mergeObservedMessage", () => {
-  it("replaces the message with the same id in place (per-step growth)", () => {
-    const prev = [makeMsg("u1", "hi"), makeMsg("a1", "step 1")];
-    const observed = makeMsg("a1", "step 1\nstep 2");
-    const next = mergeObservedMessage(prev, observed);
-    expect(next).toHaveLength(2);
-    expect(next[1]).toBe(observed);
-    expect(next[0]).toBe(prev[0]); // untouched
-    expect(next).not.toBe(prev); // new array (never mutates input)
-  });
-
-  it("appends when the observed message is not yet present", () => {
-    const prev = [makeMsg("u1", "hi")];
-    const observed = makeMsg("a1", "first token");
-    const next = mergeObservedMessage(prev, observed);
-    expect(next).toHaveLength(2);
-    expect(next[1]).toBe(observed);
-  });
-
-  it("returns the original list unchanged when there is nothing to merge", () => {
-    const prev = [makeMsg("u1", "hi")];
-    expect(mergeObservedMessage(prev, null)).toBe(prev);
-    expect(mergeObservedMessage(prev, undefined)).toBe(prev);
-  });
-});

apps/client/src/features/ai-chat/utils/run-polling.ts

@@ -1,71 +0,0 @@
-import type { UIMessage } from "@ai-sdk/react";
-import type { IAiChatRun } from "@/features/ai-chat/types/ai-chat.types.ts";
-
-/**
- * Reconnect-and-live-follow helpers (#184). When a chat is reopened while its
- * agent run is STILL going, this tab is a PASSIVE OBSERVER: it did not start the
- * run here (no local SSE stream), so it catches up by POLLING the reconnect
- * endpoint (`POST /ai-chat/run`) and merging the run's incrementally-persisted
- * assistant message into the rendered thread. These are the small pure decisions
- * that machinery hangs off, extracted so they can be unit-tested in isolation
- * (mirrors how reindex polling / editor-sync-state are tested).
- */
-
-/** How often to re-poll the reconnect endpoint while a run is ACTIVE. */
-export const RUN_POLL_INTERVAL_MS = 2000;
-
-// 'pending' and 'running' are the two ACTIVE statuses; 'succeeded' | 'failed' |
-// 'aborted' are TERMINAL (and any unknown future status is treated as terminal,
-// so a stale/odd value never polls forever).
-const ACTIVE_STATUSES = new Set(["pending", "running"]);
-
-/** Whether a run is still going (worth polling / merging live updates from). */
-export function isRunActive(run: IAiChatRun | null | undefined): boolean {
-  return !!run && ACTIVE_STATUSES.has(run.status);
-}
-
-/**
- * The TanStack Query `refetchInterval` value for the run query: poll every
- * {@link RUN_POLL_INTERVAL_MS} while the run is active, and `false` (stop) once
- * it is terminal or there is no run. Polling is thus naturally bounded by the run
- * reaching a terminal status — no separate timeout cap is needed.
- */
-export function runPollInterval(
-  run: IAiChatRun | null | undefined,
-): number | false {
-  return isRunActive(run) ? RUN_POLL_INTERVAL_MS : false;
-}
-
-/**
- * Observer-vs-streamer decision. We render the polled run message (catch up +
- * keep advancing) ONLY when this tab is a passive observer: there IS a run AND
- * this tab is NOT the one locally streaming it (we reconnected, we didn't start
- * it here). When this tab is the streamer, the live SSE stream owns the view, so
- * we neither poll nor merge — avoiding a double-render fight. Terminal runs still
- * merge (so the final persisted output is shown on reopen); the poll itself is
- * stopped separately by {@link runPollInterval}.
- */
-export function shouldObserveRun(
-  run: IAiChatRun | null | undefined,
-  localStreaming: boolean,
-): boolean {
-  return !!run && !localStreaming;
-}
-
-/**
- * Merge an observed assistant message into the rendered list: replace the message
- * with the same id in place (the in-progress assistant row is already seeded from
- * history, so per-step growth replaces it), or append it when absent. Returns a
- * new array; the input is never mutated.
- */
-export function mergeObservedMessage(
-  messages: UIMessage[],
-  observed: UIMessage | null | undefined,
-): UIMessage[] {
-  if (!observed) return messages;
-  const idx = messages.findIndex((m) => m.id === observed.id);
-  if (idx === -1) return [...messages, observed];
-  const next = messages.slice();
-  next[idx] = observed;
-  return next;
-}

apps/client/src/features/auth/hooks/use-auth.test.ts

@@ -0,0 +1,154 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
+import { renderHook, act } from "@testing-library/react";
+
+// react-i18next: identity t() so the hook renders without an i18n provider.
+vi.mock("react-i18next", () => ({
+  useTranslation: () => ({ t: (k: string) => k }),
+}));
+
+// react-router-dom: only useNavigate is used by the hook.
+const navigateMock = vi.fn();
+vi.mock("react-router-dom", () => ({
+  useNavigate: () => navigateMock,
+}));
+
+// The auth service is the network boundary; stub login/logout per test.
+const loginMock = vi.fn();
+const logoutMock = vi.fn();
+vi.mock("@/features/auth/services/auth-service", () => ({
+  login: (...args: unknown[]) => loginMock(...args),
+  logout: (...args: unknown[]) => logoutMock(...args),
+  forgotPassword: vi.fn(),
+  passwordReset: vi.fn(),
+  setupWorkspace: vi.fn(),
+  verifyUserToken: vi.fn(),
+}));
+
+vi.mock("@/features/workspace/services/workspace-service.ts", () => ({
+  acceptInvitation: vi.fn(),
+}));
+
+// The offline cache purge is the unit under test — assert it is invoked.
+const clearOfflineCacheMock = vi.fn();
+vi.mock("@/features/offline/clear-offline-cache", () => ({
+  clearOfflineCache: () => clearOfflineCacheMock(),
+}));
+
+// app-route helpers are pure config; provide deterministic values.
+vi.mock("@/lib/app-route.ts", () => ({
+  default: { AUTH: { LOGIN: "/login" }, HOME: "/home" },
+  getPostLoginRedirect: () => "/home",
+}));
+
+// Mantine notifications: avoid touching the DOM-bound notification system.
+vi.mock("@mantine/notifications", () => ({
+  notifications: { show: vi.fn() },
+}));
+
+import useAuth from "./use-auth";
+
+beforeEach(() => {
+  navigateMock.mockReset();
+  loginMock.mockReset();
+  loginMock.mockResolvedValue(undefined);
+  logoutMock.mockReset();
+  logoutMock.mockResolvedValue(undefined);
+  clearOfflineCacheMock.mockReset();
+  clearOfflineCacheMock.mockResolvedValue(undefined);
+});
+
+describe("useAuth.handleSignIn", () => {
+  it("clears the offline cache BEFORE logging in (cross-user leak guard)", async () => {
+    const order: string[] = [];
+    clearOfflineCacheMock.mockImplementation(async () => {
+      order.push("clear");
+    });
+    loginMock.mockImplementation(async () => {
+      order.push("login");
+    });
+
+    const { result } = renderHook(() => useAuth());
+    await act(async () => {
+      await result.current.signIn({ email: "b@x", password: "pw" } as any);
+    });
+
+    expect(clearOfflineCacheMock).toHaveBeenCalledTimes(1);
+    expect(loginMock).toHaveBeenCalledTimes(1);
+    // The purge must run before the new session's login resolves.
+    expect(order).toEqual(["clear", "login"]);
+    expect(navigateMock).toHaveBeenCalledWith("/home");
+  });
+
+  it("does not block sign-in when the cache purge throws (best-effort)", async () => {
+    clearOfflineCacheMock.mockRejectedValue(new Error("idb unavailable"));
+
+    const { result } = renderHook(() => useAuth());
+    await act(async () => {
+      await result.current.signIn({ email: "b@x", password: "pw" } as any);
+    });
+
+    // Login still proceeds despite the cleanup failure.
+    expect(loginMock).toHaveBeenCalledTimes(1);
+    expect(navigateMock).toHaveBeenCalledWith("/home");
+  });
+});
+
+describe("useAuth.handleLogout", () => {
+  const replaceMock = vi.fn();
+  let originalLocation: Location;
+
+  beforeEach(() => {
+    replaceMock.mockReset();
+    // window.location.replace is the post-logout redirect. jsdom's real `replace`
+    // is a non-configurable method that warns "not implemented", so swap the
+    // whole location object for one whose `replace` we can capture.
+    originalLocation = window.location;
+    Object.defineProperty(window, "location", {
+      configurable: true,
+      writable: true,
+      value: { replace: replaceMock },
+    });
+  });
+
+  afterEach(() => {
+    Object.defineProperty(window, "location", {
+      configurable: true,
+      writable: true,
+      value: originalLocation,
+    });
+  });
+
+  it("purges the offline cache exactly once BEFORE redirecting (cross-user leak guard)", async () => {
+    const order: string[] = [];
+    clearOfflineCacheMock.mockImplementation(async () => {
+      order.push("clear");
+    });
+    replaceMock.mockImplementation((url: string) => {
+      order.push(`replace:${url}`);
+    });
+
+    const { result } = renderHook(() => useAuth());
+    await act(async () => {
+      await result.current.logout();
+    });
+
+    expect(clearOfflineCacheMock).toHaveBeenCalledTimes(1);
+    // Purge must complete before the redirect (which would otherwise interrupt
+    // the async cleanup).
+    expect(order).toEqual(["clear", "replace:/login?logout=1"]);
+  });
+
+  it("still redirects when the cache purge throws (best-effort, never blocks logout)", async () => {
+    clearOfflineCacheMock.mockRejectedValue(new Error("idb unavailable"));
+
+    const { result } = renderHook(() => useAuth());
+    await act(async () => {
+      await result.current.logout();
+    });
+
+    // The thrown purge error is swallowed and the redirect still fires.
+    expect(clearOfflineCacheMock).toHaveBeenCalledTimes(1);
+    expect(replaceMock).toHaveBeenCalledTimes(1);
+    expect(replaceMock).toHaveBeenCalledWith("/login?logout=1");
+  });
+});

apps/client/src/features/auth/hooks/use-auth.ts

@@ -23,6 +23,7 @@ import { acceptInvitation } from "@/features/workspace/services/workspace-servic
 import APP_ROUTE, { getPostLoginRedirect } from "@/lib/app-route.ts";
 import { RESET } from "jotai/utils";
 import { useTranslation } from "react-i18next";
+import { clearOfflineCache } from "@/features/offline/clear-offline-cache";
 
 export default function useAuth() {
   const { t } = useTranslation();
@@ -33,6 +34,20 @@ export default function useAuth() {
   const handleSignIn = async (data: ILogin) => {
     setIsLoading(true);
 
+    // Purge any previous user's offline data BEFORE signing in (mirrors logout).
+    // On a shared/kiosk device the prior session may have ended WITHOUT an
+    // explicit logout (cookie/JWT expiry, tab close, force-quit), leaving user
+    // A's persisted query cache (gitmost-rq-cache) and Yjs page bodies
+    // (page.<id>) in IndexedDB. Without this purge user B would briefly read A's
+    // cached currentUser/pages/comments on first render (UserProvider serves the
+    // cached user) and A's page bodies would stay readable offline. Best-effort:
+    // never block sign-in on cache cleanup.
+    try {
+      await clearOfflineCache();
+    } catch {
+      // best-effort: never block sign-in on cache cleanup
+    }
+
     try {
       await login(data);
       setIsLoading(false);
@@ -123,6 +138,13 @@ export default function useAuth() {
   const handleLogout = async () => {
     setCurrentUser(RESET);
     await logout();
+    // Purge the previous user's offline data while the page is still alive —
+    // window.location.replace below would otherwise interrupt async cleanup.
+    try {
+      await clearOfflineCache();
+    } catch {
+      // best-effort: never block logout on cache cleanup
+    }
     window.location.replace(`${APP_ROUTE.AUTH.LOGIN}?logout=1`);
   };
 

apps/client/src/features/auth/queries/auth-query.test.ts

@@ -0,0 +1,43 @@
+import { describe, it, expect } from "vitest";
+import { AxiosError } from "axios";
+import { collabTokenRetry } from "./auth-query";
+
+// Regression for the offline white-screen (#237/#238): offline the collab-token
+// POST rejects as an axios NETWORK error (isAxiosError === true but
+// error.response === undefined). The old predicate read `error.response.status`
+// without a guard and threw an uncaught TypeError inside the React Query retryer
+// BEFORE React mounted, blanking the whole app. The predicate must stay total.
+describe("collabTokenRetry", () => {
+  it("does NOT throw and returns a retryable value for a network error with no response (offline)", () => {
+    // An axios error with no `response` is exactly the offline/network-failure shape.
+    const networkError = new AxiosError("Network Error");
+    expect(networkError.response).toBeUndefined();
+
+    let result: boolean | number = false;
+    expect(() => {
+      result = collabTokenRetry(0, networkError);
+    }).not.toThrow();
+    // Network failures stay retryable (truthy), matching the original intent.
+    expect(result).toBe(true);
+  });
+
+  it("returns false (no retry) for a real 404 response", () => {
+    const notFound = new AxiosError("Not Found");
+    notFound.response = { status: 404 } as AxiosError["response"];
+    expect(collabTokenRetry(0, notFound)).toBe(false);
+  });
+
+  it("retries for a non-404 response (e.g. 500)", () => {
+    const serverError = new AxiosError("Server Error");
+    serverError.response = { status: 500 } as AxiosError["response"];
+    expect(collabTokenRetry(0, serverError)).toBe(true);
+  });
+
+  it("does not throw and retries for a non-axios error", () => {
+    let result: boolean | number = false;
+    expect(() => {
+      result = collabTokenRetry(0, new Error("boom"));
+    }).not.toThrow();
+    expect(result).toBe(true);
+  });
+});

apps/client/src/features/auth/queries/auth-query.tsx

@@ -3,6 +3,27 @@ import { getCollabToken, verifyUserToken } from "../services/auth-service";
 import { ICollabToken, IVerifyUserToken } from "../types/auth.types";
 import { isAxiosError } from "axios";
 
+/**
+ * Retry predicate for the collab-token query.
+ *
+ * Offline (or any network failure) the POST rejects as an axios NETWORK error:
+ * `isAxiosError(error) === true` but `error.response === undefined`. Reading
+ * `error.response.status` without a guard threw an uncaught TypeError inside the
+ * React Query retryer BEFORE React mounted, white-screening the whole app on an
+ * offline cold boot (#237/#238). Optional-chaining `error.response?.status`
+ * keeps the predicate total: a network error (no response) is retryable, a real
+ * 404 is not. Extracted (and exported) so it can be unit-tested in isolation.
+ */
+export function collabTokenRetry(
+  _failureCount: number,
+  error: Error,
+): boolean {
+  if (isAxiosError(error) && error.response?.status === 404) {
+    return false;
+  }
+  return true;
+}
+
 export function useVerifyUserTokenQuery(
   verify: IVerifyUserToken,
 ): UseQueryResult<any, Error> {
@@ -22,13 +43,7 @@ export function useCollabToken(): UseQueryResult<ICollabToken, Error> {
     //refetchInterval: 12 * 60 * 60 * 1000, // 12hrs
     //refetchIntervalInBackground: true,
     refetchOnMount: true,
-    //@ts-ignore
-    retry: (failureCount, error) => {
-      if (isAxiosError(error) && error.response.status === 404) {
-        return false;
-      }
-      return 10;
-    },
+    retry: collabTokenRetry,
     retryDelay: (retryAttempt) => {
       // Exponential backoff: 5s, 10s, 20s, etc.
       return 5000 * Math.pow(2, retryAttempt - 1);

apps/client/src/features/comment/queries/comment-query.ts

@@ -20,6 +20,7 @@ import { notifications } from "@mantine/notifications";
 import { IPagination } from "@/lib/types.ts";
 import { useTranslation } from "react-i18next";
 import { useEffect, useMemo } from "react";
+import { offlineMutationKeys } from "@/features/offline/offline-mutations";
 
 export const RQ_KEY = (pageId: string) => ["comments", pageId];
 
@@ -60,6 +61,9 @@ export function useCreateCommentMutation() {
   const { t } = useTranslation();
 
   return useMutation<IComment, Error, Partial<IComment>>({
+    // Stable key so a paused comment-create restored from IndexedDB after an
+    // offline reload finds its default mutationFn and is replayed on reconnect.
+    mutationKey: offlineMutationKeys.createComment,
     mutationFn: (data) => createComment(data),
     onSuccess: (newComment) => {
       const cache = queryClient.getQueryData(

apps/client/src/features/editor/atoms/editor-atoms.ts

@@ -10,6 +10,12 @@ export const readOnlyEditorAtom = atom<Editor | null>(null);
 
 export const yjsConnectionStatusAtom = atom<string>("");
 
+// Local (IndexedDB) persistence sync state for the current page's Y.Doc.
+export const isLocalSyncedAtom = atom<boolean>(false);
+
+// Remote (Hocuspocus) sync state for the current page's Y.Doc.
+export const isRemoteSyncedAtom = atom<boolean>(false);
+
 export const showLinkMenuAtom = atom(false);
 
 // Current page's edit mode — initialized from the user's saved preference on

apps/client/src/features/editor/contexts/editor-providers-context.tsx

@@ -0,0 +1,21 @@
+import { createContext, useContext } from "react";
+import type { HocuspocusProvider } from "@hocuspocus/provider";
+import type * as Y from "yjs";
+
+// Shared collaboration providers lifted above the title/body editors so that
+// both siblings bind to the SAME Y.Doc and HocuspocusProvider. The title lives
+// in a dedicated 'title' fragment of the same doc as the body.
+export interface EditorProvidersContextValue {
+  ydoc: Y.Doc;
+  remote: HocuspocusProvider;
+  providersReady: boolean;
+}
+
+export const EditorProvidersContext =
+  createContext<EditorProvidersContextValue | null>(null);
+
+// Returns the shared providers, or null when rendered outside of a provider.
+// Consumers must be null-safe (the body editor falls back to a non-collab mode).
+export function useEditorProviders(): EditorProvidersContextValue | null {
+  return useContext(EditorProvidersContext);
+}

apps/client/src/features/editor/full-editor.tsx

@@ -34,6 +34,8 @@ import {
 } from "@/features/editor/atoms/editor-atoms.ts";
 import { DictationGroup } from "@/features/editor/components/fixed-toolbar/groups/dictation-group";
 import { GenerateTitleGroup } from "@/features/editor/components/fixed-toolbar/groups/generate-title-group";
+import { usePageCollabProviders } from "@/features/editor/hooks/use-page-collab-providers";
+import { EditorProvidersContext } from "@/features/editor/contexts/editor-providers-context";
 
 const MemoizedTitleEditor = React.memo(TitleEditor);
 const MemoizedPageEditor = React.memo(PageEditor);
@@ -80,16 +82,24 @@ export function FullEditor({
   // AI title generation is gated by the general AI chat flag (the same toggle
   // that enables the chat agent); the server enforces it too (#199).
   const isTitleGenEnabled = workspace?.settings?.ai?.chat === true;
-  const fullPageWidth = user.settings?.preferences?.fullPageWidth;
+  // `user` can momentarily be null during logout teardown (the currentUser atom
+  // is reset before this subtree unmounts). Optional-chain every access so the
+  // teardown render does not throw "Cannot read properties of null (reading
+  // 'settings')".
+  const fullPageWidth = user?.settings?.preferences?.fullPageWidth;
   const editorToolbarEnabled =
-    user.settings?.preferences?.editorToolbar ?? false;
+    user?.settings?.preferences?.editorToolbar ?? false;
   const [currentPageEditMode, setCurrentPageEditMode] = useAtom(
     currentPageEditModeAtom,
   );
   const userPageEditMode =
-    user.settings?.preferences?.pageEditMode ?? PageEditMode.Edit;
+    user?.settings?.preferences?.pageEditMode ?? PageEditMode.Edit;
   const isEditMode = currentPageEditMode === PageEditMode.Edit;
 
+  // Single shared Y.Doc + HocuspocusProvider for both the title and body
+  // editors (title lives in the 'title' fragment of the same doc).
+  const { ydoc, remote, providersReady } = usePageCollabProviders(pageId);
+
   // Apply the user's saved preference only once on initial load, not on every
   // page navigation — so the mode sticks across navigations within a session.
   useEffect(() => {
@@ -110,28 +120,32 @@ export function FullEditor({
       )}
       <MemoizedDeletedPageBanner slugId={slugId} />
       <MemoizedTemporaryNoteBanner slugId={slugId} />
-      <MemoizedTitleEditor
-        pageId={pageId}
-        slugId={slugId}
-        title={title}
-        spaceSlug={spaceSlug}
-        editable={editable}
-      />
-      <PageByline
-        pageId={pageId}
-        creator={creator}
-        contributors={contributors}
-        editable={editable}
-        isEditMode={isEditMode}
-        isDictationEnabled={isDictationEnabled}
-        isTitleGenEnabled={isTitleGenEnabled}
-      />
-      <MemoizedPageEditor
-        pageId={pageId}
-        editable={editable}
-        content={content}
-        canComment={canComment}
-      />
+      <EditorProvidersContext.Provider
+        value={ydoc && remote ? { ydoc, remote, providersReady } : null}
+      >
+        <MemoizedTitleEditor
+          pageId={pageId}
+          slugId={slugId}
+          title={title}
+          spaceSlug={spaceSlug}
+          editable={editable}
+        />
+        <PageByline
+          pageId={pageId}
+          creator={creator}
+          contributors={contributors}
+          editable={editable}
+          isEditMode={isEditMode}
+          isDictationEnabled={isDictationEnabled}
+          isTitleGenEnabled={isTitleGenEnabled}
+        />
+        <MemoizedPageEditor
+          pageId={pageId}
+          editable={editable}
+          content={content}
+          canComment={canComment}
+        />
+      </EditorProvidersContext.Provider>
     </Container>
   );
 }

apps/client/src/features/editor/hooks/collab-token.test.ts

@@ -0,0 +1,48 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+
+// jwt-decode is mocked so we can drive the four token states deterministically
+// (decode success with a chosen exp, or a thrown decode error).
+const decodeMock = vi.hoisted(() => vi.fn());
+vi.mock("jwt-decode", () => ({
+  jwtDecode: decodeMock,
+}));
+
+import { collabTokenNeedsRefresh } from "./collab-token";
+
+const NOW_MS = 1_000_000_000; // fixed "now" in ms (so NOW_MS/1000 seconds)
+
+beforeEach(() => {
+  decodeMock.mockReset();
+});
+
+describe("collabTokenNeedsRefresh", () => {
+  it("returns true when there is no token (fetch a fresh one)", () => {
+    expect(collabTokenNeedsRefresh(undefined, NOW_MS)).toBe(true);
+    // jwtDecode must not even be called for a missing token.
+    expect(decodeMock).not.toHaveBeenCalled();
+  });
+
+  it("returns true when the token is malformed (jwtDecode throws)", () => {
+    decodeMock.mockImplementation(() => {
+      throw new Error("invalid token");
+    });
+    expect(collabTokenNeedsRefresh("garbage", NOW_MS)).toBe(true);
+  });
+
+  it("returns false for a valid, not-yet-expired token (no reconnect)", () => {
+    // exp is in the future relative to NOW.
+    decodeMock.mockReturnValue({ exp: NOW_MS / 1000 + 60 });
+    expect(collabTokenNeedsRefresh("good", NOW_MS)).toBe(false);
+  });
+
+  it("returns true for a valid but expired token (refresh + reconnect)", () => {
+    // exp is in the past relative to NOW.
+    decodeMock.mockReturnValue({ exp: NOW_MS / 1000 - 60 });
+    expect(collabTokenNeedsRefresh("expired", NOW_MS)).toBe(true);
+  });
+
+  it("treats exp exactly equal to now as expired (>= boundary)", () => {
+    decodeMock.mockReturnValue({ exp: NOW_MS / 1000 });
+    expect(collabTokenNeedsRefresh("boundary", NOW_MS)).toBe(true);
+  });
+});

apps/client/src/features/editor/hooks/collab-token.ts

@@ -0,0 +1,26 @@
+import { jwtDecode } from "jwt-decode";
+
+/**
+ * Decide whether a collab token must be refreshed before reconnecting after an
+ * onAuthenticationFailed event. Pure and side-effect free so the four token
+ * states can be unit-tested directly:
+ *   - no token              -> true  (fetch a fresh one and reconnect)
+ *   - undecodable/malformed -> true  (jwtDecode throws -> refresh)
+ *   - valid, not expired    -> false (token is still good; do NOT reconnect)
+ *   - valid, expired        -> true  (refresh + reconnect)
+ *
+ * `nowMs` is injectable for deterministic tests; it defaults to `Date.now()`.
+ */
+export function collabTokenNeedsRefresh(
+  token: string | undefined,
+  nowMs: number = Date.now(),
+): boolean {
+  if (!token) return true;
+  try {
+    const payload = jwtDecode<{ exp: number }>(token);
+    return nowMs / 1000 >= payload.exp;
+  } catch {
+    // malformed/undecodable token -> refresh
+    return true;
+  }
+}

apps/client/src/features/editor/hooks/use-generate-page-title.test.tsx

@@ -139,7 +139,7 @@ describe("useGeneratePageTitle", () => {
     );
   });
 
-  it("happy path: applies the title, refreshes cache, writes the field, broadcasts", async () => {
+  it("happy path: applies the title, refreshes cache, broadcasts, and does NOT write the editor", async () => {
     const store = createStore();
     const titleEditor = makeTitleEditor();
     store.set(pageEditorAtom as never, makePageEditor("pageA"));
@@ -157,9 +157,11 @@ describe("useGeneratePageTitle", () => {
       title: "Generated Title",
     });
     expect(updatePageDataMock).toHaveBeenCalledWith(PAGE_A);
-    expect(titleEditor.commands.setContent).toHaveBeenCalledWith(
-      "Generated Title",
-    );
+    // The title editor is bound to the Yjs `title` fragment; the server REST
+    // update reseeds that fragment and the reseed reaches the bound editor on
+    // its own. Writing here too would double/garble the title, so the hook must
+    // NOT touch the editor (regression guard for the Yjs duplication trap).
+    expect(titleEditor.commands.setContent).not.toHaveBeenCalled();
     expect(localEmitMock).toHaveBeenCalled();
     expect(emitMock).toHaveBeenCalled();
     expect(notificationsShowMock).toHaveBeenCalledWith(
@@ -167,7 +169,7 @@ describe("useGeneratePageTitle", () => {
     );
   });
 
-  it("does NOT write the visible title field when the user navigated away during generation", async () => {
+  it("keeps the DB write keyed by the captured pageId and still broadcasts after navigation", async () => {
     const store = createStore();
     const titleEditor = makeTitleEditor(); // persistent across navigation
     store.set(pageEditorAtom as never, makePageEditor("pageA"));
@@ -203,55 +205,9 @@ describe("useGeneratePageTitle", () => {
       pageId: "pageA",
       title: "Generated Title",
     });
-    // ...but we must NOT stamp page A's title into page B's visible field.
+    // ...the hook never writes the editor regardless of navigation...
     expect(titleEditor.commands.setContent).not.toHaveBeenCalled();
-    // The change is still broadcast to other clients.
-    expect(emitMock).toHaveBeenCalled();
-  });
-
-  it("does NOT write the visible title field when the title editor is focused", async () => {
-    const store = createStore();
-    const titleEditor = makeTitleEditor();
-    store.set(pageEditorAtom as never, makePageEditor("pageA"));
-    store.set(titleEditorAtom as never, titleEditor);
-
-    // Resolve generation under our control so we can mark the live title editor
-    // as focused before the post-generation write runs.
-    let resolveTitle!: (t: string) => void;
-    generatePageTitleMock.mockReturnValue(
-      new Promise<string>((res) => {
-        resolveTitle = res;
-      }),
-    );
-    updateTitleMock.mockResolvedValue(PAGE_A);
-    const { result } = setup("pageA", store);
-
-    let pending!: Promise<void>;
-    act(() => {
-      pending = result.current.mutateAsync();
-    });
-
-    // The user clicked into the title field while the model ran — overwriting it
-    // now would clobber what they are actively typing.
-    act(() => {
-      (titleEditor as { isFocused: boolean }).isFocused = true;
-    });
-
-    await act(async () => {
-      resolveTitle("Generated Title");
-      await pending;
-    });
-
-    // The DB write still persists the value...
-    expect(updateTitleMock).toHaveBeenCalledWith({
-      pageId: "pageA",
-      title: "Generated Title",
-    });
-    expect(updatePageDataMock).toHaveBeenCalledWith(PAGE_A);
-    // ...but the visible field is left alone while it is focused.
-    expect(titleEditor.commands.setContent).not.toHaveBeenCalled();
-    // The change is still broadcast to other clients.
-    expect(localEmitMock).toHaveBeenCalled();
+    // ...and the change is still broadcast to other clients.
     expect(emitMock).toHaveBeenCalled();
   });
 

apps/client/src/features/editor/hooks/use-generate-page-title.ts

@@ -1,13 +1,9 @@
-import { useRef } from "react";
 import { useMutation } from "@tanstack/react-query";
 import { useAtomValue } from "jotai";
 import { notifications } from "@mantine/notifications";
 import { useTranslation } from "react-i18next";
 import { htmlToMarkdown } from "@docmost/editor-ext";
-import {
-  pageEditorAtom,
-  titleEditorAtom,
-} from "@/features/editor/atoms/editor-atoms.ts";
+import { pageEditorAtom } from "@/features/editor/atoms/editor-atoms.ts";
 import {
   updatePageData,
   useUpdateTitlePageMutation,
@@ -33,18 +29,9 @@ const MAX_CONTENT_CHARS = 20000;
 export function useGeneratePageTitle(pageId: string) {
   const { t } = useTranslation();
   const pageEditor = useAtomValue(pageEditorAtom);
-  const titleEditor = useAtomValue(titleEditorAtom);
   const { mutateAsync: updateTitle } = useUpdateTitlePageMutation();
   const emit = useQueryEmit();
 
-  // The page/title editors come from GLOBAL atoms that re-point when the user
-  // navigates to another page. The mutation below awaits the model for 1-3s, and
-  // its closure captures the editors from the render that started it. Keep a live
-  // reference so the post-generation write targets whatever page is on screen
-  // *now*, not the page the generation was started from.
-  const editorsRef = useRef({ pageEditor, titleEditor });
-  editorsRef.current = { pageEditor, titleEditor };
-
   return useMutation<void, Error, void>({
     mutationFn: async () => {
       if (!pageEditor || pageEditor.isDestroyed) return;
@@ -70,33 +57,15 @@ export function useGeneratePageTitle(pageId: string) {
       const page = await updateTitle({ pageId, title }); // POST /pages/update
       updatePageData(page); // refresh the react-query cache
 
-      // Reflect the new title in the field immediately. The button lives in the
-      // byline, so the title editor is not focused — setContent is safe and stays
-      // undoable through its History extension (Ctrl/Cmd+Z reverts the change).
-      //
-      // Guard against navigation during generation: if the user switched pages
-      // while the model ran, the (persistent) title editor now shows ANOTHER
-      // page, so writing here would drop page A's title into page B's visible
-      // field. page-editor.tsx stamps the live page editor with its pageId
-      // (`editor.storage.pageId`), mirroring TitleEditor's `activePageId !==
-      // pageId` guard — bail the visible write unless that live editor still
-      // belongs to the page this title was generated for. The DB write above is
-      // already correct (keyed by the captured `pageId`), and the broadcast below
-      // still propagates page A's change to other clients.
-      const livePageEditor = editorsRef.current.pageEditor;
-      const liveTitleEditor = editorsRef.current.titleEditor;
-      // `storage.pageId` is stamped untyped in page-editor.tsx's onCreate.
-      const livePageId = (livePageEditor?.storage as { pageId?: string })
-        ?.pageId;
-      const stillOnPage = livePageId === pageId;
-      if (
-        stillOnPage &&
-        liveTitleEditor &&
-        !liveTitleEditor.isDestroyed &&
-        !liveTitleEditor.isFocused
-      ) {
-        liveTitleEditor.commands.setContent(page.title);
-      }
+      // Do NOT write the title into the editor here. The title editor is bound to
+      // the Yjs `title` fragment and Yjs is the source of truth. The server REST
+      // /pages/update reseeds that fragment (writePageTitle → writeTitleFragment,
+      // a full clear+replace) and the reseed reaches the bound title editor on
+      // its own as a remote provider update. The old REST-era setContent here
+      // would race that reseed and double/garble the title (the "Yjs duplication
+      // trap"), so it is intentionally omitted. The DB write above is keyed by
+      // the captured `pageId`, so it stays correct even if the user navigated
+      // away during generation.
 
       // Broadcast to other clients, mirroring TitleEditor.saveTitle's event shape.
       const event: UpdateEvent = {

apps/client/src/features/editor/hooks/use-page-collab-providers.ts

@@ -0,0 +1,189 @@
+import { useEffect, useRef, useState } from "react";
+import { IndexeddbPersistence } from "y-indexeddb";
+import * as Y from "yjs";
+import {
+  HocuspocusProvider,
+  onStatusParameters,
+  WebSocketStatus,
+  HocuspocusProviderWebsocket,
+  onSyncedParameters,
+  onStatelessParameters,
+} from "@hocuspocus/provider";
+import { useAtom, useSetAtom } from "jotai";
+import useCollaborationUrl from "@/features/editor/hooks/use-collaboration-url";
+import {
+  isLocalSyncedAtom,
+  isRemoteSyncedAtom,
+  yjsConnectionStatusAtom,
+} from "@/features/editor/atoms/editor-atoms";
+import { useCollabToken } from "@/features/auth/queries/auth-query.tsx";
+import { useDocumentVisibility } from "@mantine/hooks";
+import { useIdle } from "@/hooks/use-idle.ts";
+import { queryClient } from "@/main.tsx";
+import { IPage } from "@/features/page/types/page.types.ts";
+import { useParams } from "react-router-dom";
+import { extractPageSlugId } from "@/lib";
+import { FIVE_MINUTES } from "@/lib/constants.ts";
+import { collabTokenNeedsRefresh } from "@/features/editor/hooks/collab-token";
+import { pageYdocName } from "@/features/editor/page-ydoc-name";
+import { pageKeys } from "@/features/page/queries/page-query";
+
+export interface PageCollabProviders {
+  ydoc: Y.Doc | null;
+  remote: HocuspocusProvider | null;
+  socket: HocuspocusProviderWebsocket | null;
+  providersReady: boolean;
+}
+
+/**
+ * Owns the full collaboration provider lifecycle for a page so that the title
+ * and body editors can share a single Y.Doc + HocuspocusProvider. The behavior
+ * is relocated verbatim from page-editor.tsx: it creates the providers once per
+ * pageId, connects/disconnects on idle/visibility, attaches each render,
+ * destroys on unmount, refreshes the collab token on auth failure, and applies
+ * the onStateless 'page.updated' cache update.
+ */
+export function usePageCollabProviders(pageId: string): PageCollabProviders {
+  const collaborationURL = useCollaborationUrl();
+  const [yjsConnectionStatus, setYjsConnectionStatus] = useAtom(
+    yjsConnectionStatusAtom,
+  );
+  const setIsLocalSyncedAtom = useSetAtom(isLocalSyncedAtom);
+  const setIsRemoteSyncedAtom = useSetAtom(isRemoteSyncedAtom);
+  const { data: collabQuery, refetch: refetchCollabToken } = useCollabToken();
+  // The provider-creating effect runs only once per pageId, so any token read
+  // inside its handlers would be captured STALE (the old token at first render).
+  // Mirror the latest token into a ref the auth-failure handler can read live.
+  const collabTokenRef = useRef<string | undefined>(undefined);
+  useEffect(() => {
+    collabTokenRef.current = collabQuery?.token;
+  }, [collabQuery?.token]);
+  const { isIdle, resetIdle } = useIdle(FIVE_MINUTES, { initialState: false });
+  const documentState = useDocumentVisibility();
+  const { pageSlug } = useParams();
+  const slugId = extractPageSlugId(pageSlug);
+
+  // Providers only created once per pageId
+  const providersRef = useRef<{
+    ydoc: Y.Doc;
+    local: IndexeddbPersistence;
+    remote: HocuspocusProvider;
+    socket: HocuspocusProviderWebsocket;
+  } | null>(null);
+  const [providersReady, setProvidersReady] = useState(false);
+
+  useEffect(() => {
+    if (!providersRef.current) {
+      const documentName = pageYdocName(pageId);
+      const ydoc = new Y.Doc();
+      const local = new IndexeddbPersistence(documentName, ydoc);
+      const socket = new HocuspocusProviderWebsocket({
+        url: collaborationURL,
+      });
+      const onLocalSyncedHandler = () => {
+        setIsLocalSyncedAtom(true);
+      };
+      const onStatusHandler = (event: onStatusParameters) => {
+        setYjsConnectionStatus(event.status);
+      };
+      const onSyncedHandler = (event: onSyncedParameters) => {
+        setIsRemoteSyncedAtom(event.state);
+      };
+      const onStatelessHandler = ({ payload }: onStatelessParameters) => {
+        try {
+          const message = JSON.parse(payload);
+          if (message?.type !== "page.updated" || !message.updatedAt) return;
+          const pageData = queryClient.getQueryData<IPage>(
+            pageKeys.detail(slugId),
+          );
+          if (pageData) {
+            queryClient.setQueryData(pageKeys.detail(slugId), {
+              ...pageData,
+              updatedAt: message.updatedAt,
+              ...(message.lastUpdatedBy && {
+                lastUpdatedBy: message.lastUpdatedBy,
+              }),
+            });
+          }
+        } catch {
+          // ignore unrelated stateless messages
+        }
+      };
+      const onAuthenticationFailedHandler = () => {
+        // Read the token from the ref, not the closed-over `collabQuery`: this
+        // handler is created once and would otherwise decode a stale token after
+        // a refetch. A missing/malformed token must NOT crash the handler —
+        // jwtDecode(undefined) throws — so treat any decode failure as "needs
+        // refresh" and proceed to refetch + reconnect instead of getting stuck.
+        if (!collabTokenNeedsRefresh(collabTokenRef.current)) return;
+        refetchCollabToken().then((result) => {
+          if (result.data?.token) {
+            socket.disconnect();
+            setTimeout(() => {
+              remote.configuration.token = result.data.token;
+              socket.connect();
+            }, 100);
+          }
+        });
+      };
+      const remote = new HocuspocusProvider({
+        websocketProvider: socket,
+        name: documentName,
+        document: ydoc,
+        token: collabQuery?.token,
+        onAuthenticationFailed: onAuthenticationFailedHandler,
+        onStatus: onStatusHandler,
+        onSynced: onSyncedHandler,
+        onStateless: onStatelessHandler,
+      });
+
+      local.on("synced", onLocalSyncedHandler);
+      providersRef.current = { ydoc, socket, local, remote };
+      setProvidersReady(true);
+    } else {
+      setProvidersReady(true);
+    }
+    // Only destroy on final unmount
+    return () => {
+      providersRef.current?.socket.destroy();
+      providersRef.current?.remote.destroy();
+      providersRef.current?.local.destroy();
+      providersRef.current = null;
+      // Reset shared sync state on page change/unmount.
+      setIsLocalSyncedAtom(false);
+      setIsRemoteSyncedAtom(false);
+    };
+  }, [pageId]);
+
+  // Only connect/disconnect on tab/idle, not destroy
+  useEffect(() => {
+    if (!providersReady || !providersRef.current) return;
+    const socket = providersRef.current.socket;
+
+    if (
+      isIdle &&
+      documentState === "hidden" &&
+      yjsConnectionStatus === WebSocketStatus.Connected
+    ) {
+      socket.disconnect();
+      return;
+    }
+    if (
+      documentState === "visible" &&
+      yjsConnectionStatus === WebSocketStatus.Disconnected
+    ) {
+      resetIdle();
+      socket.connect();
+    }
+  }, [isIdle, documentState, providersReady, resetIdle]);
+
+  // Attach here, to make sure the connection gets properly established
+  providersRef.current?.remote.attach();
+
+  return {
+    ydoc: providersRef.current?.ydoc ?? null,
+    remote: providersRef.current?.remote ?? null,
+    socket: providersRef.current?.socket ?? null,
+    providersReady,
+  };
+}

apps/client/src/features/editor/page-editor.tsx

@@ -6,16 +6,7 @@ import React, {
   useRef,
   useState,
 } from "react";
-import { IndexeddbPersistence } from "y-indexeddb";
-import * as Y from "yjs";
-import {
-  HocuspocusProvider,
-  onStatusParameters,
-  WebSocketStatus,
-  HocuspocusProviderWebsocket,
-  onSyncedParameters,
-  onStatelessParameters,
-} from "@hocuspocus/provider";
+import { WebSocketStatus } from "@hocuspocus/provider";
 import {
   Editor,
   EditorContent,
@@ -28,13 +19,15 @@ import {
   mainExtensions,
 } from "@/features/editor/extensions/extensions";
 import { useAtom, useAtomValue } from "jotai";
-import useCollaborationUrl from "@/features/editor/hooks/use-collaboration-url";
 import { currentUserAtom } from "@/features/user/atoms/current-user-atom";
 import {
   currentPageEditModeAtom,
+  isLocalSyncedAtom,
+  isRemoteSyncedAtom,
   pageEditorAtom,
   yjsConnectionStatusAtom,
 } from "@/features/editor/atoms/editor-atoms";
+import { useEditorProviders } from "@/features/editor/contexts/editor-providers-context";
 import { asideStateAtom } from "@/components/layouts/global/hooks/atoms/sidebar-atom";
 import {
   activeCommentIdAtom,
@@ -58,10 +51,8 @@ import {
 } from "@/features/editor/components/common/editor-paste-handler.tsx";
 import ExcalidrawMenu from "./components/excalidraw/excalidraw-menu-lazy";
 import DrawioMenu from "./components/drawio/drawio-menu";
-import { useCollabToken } from "@/features/auth/queries/auth-query.tsx";
 import SearchAndReplaceDialog from "@/features/editor/components/search-and-replace/search-and-replace-dialog.tsx";
-import { useDebouncedCallback, useDocumentVisibility } from "@mantine/hooks";
-import { useIdle } from "@/hooks/use-idle.ts";
+import { useDebouncedCallback } from "@mantine/hooks";
 import { queryClient } from "@/main.tsx";
 import { IPage } from "@/features/page/types/page.types.ts";
 import { useParams } from "react-router-dom";
@@ -72,9 +63,7 @@ import {
   GitmostInsertRecordingResult,
   gitmostInsertRecordingIntoEditor,
 } from "@/features/editor/gitmost/gitmost-recording.ts";
-import { FIVE_MINUTES } from "@/lib/constants.ts";
 import { PageEditMode } from "@/features/user/types/user.types.ts";
-import { jwtDecode } from "jwt-decode";
 import { searchSpotlight } from "@/features/search/constants.ts";
 import { useEditorScroll } from "./hooks/use-editor-scroll";
 import { EditorLinkMenu } from "@/features/editor/components/link/link-menu";
@@ -103,7 +92,6 @@ export default function PageEditor({
   canComment,
 }: PageEditorProps) {
   const { t } = useTranslation();
-  const collaborationURL = useCollaborationUrl();
   const isComponentMounted = useRef(false);
   const editorRef = useRef<Editor | null>(null);
 
@@ -117,22 +105,10 @@ export default function PageEditor({
   const [, setActiveCommentId] = useAtom(activeCommentIdAtom);
   const [showCommentPopup, setShowCommentPopup] = useAtom(showCommentPopupAtom);
   const [showReadOnlyCommentPopup] = useAtom(showReadOnlyCommentPopupAtom);
-  const [isLocalSynced, setIsLocalSynced] = useState(false);
-  const [isRemoteSynced, setIsRemoteSynced] = useState(false);
   const [yjsConnectionStatus, setYjsConnectionStatus] = useAtom(
     yjsConnectionStatusAtom,
   );
   const menuContainerRef = useRef(null);
-  const { data: collabQuery, refetch: refetchCollabToken } = useCollabToken();
-  // Always holds the latest collab token. The provider effect below runs once
-  // per pageId, so a handler created inside it would otherwise close over a
-  // stale `collabQuery`. Reading the ref gives the current token instead.
-  const collabTokenRef = useRef<string | undefined>(undefined);
-  useEffect(() => {
-    collabTokenRef.current = collabQuery?.token;
-  }, [collabQuery?.token]);
-  const { isIdle, resetIdle } = useIdle(FIVE_MINUTES, { initialState: false });
-  const documentState = useDocumentVisibility();
   const { pageSlug } = useParams();
   const slugId = extractPageSlugId(pageSlug);
   const currentPageEditMode = useAtomValue(currentPageEditModeAtom);
@@ -141,141 +117,27 @@ export default function PageEditor({
     [isComponentMounted],
   );
   const { handleScrollTo } = useEditorScroll({ canScroll });
-  // Providers only created once per pageId
-  const providersRef = useRef<{
-    local: IndexeddbPersistence;
-    remote: HocuspocusProvider;
-    socket: HocuspocusProviderWebsocket;
-  } | null>(null);
-  const [providersReady, setProvidersReady] = useState(false);
 
-  useEffect(() => {
-    if (!providersRef.current) {
-      const documentName = `page.${pageId}`;
-      const ydoc = new Y.Doc();
-      const local = new IndexeddbPersistence(documentName, ydoc);
-      const socket = new HocuspocusProviderWebsocket({
-        url: collaborationURL,
-      });
-      const onLocalSyncedHandler = () => {
-        setIsLocalSynced(true);
-      };
-      const onStatusHandler = (event: onStatusParameters) => {
-        setYjsConnectionStatus(event.status);
-      };
-      const onSyncedHandler = (event: onSyncedParameters) => {
-        setIsRemoteSynced(event.state);
-      };
-      const onStatelessHandler = ({ payload }: onStatelessParameters) => {
-        try {
-          const message = JSON.parse(payload);
-          if (message?.type !== "page.updated" || !message.updatedAt) return;
-          const pageData = queryClient.getQueryData<IPage>(["pages", slugId]);
-          if (pageData) {
-            queryClient.setQueryData(["pages", slugId], {
-              ...pageData,
-              updatedAt: message.updatedAt,
-              ...(message.lastUpdatedBy && {
-                lastUpdatedBy: message.lastUpdatedBy,
-              }),
-            });
-          }
-        } catch {
-          // ignore unrelated stateless messages
-        }
-      };
-      const onAuthenticationFailedHandler = () => {
-        // Read the latest token via the ref (the closure-captured `collabQuery`
-        // may be stale). Guard the decode: a missing or unparseable token must
-        // not throw "Invalid token specified" and should trigger a refresh so
-        // the editor reconnects even when the initial token fetch failed.
-        const token = collabTokenRef.current;
-        let needsRefresh = true; // no/unparseable token -> fetch a fresh one and reconnect
-        if (token) {
-          try {
-            // A token that decodes but lacks a numeric `exp` must be treated as
-            // expired (`Date.now()/1000 >= undefined` is `false`, which would
-            // otherwise skip the reconnect), so refresh on any missing/non-number exp.
-            const exp = jwtDecode<{ exp?: number }>(token).exp;
-            needsRefresh = typeof exp !== "number" || Date.now() / 1000 >= exp;
-          } catch {
-            needsRefresh = true;
-          }
-        }
-        if (!needsRefresh) return;
-        refetchCollabToken().then((result) => {
-          if (result.data?.token) {
-            socket.disconnect();
-            setTimeout(() => {
-              remote.configuration.token = result.data.token;
-              socket.connect();
-            }, 100);
-          }
-        });
-      };
-      const remote = new HocuspocusProvider({
-        websocketProvider: socket,
-        name: documentName,
-        document: ydoc,
-        token: collabQuery?.token,
-        onAuthenticationFailed: onAuthenticationFailedHandler,
-        onStatus: onStatusHandler,
-        onSynced: onSyncedHandler,
-        onStateless: onStatelessHandler,
-      });
-
-      local.on("synced", onLocalSyncedHandler);
-      providersRef.current = { socket, local, remote };
-      setProvidersReady(true);
-    } else {
-      setProvidersReady(true);
-    }
-    // Only destroy on final unmount
-    return () => {
-      providersRef.current?.socket.destroy();
-      providersRef.current?.remote.destroy();
-      providersRef.current?.local.destroy();
-      providersRef.current = null;
-    };
-  }, [pageId]);
-
-  // Only connect/disconnect on tab/idle, not destroy
-  useEffect(() => {
-    if (!providersReady || !providersRef.current) return;
-    const socket = providersRef.current.socket;
-
-    if (
-      isIdle &&
-      documentState === "hidden" &&
-      yjsConnectionStatus === WebSocketStatus.Connected
-    ) {
-      socket.disconnect();
-      return;
-    }
-    if (
-      documentState === "visible" &&
-      yjsConnectionStatus === WebSocketStatus.Disconnected
-    ) {
-      resetIdle();
-      socket.connect();
-    }
-  }, [isIdle, documentState, providersReady, resetIdle]);
-
-  // Attach here, to make sure the connection gets properly established
-  providersRef.current?.remote.attach();
+  // Shared providers + Y.Doc lifted into full-editor via context. The provider
+  // lifecycle (creation, idle/visibility connect, attach, destroy, token
+  // refresh) lives in usePageCollabProviders. Null-safe when rendered without
+  // the context (defensive) — in practice full-editor always provides it.
+  const editorProviders = useEditorProviders();
+  const remote = editorProviders?.remote ?? null;
+  const providersReady = editorProviders?.providersReady ?? false;
+  const isLocalSynced = useAtomValue(isLocalSyncedAtom);
+  const isRemoteSynced = useAtomValue(isRemoteSyncedAtom);
 
   const extensions = useMemo(() => {
-    if (!providersReady || !providersRef.current || !currentUser?.user) {
+    if (!providersReady || !remote || !currentUser?.user) {
       return mainExtensions;
     }
 
-    const remoteProvider = providersRef.current.remote;
-
     return [
       ...mainExtensions,
-      ...collabExtensions(remoteProvider, currentUser?.user),
+      ...collabExtensions(remote, currentUser?.user),
     ];
-  }, [providersReady, currentUser?.user]);
+  }, [providersReady, remote, currentUser?.user]);
 
   const editor = useEditor(
     {
@@ -550,7 +412,7 @@ export default function PageEditor({
             {editor &&
               !editorIsEditable &&
               (editable || canComment) &&
-              providersRef.current && <ReadonlyBubbleMenu editor={editor} />}
+              remote && <ReadonlyBubbleMenu editor={editor} />}
             {showCommentPopup && (
               <CommentDialog editor={editor} pageId={pageId} />
             )}

apps/client/src/features/editor/page-ydoc-name.ts

@@ -0,0 +1,14 @@
+/**
+ * Single source of truth for the IndexedDB / Hocuspocus document name of a
+ * page's collaborative Yjs doc.
+ *
+ * The `page.<id>` convention is shared knowledge across three call sites: the
+ * live editor providers (`use-page-collab-providers`), the offline warm path
+ * (`make-offline`), and the offline purge (`clear-offline-cache`, which matches
+ * the databases to delete by this prefix). Centralizing it here stops those
+ * sites from silently drifting apart.
+ */
+export const PAGE_YDOC_NAME_PREFIX = "page.";
+
+export const pageYdocName = (pageId: string): string =>
+  `${PAGE_YDOC_NAME_PREFIX}${pageId}`;

apps/client/src/features/editor/title-collab.test.ts

@@ -0,0 +1,33 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+
+// isChangeOrigin is mocked so we can simulate local vs remote/collab-origin
+// transactions without constructing a real ProseMirror/Yjs transaction.
+const isChangeOriginMock = vi.hoisted(() => vi.fn());
+vi.mock("@tiptap/extension-collaboration", () => ({
+  isChangeOrigin: isChangeOriginMock,
+}));
+
+import { shouldPropagateTitleChange } from "./title-collab";
+
+beforeEach(() => {
+  isChangeOriginMock.mockReset();
+});
+
+describe("shouldPropagateTitleChange", () => {
+  it("propagates a genuine local edit (isChangeOrigin false)", () => {
+    isChangeOriginMock.mockReturnValue(false);
+    expect(shouldPropagateTitleChange({ local: true })).toBe(true);
+    expect(isChangeOriginMock).toHaveBeenCalledWith({ local: true });
+  });
+
+  it("skips a remote/collab-origin update (isChangeOrigin true)", () => {
+    isChangeOriginMock.mockReturnValue(true);
+    expect(shouldPropagateTitleChange({ remote: true })).toBe(false);
+  });
+
+  it("propagates when there is no transaction (treated as local)", () => {
+    expect(shouldPropagateTitleChange(undefined)).toBe(true);
+    // isChangeOrigin must not be called for a missing transaction.
+    expect(isChangeOriginMock).not.toHaveBeenCalled();
+  });
+});

apps/client/src/features/editor/title-collab.ts

@@ -0,0 +1,19 @@
+import { isChangeOrigin } from "@tiptap/extension-collaboration";
+
+/**
+ * Whether a TitleEditor `onUpdate` should drive URL + tree propagation.
+ *
+ * Only genuine LOCAL edits propagate. Remote/collab-origin Yjs updates
+ * (detected via `isChangeOrigin`) are skipped so a remote title change is not
+ * re-broadcast back, which would create a feedback loop. A missing transaction
+ * is treated as a local edit (propagate).
+ *
+ * Extracted as a pure helper so the skip decision is unit-testable without
+ * mounting the full collaborative editor.
+ */
+export function shouldPropagateTitleChange(transaction: unknown): boolean {
+  return !(
+    transaction &&
+    isChangeOrigin(transaction as Parameters<typeof isChangeOrigin>[0])
+  );
+}

apps/client/src/features/editor/title-editor.test.tsx

@@ -0,0 +1,87 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+import { render, screen } from "@testing-library/react";
+
+// Drive the fallback-vs-collaborative switch (titleReady = providersReady &&
+// !!ydoc) by controlling what the editor-providers context returns.
+const editorProvidersValue: { ydoc: unknown; providersReady: boolean } = {
+  ydoc: null,
+  providersReady: false,
+};
+vi.mock("@/features/editor/contexts/editor-providers-context", () => ({
+  useEditorProviders: () => editorProvidersValue,
+}));
+
+// Mock the tiptap React bindings so the test does not mount a real editor:
+// useEditor returns a minimal stub and EditorContent renders a marker.
+vi.mock("@tiptap/react", () => ({
+  useEditor: () => ({
+    isInitialized: true,
+    commands: { focus: vi.fn() },
+    setEditable: vi.fn(),
+    getText: () => "",
+  }),
+  EditorContent: () => <div data-testid="collab-editor" />,
+}));
+
+vi.mock("react-i18next", () => ({
+  useTranslation: () => ({ t: (k: string) => k }),
+}));
+
+const navigateMock = vi.fn();
+vi.mock("react-router-dom", () => ({
+  useNavigate: () => navigateMock,
+}));
+
+vi.mock("@/features/websocket/use-query-emit.ts", () => ({
+  useQueryEmit: () => vi.fn(),
+}));
+
+// page-query transitively imports @/main.tsx; mock it to a pure stub.
+vi.mock("@/features/page/queries/page-query", () => ({
+  updatePageData: vi.fn(),
+}));
+vi.mock("@/main.tsx", () => ({
+  queryClient: { getQueryData: vi.fn(), setQueryData: vi.fn() },
+}));
+
+import { TitleEditor } from "./title-editor";
+
+const baseProps = {
+  pageId: "p1",
+  slugId: "slug-1",
+  title: "My Page Title",
+  spaceSlug: "space",
+  editable: true,
+};
+
+beforeEach(() => {
+  navigateMock.mockReset();
+  editorProvidersValue.ydoc = null;
+  editorProvidersValue.providersReady = false;
+});
+
+describe("TitleEditor fallback vs collaborative switch", () => {
+  it("renders a static <h1> with the title before the shared doc is ready", () => {
+    editorProvidersValue.ydoc = null;
+    editorProvidersValue.providersReady = false;
+
+    render(<TitleEditor {...baseProps} />);
+
+    const heading = screen.getByRole("heading", { level: 1 });
+    expect(heading.textContent).toBe("My Page Title");
+    // The collaborative editor must NOT mount until the doc is ready.
+    expect(screen.queryByTestId("collab-editor")).toBeNull();
+  });
+
+  it("renders the collaborative editor once the shared doc is ready", () => {
+    editorProvidersValue.ydoc = {}; // truthy shared doc
+    editorProvidersValue.providersReady = true;
+
+    render(<TitleEditor {...baseProps} />);
+
+    expect(screen.getByTestId("collab-editor")).toBeDefined();
+    // The static fallback <h1> is gone — Yjs is the single source of truth and
+    // the prop is never seeded into the collaborative editor.
+    expect(screen.queryByRole("heading", { level: 1 })).toBeNull();
+  });
+});

apps/client/src/features/editor/title-editor.tsx

@@ -1,5 +1,5 @@
 import "@/features/editor/styles/index.css";
-import React, { useCallback, useEffect, useState } from "react";
+import { useEffect } from "react";
 import { EditorContent, useEditor } from "@tiptap/react";
 import { Document } from "@tiptap/extension-document";
 import { Heading } from "@tiptap/extension-heading";
@@ -11,14 +11,11 @@ import {
   pageEditorAtom,
   titleEditorAtom,
 } from "@/features/editor/atoms/editor-atoms";
-import {
-  updatePageData,
-  useUpdateTitlePageMutation,
-} from "@/features/page/queries/page-query";
+import { pageKeys, updatePageData } from "@/features/page/queries/page-query";
 import { useDebouncedCallback, getHotkeyHandler } from "@mantine/hooks";
 import { useAtom } from "jotai";
-import { useQueryEmit } from "@/features/websocket/use-query-emit.ts";
-import { History } from "@tiptap/extension-history";
+import { Collaboration } from "@tiptap/extension-collaboration";
+import { shouldPropagateTitleChange } from "@/features/editor/title-collab";
 import { buildPageUrl } from "@/features/page/page.utils.ts";
 import { useNavigate } from "react-router-dom";
 import { useTranslation } from "react-i18next";
@@ -28,6 +25,9 @@ import localEmitter from "@/lib/local-emitter.ts";
 import { PageEditMode } from "@/features/user/types/user.types.ts";
 import { searchSpotlight } from "@/features/search/constants.ts";
 import { platformModifierKey } from "@/lib";
+import { useEditorProviders } from "@/features/editor/contexts/editor-providers-context";
+import { queryClient } from "@/main.tsx";
+import { IPage } from "@/features/page/types/page.types.ts";
 
 export interface TitleEditorProps {
   pageId: string;
@@ -45,65 +45,82 @@ export function TitleEditor({
   editable,
 }: TitleEditorProps) {
   const { t } = useTranslation();
-  const { mutateAsync: updateTitlePageMutationAsync } =
-    useUpdateTitlePageMutation();
   const pageEditor = useAtomValue(pageEditorAtom);
   const [, setTitleEditor] = useAtom(titleEditorAtom);
-  const emit = useQueryEmit();
   const navigate = useNavigate();
-  const [activePageId, setActivePageId] = useState(pageId);
   const currentPageEditMode = useAtomValue(currentPageEditModeAtom);
 
-  const titleEditor = useEditor({
-    extensions: [
-      Document.extend({
-        content: "heading",
-      }),
-      Heading.configure({
-        levels: [1],
-      }),
-      Text,
-      Placeholder.configure({
-        placeholder: t("Untitled"),
-        showOnlyWhenEditable: false,
-      }),
-      History.configure({
-        depth: 20,
-      }),
-      EmojiCommand,
-    ],
-    onCreate({ editor }) {
-      if (editor) {
-        // @ts-ignore
-        setTitleEditor(editor);
-        setActivePageId(pageId);
-      }
-    },
-    onUpdate({ editor }) {
-      debounceUpdate();
-    },
-    editable: editable,
-    content: title,
-    immediatelyRender: true,
-    shouldRerenderOnTransaction: false,
-    editorProps: {
-      attributes: {
-        "aria-label": t("Page title"),
+  // Shared Y.Doc (title lives in its own 'title' fragment of the same doc as
+  // the body). Yjs is the source of truth for the title content.
+  const editorProviders = useEditorProviders();
+  const ydoc = editorProviders?.ydoc ?? null;
+  const providersReady = editorProviders?.providersReady ?? false;
+
+  // Until the shared doc is ready, the collaborative editor binds nothing and
+  // would render an empty heading until the Yjs 'title' fragment hydrates. Show
+  // a non-editable static <h1> with the `title` prop in the meantime. The prop
+  // is NEVER fed into the collaborative editor (Yjs stays the single source of
+  // truth — seeding it would duplicate the title).
+  const titleReady = providersReady && !!ydoc;
+
+  const titleEditor = useEditor(
+    {
+      extensions: [
+        Document.extend({
+          content: "heading",
+        }),
+        Heading.configure({
+          levels: [1],
+        }),
+        Text,
+        Placeholder.configure({
+          placeholder: t("Untitled"),
+          showOnlyWhenEditable: false,
+        }),
+        // Bind the title to the dedicated 'title' fragment of the shared doc.
+        // Collaboration also manages undo/redo, so the History extension is
+        // intentionally omitted (it would conflict with Yjs). When the doc is
+        // not ready yet the editor renders empty until the doc arrives.
+        ...(ydoc
+          ? [Collaboration.configure({ document: ydoc, field: "title" })]
+          : []),
+        EmojiCommand,
+      ],
+      onCreate({ editor }) {
+        if (editor) {
+          // @ts-ignore
+          setTitleEditor(editor);
+        }
       },
-      handleDOMEvents: {
-        keydown: (_view, event) => {
-          if (platformModifierKey(event) && event.code === "KeyS") {
-            event.preventDefault();
-            return true;
-          }
-          if (platformModifierKey(event) && event.code === "KeyK") {
-            searchSpotlight.open();
-            return true;
-          }
+      onUpdate({ editor, transaction }) {
+        // Drive URL + tree propagation only on genuine local edits; skip
+        // remote/collab-origin Yjs updates to avoid feedback loops.
+        if (!shouldPropagateTitleChange(transaction)) return;
+        debouncedPropagateTitle(editor.getText());
+      },
+      editable: editable,
+      immediatelyRender: true,
+      shouldRerenderOnTransaction: false,
+      editorProps: {
+        attributes: {
+          "aria-label": t("Page title"),
+        },
+        handleDOMEvents: {
+          keydown: (_view, event) => {
+            if (platformModifierKey(event) && event.code === "KeyS") {
+              event.preventDefault();
+              return true;
+            }
+            if (platformModifierKey(event) && event.code === "KeyK") {
+              searchSpotlight.open();
+              return true;
+            }
+          },
         },
       },
     },
-  });
+    [pageId, ydoc],
+  );
 
   useEffect(() => {
     const anchorId = window.location.hash
@@ -113,59 +130,45 @@ export function TitleEditor({
     navigate(pageSlug, { replace: true });
   }, [title]);
 
-  const saveTitle = useCallback(() => {
-    if (!titleEditor || activePageId !== pageId) return;
-
-    if (
-      titleEditor.getText() === title ||
-      (titleEditor.getText() === "" && title === null)
-    ) {
-      return;
-    }
-
-    updateTitlePageMutationAsync({
-      pageId: pageId,
-      title: titleEditor.getText(),
-    }).then((page) => {
-      const event: UpdateEvent = {
-        operation: "updateOne",
-        spaceId: page.spaceId,
-        entity: ["pages"],
-        id: page.id,
-        payload: {
-          title: page.title,
-          slugId: page.slugId,
-          parentPageId: page.parentPageId,
-          icon: page.icon,
-        },
-      };
-
-      if (page.title !== titleEditor.getText()) return;
-
-      updatePageData(page);
-
-      localEmitter.emit("message", event);
-      emit(event);
+  // On a local title change: update the URL slug and propagate the change to
+  // the live tree/breadcrumbs for online users. No REST round-trip — the title
+  // itself is persisted through Yjs. Offline this simply no-ops the socket
+  // emit and the title syncs on reconnect.
+  const debouncedPropagateTitle = useDebouncedCallback((titleText: string) => {
+    const anchorId = window.location.hash
+      ? window.location.hash.substring(1)
+      : undefined;
+    navigate(buildPageUrl(spaceSlug, slugId, titleText, anchorId), {
+      replace: true,
     });
-  }, [pageId, title, titleEditor]);
 
-  const debounceUpdate = useDebouncedCallback(saveTitle, 500);
+    const page =
+      queryClient.getQueryData<IPage>(pageKeys.detail(slugId)) ??
+      queryClient.getQueryData<IPage>(pageKeys.detail(pageId));
+    if (!page) return;
 
-  useEffect(() => {
-    // Do not overwrite the title while the user is actively editing it. The
-    // server rebroadcasts PAGE_UPDATED to the author too, and that echo can
-    // carry a title that lags behind what the user has just typed; resetting
-    // content from it here would drop in-progress characters and jump the
-    // cursor. Apply external title changes only when the field is not focused.
-    if (
-      titleEditor &&
-      !titleEditor.isDestroyed &&
-      !titleEditor.isFocused &&
-      title !== titleEditor.getText()
-    ) {
-      titleEditor.commands.setContent(title);
-    }
-  }, [pageId, title, titleEditor]);
+    const updatedPage: IPage = { ...page, title: titleText };
+
+    const event: UpdateEvent = {
+      operation: "updateOne",
+      spaceId: page.spaceId,
+      entity: ["pages"],
+      id: page.id,
+      payload: {
+        title: titleText,
+        slugId: page.slugId,
+        parentPageId: page.parentPageId,
+        icon: page.icon,
+      },
+    };
+
+    updatePageData(updatedPage);
+    // Drive the local (same-tab) tree/breadcrumb update. The cross-user tree
+    // refresh is handled server-side: the collab process extracts the renamed
+    // 'title' Yjs fragment and broadcasts a treeUpdate. The previous socket
+    // `emit(event)` here was a no-op (the gateway ignores it) and was removed.
+    localEmitter.emit("message", event);
+  }, 500);
 
   useEffect(() => {
     setTimeout(() => {
@@ -175,13 +178,6 @@ export function TitleEditor({
     }, 300);
   }, [titleEditor]);
 
-  useEffect(() => {
-    return () => {
-      // force-save title on navigation
-      saveTitle();
-    };
-  }, [pageId]);
-
   useEffect(() => {
     if (!titleEditor) return;
     titleEditor.setEditable(editable && currentPageEditMode === PageEditMode.Edit);
@@ -248,16 +244,22 @@ export function TitleEditor({
 
   return (
     <div className="page-title">
-      <EditorContent
-        editor={titleEditor}
-        onKeyDown={(event) => {
-          // First handle the search hotkey
-          getHotkeyHandler([["mod+F", openSearchDialog]])(event);
+      {titleReady ? (
+        <EditorContent
+          editor={titleEditor}
+          onKeyDown={(event) => {
+            // First handle the search hotkey
+            getHotkeyHandler([["mod+F", openSearchDialog]])(event);
 
-          // Then handle other key events
-          handleTitleKeyDown(event);
-        }}
-      />
+            // Then handle other key events
+            handleTitleKeyDown(event);
+          }}
+        />
+      ) : (
+        // Static, non-editable fallback so the title is visible before Yjs
+        // hydrates the 'title' fragment. Not wired into the collaborative editor.
+        <h1>{title}</h1>
+      )}
     </div>
   );
 }

apps/client/src/features/home/components/new-note-button.tsx

@@ -3,6 +3,8 @@ import { IconHourglass, IconPlus } from "@tabler/icons-react";
 import { ReactNode } from "react";
 import { useNavigate } from "react-router-dom";
 import { useTranslation } from "react-i18next";
+import { onlineManager } from "@tanstack/react-query";
+import { notifications } from "@mantine/notifications";
 import { useGetSpacesQuery } from "@/features/space/queries/space-query.ts";
 import { useCreatePageMutation } from "@/features/page/queries/page-query.ts";
 import { buildPageUrl } from "@/features/page/page.utils.ts";
@@ -36,21 +38,39 @@ function CreateNoteButton({
   const createPageMutation = useCreatePageMutation();
 
   const createNote = async (space: ISpace) => {
+    // `spaceId`/`temporary` are accepted by the create-page endpoint but are
+    // not part of the shared `IPageInput` type; cast to satisfy the mutation
+    // signature.
+    const variables = {
+      spaceId: space.id,
+      ...(temporary ? { temporary: true } : {}),
+    } as any;
+
+    if (!onlineManager.isOnline()) {
+      // Offline: the create is PAUSED and queued — its promise will not resolve
+      // until we are back online, so awaiting it here would spin the button
+      // forever. Fire it without awaiting (it persists and replays on reconnect)
+      // and tell the user it was saved offline instead of leaving a dead spinner.
+      createPageMutation.mutate(variables);
+      notifications.show({
+        color: "blue",
+        message: t("You're offline. This note will be created once you reconnect."),
+      });
+      return;
+    }
+
     try {
-      // `spaceId`/`temporary` are accepted by the create-page endpoint but are
-      // not part of the shared `IPageInput` type; cast to satisfy the mutation
-      // signature.
-      const createdPage = await createPageMutation.mutateAsync({
-        spaceId: space.id,
-        ...(temporary ? { temporary: true } : {}),
-      } as any);
+      const createdPage = await createPageMutation.mutateAsync(variables);
       navigate(buildPageUrl(space.slug, createdPage.slugId, createdPage.title));
     } catch {
       // useCreatePageMutation already surfaces a red notification on error.
     }
   };
 
-  const isPending = createPageMutation.isPending;
+  // A paused (offline) mutation stays `isPending`, so gate the spinner on it NOT
+  // being paused — otherwise the button would spin forever after an offline
+  // create. The offline path above gives its own "saved offline" feedback.
+  const isPending = createPageMutation.isPending && !createPageMutation.isPaused;
 
   // Exactly one writable space → create directly, no picker needed.
   if (writableSpaces.length === 1) {

apps/client/src/features/offline/clear-offline-cache.test.ts

@@ -0,0 +1,107 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
+
+// vi.mock factories are hoisted above imports, so the spies they reference must
+// be declared via vi.hoisted (also hoisted). These are inspected by assertions.
+const h = vi.hoisted(() => ({
+  clear: vi.fn(),
+  del: vi.fn(),
+}));
+
+// The module under test imports the app entry at load time — it must be mocked.
+vi.mock("@/main.tsx", () => ({
+  queryClient: { clear: h.clear },
+}));
+vi.mock("idb-keyval", () => ({
+  del: h.del,
+}));
+
+import { clearOfflineCache } from "./clear-offline-cache";
+import { OFFLINE_CACHE_KEY } from "./query-persister";
+
+// jsdom does not provide indexedDB.databases() or Cache Storage, so the browser
+// globals are stubbed per-test. We restore them afterwards.
+const originalIndexedDB = (globalThis as any).indexedDB;
+const originalCaches = (globalThis as any).caches;
+
+beforeEach(() => {
+  h.clear.mockClear();
+  h.del.mockClear();
+});
+
+afterEach(() => {
+  (globalThis as any).indexedDB = originalIndexedDB;
+  (globalThis as any).caches = originalCaches;
+  vi.restoreAllMocks();
+});
+
+describe("clearOfflineCache", () => {
+  it("resolves without throwing when the browser globals are absent", async () => {
+    (globalThis as any).indexedDB = undefined;
+    delete (globalThis as any).caches;
+
+    await expect(clearOfflineCache()).resolves.toBeUndefined();
+
+    // The two store-agnostic steps still run.
+    expect(h.clear).toHaveBeenCalledTimes(1);
+    expect(h.del).toHaveBeenCalledWith(OFFLINE_CACHE_KEY);
+  });
+
+  it("deletes only `page.*` IndexedDB databases and only `api-get-cache` caches", async () => {
+    const deleteDatabase = vi.fn((_name: string) => {
+      const request: any = {};
+      // Resolve the deletion on the next microtask, like a real IDBRequest.
+      queueMicrotask(() => request.onsuccess && request.onsuccess());
+      return request;
+    });
+    (globalThis as any).indexedDB = {
+      databases: vi
+        .fn()
+        .mockResolvedValue([
+          { name: "page.aaa" },
+          { name: "page.bbb" },
+          { name: "keyval-store" },
+          { name: undefined },
+        ]),
+      deleteDatabase,
+    };
+
+    const cacheDelete = vi.fn().mockResolvedValue(true);
+    (globalThis as any).caches = {
+      keys: vi
+        .fn()
+        .mockResolvedValue([
+          "workbox-runtime-https://app/api-get-cache",
+          "other-cache",
+        ]),
+      delete: cacheDelete,
+    };
+
+    await expect(clearOfflineCache()).resolves.toBeUndefined();
+
+    // Only the two page.* databases are deleted.
+    expect(deleteDatabase).toHaveBeenCalledTimes(2);
+    expect(deleteDatabase).toHaveBeenCalledWith("page.aaa");
+    expect(deleteDatabase).toHaveBeenCalledWith("page.bbb");
+
+    // Only the api-get-cache entry is deleted.
+    expect(cacheDelete).toHaveBeenCalledTimes(1);
+    expect(cacheDelete).toHaveBeenCalledWith(
+      "workbox-runtime-https://app/api-get-cache",
+    );
+  });
+
+  it("never throws even if a step rejects (best-effort)", async () => {
+    h.del.mockRejectedValueOnce(new Error("idb boom"));
+    (globalThis as any).indexedDB = {
+      databases: vi.fn().mockRejectedValue(new Error("databases boom")),
+      deleteDatabase: vi.fn(),
+    };
+    (globalThis as any).caches = {
+      keys: vi.fn().mockRejectedValue(new Error("caches boom")),
+      delete: vi.fn(),
+    };
+
+    await expect(clearOfflineCache()).resolves.toBeUndefined();
+    expect(h.clear).toHaveBeenCalledTimes(1);
+  });
+});

apps/client/src/features/offline/clear-offline-cache.ts

@@ -0,0 +1,112 @@
+import { del } from "idb-keyval";
+
+import { queryClient } from "@/main.tsx";
+import {
+  OFFLINE_CACHE_KEY,
+  freezeOfflinePersistence,
+  unfreezeOfflinePersistence,
+} from "./query-persister";
+import { PAGE_YDOC_NAME_PREFIX } from "@/features/editor/page-ydoc-name";
+
+/**
+ * Best-effort purge of all of the current user's offline data from the browser.
+ *
+ * On logout the previous user's private data would otherwise linger locally and
+ * be readable by the next person on the device. This clears the three offline
+ * stores the app writes:
+ *   1. the in-memory + IndexedDB-persisted TanStack Query cache (idb-keyval key
+ *      `OFFLINE_CACHE_KEY`),
+ *   2. the Yjs page documents (IndexedDB databases named `page.<id>` created by
+ *      y-indexeddb in make-offline.ts), and
+ *   3. any legacy service worker `api-get-cache` Cache Storage entry. The
+ *      Workbox runtime no longer creates this cache (the GET /api NetworkFirst
+ *      rule was removed — offline reads come from the persisted RQ cache), so
+ *      this is now a defensive cleanup for caches left by older app versions.
+ *
+ * Fully best-effort: every step is isolated so a single failure neither blocks
+ * the remaining steps nor throws to the caller (logout must never be blocked on
+ * cache cleanup). Callers may ignore the resolved value.
+ *
+ * Limitations:
+ *   - Deleting the Yjs page databases relies on `indexedDB.databases()`, which
+ *     is unavailable in some browsers (notably Firefox). There we skip silently;
+ *     those `page.<id>` databases are then left in place.
+ *   - Cache Storage clearing only runs where `caches` exists (secure contexts /
+ *     service-worker-capable browsers).
+ */
+export async function clearOfflineCache(): Promise<void> {
+  // Freeze the throttled persister BEFORE touching the cache so the
+  // queryClient.clear() below cannot trigger a late re-write of the (still
+  // nearly-full) dehydrated snapshot after we del() the key — which would
+  // otherwise resurrect the previous user's persisted data in IndexedDB.
+  // Re-enabled in `finally` so the next (sign-in) session persists normally.
+  freezeOfflinePersistence();
+
+  try {
+    // 1a. Drop the in-memory query cache immediately.
+    try {
+      queryClient.clear();
+    } catch {
+      // best-effort: ignore in-memory cache reset failures
+    }
+
+    // 1b. Delete the persisted RQ cache from IndexedDB.
+    try {
+      await del(OFFLINE_CACHE_KEY);
+    } catch {
+      // best-effort: ignore persisted-cache deletion failures
+    }
+
+  // 2. Delete the Yjs page IndexedDB databases (`page.<id>`).
+  // `indexedDB.databases()` is not implemented everywhere (e.g. Firefox); when
+  // it is missing we cannot enumerate the page databases, so we skip silently.
+  try {
+    if (
+      typeof indexedDB !== "undefined" &&
+      typeof indexedDB.databases === "function"
+    ) {
+      const dbs = await indexedDB.databases();
+      for (const db of dbs) {
+        const name = db?.name;
+        if (typeof name !== "string" || !name.startsWith(PAGE_YDOC_NAME_PREFIX))
+          continue;
+        try {
+          // Fire-and-forget delete; await a thin wrapper so a slow delete does
+          // not race the page teardown, but never reject on it.
+          await new Promise<void>((resolve) => {
+            const request = indexedDB.deleteDatabase(name);
+            request.onsuccess = () => resolve();
+            request.onerror = () => resolve();
+            request.onblocked = () => resolve();
+          });
+        } catch {
+          // best-effort per database
+        }
+      }
+    }
+  } catch {
+    // best-effort: ignore enumeration/deletion failures
+  }
+
+  // 3. Clear any legacy service worker API cache. Current builds no longer
+  // create it, but an older client may have left an "api-get-cache" entry
+  // (Workbox may prefix the name), so match by substring rather than exact name.
+  try {
+    if ("caches" in window) {
+      const keys = await caches.keys();
+      await Promise.all(
+        keys
+          .filter((key) => key.includes("api-get-cache"))
+          .map((key) => caches.delete(key)),
+      );
+    }
+  } catch {
+    // best-effort: ignore Cache Storage failures
+  }
+  } finally {
+    // Re-enable persistence for the next session (sign-in continues running in
+    // the same tab; logout reloads via window.location.replace, so this is a
+    // harmless no-op there).
+    unfreezeOfflinePersistence();
+  }
+}

apps/client/src/features/offline/make-offline.test.ts

@@ -0,0 +1,373 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
+
+// vi.mock factories are hoisted above imports, so any spy they reference must be
+// declared with vi.hoisted (which is hoisted as well). These shared spies are
+// inspected by the assertions below.
+const h = vi.hoisted(() => ({
+  ydocDestroy: vi.fn(),
+  idbDestroy: vi.fn(),
+  providerOn: vi.fn(),
+  providerOff: vi.fn(),
+  providerDestroy: vi.fn(),
+}));
+
+// The module under test imports the app entry at load time — it must be mocked.
+vi.mock("@/main.tsx", () => ({
+  queryClient: { setQueryData: vi.fn(), prefetchQuery: vi.fn() },
+}));
+vi.mock("@/features/page/services/page-service", () => ({
+  getPageById: vi.fn(),
+  getPageBreadcrumbs: vi.fn(),
+  getSidebarPages: vi.fn(),
+  getAllSidebarPages: vi.fn(),
+}));
+vi.mock("@/features/space/services/space-service.ts", () => ({
+  getSpaceById: vi.fn(),
+}));
+vi.mock("@/features/comment/services/comment-service", () => ({
+  getPageComments: vi.fn(),
+}));
+
+// Use the `function` form (not an arrow) so Vitest binds the constructor return
+// value when the module under test calls `new Y.Doc()` etc.
+vi.mock("yjs", () => ({
+  Doc: vi.fn(function () {
+    return { destroy: h.ydocDestroy };
+  }),
+}));
+vi.mock("y-indexeddb", () => ({
+  IndexeddbPersistence: vi.fn(function () {
+    return { destroy: h.idbDestroy };
+  }),
+}));
+vi.mock("@hocuspocus/provider", () => ({
+  HocuspocusProvider: vi.fn(function () {
+    return { on: h.providerOn, off: h.providerOff, destroy: h.providerDestroy };
+  }),
+}));
+
+import {
+  warmInfiniteAll,
+  warmPageYdoc,
+  makePageAvailableOffline,
+} from "./make-offline";
+import { queryClient } from "@/main.tsx";
+import {
+  getPageById,
+  getPageBreadcrumbs,
+  getSidebarPages,
+} from "@/features/page/services/page-service";
+import { getPageComments } from "@/features/comment/services/comment-service";
+
+const setQueryData = (queryClient as any).setQueryData as ReturnType<
+  typeof vi.fn
+>;
+const prefetchQuery = (queryClient as any).prefetchQuery as ReturnType<
+  typeof vi.fn
+>;
+
+beforeEach(() => {
+  // Clear call history WITHOUT wiping the mock implementations the vi.mock
+  // factories installed (vi.clearAllMocks would drop the constructor return
+  // objects and break the provider/idb/yjs spies).
+  setQueryData.mockClear();
+  prefetchQuery.mockReset();
+  prefetchQuery.mockResolvedValue(undefined);
+  (getPageById as ReturnType<typeof vi.fn>).mockReset();
+  (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockReset();
+  (getSidebarPages as ReturnType<typeof vi.fn>).mockReset();
+  (getPageComments as ReturnType<typeof vi.fn>).mockReset();
+  h.ydocDestroy.mockClear();
+  h.idbDestroy.mockClear();
+  h.providerOn.mockClear();
+  h.providerOff.mockClear();
+  h.providerDestroy.mockClear();
+});
+
+describe("warmInfiniteAll", () => {
+  it("warms a single page and writes the InfiniteData cache shape", async () => {
+    const res = { items: [{ id: 1 }], meta: { nextCursor: null } };
+    const fetchPage = vi.fn().mockResolvedValue(res);
+
+    await warmInfiniteAll(["comments", "p1"], fetchPage);
+
+    expect(fetchPage).toHaveBeenCalledTimes(1);
+    expect(fetchPage).toHaveBeenCalledWith(undefined);
+    expect(setQueryData).toHaveBeenCalledTimes(1);
+    expect(setQueryData).toHaveBeenCalledWith(["comments", "p1"], {
+      pages: [res],
+      pageParams: [undefined],
+    });
+  });
+
+  it("walks the cursor chain across multiple pages", async () => {
+    const r0 = { items: [], meta: { nextCursor: "c1" } };
+    const r1 = { items: [], meta: { nextCursor: "c2" } };
+    const r2 = { items: [], meta: { nextCursor: null } };
+    const fetchPage = vi
+      .fn()
+      .mockResolvedValueOnce(r0)
+      .mockResolvedValueOnce(r1)
+      .mockResolvedValueOnce(r2);
+
+    await warmInfiniteAll(["comments", "p1"], fetchPage);
+
+    expect(fetchPage).toHaveBeenCalledTimes(3);
+    expect(fetchPage.mock.calls.map((c) => c[0])).toEqual([
+      undefined,
+      "c1",
+      "c2",
+    ]);
+    const payload = setQueryData.mock.calls[0][1];
+    expect(payload.pages).toEqual([r0, r1, r2]);
+    expect(payload.pageParams).toEqual([undefined, "c1", "c2"]);
+  });
+
+  it("caps pagination at maxPages and reports the truncation (returns false)", async () => {
+    // Always returns a non-null cursor — the cap is the only thing that stops it.
+    const fetchPage = vi
+      .fn()
+      .mockResolvedValue({ items: [], meta: { nextCursor: "more" } });
+    const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+
+    // Hitting maxPages with a cursor still pending is a truncated warm: the
+    // (partial) cache is still written, but the result is reported as false.
+    await expect(
+      warmInfiniteAll(["comments", "p1"], fetchPage, 2),
+    ).resolves.toBe(false);
+
+    expect(fetchPage).toHaveBeenCalledTimes(2);
+    const payload = setQueryData.mock.calls[0][1];
+    expect(payload.pages).toHaveLength(2);
+    expect(errorSpy).toHaveBeenCalled();
+
+    errorSpy.mockRestore();
+  });
+
+  it("returns true on success", async () => {
+    const fetchPage = vi
+      .fn()
+      .mockResolvedValue({ items: [], meta: { nextCursor: null } });
+
+    await expect(
+      warmInfiniteAll(["comments", "p1"], fetchPage),
+    ).resolves.toBe(true);
+  });
+
+  it("reports errors (returns false) and never writes the cache on failure", async () => {
+    const fetchPage = vi.fn().mockRejectedValue(new Error("network"));
+    const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+
+    await expect(
+      warmInfiniteAll(["comments", "p1"], fetchPage),
+    ).resolves.toBe(false);
+    expect(setQueryData).not.toHaveBeenCalled();
+    expect(errorSpy).toHaveBeenCalled();
+
+    errorSpy.mockRestore();
+  });
+});
+
+describe("makePageAvailableOffline", () => {
+  const okPage = {
+    id: "uuid-1",
+    slugId: "slug-1",
+    space: { slug: "space-slug" },
+  };
+
+  it("returns ok:true with no failures when every step succeeds", async () => {
+    (getPageById as ReturnType<typeof vi.fn>).mockResolvedValue(okPage);
+    (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockResolvedValue([]);
+    (getSidebarPages as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    (getPageComments as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+
+    const result = await makePageAvailableOffline({
+      pageId: "uuid-1",
+      spaceId: "space-uuid",
+    });
+
+    expect(result).toEqual({ ok: true, failed: [] });
+  });
+
+  it("returns ok:false with the failed step label when a warm step fails", async () => {
+    const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+    (getPageById as ReturnType<typeof vi.fn>).mockResolvedValue(okPage);
+    (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockResolvedValue([]);
+    (getSidebarPages as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    // Comments warm fails -> labeled "comments".
+    (getPageComments as ReturnType<typeof vi.fn>).mockRejectedValue(
+      new Error("network"),
+    );
+
+    const result = await makePageAvailableOffline({
+      pageId: "uuid-1",
+      spaceId: "space-uuid",
+    });
+
+    expect(result.ok).toBe(false);
+    expect(result.failed).toContain("comments");
+    expect(errorSpy).toHaveBeenCalled();
+
+    errorSpy.mockRestore();
+  });
+
+  // Helper: the page-ids passed to the sidebar-children warm (its query key is
+  // ["sidebar-pages", { pageId, spaceId }]) — i.e. which nodes were prefetched.
+  const warmedSidebarIds = () =>
+    prefetchQuery.mock.calls
+      .map((c) => c[0])
+      .filter((opts: any) => opts?.queryKey?.[0] === "sidebar-pages")
+      .map((opts: any) => opts.queryKey[1]?.pageId);
+
+  it("warms the page + every ancestor's children once and skips the self-ancestor guard", async () => {
+    (getPageById as ReturnType<typeof vi.fn>).mockResolvedValue(okPage);
+    // Breadcrumbs include two real ancestors, the page's OWN id (must be skipped
+    // by the ancestorId === pageId guard so it is not warmed twice), and a
+    // malformed entry with no id (also skipped).
+    (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockResolvedValue([
+      { id: "anc-1" },
+      { id: "uuid-1" }, // === pageId -> guard
+      { id: "anc-2" },
+      {}, // no id -> skipped
+    ]);
+    (getSidebarPages as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    (getPageComments as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+
+    const result = await makePageAvailableOffline({
+      pageId: "uuid-1",
+      spaceId: "space-uuid",
+    });
+
+    const ids = warmedSidebarIds();
+    // The page's own children (warmSidebarChildren(pageId)) plus each real
+    // ancestor — exactly once each. The self-ancestor (uuid-1 in breadcrumbs) is
+    // NOT a second warm: uuid-1 appears once (from the page's own children call).
+    expect(ids).toEqual(["uuid-1", "anc-1", "anc-2"]);
+    expect(ids.filter((id: string) => id === "uuid-1")).toHaveLength(1);
+    expect(result).toEqual({ ok: true, failed: [] });
+  });
+
+  it("dedupes repeated tree failures into a single 'tree' label", async () => {
+    const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+    (getPageById as ReturnType<typeof vi.fn>).mockResolvedValue(okPage);
+    (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockResolvedValue([
+      { id: "anc-1" },
+      { id: "anc-2" },
+    ]);
+    (getSidebarPages as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    (getPageComments as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    // Fail ONLY the sidebar-children prefetches (page-own + both ancestors = 3
+    // failures); the currentUser/space prefetches still resolve.
+    prefetchQuery.mockImplementation(async (opts: any) => {
+      if (opts?.queryKey?.[0] === "sidebar-pages") throw new Error("network");
+      return undefined;
+    });
+
+    const result = await makePageAvailableOffline({
+      pageId: "uuid-1",
+      spaceId: "space-uuid",
+    });
+
+    // Three node warms failed but the contract collapses them to one "tree".
+    expect(result.ok).toBe(false);
+    expect(result.failed).toEqual(["tree"]);
+    expect(errorSpy).toHaveBeenCalled();
+
+    errorSpy.mockRestore();
+  });
+
+  it("records 'breadcrumbs' (not 'tree') when the breadcrumbs lookup rejects", async () => {
+    const errorSpy = vi.spyOn(console, "error").mockImplementation(() => {});
+    (getPageById as ReturnType<typeof vi.fn>).mockResolvedValue(okPage);
+    // Ancestor discovery fails -> the ancestor-walk is recorded as "breadcrumbs".
+    (getPageBreadcrumbs as ReturnType<typeof vi.fn>).mockRejectedValue(
+      new Error("network"),
+    );
+    (getSidebarPages as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+    (getPageComments as ReturnType<typeof vi.fn>).mockResolvedValue({
+      items: [],
+      meta: { nextCursor: null },
+    });
+
+    const result = await makePageAvailableOffline({
+      pageId: "uuid-1",
+      spaceId: "space-uuid",
+    });
+
+    // The page's own children still warmed fine (prefetch resolves), so the only
+    // failure is the breadcrumbs lookup.
+    expect(result.ok).toBe(false);
+    expect(result.failed).toEqual(["breadcrumbs"]);
+    expect(errorSpy).toHaveBeenCalled();
+
+    errorSpy.mockRestore();
+  });
+});
+
+describe("warmPageYdoc", () => {
+  afterEach(() => {
+    vi.useRealTimers();
+  });
+
+  it("resolves on synced, detaches the listener once, and tears everything down (settle-once)", async () => {
+    const promise = warmPageYdoc("p1", "ws://x");
+
+    // Grab the synced handler the provider registered.
+    expect(h.providerOn).toHaveBeenCalledWith("synced", expect.any(Function));
+    const handler = h.providerOn.mock.calls.find(
+      (c) => c[0] === "synced",
+    )![1] as () => void;
+
+    handler();
+    await expect(promise).resolves.toBeUndefined();
+
+    // Listener detached and everything cleaned up.
+    expect(h.providerOff).toHaveBeenCalledWith("synced", expect.any(Function));
+    expect(h.providerDestroy).toHaveBeenCalledTimes(1);
+    expect(h.idbDestroy).toHaveBeenCalledTimes(1);
+    expect(h.ydocDestroy).toHaveBeenCalledTimes(1);
+
+    // Firing the handler again must NOT re-run cleanup (settled guard).
+    handler();
+    expect(h.providerDestroy).toHaveBeenCalledTimes(1);
+    expect(h.idbDestroy).toHaveBeenCalledTimes(1);
+    expect(h.ydocDestroy).toHaveBeenCalledTimes(1);
+  });
+
+  it("resolves and cleans up after the timeout when synced never fires", async () => {
+    vi.useFakeTimers();
+    const promise = warmPageYdoc("p1", "ws://x");
+
+    // Do not fire "synced"; let the 8s safety timeout settle it.
+    await vi.advanceTimersByTimeAsync(8000);
+    await expect(promise).resolves.toBeUndefined();
+
+    expect(h.providerDestroy).toHaveBeenCalledTimes(1);
+    expect(h.idbDestroy).toHaveBeenCalledTimes(1);
+    expect(h.ydocDestroy).toHaveBeenCalledTimes(1);
+  });
+});

apps/client/src/features/offline/make-offline.ts

@@ -0,0 +1,315 @@
+import * as Y from "yjs";
+import { IndexeddbPersistence } from "y-indexeddb";
+import { HocuspocusProvider } from "@hocuspocus/provider";
+
+import { queryClient } from "@/main.tsx";
+import {
+  getPageById,
+  getPageBreadcrumbs,
+  getSidebarPages,
+} from "@/features/page/services/page-service";
+import {
+  pageKeys,
+  sidebarPagesQueryOptions,
+} from "@/features/page/queries/page-query";
+import { spaceByIdQueryOptions } from "@/features/space/queries/space-query";
+import { RQ_KEY } from "@/features/comment/queries/comment-query";
+import { getPageComments } from "@/features/comment/services/comment-service";
+import { getMyInfo } from "@/features/user/services/user-service";
+import { userKeys } from "@/features/user/hooks/use-current-user";
+import { IPage } from "@/features/page/types/page.types";
+import { IPagination } from "@/lib/types.ts";
+import { pageYdocName } from "@/features/editor/page-ydoc-name";
+
+/**
+ * Fully paginate an infinite query and write the @tanstack InfiniteData cache
+ * shape ({ pages, pageParams }) that the matching useInfiniteQuery hook reads.
+ *
+ * The default prefetchInfiniteQuery only warms the FIRST page, which leaves
+ * hooks that treat hasNextPage as still-loading (e.g. the comments panel)
+ * spinning forever offline, and silently truncates large lists. This walks the
+ * cursor chain until it runs out (or hits maxPages) so the whole list is cached.
+ *
+ * Best-effort: a failure does not throw (a partial/failed warm is still useful),
+ * but it is reported — the error is logged with context and `false` is returned
+ * so the caller can record the failed step instead of silently succeeding.
+ *
+ * Returns true ONLY if the cursor chain was fully exhausted and written. If the
+ * walk stops because it hit `maxPages` while a `nextCursor` is still pending,
+ * the cached list is truncated AND its last page keeps a nextCursor that cannot
+ * be re-fetched offline (hooks that gate on hasNextPage would spin forever), so
+ * that case is logged and returns false too — the caller records it as a failed
+ * warm instead of a silent truncated success. The (partial) cache is still
+ * written so what we did fetch is usable.
+ *
+ * Exported for unit testing of the cursor-walk / cache-write behavior.
+ */
+export async function warmInfiniteAll<T>(
+  queryKey: readonly unknown[],
+  fetchPage: (cursor: string | undefined) => Promise<IPagination<T>>,
+  maxPages = 50,
+): Promise<boolean> {
+  try {
+    const pages: IPagination<T>[] = [];
+    const pageParams: (string | undefined)[] = [];
+    let cursor: string | undefined = undefined;
+    let exhausted = false;
+
+    for (let i = 0; i < maxPages; i++) {
+      const res = await fetchPage(cursor);
+      pages.push(res);
+      pageParams.push(cursor);
+      cursor = res?.meta?.nextCursor ?? undefined;
+      if (!cursor) {
+        exhausted = true;
+        break;
+      }
+    }
+
+    queryClient.setQueryData(queryKey, { pages, pageParams });
+
+    if (!exhausted) {
+      // Stopped at maxPages with a cursor still pending: the list is truncated
+      // and the last cached page's nextCursor is un-fetchable offline. Report it
+      // as a failed warm rather than a silent truncated success.
+      console.error("warmInfiniteAll truncated at maxPages", {
+        queryKey,
+        maxPages,
+      });
+      return false;
+    }
+    return true;
+  } catch (error) {
+    console.error("warmInfiniteAll failed", { queryKey, error });
+    return false;
+  }
+}
+
+export interface MakePageAvailableOfflineParams {
+  pageId: string;
+  spaceId?: string;
+}
+
+/**
+ * Outcome of {@link makePageAvailableOffline}. `ok` is true only when every warm
+ * step succeeded; `failed` lists the labels of the steps that failed (a subset
+ * of: "currentUser", "page", "space", "tree", "breadcrumbs", "comments").
+ */
+export interface MakePageAvailableOfflineResult {
+  ok: boolean;
+  failed: string[];
+}
+
+/**
+ * Best-effort prefetch of a page's read queries so they get persisted to
+ * IndexedDB and become readable offline.
+ *
+ * Each step is isolated and this function does NOT throw — a partial warm is
+ * still useful. Instead of silently succeeding, every failed step is logged
+ * with a label and recorded in the returned result: `{ ok, failed }` where
+ * `ok` is true only if no step failed and `failed` lists the failed step
+ * labels. Only meaningful while online (the underlying requests must succeed).
+ */
+export async function makePageAvailableOffline({
+  pageId,
+  spaceId,
+}: MakePageAvailableOfflineParams): Promise<MakePageAvailableOfflineResult> {
+  const failed: string[] = [];
+
+  // Warm the current user (['currentUser']) so the auth-gated <Layout> can
+  // hydrate offline. UserProvider blanks the whole app while useCurrentUser has
+  // no data, and the offline POST /api/users/me fails as a network error, so
+  // without a persisted user a pinned page still white-screens after relaunch
+  // (#238). Persisted via OFFLINE_PERSIST_ROOTS; warmed here so the persisted
+  // cache actually has an entry to restore.
+  try {
+    await queryClient.prefetchQuery({
+      queryKey: userKeys.currentUser(),
+      queryFn: () => getMyInfo(),
+    });
+  } catch (error) {
+    console.error("makePageAvailableOffline: currentUser step failed", {
+      pageId,
+      error,
+    });
+    failed.push("currentUser");
+  }
+
+  // Fetch the page document ONCE and write it under BOTH cache keys, exactly
+  // like usePageQuery's onData effect. Every page consumer reads
+  // pageKeys.detail(slugId) (usePageQuery keys on the slugId for routed reads),
+  // so warming only the uuid key would leave the offline page blank.
+  let page: IPage | undefined;
+  try {
+    page = await getPageById({ pageId });
+    queryClient.setQueryData(pageKeys.detail(page.slugId), page);
+    queryClient.setQueryData(pageKeys.detail(page.id), page);
+  } catch (error) {
+    console.error("makePageAvailableOffline: page step failed", {
+      pageId,
+      error,
+    });
+    failed.push("page");
+  }
+
+  // Warm the space — page.tsx renders nothing until the space query resolves
+  // (useGetSpaceBySlugQuery). Awaited (not the fire-and-forget prefetchSpace) so
+  // the space is actually persisted before the caller fires its toast. Shares
+  // spaceByIdQueryOptions so the key/fn cannot drift from the hook.
+  try {
+    const spaceSlug = page?.space?.slug;
+    if (spaceSlug) {
+      await queryClient.prefetchQuery(spaceByIdQueryOptions(spaceSlug));
+    }
+  } catch (error) {
+    console.error("makePageAvailableOffline: space step failed", {
+      pageId,
+      error,
+    });
+    failed.push("space");
+  }
+
+  // Warm the sidebar tree root so the WHOLE root level renders offline (matches
+  // useGetRootSidebarPagesQuery's pageKeys.rootSidebar(spaceId) infinite cache).
+  // Fully paginated so large root levels are not truncated at 100.
+  if (spaceId) {
+    const ok = await warmInfiniteAll(pageKeys.rootSidebar(spaceId), (cursor) =>
+      getSidebarPages({ spaceId, cursor, limit: 100 }),
+    );
+    if (!ok) failed.push("tree");
+  }
+
+  // Warm the children of the page and of every ancestor so the path to this
+  // page is expandable offline. We MIRROR fetchAllAncestorChildren exactly via
+  // sidebarPagesQueryOptions — same pageKeys.sidebar({ pageId, spaceId }) key,
+  // same getAllSidebarPages fn (which aggregates ALL children pages, so nothing
+  // is truncated at 100), same 30min staleTime — otherwise the warmed cache
+  // would never be read by the offline tree.
+  const warmSidebarChildren = async (id: string): Promise<boolean> => {
+    try {
+      // Keep EXACTLY { pageId, spaceId } so the key hashes identically to
+      // fetchAllAncestorChildren's (no parentPageId, no extra fields).
+      const params = { pageId: id, spaceId };
+      await queryClient.prefetchQuery(sidebarPagesQueryOptions(params));
+      return true;
+    } catch (error) {
+      console.error("makePageAvailableOffline: tree node step failed", {
+        pageId: id,
+        error,
+      });
+      return false;
+    }
+  };
+
+  // The page's own children.
+  if (!(await warmSidebarChildren(pageId))) failed.push("tree");
+
+  // Each ancestor's children. Use the breadcrumbs endpoint ONLY to discover the
+  // ancestor ids — we intentionally do NOT cache the breadcrumbs themselves
+  // (the UI derives the path from the tree).
+  try {
+    const ancestors = (await getPageBreadcrumbs(pageId)) as
+      | Array<{ id?: string }>
+      | undefined;
+    for (const ancestor of ancestors ?? []) {
+      const ancestorId = ancestor?.id;
+      if (!ancestorId || ancestorId === pageId) continue;
+      if (!(await warmSidebarChildren(ancestorId))) failed.push("tree");
+    }
+  } catch (error) {
+    console.error("makePageAvailableOffline: breadcrumbs step failed", {
+      pageId,
+      error,
+    });
+    failed.push("breadcrumbs");
+  }
+
+  // Comments (matches useCommentsQuery's RQ_KEY(pageId) infinite cache).
+  // useCommentsQuery reports isLoading while hasNextPage is true, so warming
+  // only the first page leaves the offline comments panel spinning forever on
+  // pages with >100 comments. Fully paginate so the last cached page has no
+  // nextCursor and the panel settles offline.
+  const commentsOk = await warmInfiniteAll(RQ_KEY(pageId), (cursor) =>
+    getPageComments({ pageId, cursor, limit: 100 }),
+  );
+  if (!commentsOk) failed.push("comments");
+
+  // Dedupe — the tree label can be recorded once per failed node/ancestor.
+  const uniqueFailed = [...new Set(failed)];
+  return { ok: uniqueFailed.length === 0, failed: uniqueFailed };
+}
+
+/**
+ * Best-effort warm-up of the page's Yjs document into IndexedDB so the editor
+ * can open offline.
+ *
+ * Opens a local IndexeddbPersistence plus a transient HocuspocusProvider to
+ * pull the server state into IndexedDB, then tears both down once synced (or
+ * after a timeout). Entirely wrapped in try/catch — NEVER throws.
+ *
+ * Only meaningful when online at warm time; offline it is a no-op that resolves.
+ */
+export async function warmPageYdoc(
+  pageId: string,
+  collabUrl: string,
+  token?: string,
+): Promise<void> {
+  let ydoc: Y.Doc | null = null;
+  let local: IndexeddbPersistence | null = null;
+  let remote: HocuspocusProvider | null = null;
+
+  try {
+    const documentName = pageYdocName(pageId);
+    ydoc = new Y.Doc();
+    local = new IndexeddbPersistence(documentName, ydoc);
+    remote = new HocuspocusProvider({
+      url: collabUrl,
+      name: documentName,
+      document: ydoc,
+      token,
+    });
+
+    const provider = remote;
+
+    await new Promise<void>((resolve) => {
+      let settled = false;
+      let timeoutId: ReturnType<typeof setTimeout> | undefined;
+      const finish = () => {
+        if (settled) return;
+        settled = true;
+        // Clear the pending timeout and detach the listener so neither leaks
+        // after we resolve.
+        if (timeoutId !== undefined) clearTimeout(timeoutId);
+        try {
+          provider.off("synced", finish);
+        } catch {
+          // best-effort
+        }
+        resolve();
+      };
+
+      // Resolve once the server state has synced into the local doc...
+      provider.on("synced", finish);
+      // ...or give up after a short timeout so we never hang.
+      timeoutId = setTimeout(finish, 8000);
+    });
+  } catch {
+    // best-effort
+  } finally {
+    try {
+      remote?.destroy();
+    } catch {
+      // best-effort
+    }
+    try {
+      local?.destroy();
+    } catch {
+      // best-effort
+    }
+    try {
+      ydoc?.destroy();
+    } catch {
+      // best-effort
+    }
+  }
+}

apps/client/src/features/offline/offline-fallback.tsx

@@ -0,0 +1,45 @@
+import { Button, Container, Group, Stack, Text, Title } from "@mantine/core";
+import { Helmet } from "react-helmet-async";
+import { useTranslation } from "react-i18next";
+import { getAppName } from "@/lib/config";
+
+/**
+ * Shown when the authenticated app shell cannot hydrate because the current
+ * user is unavailable AND there is no cached user to fall back on (e.g. an
+ * offline cold boot of a page that was never warmed for offline).
+ *
+ * Previously UserProvider returned a bare `<></>` in this situation, which
+ * white-screened the whole app on any offline reload (#237/#238). Rendering an
+ * explicit "you're offline" state with a retry instead gives the user a clear,
+ * non-blank fallback and a way to recover once the network returns.
+ */
+export function OfflineFallback() {
+  const { t } = useTranslation();
+
+  return (
+    <>
+      <Helmet>
+        <title>
+          {t("You're offline")} - {getAppName()}
+        </title>
+      </Helmet>
+      <Container size="sm" py={80}>
+        <Stack align="center" gap="md">
+          <Title order={2} ta="center">
+            {t("You're offline")}
+          </Title>
+          <Text c="dimmed" size="lg" ta="center">
+            {t(
+              "This page hasn't been saved for offline use, so it can't be loaded right now. Reconnect to the internet and try again.",
+            )}
+          </Text>
+          <Group justify="center">
+            <Button onClick={() => window.location.reload()} variant="subtle">
+              {t("Retry")}
+            </Button>
+          </Group>
+        </Stack>
+      </Container>
+    </>
+  );
+}

apps/client/src/features/offline/offline-mutations.test.ts

@@ -0,0 +1,114 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+import { QueryClient, hydrate, dehydrate } from "@tanstack/react-query";
+
+// Stub the network services so a replayed mutation hits a spy, not the network.
+const h = vi.hoisted(() => ({
+  createPage: vi.fn(),
+  movePage: vi.fn(),
+  createComment: vi.fn(),
+}));
+
+vi.mock("@/features/page/services/page-service", () => ({
+  createPage: h.createPage,
+  movePage: h.movePage,
+}));
+vi.mock("@/features/comment/services/comment-service", () => ({
+  createComment: h.createComment,
+}));
+// page-query pulls in the app entry (queryClient) and a lot of UI deps via its
+// cache helpers; we only need invalidateOnCreatePage to be a no-op here.
+vi.mock("@/features/page/queries/page-query", () => ({
+  invalidateOnCreatePage: vi.fn(),
+}));
+
+import {
+  offlineMutationKeys,
+  registerOfflineMutationDefaults,
+} from "./offline-mutations";
+
+beforeEach(() => {
+  h.createPage.mockReset().mockResolvedValue({ id: "new-page" });
+  h.movePage.mockReset().mockResolvedValue(undefined);
+  h.createComment.mockReset().mockResolvedValue({ id: "new-comment" });
+});
+
+describe("registerOfflineMutationDefaults", () => {
+  it("registers a default mutationFn for every offline mutation key", () => {
+    const qc = new QueryClient();
+    registerOfflineMutationDefaults(qc);
+
+    for (const key of Object.values(offlineMutationKeys)) {
+      const defaults = qc.getMutationDefaults(key);
+      expect(typeof defaults?.mutationFn).toBe("function");
+    }
+  });
+
+  // The headline durability guarantee: a paused mutation dehydrated into
+  // IndexedDB while offline must, after a reload, have a mutationFn so
+  // resumePausedMutations() actually replays the write on reconnect.
+  it("makes a rehydrated paused create replayable by resumePausedMutations", async () => {
+    // 1) Simulate the offline tab: a paused create mutation gets dehydrated.
+    const offlineClient = new QueryClient();
+    const observer = offlineClient.getMutationCache().build(offlineClient, {
+      mutationKey: offlineMutationKeys.createPage,
+    });
+    // Force the dehydrate-worthy paused state (offline = isPaused) with the
+    // payload the user submitted before losing connectivity.
+    observer.state.isPaused = true;
+    observer.state.status = "pending";
+    observer.state.variables = { spaceId: "s1", title: "Offline page" };
+
+    const dehydrated = dehydrate(offlineClient, {
+      shouldDehydrateMutation: () => true,
+    });
+    expect(dehydrated.mutations).toHaveLength(1);
+    // The dehydrated mutation carries NO mutationFn (functions aren't
+    // serializable) — only its key + variables survive the reload.
+    expect((dehydrated.mutations[0] as any).mutationFn).toBeUndefined();
+
+    // 2) Simulate the fresh page after reload: register defaults, then hydrate
+    //    the persisted paused mutation back in.
+    const freshClient = new QueryClient();
+    registerOfflineMutationDefaults(freshClient);
+    hydrate(freshClient, dehydrated);
+
+    expect(freshClient.getMutationCache().getAll()).toHaveLength(1);
+
+    // 3) Reconnect: replay the paused mutations.
+    await freshClient.resumePausedMutations();
+
+    // The default mutationFn ran with the persisted variables — the write is
+    // NOT silently dropped.
+    expect(h.createPage).toHaveBeenCalledTimes(1);
+    expect(h.createPage).toHaveBeenCalledWith({
+      spaceId: "s1",
+      title: "Offline page",
+    });
+  });
+
+  it("makes a rehydrated paused move replayable by resumePausedMutations", async () => {
+    const offlineClient = new QueryClient();
+    const observer = offlineClient.getMutationCache().build(offlineClient, {
+      mutationKey: offlineMutationKeys.movePage,
+    });
+    observer.state.isPaused = true;
+    observer.state.status = "pending";
+    observer.state.variables = { pageId: "p1", parentPageId: null, position: "a" };
+
+    const dehydrated = dehydrate(offlineClient, {
+      shouldDehydrateMutation: () => true,
+    });
+
+    const freshClient = new QueryClient();
+    registerOfflineMutationDefaults(freshClient);
+    hydrate(freshClient, dehydrated);
+    await freshClient.resumePausedMutations();
+
+    expect(h.movePage).toHaveBeenCalledTimes(1);
+    expect(h.movePage).toHaveBeenCalledWith({
+      pageId: "p1",
+      parentPageId: null,
+      position: "a",
+    });
+  });
+});

apps/client/src/features/offline/offline-mutations.ts

@@ -0,0 +1,64 @@
+import type { QueryClient } from "@tanstack/react-query";
+import { createPage, movePage } from "@/features/page/services/page-service";
+import { createComment } from "@/features/comment/services/comment-service";
+import { invalidateOnCreatePage } from "@/features/page/queries/page-query";
+import type {
+  IMovePage,
+  IPage,
+  IPageInput,
+} from "@/features/page/types/page.types";
+import type { IComment } from "@/features/comment/types/comment.types";
+
+/**
+ * Stable mutation keys for the offline-relevant structural mutations.
+ *
+ * When the browser goes offline, React Query PAUSES these mutations and the
+ * PersistQueryClientProvider dehydrates the paused mutation into IndexedDB. On a
+ * reload-while-offline the mutation is restored, but a restored mutation has NO
+ * observer (no component is mounted) — so its replay relies entirely on the
+ * `mutationFn` registered via `setMutationDefaults` for its `mutationKey`.
+ * Without that, `resumePausedMutations()` finds a paused mutation with no
+ * `mutationFn` and silently no-ops, dropping the offline create/move/comment
+ * (#237/#238). Each offline mutation hook tags itself with the matching key so
+ * the rehydrated paused mutation can find its default `mutationFn` and replay.
+ */
+export const offlineMutationKeys = {
+  createPage: ["create-page"] as const,
+  movePage: ["move-page"] as const,
+  createComment: ["create-comment"] as const,
+};
+
+/**
+ * Register default `mutationFn`s (and the minimal success side effects safe to
+ * run without a mounted component) for the offline-relevant mutation keys, so a
+ * paused mutation restored from IndexedDB after an offline reload is replayable
+ * by `resumePausedMutations()` on reconnect.
+ *
+ * Called once when the QueryClient is created (see main.tsx). The hooks still
+ * carry their own inline `mutationFn`/`onSuccess` for the live in-session path;
+ * these defaults only take over for a rehydrated paused mutation that lost its
+ * observer across the reload.
+ */
+export function registerOfflineMutationDefaults(queryClient: QueryClient): void {
+  queryClient.setMutationDefaults(offlineMutationKeys.createPage, {
+    mutationFn: (data: Partial<IPageInput>) => createPage(data),
+    // Re-converge the sidebar tree / recent-changes from the authoritative
+    // create response. Pure cache writes — safe with no component mounted.
+    onSuccess: (data: IPage) => {
+      invalidateOnCreatePage(data);
+    },
+  });
+
+  queryClient.setMutationDefaults(offlineMutationKeys.movePage, {
+    // Replay the server-side move. The tree re-converges from the next online
+    // sidebar fetch / websocket `moveTreeNode` echo, so no cache write is
+    // needed here (the optimistic tree state was local-only anyway).
+    mutationFn: (data: IMovePage) => movePage(data),
+  });
+
+  queryClient.setMutationDefaults(offlineMutationKeys.createComment, {
+    // Replay the server-side comment create. The comments list refetches on the
+    // online reload, so the replay only needs to persist the write.
+    mutationFn: (data: Partial<IComment>) => createComment(data),
+  });
+}

apps/client/src/features/offline/offline-resume.test.ts

@@ -0,0 +1,128 @@
+import { describe, it, expect, vi, beforeEach, afterEach } from "vitest";
+import { QueryClient, onlineManager } from "@tanstack/react-query";
+import {
+  persistQueryClientRestore,
+  persistQueryClientSave,
+} from "@tanstack/react-query-persist-client";
+
+// Stub the network services so a replayed mutation hits a spy, not the network.
+const h = vi.hoisted(() => ({
+  createPage: vi.fn(),
+  movePage: vi.fn(),
+  createComment: vi.fn(),
+}));
+
+vi.mock("@/features/page/services/page-service", () => ({
+  createPage: h.createPage,
+  movePage: h.movePage,
+}));
+vi.mock("@/features/comment/services/comment-service", () => ({
+  createComment: h.createComment,
+}));
+vi.mock("@/features/page/queries/page-query", () => ({
+  invalidateOnCreatePage: vi.fn(),
+}));
+
+// In-memory idb-keyval so the REAL queryPersister round-trips through a fake
+// store (the actual persist -> reload -> restore path, not a hand-built blob).
+const store = new Map<string, string>();
+vi.mock("idb-keyval", () => ({
+  get: vi.fn((k: string) => Promise.resolve(store.get(k) ?? undefined)),
+  set: vi.fn((k: string, v: string) => {
+    store.set(k, v);
+    return Promise.resolve();
+  }),
+  del: vi.fn((k: string) => {
+    store.delete(k);
+    return Promise.resolve();
+  }),
+}));
+
+import { queryPersister } from "./query-persister";
+import {
+  offlineMutationKeys,
+  registerOfflineMutationDefaults,
+} from "./offline-mutations";
+
+const BUSTER = "test-buster";
+
+beforeEach(() => {
+  store.clear();
+  h.createPage.mockReset().mockResolvedValue({ id: "new-page" });
+  h.movePage.mockReset().mockResolvedValue(undefined);
+  h.createComment.mockReset().mockResolvedValue({ id: "new-comment" });
+});
+
+afterEach(() => {
+  // onlineManager is a global singleton; leave it in the default online state.
+  onlineManager.setOnline(true);
+});
+
+describe("offline paused-mutation resume across a reload", () => {
+  // This is the #120 silent-data-loss reproduction: a paused mutation persisted
+  // to IndexedDB while offline, then the tab RELOADS while still offline, must
+  // resume on reconnect. It exercises the real persister round-trip plus the two
+  // boot-time fixes the app wiring relies on:
+  //   (a) onlineManager seeded to the real offline state so the later reconnect
+  //       is a true offline->online transition that auto-resumes, and
+  //   (b) resumePausedMutations() called after the persister restores (what the
+  //       PersistQueryClientProvider onSuccess does), with mutation defaults
+  //       registered BEFORE the resume so the rehydrated mutation has a fn.
+  it("replays a rehydrated paused create on reconnect (mutationFn fires)", async () => {
+    // --- Tab 1, OFFLINE: user creates a page; it pauses and gets persisted. ---
+    onlineManager.setOnline(false); // (a) boot seeded offline
+
+    const client1 = new QueryClient();
+    registerOfflineMutationDefaults(client1);
+    const observer = client1.getMutationCache().build(client1, {
+      mutationKey: offlineMutationKeys.createPage,
+    });
+    observer.state.isPaused = true;
+    observer.state.status = "pending";
+    observer.state.variables = { spaceId: "s1", title: "Offline page" };
+
+    await persistQueryClientSave({
+      // Cast: persist-client-core and react-query may resolve to different
+      // @tanstack/query-core copies whose QueryClient brands are nominally
+      // incompatible (see query-persister.ts). Structurally identical at runtime.
+      queryClient: client1 as any,
+      persister: queryPersister,
+      buster: BUSTER,
+      dehydrateOptions: { shouldDehydrateMutation: () => true },
+    });
+    // The paused mutation is now in the persisted store.
+    expect(store.size).toBe(1);
+
+    // --- RELOAD while still offline: fresh client restores from the SAME
+    //     persister. Defaults are registered BEFORE restore/resume. ---
+    const client2 = new QueryClient();
+    registerOfflineMutationDefaults(client2);
+    client2.mount(); // subscribes to onlineManager (auto-resume on reconnect)
+
+    await persistQueryClientRestore({
+      queryClient: client2 as any,
+      persister: queryPersister,
+      buster: BUSTER,
+    });
+    expect(client2.getMutationCache().getAll()).toHaveLength(1);
+
+    // (b) onSuccess wiring resumes after restore — but we are still OFFLINE, so
+    // the mutation must stay paused and NOT fire yet.
+    await client2.resumePausedMutations();
+    expect(h.createPage).not.toHaveBeenCalled();
+
+    // --- RECONNECT: the offline->online transition auto-resumes the paused
+    //     mutation and its registered default mutationFn finally fires. ---
+    onlineManager.setOnline(true);
+
+    await vi.waitFor(() => {
+      expect(h.createPage).toHaveBeenCalledTimes(1);
+    });
+    expect(h.createPage).toHaveBeenCalledWith({
+      spaceId: "s1",
+      title: "Offline page",
+    });
+
+    client2.unmount();
+  });
+});

apps/client/src/features/offline/persist-roots.guard.test.ts

@@ -0,0 +1,48 @@
+import { describe, it, expect } from "vitest";
+
+// The query modules transitively import the app entry (@/main.tsx) for the
+// shared queryClient; mock it so importing the key factories has no side effects.
+import { vi } from "vitest";
+vi.mock("@/main.tsx", () => ({
+  queryClient: { setQueryData: vi.fn(), getQueryData: vi.fn() },
+}));
+
+import { OFFLINE_PERSIST_ROOTS } from "./query-persister";
+import { pageKeys } from "@/features/page/queries/page-query";
+import { spaceKeys } from "@/features/space/queries/space-query";
+import { RQ_KEY } from "@/features/comment/queries/comment-query";
+import { userKeys } from "@/features/user/hooks/use-current-user";
+
+/**
+ * Architecture guard (#13): every string persisted via OFFLINE_PERSIST_ROOTS
+ * must be the ROOT (queryKey[0]) of some exported query-key factory. If a
+ * factory's root is renamed without updating the persist registry — or vice
+ * versa — offline persist/warm silently breaks (persisted keys never match the
+ * live queries). This turns that silent regression into a red build.
+ *
+ * Each factory is invoked with throwaway args; only queryKey[0] is inspected.
+ */
+function rootOf(key: readonly unknown[]): string {
+  return String(key[0]);
+}
+
+const FACTORY_ROOTS = new Set<string>([
+  rootOf(pageKeys.detail("x")),
+  rootOf(pageKeys.sidebar({})),
+  rootOf(pageKeys.rootSidebar("x")),
+  rootOf(pageKeys.breadcrumbs("x")),
+  rootOf(pageKeys.recentChanges("x")),
+  rootOf(spaceKeys.detail("x")),
+  rootOf(spaceKeys.list()),
+  rootOf(RQ_KEY("x")),
+  rootOf(userKeys.currentUser()),
+]);
+
+describe("OFFLINE_PERSIST_ROOTS is backed by real query-key factories", () => {
+  it("maps every persisted root to an exported factory root", () => {
+    const unbacked = [...OFFLINE_PERSIST_ROOTS].filter(
+      (root) => !FACTORY_ROOTS.has(root),
+    );
+    expect(unbacked).toEqual([]);
+  });
+});

apps/client/src/features/offline/query-persister.test.ts

@@ -0,0 +1,128 @@
+import { describe, it, expect, vi, afterEach } from "vitest";
+
+// In-memory idb-keyval so we can observe whether the persister actually writes.
+const h = vi.hoisted(() => ({
+  get: vi.fn(() => Promise.resolve(undefined)),
+  set: vi.fn(() => Promise.resolve()),
+  del: vi.fn(() => Promise.resolve()),
+}));
+vi.mock("idb-keyval", () => h);
+
+import {
+  shouldDehydrateOfflineQuery,
+  OFFLINE_PERSIST_ROOTS,
+  queryPersister,
+  freezeOfflinePersistence,
+  unfreezeOfflinePersistence,
+} from "./query-persister";
+
+// Small helper to build the structural query shape the predicate reads.
+const makeQuery = (status: string, queryKey: readonly unknown[]) =>
+  ({ state: { status }, queryKey }) as any;
+
+describe("shouldDehydrateOfflineQuery", () => {
+  it("returns true for a successful query whose root is in the allowlist", () => {
+    expect(shouldDehydrateOfflineQuery(makeQuery("success", ["pages", "abc"]))).toBe(
+      true,
+    );
+    expect(
+      shouldDehydrateOfflineQuery(
+        makeQuery("success", ["sidebar-pages", { pageId: "p", spaceId: "s" }]),
+      ),
+    ).toBe(true);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["comments", "p1"])),
+    ).toBe(true);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["space", "s"])),
+    ).toBe(true);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["recent-changes"])),
+    ).toBe(true);
+    // currentUser is persisted so the auth-gated Layout can hydrate offline.
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["currentUser"])),
+    ).toBe(true);
+  });
+
+  it("returns false when the status is not success (status gate)", () => {
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("pending", ["pages", "abc"])),
+    ).toBe(false);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("error", ["pages", "abc"])),
+    ).toBe(false);
+  });
+
+  it("returns false for a successful query whose root is NOT in the allowlist (privacy gate)", () => {
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["collab-token", "ws"])),
+    ).toBe(false);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["trash", "s"])),
+    ).toBe(false);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", ["unknown"])),
+    ).toBe(false);
+  });
+
+  it("returns false for an empty/undefined queryKey", () => {
+    // String(undefined) is not a member of the allowlist.
+    expect(shouldDehydrateOfflineQuery(makeQuery("success", []))).toBe(false);
+    expect(
+      shouldDehydrateOfflineQuery(makeQuery("success", undefined as any)),
+    ).toBe(false);
+  });
+});
+
+describe("OFFLINE_PERSIST_ROOTS", () => {
+  it("contains exactly the expected 9 navigation/read roots", () => {
+    const expected = [
+      "pages",
+      "sidebar-pages",
+      "root-sidebar-pages",
+      "breadcrumbs",
+      "comments",
+      "space",
+      "spaces",
+      "recent-changes",
+      "currentUser",
+    ];
+    expect(OFFLINE_PERSIST_ROOTS.size).toBe(9);
+    for (const root of expected) {
+      expect(OFFLINE_PERSIST_ROOTS.has(root)).toBe(true);
+    }
+  });
+
+  it("does NOT contain volatile/auth keys", () => {
+    expect(OFFLINE_PERSIST_ROOTS.has("collab-token")).toBe(false);
+    expect(OFFLINE_PERSIST_ROOTS.has("trash")).toBe(false);
+  });
+});
+
+describe("freeze/unfreeze persistence (logout no-late-write guard)", () => {
+  const dummyClient = {
+    timestamp: Date.now(),
+    buster: "",
+    clientState: { mutations: [], queries: [] },
+  } as any;
+
+  afterEach(() => {
+    // Always leave persistence enabled so other tests/sessions persist normally.
+    unfreezeOfflinePersistence();
+    h.set.mockClear();
+  });
+
+  it("does NOT write to storage while frozen", async () => {
+    freezeOfflinePersistence();
+    await queryPersister.persistClient(dummyClient);
+    expect(h.set).not.toHaveBeenCalled();
+  });
+
+  it("resumes writing to storage once unfrozen", async () => {
+    freezeOfflinePersistence();
+    unfreezeOfflinePersistence();
+    await queryPersister.persistClient(dummyClient);
+    expect(h.set).toHaveBeenCalledTimes(1);
+  });
+});

apps/client/src/features/offline/query-persister.ts

@@ -0,0 +1,84 @@
+import { get, set, del } from "idb-keyval";
+import { createAsyncStoragePersister } from "@tanstack/query-async-storage-persister";
+
+// Structural subset of a TanStack Query we read when deciding what to persist.
+// We avoid importing the branded `Query` class because the persist-client and
+// react-query may resolve to different `@tanstack/query-core` copies, whose
+// `Query` types are nominally incompatible (private brand). This structural
+// shape stays assignable to whichever copy the persister expects.
+type DehydratableQuery = {
+  state: { status: string };
+  queryKey: readonly unknown[];
+};
+
+// idb-keyval key under which TanStack Query persists its dehydrated cache.
+// Exported so the logout cache-clear logic deletes the exact same key (no
+// magic-string drift between persist and purge).
+export const OFFLINE_CACHE_KEY = "gitmost-rq-cache";
+
+// IndexedDB-backed storage adapter for TanStack Query's async persister.
+const idbStorage = {
+  getItem: (key: string) => get<string>(key).then((v) => v ?? null),
+  setItem: (key: string, value: string) => set(key, value),
+  removeItem: (key: string) => del(key),
+};
+
+const basePersister = createAsyncStoragePersister({
+  storage: idbStorage,
+  key: OFFLINE_CACHE_KEY,
+  throttleTime: 1000,
+});
+
+// When frozen, persistClient becomes a no-op so no new dehydrated snapshot is
+// written to IndexedDB. This closes a logout data-leak race: clearing the cache
+// (queryClient.clear()) fires `removed` cache events, each of which the persist
+// subscription turns into a throttled persistClient call. The FIRST such call
+// dehydrates a still-nearly-full snapshot and its async write can land AFTER the
+// del() that clears the key, resurrecting the previous user's data (~180KB) in
+// IndexedDB. Freezing before clear()/del() prevents any such rewrite. Re-enabled
+// afterwards so the next (sign-in) session persists normally. See
+// clear-offline-cache.ts.
+let persistFrozen = false;
+
+export function freezeOfflinePersistence(): void {
+  persistFrozen = true;
+}
+
+export function unfreezeOfflinePersistence(): void {
+  persistFrozen = false;
+}
+
+export const queryPersister = {
+  persistClient: (persistedClient: Parameters<typeof basePersister.persistClient>[0]) =>
+    persistFrozen ? Promise.resolve() : basePersister.persistClient(persistedClient),
+  restoreClient: () => basePersister.restoreClient(),
+  removeClient: () => basePersister.removeClient(),
+};
+
+// Only navigation/read query roots are persisted for offline reading.
+// Volatile/auth queries (collab tokens, trash lists) are intentionally excluded.
+//
+// `currentUser` IS persisted: UserProvider gates the entire <Layout> subtree on
+// useCurrentUser(), and offline the POST /api/users/me fails as a no-response
+// network error. Without the persisted/hydrated user the gate blanked every
+// authenticated route on an offline cold boot (#237/#238). It is the logged-in
+// user's own profile (already mirrored to localStorage["currentUser"]), so
+// persisting it to IndexedDB leaks nothing new while unlocking offline reads.
+export const OFFLINE_PERSIST_ROOTS = new Set<string>([
+  "pages",
+  "sidebar-pages",
+  "root-sidebar-pages",
+  "breadcrumbs",
+  "comments",
+  "space",
+  "spaces",
+  "recent-changes",
+  "currentUser",
+]);
+
+export function shouldDehydrateOfflineQuery(query: DehydratableQuery): boolean {
+  return (
+    query.state.status === "success" &&
+    OFFLINE_PERSIST_ROOTS.has(String(query.queryKey?.[0]))
+  );
+}

apps/client/src/features/page/components/header/page-header-menu.tsx

@@ -12,6 +12,8 @@ import {
   IconList,
   IconMarkdown,
   IconPrinter,
+  IconCloud,
+  IconCloudCheck,
   IconStar,
   IconStarFilled,
   IconTrash,
@@ -39,6 +41,8 @@ import { Trans, useTranslation } from "react-i18next";
 import ExportModal from "@/components/common/export-modal";
 import { htmlToMarkdown } from "@docmost/editor-ext";
 import {
+  isLocalSyncedAtom,
+  isRemoteSyncedAtom,
   pageEditorAtom,
   yjsConnectionStatusAtom,
 } from "@/features/editor/atoms/editor-atoms.ts";
@@ -411,14 +415,16 @@ function PageActionMenu({ readOnly }: PageActionMenuProps) {
 function ConnectionWarning() {
   const { t } = useTranslation();
   const yjsConnectionStatus = useAtomValue(yjsConnectionStatusAtom);
+  const isLocalSynced = useAtomValue(isLocalSyncedAtom);
+  const isRemoteSynced = useAtomValue(isRemoteSyncedAtom);
   const [showWarning, setShowWarning] = useState(false);
   const timeoutRef = useRef<ReturnType<typeof setTimeout> | null>(null);
 
-  useEffect(() => {
-    const isDisconnected = ["disconnected", "connecting"].includes(
-      yjsConnectionStatus,
-    );
+  const isDisconnected = ["disconnected", "connecting"].includes(
+    yjsConnectionStatus,
+  );
 
+  useEffect(() => {
     if (isDisconnected) {
       if (!timeoutRef.current) {
         timeoutRef.current = setTimeout(() => setShowWarning(true), 5000);
@@ -430,7 +436,7 @@ function ConnectionWarning() {
       }
       setShowWarning(false);
     }
-  }, [yjsConnectionStatus]);
+  }, [isDisconnected]);
 
   // Cleanup only on unmount
   useEffect(() => {
@@ -441,22 +447,59 @@ function ConnectionWarning() {
     };
   }, []);
 
-  if (!showWarning) return null;
+  // State (1): offline/disconnected — changes are kept locally. Preserve the
+  // existing >5s debounce before surfacing this state.
+  if (isDisconnected) {
+    if (!showWarning) return null;
 
+    const offlineLabel = t(
+      "Offline — changes are saved locally and will sync when you reconnect",
+    );
+    return (
+      <Tooltip label={offlineLabel} openDelay={250} withArrow>
+        <ThemeIcon
+          variant="default"
+          c="red"
+          role="status"
+          aria-label={offlineLabel}
+          style={{ border: "none" }}
+        >
+          <IconWifiOff size={20} stroke={2} />
+        </ThemeIcon>
+      </Tooltip>
+    );
+  }
+
+  // State (2): connected but the remote replica is not fully caught up yet.
+  if (!isRemoteSynced || !isLocalSynced) {
+    const syncingLabel = t("Syncing changes…");
+    return (
+      <Tooltip label={syncingLabel} openDelay={250} withArrow>
+        <ThemeIcon
+          variant="default"
+          c="dimmed"
+          role="status"
+          aria-label={syncingLabel}
+          style={{ border: "none" }}
+        >
+          <IconCloud size={20} stroke={2} />
+        </ThemeIcon>
+      </Tooltip>
+    );
+  }
+
+  // State (3): fully synced — subtle confirmation indicator.
+  const syncedLabel = t("All changes synced");
   return (
-    <Tooltip
-      label={t("Real-time editor connection lost. Retrying...")}
-      openDelay={250}
-      withArrow
-    >
+    <Tooltip label={syncedLabel} openDelay={250} withArrow>
       <ThemeIcon
         variant="default"
-        c="red"
+        c="dimmed"
         role="status"
-        aria-label={t("Real-time editor connection lost. Retrying...")}
+        aria-label={syncedLabel}
         style={{ border: "none" }}
       >
-        <IconWifiOff size={20} stroke={2} />
+        <IconCloudCheck size={20} stroke={2} />
       </ThemeIcon>
     </Tooltip>
   );

apps/client/src/features/page/queries/page-query.ts

@@ -1,6 +1,7 @@
 import {
   InfiniteData,
   QueryKey,
+  queryOptions,
   useInfiniteQuery,
   UseInfiniteQueryResult,
   useMutation,
@@ -42,12 +43,38 @@ import { treeModel } from "@/features/page/tree/model/tree-model";
 import { SpaceTreeNode } from "@/features/page/tree/types";
 import { useQueryEmit } from "@/features/websocket/use-query-emit";
 import { moveToTrashNotificationMessage } from "@/features/page/components/move-to-trash-notification";
+import { offlineMutationKeys } from "@/features/offline/offline-mutations";
+
+/**
+ * Centralized React Query key factories for page queries. The hooks below and
+ * the offline warm path (features/offline/make-offline.ts) share these so the
+ * runtime keys can never silently drift apart.
+ */
+export const pageKeys = {
+  detail: (idOrSlug: string) => ["pages", idOrSlug] as const,
+  sidebar: (data: unknown) => ["sidebar-pages", data] as const,
+  rootSidebar: (spaceId: string) => ["root-sidebar-pages", spaceId] as const,
+  breadcrumbs: (pageId: string) => ["breadcrumbs", pageId] as const,
+  recentChanges: (spaceId?: string) => ["recent-changes", spaceId] as const,
+};
+
+/**
+ * Shared queryOptions for the sidebar-pages (ancestor children) query. Both
+ * fetchAllAncestorChildren and the offline warm path consume this so the key,
+ * queryFn and staleTime stay identical.
+ */
+export const sidebarPagesQueryOptions = (params: SidebarPagesParams) =>
+  queryOptions({
+    queryKey: pageKeys.sidebar(params),
+    queryFn: () => getAllSidebarPages(params),
+    staleTime: 30 * 60 * 1000,
+  });
 
 export function usePageQuery(
   pageInput: Partial<IPageInput>,
 ): UseQueryResult<IPage, Error> {
   const query = useQuery({
-    queryKey: ["pages", pageInput.pageId],
+    queryKey: pageKeys.detail(pageInput.pageId),
     queryFn: () => getPageById(pageInput),
     enabled: !!pageInput.pageId,
     staleTime: 5 * 60 * 1000,
@@ -56,9 +83,9 @@ export function usePageQuery(
   useEffect(() => {
     if (query.data) {
       if (isValidUuid(pageInput.pageId)) {
-        queryClient.setQueryData(["pages", query.data.slugId], query.data);
+        queryClient.setQueryData(pageKeys.detail(query.data.slugId), query.data);
       } else {
-        queryClient.setQueryData(["pages", query.data.id], query.data);
+        queryClient.setQueryData(pageKeys.detail(query.data.id), query.data);
       }
     }
   }, [query.data]);
@@ -69,6 +96,10 @@ export function usePageQuery(
 export function useCreatePageMutation() {
   const { t } = useTranslation();
   return useMutation<IPage, Error, Partial<IPageInput>>({
+    // Stable key so a paused create restored from IndexedDB after an offline
+    // reload finds its default mutationFn (registerOfflineMutationDefaults) and
+    // is replayed by resumePausedMutations() on reconnect instead of being lost.
+    mutationKey: offlineMutationKeys.createPage,
     mutationFn: (data) => createPage(data),
     onSuccess: (data) => {
       invalidateOnCreatePage(data);
@@ -80,18 +111,20 @@ export function useCreatePageMutation() {
 }
 
 export function updatePageData(data: IPage) {
-  const pageBySlug = queryClient.getQueryData<IPage>(["pages", data.slugId]);
-  const pageById = queryClient.getQueryData<IPage>(["pages", data.id]);
+  const pageBySlug = queryClient.getQueryData<IPage>(
+    pageKeys.detail(data.slugId),
+  );
+  const pageById = queryClient.getQueryData<IPage>(pageKeys.detail(data.id));
 
   if (pageBySlug) {
-    queryClient.setQueryData(["pages", data.slugId], {
+    queryClient.setQueryData(pageKeys.detail(data.slugId), {
       ...pageBySlug,
       ...data,
     });
   }
 
   if (pageById) {
-    queryClient.setQueryData(["pages", data.id], { ...pageById, ...data });
+    queryClient.setQueryData(pageKeys.detail(data.id), { ...pageById, ...data });
   }
 
   invalidateOnUpdatePage(
@@ -145,11 +178,11 @@ export function useRemovePageMutation() {
       });
 
       // Stamp deletedAt so a re-visit shows the trash banner, not stale state.
-      const cached = queryClient.getQueryData<IPage>(["pages", pageId]);
+      const cached = queryClient.getQueryData<IPage>(pageKeys.detail(pageId));
       if (cached) {
         const stamped = { ...cached, deletedAt: new Date() };
-        queryClient.setQueryData(["pages", cached.id], stamped);
-        queryClient.setQueryData(["pages", cached.slugId], stamped);
+        queryClient.setQueryData(pageKeys.detail(cached.id), stamped);
+        queryClient.setQueryData(pageKeys.detail(cached.slugId), stamped);
       }
 
       invalidateOnDeletePage(pageId);
@@ -188,6 +221,9 @@ export function useDeletePageMutation() {
 
 export function useMovePageMutation() {
   return useMutation<void, Error, IMovePage>({
+    // Stable key so a paused move restored from IndexedDB after an offline
+    // reload finds its default mutationFn and is replayed on reconnect.
+    mutationKey: offlineMutationKeys.movePage,
     mutationFn: (data) => movePage(data),
   });
 }
@@ -267,8 +303,11 @@ export function useRestorePageMutation() {
       // Replace would strip space/permissions/content and break the editor.
       const merge = (cached: IPage | undefined) =>
         cached ? { ...cached, ...restoredPage } : cached;
-      queryClient.setQueryData<IPage>(["pages", restoredPage.id], merge);
-      queryClient.setQueryData<IPage>(["pages", restoredPage.slugId], merge);
+      queryClient.setQueryData<IPage>(pageKeys.detail(restoredPage.id), merge);
+      queryClient.setQueryData<IPage>(
+        pageKeys.detail(restoredPage.slugId),
+        merge,
+      );
     },
     onError: (error) => {
       notifications.show({
@@ -283,7 +322,7 @@ export function useGetSidebarPagesQuery(
   data: SidebarPagesParams | null,
 ): UseInfiniteQueryResult<InfiniteData<IPagination<IPage>, unknown>> {
   return useInfiniteQuery({
-    queryKey: ["sidebar-pages", data],
+    queryKey: pageKeys.sidebar(data),
     enabled: !!data?.pageId || !!data?.spaceId,
     queryFn: ({ pageParam }) =>
       getSidebarPages({ ...data, cursor: pageParam, limit: 100 }),
@@ -294,7 +333,7 @@ export function useGetSidebarPagesQuery(
 
 export function useGetRootSidebarPagesQuery(data: SidebarPagesParams) {
   return useInfiniteQuery({
-    queryKey: ["root-sidebar-pages", data.spaceId],
+    queryKey: pageKeys.rootSidebar(data.spaceId),
     queryFn: async ({ pageParam }) => {
       return getSidebarPages({
         spaceId: data.spaceId,
@@ -320,7 +359,7 @@ export function usePageBreadcrumbsQuery(
   pageId: string,
 ): UseQueryResult<Partial<IPage[]>, Error> {
   return useQuery({
-    queryKey: ["breadcrumbs", pageId],
+    queryKey: pageKeys.breadcrumbs(pageId),
     queryFn: () => getPageBreadcrumbs(pageId),
     enabled: !!pageId,
   });
@@ -332,10 +371,12 @@ export async function fetchAllAncestorChildren(
   // refresh (#159 #8), which must NOT receive the 30-min-cached children.
   opts?: { fresh?: boolean },
 ) {
-  // not using a hook here, so we can call it inside a useEffect hook
+  // not using a hook here, so we can call it inside a useEffect hook. Reuse the
+  // shared sidebarPagesQueryOptions (key + queryFn) so the offline warm path and
+  // this fetch never drift, but override staleTime for the `fresh` reconnect
+  // refresh (#159 #8), which must force a server refetch (staleTime 0).
   const response = await queryClient.fetchQuery({
-    queryKey: ["sidebar-pages", params],
-    queryFn: () => getAllSidebarPages(params),
+    ...sidebarPagesQueryOptions(params),
     staleTime: opts?.fresh ? 0 : 30 * 60 * 1000,
   });
 
@@ -345,7 +386,7 @@ export async function fetchAllAncestorChildren(
 
 export function useRecentChangesQuery(spaceId?: string) {
   return useInfiniteQuery({
-    queryKey: ["recent-changes", spaceId],
+    queryKey: pageKeys.recentChanges(spaceId),
     queryFn: ({ pageParam }) =>
       getRecentChanges({ spaceId, cursor: pageParam, limit: 15 }),
     initialPageParam: undefined as string | undefined,
@@ -416,12 +457,12 @@ export function invalidateOnCreatePage(data: Partial<IPage>) {
 
   let queryKey: QueryKey = null;
   if (data.parentPageId === null) {
-    queryKey = ["root-sidebar-pages", data.spaceId];
+    queryKey = pageKeys.rootSidebar(data.spaceId);
   } else {
-    queryKey = [
-      "sidebar-pages",
-      { pageId: data.parentPageId, spaceId: data.spaceId },
-    ];
+    queryKey = pageKeys.sidebar({
+      pageId: data.parentPageId,
+      spaceId: data.spaceId,
+    });
   }
 
   //update all sidebar pages
@@ -481,7 +522,7 @@ export function invalidateOnCreatePage(data: Partial<IPage>) {
 
     //update root sidebar pages haschildern
     const rootSideBarMatches = queryClient.getQueriesData({
-      queryKey: ["root-sidebar-pages", data.spaceId],
+      queryKey: pageKeys.rootSidebar(data.spaceId),
       exact: false,
     });
 
@@ -505,7 +546,7 @@ export function invalidateOnCreatePage(data: Partial<IPage>) {
 
   //update recent changes
   queryClient.invalidateQueries({
-    queryKey: ["recent-changes", data.spaceId],
+    queryKey: pageKeys.recentChanges(data.spaceId),
   });
 }
 
@@ -519,9 +560,9 @@ export function invalidateOnUpdatePage(
   invalidatePageTree();
   let queryKey: QueryKey = null;
   if (parentPageId === null) {
-    queryKey = ["root-sidebar-pages", spaceId];
+    queryKey = pageKeys.rootSidebar(spaceId);
   } else {
-    queryKey = ["sidebar-pages", { pageId: parentPageId, spaceId: spaceId }];
+    queryKey = pageKeys.sidebar({ pageId: parentPageId, spaceId: spaceId });
   }
   //update all sidebar pages
   queryClient.setQueryData<InfiniteData<IPagination<IPage>>>(
@@ -544,7 +585,7 @@ export function invalidateOnUpdatePage(
 
   //update recent changes
   queryClient.invalidateQueries({
-    queryKey: ["recent-changes", spaceId],
+    queryKey: pageKeys.recentChanges(spaceId),
   });
 }
 
@@ -559,8 +600,8 @@ export function updateCacheOnMovePage(
   // Remove page from old parent's cache
   const oldQueryKey =
     oldParentId === null
-      ? ["root-sidebar-pages", spaceId]
-      : ["sidebar-pages", { pageId: oldParentId, spaceId }];
+      ? pageKeys.rootSidebar(spaceId)
+      : pageKeys.sidebar({ pageId: oldParentId, spaceId });
 
   queryClient.setQueryData<InfiniteData<IPagination<IPage>>>(
     oldQueryKey,
@@ -580,7 +621,7 @@ export function updateCacheOnMovePage(
   if (oldParentId !== null) {
     const oldParentCache = queryClient.getQueryData<
       InfiniteData<IPagination<IPage>>
-    >(["sidebar-pages", { pageId: oldParentId, spaceId }]);
+    >(pageKeys.sidebar({ pageId: oldParentId, spaceId }));
 
     const remainingChildren =
       oldParentCache?.pages.flatMap((p) => p.items).length ?? 0;
@@ -618,8 +659,8 @@ export function updateCacheOnMovePage(
   // Add page to new parent's cache
   const newQueryKey =
     newParentId === null
-      ? ["root-sidebar-pages", spaceId]
-      : ["sidebar-pages", { pageId: newParentId, spaceId }];
+      ? pageKeys.rootSidebar(spaceId)
+      : pageKeys.sidebar({ pageId: newParentId, spaceId });
 
   queryClient.setQueryData<InfiniteData<IPagination<Partial<IPage>>>>(
     newQueryKey,

apps/client/src/features/page/tree/components/space-tree-node-menu.tsx

@@ -7,6 +7,7 @@ import { notifications } from "@mantine/notifications";
 import {
   IconArrowRight,
   IconClockHour4,
+  IconCloudDownload,
   IconCopy,
   IconDotsVertical,
   IconFileExport,
@@ -35,6 +36,12 @@ import {
   useToggleTemplateMutation,
   useToggleTemporaryMutation,
 } from "@/features/page-embed/queries/page-embed-query";
+import { useCollabToken } from "@/features/auth/queries/auth-query.tsx";
+import { getCollaborationUrl } from "@/lib/config.ts";
+import {
+  makePageAvailableOffline,
+  warmPageYdoc,
+} from "@/features/offline/make-offline";
 import { treeDataAtom } from "@/features/page/tree/atoms/tree-data-atom.ts";
 import { treeModel } from "@/features/page/tree/model/tree-model";
 import { pageToTreeNode } from "@/features/page/tree/utils";
@@ -72,6 +79,46 @@ export function NodeMenu({ node, canEdit }: NodeMenuProps) {
   const isTemplate = !!node.isTemplate;
   const toggleTemporary = useToggleTemporaryMutation();
   const isTemporary = !!node.temporaryExpiresAt;
+  const { data: collabQuery } = useCollabToken();
+
+  const handleMakeAvailableOffline = async () => {
+    notifications.show({ message: t("Saving page for offline use...") });
+    try {
+      // Prefetch read queries so they get persisted to IndexedDB. The result
+      // reports whether every warm step succeeded.
+      const result = await makePageAvailableOffline({
+        pageId: node.id,
+        spaceId: node.spaceId,
+      });
+      // Best-effort: warm the page's Yjs document into IndexedDB.
+      await warmPageYdoc(node.id, getCollaborationUrl(), collabQuery?.token);
+
+      if (result.ok) {
+        notifications.show({ message: t("Page is now available offline") });
+      } else {
+        // Partial warm — the page may still be partly usable offline, but some
+        // queries failed to cache, so surface it as an error rather than a
+        // silent success. Name the failed step(s) (AGENTS.md: errors must be
+        // specific, never a bare generic string); `result.failed` carries them.
+        notifications.show({
+          message: `${t("Failed to make page available offline")}: ${result.failed.join(", ")}`,
+          color: "red",
+        });
+      }
+    } catch (err) {
+      // makePageAvailableOffline no longer throws, but warmPageYdoc and other
+      // unexpected failures stay guarded here. Log the raw error and surface the
+      // real cause to the user instead of a bare generic string (AGENTS.md).
+      console.error("handleMakeAvailableOffline failed", err);
+      const reason =
+        (err as { response?: { data?: { message?: string } } })?.response?.data
+          ?.message ?? (err instanceof Error ? err.message : String(err));
+      notifications.show({
+        message: `${t("Failed to make page available offline")}: ${reason}`,
+        color: "red",
+      });
+    }
+  };
 
   const handleToggleTemplate = async () => {
     const next = !isTemplate;
@@ -228,6 +275,17 @@ export function NodeMenu({ node, canEdit }: NodeMenuProps) {
             {t("Export")}
           </Menu.Item>
 
+          <Menu.Item
+            leftSection={<IconCloudDownload size={16} />}
+            onClick={(e) => {
+              e.preventDefault();
+              e.stopPropagation();
+              handleMakeAvailableOffline();
+            }}
+          >
+            {t("Make available offline")}
+          </Menu.Item>
+
           {canEdit && (
             <>
               <Menu.Item

apps/client/src/features/page/tree/hooks/use-tree-mutation.ts

@@ -15,7 +15,6 @@ import {
   useCreatePageMutation,
   useRemovePageMutation,
   useMovePageMutation,
-  useUpdatePageMutation,
   updateCacheOnMovePage,
 } from "@/features/page/queries/page-query.ts";
 import { buildPageUrl } from "@/features/page/page.utils.ts";
@@ -27,7 +26,6 @@ export type UseTreeMutation = {
     parentId: string | null,
     opts?: { temporary?: boolean },
   ) => Promise<void>;
-  handleRename: (id: string, name: string) => Promise<void>;
   handleDelete: (id: string) => Promise<void>;
 };
 
@@ -39,7 +37,6 @@ export function useTreeMutation(spaceId: string): UseTreeMutation {
   // children) and then immediately invokes a handler.
   const store = useStore();
   const createPageMutation = useCreatePageMutation();
-  const updatePageMutation = useUpdatePageMutation();
   const removePageMutation = useRemovePageMutation();
   const movePageMutation = useMovePageMutation();
   const navigate = useNavigate();
@@ -205,20 +202,6 @@ export function useTreeMutation(spaceId: string): UseTreeMutation {
     [spaceId, createPageMutation, setData, store, navigate, spaceSlug],
   );
 
-  const handleRename = useCallback(
-    async (id: string, name: string) => {
-      setData((prev) =>
-        treeModel.update(prev, id, { name } as Partial<SpaceTreeNode>),
-      );
-      try {
-        await updatePageMutation.mutateAsync({ pageId: id, title: name });
-      } catch (error) {
-        console.error("Error updating page title:", error);
-      }
-    },
-    [updatePageMutation, setData],
-  );
-
   const handleDelete = useCallback(
     async (id: string) => {
       const node = treeModel.find(
@@ -264,7 +247,7 @@ export function useTreeMutation(spaceId: string): UseTreeMutation {
     [removePageMutation, setData, store, pageSlug, navigate, spaceSlug],
   );
 
-  return { handleMove, handleCreate, handleRename, handleDelete };
+  return { handleMove, handleCreate, handleDelete };
 }
 
 function isPageInNode(node: SpaceTreeNode, pageSlug: string): boolean {

apps/client/src/features/space/queries/space-query.ts

@@ -1,5 +1,6 @@
 import {
   keepPreviousData,
+  queryOptions,
   useInfiniteQuery,
   useMutation,
   useQuery,
@@ -31,11 +32,37 @@ import { getRecentChanges } from "@/features/page/services/page-service.ts";
 import { useEffect } from "react";
 import { validate as isValidUuid } from "uuid";
 
+/**
+ * Centralized React Query key factories for space queries. The hooks below and
+ * the offline warm path (features/offline/make-offline.ts) share these so the
+ * runtime keys can never silently drift apart.
+ */
+export const spaceKeys = {
+  detail: (idOrSlug: string) => ["space", idOrSlug] as const,
+  list: (params?: QueryParams) => ["spaces", params] as const,
+  members: (spaceId: string, query?: string) =>
+    ["spaceMembers", spaceId, query] as const,
+};
+
+/**
+ * Shared queryOptions for fetching a space by id/slug. Both
+ * useGetSpaceBySlugQuery and the offline warm path consume this so the key,
+ * queryFn and staleTime stay identical. (`enabled` is intentionally omitted —
+ * prefetchQuery ignores it anyway and the warm path always passes a real id;
+ * the hook reapplies `enabled` itself.)
+ */
+export const spaceByIdQueryOptions = (spaceId: string) =>
+  queryOptions({
+    queryKey: spaceKeys.detail(spaceId),
+    queryFn: () => getSpaceById(spaceId),
+    staleTime: 5 * 60 * 1000,
+  });
+
 export function useGetSpacesQuery(
   params?: QueryParams,
 ): UseQueryResult<IPagination<ISpace>, Error> {
   return useQuery({
-    queryKey: ["spaces", params],
+    queryKey: spaceKeys.list(params),
     queryFn: () => getSpaces(params),
     placeholderData: keepPreviousData,
     refetchOnMount: true,
@@ -44,16 +71,16 @@ export function useGetSpacesQuery(
 
 export function useSpaceQuery(spaceId: string): UseQueryResult<ISpace, Error> {
   const query = useQuery({
-    queryKey: ["space", spaceId],
+    queryKey: spaceKeys.detail(spaceId),
     queryFn: () => getSpaceById(spaceId),
     enabled: !!spaceId,
   });
   useEffect(() => {
     if (query.data) {
       if (isValidUuid(spaceId)) {
-        queryClient.setQueryData(["space", query.data.slug], query.data);
+        queryClient.setQueryData(spaceKeys.detail(query.data.slug), query.data);
       } else {
-        queryClient.setQueryData(["space", query.data.id], query.data);
+        queryClient.setQueryData(spaceKeys.detail(query.data.id), query.data);
       }
     }
   }, [query.data]);
@@ -62,8 +89,11 @@ export function useSpaceQuery(spaceId: string): UseQueryResult<ISpace, Error> {
 }
 
 export const prefetchSpace = (spaceSlug: string, spaceId?: string) => {
+  // Note: intentionally NOT using spaceByIdQueryOptions here — that factory sets
+  // a 5min staleTime which would let this prefetch skip fetching fresh data;
+  // prefetchSpace must always refetch (default staleTime: 0).
   queryClient.prefetchQuery({
-    queryKey: ["space", spaceSlug],
+    queryKey: spaceKeys.detail(spaceSlug),
     queryFn: () => getSpaceById(spaceSlug),
   });
 
@@ -100,10 +130,8 @@ export function useGetSpaceBySlugQuery(
   spaceId: string,
 ): UseQueryResult<ISpace, Error> {
   return useQuery({
-    queryKey: ["space", spaceId],
-    queryFn: () => getSpaceById(spaceId),
+    ...spaceByIdQueryOptions(spaceId),
     enabled: !!spaceId,
-    staleTime: 5 * 60 * 1000,
   });
 }
 
@@ -116,14 +144,16 @@ export function useUpdateSpaceMutation() {
     onSuccess: (data, variables) => {
       notifications.show({ message: t("Space updated successfully") });
 
-      const space = queryClient.getQueryData([
-        "space",
-        variables.spaceId,
-      ]) as ISpace;
+      const space = queryClient.getQueryData(
+        spaceKeys.detail(variables.spaceId),
+      ) as ISpace;
       if (space) {
         const updatedSpace = { ...space, ...data };
-        queryClient.setQueryData(["space", variables.spaceId], updatedSpace);
-        queryClient.setQueryData(["space", data.slug], updatedSpace);
+        queryClient.setQueryData(
+          spaceKeys.detail(variables.spaceId),
+          updatedSpace,
+        );
+        queryClient.setQueryData(spaceKeys.detail(data.slug), updatedSpace);
       }
 
       queryClient.invalidateQueries({
@@ -148,7 +178,7 @@ export function useDeleteSpaceMutation() {
 
       if (variables.slug) {
         queryClient.removeQueries({
-          queryKey: ["space", variables.slug],
+          queryKey: spaceKeys.detail(variables.slug),
           exact: true,
         });
       }
@@ -156,7 +186,7 @@ export function useDeleteSpaceMutation() {
       // Remove space-specific queries
       if (variables.id) {
         queryClient.removeQueries({
-          queryKey: ["space", variables.id],
+          queryKey: spaceKeys.detail(variables.id),
           exact: true,
         });
 
@@ -196,7 +226,7 @@ export function useSpaceMembersInfiniteQuery(
   query?: string,
 ) {
   return useInfiniteQuery({
-    queryKey: ["spaceMembers", spaceId, query],
+    queryKey: spaceKeys.members(spaceId, query),
     queryFn: ({ pageParam }) =>
       getSpaceMembers(spaceId, { cursor: pageParam, limit: 50, query }),
     enabled: !!spaceId,

apps/client/src/features/user/hooks/use-current-user.ts

@@ -2,9 +2,19 @@ import { useQuery, UseQueryResult } from "@tanstack/react-query";
 import { getMyInfo } from "@/features/user/services/user-service";
 import { ICurrentUser } from "@/features/user/types/user.types";
 
+/**
+ * Centralized React Query key factory for current-user queries. This hook and
+ * the offline warm path (features/offline/make-offline.ts) share it so the
+ * runtime key can never silently drift, and the OFFLINE_PERSIST_ROOTS guard
+ * test can assert the persisted "currentUser" root maps to a real factory.
+ */
+export const userKeys = {
+  currentUser: () => ["currentUser"] as const,
+};
+
 export default function useCurrentUser(): UseQueryResult<ICurrentUser> {
   return useQuery({
-    queryKey: ["currentUser"],
+    queryKey: userKeys.currentUser(),
     queryFn: async () => {
       return await getMyInfo();
     },

apps/client/src/features/user/user-provider.test.tsx

@@ -0,0 +1,118 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+import { render, screen } from "@testing-library/react";
+import { MantineProvider } from "@mantine/core";
+import { MemoryRouter } from "react-router-dom";
+import { HelmetProvider } from "react-helmet-async";
+
+// Control useCurrentUser per test; stub the rest of UserProvider's network/
+// socket dependencies so we only exercise its render-gating logic.
+const h = vi.hoisted(() => ({ useCurrentUser: vi.fn() }));
+
+vi.mock("@/features/user/hooks/use-current-user", () => ({
+  default: h.useCurrentUser,
+}));
+vi.mock("@/features/auth/queries/auth-query.tsx", () => ({
+  useCollabToken: () => ({ data: undefined }),
+}));
+vi.mock("@/features/websocket/use-query-subscription.ts", () => ({
+  useQuerySubscription: () => {},
+}));
+vi.mock("@/features/websocket/use-tree-socket.ts", () => ({
+  useTreeSocket: () => {},
+}));
+vi.mock("@/features/notification/hooks/use-notification-socket.ts", () => ({
+  useNotificationSocket: () => {},
+}));
+vi.mock("@/main.tsx", () => ({ queryClient: {} }));
+vi.mock("@/features/user/connect-resync.ts", () => ({
+  makeConnectHandler: () => () => {},
+}));
+vi.mock("socket.io-client", () => ({
+  io: () => ({ on: vi.fn(), disconnect: vi.fn() }),
+}));
+vi.mock("react-i18next", () => ({
+  useTranslation: () => ({
+    t: (k: string) => k,
+    i18n: {
+      changeLanguage: vi.fn(),
+      language: "en-US",
+      resolvedLanguage: "en-US",
+    },
+  }),
+}));
+
+import { UserProvider } from "./user-provider";
+
+const networkError = { message: "Network Error" }; // axios network error: no `response`
+
+function renderProvider() {
+  return render(
+    <HelmetProvider>
+      <MemoryRouter>
+        <MantineProvider>
+          <UserProvider>
+            <div data-testid="app-child">app content</div>
+          </UserProvider>
+        </MantineProvider>
+      </MemoryRouter>
+    </HelmetProvider>,
+  );
+}
+
+beforeEach(() => {
+  h.useCurrentUser.mockReset();
+});
+
+describe("UserProvider offline render-gating", () => {
+  it("renders the app (cached children) when useCurrentUser errors offline but a cached user exists", () => {
+    // Offline reload: the persisted ['currentUser'] cache hydrates `data`, but
+    // the background POST /api/users/me refetch fails as a network error.
+    h.useCurrentUser.mockReturnValue({
+      data: {
+        user: { id: "u1", locale: "en" },
+        workspace: { id: "w1" },
+      },
+      isLoading: false,
+      error: networkError,
+      isError: true,
+    });
+
+    renderProvider();
+
+    // The cached app must render — NOT a blank fragment (#237/#238).
+    expect(screen.getByTestId("app-child")).toBeDefined();
+    expect(screen.queryByText("You're offline")).toBeNull();
+  });
+
+  it("renders the offline fallback (not a blank fragment) when erroring with no cached user", () => {
+    h.useCurrentUser.mockReturnValue({
+      data: undefined,
+      isLoading: false,
+      error: networkError,
+      isError: true,
+    });
+
+    const { container } = renderProvider();
+
+    // Previously this returned `<></>` — a blank white screen. Now it must show
+    // an explicit offline fallback.
+    expect(screen.getByText("You're offline")).toBeDefined();
+    expect(screen.queryByTestId("app-child")).toBeNull();
+    expect(container.textContent?.length).toBeGreaterThan(0);
+  });
+
+  it("renders the app normally on a successful currentUser load", () => {
+    h.useCurrentUser.mockReturnValue({
+      data: {
+        user: { id: "u1", locale: "en" },
+        workspace: { id: "w1" },
+      },
+      isLoading: false,
+      error: null,
+      isError: false,
+    });
+
+    renderProvider();
+    expect(screen.getByTestId("app-child")).toBeDefined();
+  });
+});

apps/client/src/features/user/user-provider.tsx

@@ -11,6 +11,7 @@ import { useTreeSocket } from "@/features/websocket/use-tree-socket.ts";
 import { useNotificationSocket } from "@/features/notification/hooks/use-notification-socket.ts";
 import { useCollabToken } from "@/features/auth/queries/auth-query.tsx";
 import { Error404 } from "@/components/ui/error-404.tsx";
+import { OfflineFallback } from "@/features/offline/offline-fallback.tsx";
 import { queryClient } from "@/main.tsx";
 import { makeConnectHandler } from "@/features/user/connect-resync.ts";
 
@@ -70,14 +71,30 @@ export function UserProvider({ children }: React.PropsWithChildren) {
     document.documentElement.lang = i18n.resolvedLanguage || i18n.language || "en-US";
   }, [i18n.language, i18n.resolvedLanguage]);
 
-  if (isLoading) return <></>;
+  // First load with no cached user yet: render nothing briefly while the
+  // persisted ['currentUser'] cache hydrates (avoids flashing the offline
+  // fallback before restore). Once we have a user we render the app even if a
+  // refetch is still in flight.
+  if (isLoading && !data) return <></>;
 
   if (isError && error?.["response"]?.status === 404) {
     return <Error404 />;
   }
 
+  // We have a (possibly cached/stale) user — render the app. Offline, the
+  // POST /api/users/me refetch fails as a network error, but the persisted/
+  // hydrated user is enough to render the cached UI. Previously `if (error)
+  // return <></>` blanked every authenticated route on an offline reload even
+  // though the cached data was present (#237/#238).
+  if (data) {
+    return <>{children}</>;
+  }
+
+  // No user AND an error (offline cold boot of a page never warmed for offline,
+  // or no persisted cache to restore): show an explicit offline fallback rather
+  // than a blank white screen.
   if (error) {
-    return <></>;
+    return <OfflineFallback />;
   }
 
   return <>{children}</>;

apps/client/src/features/workspace/types/workspace.types.ts

@@ -65,9 +65,6 @@ export interface IWorkspaceAiSettings {
   dictation?: boolean;
   dictationStreaming?: boolean;
   publicShareAssistant?: boolean;
-  // #184: detached agent runs (a run survives a browser disconnect and can be
-  // reconnected to / live-followed on reopen). Gates the run-reconnect polling.
-  autonomousRuns?: boolean;
 }
 
 export interface IWorkspaceSharingSettings {

apps/client/src/main.tsx

@@ -11,7 +11,8 @@ import { MantineProvider } from "@mantine/core";
 import { BrowserRouter } from "react-router-dom";
 import { ModalsProvider } from "@mantine/modals";
 import { Notifications } from "@mantine/notifications";
-import { QueryClient, QueryClientProvider } from "@tanstack/react-query";
+import { QueryClient, onlineManager } from "@tanstack/react-query";
+import { PersistQueryClientProvider } from "@tanstack/react-query-persist-client";
 import { HelmetProvider } from "react-helmet-async";
 import "./i18n";
 import { PostHogProvider } from "posthog-js/react";
@@ -21,6 +22,13 @@ import {
   isCloud,
   isPostHogEnabled,
 } from "@/lib/config.ts";
+import {
+  queryPersister,
+  shouldDehydrateOfflineQuery,
+} from "@/features/offline/query-persister";
+import { registerOfflineMutationDefaults } from "@/features/offline/offline-mutations";
+import { PwaUpdatePrompt } from "@/pwa/pwa-update-prompt";
+import { isCapacitorNativePlatform } from "@/pwa/is-capacitor";
 import posthog from "posthog-js";
 
 export const queryClient = new QueryClient({
@@ -30,10 +38,30 @@ export const queryClient = new QueryClient({
       refetchOnWindowFocus: false,
       retry: false,
       staleTime: 5 * 60 * 1000,
+      // Keep cached read data around long enough to be persisted/restored for offline use.
+      gcTime: 1000 * 60 * 60 * 24,
     },
   },
 });
 
+// Register default mutationFns for the offline-relevant structural mutations so
+// a paused mutation restored from IndexedDB after an offline reload still has a
+// mutationFn and is replayed by resumePausedMutations() on reconnect (instead
+// of silently no-op'ing and dropping the offline create/move/comment). MUST run
+// before any resumePausedMutations() so rehydrated paused mutations have a fn.
+registerOfflineMutationDefaults(queryClient);
+
+// Seed TanStack Query's onlineManager from the REAL connectivity state at boot.
+// It defaults to `online: true` and only flips on window online/offline events,
+// so a tab that COLD-BOOTS offline would wrongly believe it is online: paused
+// mutations restored from IndexedDB would never get a later offline->online
+// transition to trigger their replay, and the offline UI affordances could not
+// tell they are offline. Seeding here makes the first real `online` event a true
+// transition that auto-resumes the rehydrated paused mutations (#120 data loss).
+if (typeof navigator !== "undefined" && "onLine" in navigator) {
+  onlineManager.setOnline(navigator.onLine);
+}
+
 if (isCloud() && isPostHogEnabled) {
   posthog.init(getPostHogKey(), {
     api_host: getPostHogHost(),
@@ -50,15 +78,44 @@ root.render(
   <BrowserRouter>
     <MantineProvider theme={theme} cssVariablesResolver={mantineCssResolver}>
       <ModalsProvider>
-        <QueryClientProvider client={queryClient}>
+        <PersistQueryClientProvider
+          client={queryClient}
+          persistOptions={{
+            persister: queryPersister,
+            maxAge: 1000 * 60 * 60 * 24,
+            buster: APP_VERSION,
+            dehydrateOptions: {
+              shouldDehydrateQuery: shouldDehydrateOfflineQuery,
+            },
+          }}
+          // After the persister finishes rehydrating, replay any paused
+          // mutations restored from IndexedDB. If we are back online this fires
+          // them immediately; if still offline they stay paused and TanStack's
+          // onlineManager auto-resumes them on the next online transition (which
+          // is now a true transition thanks to the onlineManager seeding above).
+          // Without this, a paused mutation persisted while offline and then
+          // reloaded would never resume and the user's work would be lost (#120).
+          onSuccess={() => {
+            queryClient.resumePausedMutations();
+          }}
+        >
           <Notifications position="bottom-center" limit={3} zIndex={10000} />
+          {/* Skip SW registration inside the Capacitor native WebView — the
+              native shell serves assets itself; a browser SW would conflict. */}
+          {!isCapacitorNativePlatform() && <PwaUpdatePrompt />}
           <HelmetProvider>
             <PostHogProvider client={posthog}>
               <App />
             </PostHogProvider>
           </HelmetProvider>
-        </QueryClientProvider>
+        </PersistQueryClientProvider>
       </ModalsProvider>
     </MantineProvider>
   </BrowserRouter>,
 );
+
+// Service worker registration is owned by <PwaUpdatePrompt /> above (via
+// vite-plugin-pwa's useRegisterSW: Workbox precache + prompt-based updates,
+// and skipped inside the Capacitor native WebView). The earlier hand-written
+// /sw.js registration from the mobile bootstrap was removed here to avoid a
+// double registration / competing service worker.

apps/client/src/pages/page/page.tsx

@@ -9,6 +9,7 @@ import { useGetSpaceBySlugQuery } from "@/features/space/queries/space-query.ts"
 import { useTranslation } from "react-i18next";
 import React from "react";
 import { EmptyState } from "@/components/ui/empty-state.tsx";
+import { OfflineFallback } from "@/features/offline/offline-fallback.tsx";
 import { IconAlertTriangle, IconFileOff } from "@tabler/icons-react";
 import { Button } from "@mantine/core";
 import { Link } from "react-router-dom";
@@ -62,7 +63,19 @@ function PageContent({ pageSlug }: { pageSlug: string | undefined }) {
   }
 
   if (isError || !page) {
-    if ([401, 403, 404].includes(error?.["status"])) {
+    // An offline fetch of a page that was never saved for offline use yields a
+    // network error with NO HTTP status (status is undefined), which would
+    // otherwise fall through to the generic "Error fetching page data." state.
+    // When we are offline (or the failure is a network error with no status),
+    // show the dedicated "You're offline — this page isn't saved for offline"
+    // fallback instead, so the user understands why the page won't load.
+    const httpStatus = error?.["status"];
+    const isOffline =
+      typeof navigator !== "undefined" && navigator.onLine === false;
+    if (isOffline || (isError && httpStatus == null)) {
+      return <OfflineFallback />;
+    }
+    if ([401, 403, 404].includes(httpStatus)) {
       return (
         <EmptyState
           icon={IconFileOff}

apps/client/src/pwa/is-capacitor.test.ts

@@ -0,0 +1,39 @@
+import { describe, it, expect, afterEach } from "vitest";
+import { isCapacitorNativePlatform } from "./is-capacitor";
+
+describe("isCapacitorNativePlatform", () => {
+  afterEach(() => {
+    // Keep tests isolated from each other and from the rest of the suite.
+    delete (globalThis as any).Capacitor;
+  });
+
+  it("returns false when Capacitor is undefined", () => {
+    expect(isCapacitorNativePlatform()).toBe(false);
+  });
+
+  it("uses isNativePlatform() when it is a function", () => {
+    (globalThis as any).Capacitor = { isNativePlatform: () => true };
+    expect(isCapacitorNativePlatform()).toBe(true);
+
+    (globalThis as any).Capacitor = { isNativePlatform: () => false };
+    expect(isCapacitorNativePlatform()).toBe(false);
+  });
+
+  it("falls back to the boolean property when isNativePlatform is not a function", () => {
+    (globalThis as any).Capacitor = { isNativePlatform: true };
+    expect(isCapacitorNativePlatform()).toBe(true);
+
+    (globalThis as any).Capacitor = { isNativePlatform: false };
+    expect(isCapacitorNativePlatform()).toBe(false);
+  });
+
+  it("returns false when reading Capacitor throws (try/catch)", () => {
+    Object.defineProperty(globalThis, "Capacitor", {
+      configurable: true,
+      get() {
+        throw new Error("boom");
+      },
+    });
+    expect(isCapacitorNativePlatform()).toBe(false);
+  });
+});

apps/client/src/pwa/is-capacitor.ts

@@ -0,0 +1,23 @@
+/**
+ * Detects whether the client is running inside a Capacitor native WebView
+ * (native iOS/Android shell from the feature/mobile-app-bootstrap branch).
+ *
+ * This is a pure runtime check against the global `Capacitor` object that the
+ * native bridge injects — no `@capacitor/*` dependency is added. On the plain
+ * browser / installed-PWA path `window.Capacitor` is undefined, so this returns
+ * false and the Workbox service worker registers normally.
+ *
+ * Inside the native WebView the SW must NOT register: it would layer a redundant
+ * (and conflicting) cache over Capacitor's own asset serving and interfere with
+ * the native auth/CORS flow.
+ */
+export function isCapacitorNativePlatform(): boolean {
+  try {
+    const cap = (globalThis as any)?.Capacitor;
+    return !!(cap && typeof cap.isNativePlatform === "function"
+      ? cap.isNativePlatform()
+      : cap?.isNativePlatform);
+  } catch {
+    return false;
+  }
+}

apps/client/src/pwa/pwa-update-prompt.tsx

@@ -0,0 +1,59 @@
+import { useEffect } from "react";
+import { Button } from "@mantine/core";
+import { notifications } from "@mantine/notifications";
+import { useTranslation } from "react-i18next";
+import { useRegisterSW } from "virtual:pwa-register/react";
+
+// Stable notification id so we can show/hide a single update prompt.
+const UPDATE_NOTIFICATION_ID = "pwa-update-available";
+
+/**
+ * Listens for a waiting service worker and surfaces a Mantine notification
+ * prompting the user to reload into the new version.
+ *
+ * Must be mounted inside the Mantine provider subtree (Notifications must be
+ * available). Renders nothing itself.
+ */
+export function PwaUpdatePrompt() {
+  const { t } = useTranslation();
+
+  const {
+    needRefresh: [needRefresh],
+    updateServiceWorker,
+  } = useRegisterSW({
+    onRegisterError(error) {
+      // Best-effort: a failed registration must not break the app.
+      console.error("Service worker registration error:", error);
+    },
+  });
+
+  useEffect(() => {
+    if (!needRefresh) return;
+
+    notifications.show({
+      id: UPDATE_NOTIFICATION_ID,
+      title: t("Update available"),
+      message: (
+        <Button
+          size="xs"
+          variant="light"
+          mt="xs"
+          onClick={() => updateServiceWorker(true)}
+        >
+          {t("Reload")}
+        </Button>
+      ),
+      autoClose: false,
+      withCloseButton: true,
+    });
+
+    // Hide the notification when the prompt is no longer needed / on cleanup.
+    return () => {
+      notifications.hide(UPDATE_NOTIFICATION_ID);
+    };
+  }, [needRefresh, t, updateServiceWorker]);
+
+  return null;
+}
+
+export default PwaUpdatePrompt;

apps/client/src/pwa/sw-strategy.test.ts

@@ -0,0 +1,32 @@
+import { describe, it, expect } from "vitest";
+import { isApiPath, isCollabOrSocketPath } from "./sw-strategy";
+
+describe("isApiPath", () => {
+  it("matches the /api segment and its subtree", () => {
+    expect(isApiPath("/api")).toBe(true);
+    expect(isApiPath("/api/")).toBe(true);
+    expect(isApiPath("/api/pages")).toBe(true);
+  });
+
+  it("does not over-match sibling paths", () => {
+    expect(isApiPath("/apidocs")).toBe(false);
+    expect(isApiPath("/apixyz")).toBe(false);
+    expect(isApiPath("/")).toBe(false);
+    expect(isApiPath("/pages")).toBe(false);
+  });
+});
+
+describe("isCollabOrSocketPath", () => {
+  it("matches the /collab and /socket.io segments and their subtrees", () => {
+    expect(isCollabOrSocketPath("/collab")).toBe(true);
+    expect(isCollabOrSocketPath("/collab/x")).toBe(true);
+    expect(isCollabOrSocketPath("/socket.io")).toBe(true);
+    expect(isCollabOrSocketPath("/socket.io/abc")).toBe(true);
+  });
+
+  it("does not over-match sibling paths", () => {
+    expect(isCollabOrSocketPath("/collaborators")).toBe(false);
+    expect(isCollabOrSocketPath("/collabx")).toBe(false);
+    expect(isCollabOrSocketPath("/socket.iox")).toBe(false);
+  });
+});

apps/client/src/pwa/sw-strategy.ts

@@ -0,0 +1,32 @@
+/**
+ * Canonical service-worker routing predicates.
+ *
+ * IMPORTANT: With vite-plugin-pwa using Workbox `generateSW`, the
+ * `runtimeCaching[].urlPattern` functions are serialized standalone into the
+ * generated service worker and CANNOT reference imported symbols. The matching
+ * logic is therefore duplicated as inline regex literals in
+ * apps/client/vite.config.ts. This module is the testable source of truth, and
+ * the two MUST be kept in sync. This duplication is intentional and is the
+ * documented Workbox limitation.
+ *
+ * Matching is anchored to a path SEGMENT boundary (`^/<seg>(/|$)`) so that
+ * sibling paths like `/apidocs`, `/collaborators`, `/socket.iox` are NOT
+ * wrongly treated as API/realtime traffic.
+ */
+
+/**
+ * True when `pathname` is the `/api` segment or anything beneath it.
+ * `/api` and `/api/...` -> true; `/apidocs`, `/apixyz` -> false.
+ */
+export function isApiPath(pathname: string): boolean {
+  return /^\/api(\/|$)/.test(pathname);
+}
+
+/**
+ * True when `pathname` is the `/collab` or `/socket.io` segment (or beneath it).
+ * `/collab`, `/collab/x`, `/socket.io`, `/socket.io/abc` -> true;
+ * `/collaborators`, `/collabx`, `/socket.iox` -> false.
+ */
+export function isCollabOrSocketPath(pathname: string): boolean {
+  return /^\/(collab|socket\.io)(\/|$)/.test(pathname);
+}

apps/client/src/vite-env.d.ts

@@ -1,2 +1,4 @@
 /// <reference types="vite/client" />
+/// <reference types="vite-plugin-pwa/react" />
+/// <reference types="vite-plugin-pwa/info" />
 declare const APP_VERSION: string

apps/client/vite.config.ts

@@ -1,5 +1,6 @@
 import { defineConfig, loadEnv } from "vite";
 import react from "@vitejs/plugin-react";
+import { VitePWA } from "vite-plugin-pwa";
 import * as path from "path";
 import { execSync } from "node:child_process";
 
@@ -53,7 +54,55 @@ export default defineConfig(({ mode }) => {
       },
       APP_VERSION: JSON.stringify(resolveAppVersion(envPath)),
     },
-    plugins: [react()],
+    plugins: [
+      react(),
+      VitePWA({
+        registerType: "prompt",
+        injectRegister: null,
+        strategies: "generateSW",
+        manifest: false,
+        workbox: {
+          globPatterns: ["**/*.{js,css,html,svg,png,ico,woff2,json}"],
+          navigateFallback: "index.html",
+          // Segment-anchored (`^/<seg>(/|$)`) so navigation requests to these
+          // segments are consistently excluded from the SPA fallback, mirroring
+          // the runtimeCaching urlPattern regexes below.
+          //
+          // `/share`, `/mcp`, `/l`, and `/robots.txt` mirror the server
+          // static-serve exclude list (apps/server/src/main.ts setGlobalPrefix
+          // `exclude`): robots.txt, the SEO/OG/analytics-injected public share
+          // HTML, the embedded MCP endpoint, and the `l/:alias` vanity short-link
+          // (a server 302 to a share page) are served by server controllers, so
+          // the SW must never shadow them with the precached index.html app shell.
+          // For `/l/:alias` the client router has NO matching route, so serving
+          // the app shell would dead-end on Error404 and break the public link;
+          // it must reach the server to perform the redirect.
+          navigateFallbackDenylist: [
+            /^\/api(\/|$)/,
+            /^\/collab(\/|$)/,
+            /^\/socket\.io(\/|$)/,
+            /^\/share(\/|$)/,
+            /^\/mcp(\/|$)/,
+            /^\/l(\/|$)/,
+            /^\/robots\.txt$/,
+          ],
+          cleanupOutdatedCaches: true,
+          clientsClaim: true,
+          // The urlPattern regexes below mirror apps/client/src/pwa/sw-strategy.ts
+          // and MUST be kept in sync with it. Workbox `generateSW` serializes these
+          // functions standalone into the generated service worker, so they cannot
+          // import the module — the matching logic is intentionally duplicated as
+          // self-contained inline regex literals anchored to a path segment boundary.
+          runtimeCaching: [
+            { urlPattern: ({ url }) => /^\/(collab|socket\.io)(\/|$)/.test(url.pathname), handler: "NetworkOnly" },
+            // All /api stays network-only; offline reads come from the persisted
+            // React Query cache (IndexedDB) + y-indexeddb, not the SW HTTP cache.
+            { urlPattern: ({ url }) => /^\/api(\/|$)/.test(url.pathname), handler: "NetworkOnly" },
+          ],
+        },
+        devOptions: { enabled: false },
+      }),
+    ],
     build: {
       rolldownOptions: {
         output: {

apps/server/package.json

@@ -64,6 +64,7 @@
     "@nestjs/platform-fastify": "^11.1.19",
     "@nestjs/platform-socket.io": "^11.1.19",
     "@nestjs/schedule": "^6.1.3",
+    "@nestjs/swagger": "^11.2.0",
     "@nestjs/terminus": "^11.1.1",
     "@nestjs/throttler": "^6.5.0",
     "@nestjs/websockets": "^11.1.19",

apps/server/src/collaboration/collaboration.gateway.ts

@@ -24,7 +24,10 @@ import { CollabWsAdapter } from './adapter/collab-ws.adapter';
 import {
   CollaborationHandler,
   CollabEventHandlers,
+  writeTitleFragment,
 } from './collaboration.handler';
+import { User } from '@docmost/db/types/entity.types';
+import * as Y from 'yjs';
 
 @Injectable()
 export class CollaborationGateway {
@@ -149,6 +152,70 @@ export class CollaborationGateway {
     return this.hocuspocus.openDirectConnection(documentName, context);
   }
 
+  /**
+   * Write a new page title INTO the page's Yjs 'title' fragment, Redis-INDEPENDENT.
+   *
+   * Unlike the Redis-routed `handleYjsEvent` path — which routes through
+   * `redisSync?.handleEvent` and SILENTLY no-ops when Redis is disabled
+   * (COLLAB_DISABLE_REDIS=true → redisSync === null) — this goes straight
+   * through the local Hocuspocus `openDirectConnection`. The title sync
+   * therefore works in BOTH single-process (no Redis) and Redis-clustered
+   * deployments.
+   *
+   * openDirectConnection loads the doc from persistence when no editor is
+   * connected, so this works whether or not an editor is currently open: the
+   * clear+reseed lands on the loaded doc and is persisted by onStoreDocument.
+   *
+   * Provenance: when the caller is the agent, the actor/aiChatId are threaded
+   * into the connection `context` so onStoreDocument sees `context.actor ===
+   * 'agent'` for the resulting title store (mirrors the body/REST path). The
+   * resulting title store is usually a no-op anyway — PageService already wrote
+   * the same title to the page.title column, so onStoreDocument's
+   * `titleText !== page.title` guard skips the column write — but we wire the
+   * context for correctness regardless.
+   */
+  async writePageTitle(
+    pageId: string,
+    title: string,
+    context?: { user?: User; actor?: string; aiChatId?: string },
+  ): Promise<void> {
+    const documentName = `page.${pageId}`;
+    const connection = await this.hocuspocus.openDirectConnection(
+      documentName,
+      context ?? {},
+    );
+    try {
+      // Write the new title into the in-memory 'title' fragment AND capture the
+      // resulting full doc state so we can persist it directly below.
+      let ydocState: Buffer | null = null;
+      await connection.transact((doc) => {
+        writeTitleFragment(doc, title);
+        ydocState = Buffer.from(Y.encodeStateAsUpdate(doc));
+      });
+
+      // F1 (variant C): persist the 'title' fragment to `page.ydoc` DIRECTLY,
+      // bypassing onStoreDocument. PageService.update already wrote the new title
+      // to the page.title COLUMN before calling this, so onStoreDocument's no-op
+      // fast-path (titleText === column) would NOT persist the in-memory fragment
+      // on disconnect — leaving the stored ydoc with the OLD title, which a later
+      // body edit would then revert the column back to. Writing the ydoc here
+      // makes BOTH column and persisted fragment consistent (NEW = NEW).
+      //
+      // Safe with or without a live editor: the write is idempotent and carries
+      // no tree snapshot (no double broadcast); when an editor is connected, the
+      // normal onStoreDocument flow still persists the (superset) state later and
+      // the live clients receive the title change through the transact above.
+      if (ydocState) {
+        await this.persistenceExtension.persistTitleFragmentYdoc(
+          pageId,
+          ydocState,
+        );
+      }
+    } finally {
+      await connection.disconnect();
+    }
+  }
+
   /*
    *Can be used before calling openDirectConnection directly
    */

apps/server/src/collaboration/collaboration.handler.spec.ts

@@ -0,0 +1,155 @@
+import * as Y from 'yjs';
+import { TiptapTransformer } from '@hocuspocus/transformer';
+import { writeTitleFragment } from './collaboration.handler';
+import { CollaborationGateway } from './collaboration.gateway';
+import {
+  buildTitleSeedYdoc,
+  jsonToText,
+  tiptapExtensions,
+} from './collaboration.util';
+
+// Read the plain text held in the doc's 'title' XmlFragment, the same way
+// PersistenceExtension.onStoreDocument extracts it before writing page.title.
+const readTitleText = (doc: Y.Doc): string => {
+  const titleJson = TiptapTransformer.fromYdoc(doc, 'title');
+  return titleJson ? jsonToText(titleJson).trim() : '';
+};
+
+describe('writeTitleFragment — the clear+seed title write (Bug 1)', () => {
+  it('replaces an OLD title fragment with EXACTLY the new title (no duplication)', () => {
+    // Seed the doc's 'title' fragment with an OLD title, like a real page.
+    const doc = new Y.Doc();
+    Y.applyUpdate(doc, Y.encodeStateAsUpdate(buildTitleSeedYdoc('Old Title')));
+    expect(readTitleText(doc)).toBe('Old Title');
+
+    writeTitleFragment(doc, 'New Title');
+
+    // The fragment must contain EXACTLY the new title — not "Old TitleNew Title"
+    // (append) or "New TitleNew Title" (duplication). A single heading node.
+    expect(readTitleText(doc)).toBe('New Title');
+
+    const titleJson = TiptapTransformer.fromYdoc(doc, 'title') as any;
+    expect(titleJson.content).toHaveLength(1);
+    expect(titleJson.content[0].type).toBe('heading');
+  });
+
+  it('seeds the title fragment when it started empty', () => {
+    const doc = new Y.Doc();
+    // Force the 'title' fragment to exist but be empty.
+    doc.getXmlFragment('title');
+    expect(readTitleText(doc)).toBe('');
+
+    writeTitleFragment(doc, 'First Title');
+
+    expect(readTitleText(doc)).toBe('First Title');
+  });
+
+  it('does not corrupt the body when rewriting the title', () => {
+    // A doc with both a body and an old title; the body must survive untouched.
+    const doc = new Y.Doc();
+    const bodyDoc = TiptapTransformer.toYdoc(
+      {
+        type: 'doc',
+        content: [
+          { type: 'paragraph', content: [{ type: 'text', text: 'body text' }] },
+        ],
+      },
+      'default',
+      tiptapExtensions,
+    );
+    Y.applyUpdate(doc, Y.encodeStateAsUpdate(bodyDoc));
+    Y.applyUpdate(doc, Y.encodeStateAsUpdate(buildTitleSeedYdoc('Old')));
+
+    writeTitleFragment(doc, 'New');
+
+    expect(readTitleText(doc)).toBe('New');
+    const bodyJson = TiptapTransformer.fromYdoc(doc, 'default');
+    expect(jsonToText(bodyJson)).toContain('body text');
+  });
+});
+
+describe('CollaborationGateway.writePageTitle — Redis-independent path', () => {
+  // Build a gateway with only its hocuspocus.openDirectConnection stubbed; the
+  // method must drive the clear+seed through that direct connection (NOT through
+  // redisSync), so the title write survives COLLAB_DISABLE_REDIS.
+  const makeGateway = (doc: Y.Doc) => {
+    const disconnect = jest.fn().mockResolvedValue(undefined);
+    const transact = jest.fn(async (fn: (d: Y.Doc) => void) => {
+      fn(doc);
+    });
+    const openDirectConnection = jest
+      .fn()
+      .mockResolvedValue({ transact, disconnect });
+
+    const gateway = Object.create(CollaborationGateway.prototype);
+    // redisSync is intentionally null — this is the no-Redis scenario.
+    gateway.redisSync = null;
+    gateway.hocuspocus = { openDirectConnection } as any;
+    // F1 (variant C): writePageTitle persists the 'title' fragment directly so a
+    // later body edit can't revert the rename (see title-rename-durability.spec).
+    const persistTitleFragmentYdoc = jest.fn().mockResolvedValue(undefined);
+    gateway.persistenceExtension = { persistTitleFragmentYdoc } as any;
+
+    return {
+      gateway,
+      openDirectConnection,
+      transact,
+      disconnect,
+      persistTitleFragmentYdoc,
+    };
+  };
+
+  it('writes the new title via openDirectConnection and disconnects', async () => {
+    const doc = new Y.Doc();
+    Y.applyUpdate(doc, Y.encodeStateAsUpdate(buildTitleSeedYdoc('Old Title')));
+
+    const { gateway, openDirectConnection, disconnect, persistTitleFragmentYdoc } =
+      makeGateway(doc);
+
+    await gateway.writePageTitle('page-1', 'New Title', { user: { id: 'u1' } });
+
+    expect(openDirectConnection).toHaveBeenCalledWith(
+      'page.page-1',
+      expect.objectContaining({ user: { id: 'u1' } }),
+    );
+    expect(readTitleText(doc)).toBe('New Title');
+    // The renamed fragment is persisted directly to page.ydoc (F1 variant C).
+    expect(persistTitleFragmentYdoc).toHaveBeenCalledWith(
+      'page-1',
+      expect.any(Buffer),
+    );
+    expect(disconnect).toHaveBeenCalledTimes(1);
+  });
+
+  it('threads agent provenance into the connection context', async () => {
+    const doc = new Y.Doc();
+    const { gateway, openDirectConnection } = makeGateway(doc);
+
+    await gateway.writePageTitle('page-1', 'Agent Title', {
+      user: { id: 'u1' },
+      actor: 'agent',
+      aiChatId: 'chat-1',
+    });
+
+    expect(openDirectConnection).toHaveBeenCalledWith(
+      'page.page-1',
+      expect.objectContaining({ actor: 'agent', aiChatId: 'chat-1' }),
+    );
+  });
+
+  it('disconnects even when the transaction throws', async () => {
+    const disconnect = jest.fn().mockResolvedValue(undefined);
+    const openDirectConnection = jest.fn().mockResolvedValue({
+      transact: jest.fn().mockRejectedValue(new Error('boom')),
+      disconnect,
+    });
+    const gateway = Object.create(CollaborationGateway.prototype);
+    gateway.redisSync = null;
+    gateway.hocuspocus = { openDirectConnection } as any;
+
+    await expect(
+      gateway.writePageTitle('page-1', 'X', {}),
+    ).rejects.toThrow('boom');
+    expect(disconnect).toHaveBeenCalledTimes(1);
+  });
+});

apps/server/src/collaboration/collaboration.handler.ts

@@ -2,6 +2,7 @@ import { Injectable, Logger } from '@nestjs/common';
 import { Hocuspocus, Document } from '@hocuspocus/server';
 import { TiptapTransformer } from '@hocuspocus/transformer';
 import {
+  buildTitleSeedYdoc,
   prosemirrorNodeToYElement,
   tiptapExtensions,
 } from './collaboration.util';
@@ -13,6 +14,35 @@ export type CollabEventHandlers = ReturnType<
   CollaborationHandler['getHandlers']
 >;
 
+/**
+ * Clear+reseed the 'title' XmlFragment of `doc` so it holds EXACTLY `title`.
+ *
+ * Used by the gateway's direct `writePageTitle` method to write a new page
+ * title INTO the page's Yjs 'title' fragment. The title lives in the same
+ * Y.Doc as the body; onStoreDocument extracts it on every save, so a REST/MCP
+ * rename that only updated the page.title DB column would be reverted on the
+ * next collaborative save unless the Yjs 'title' fragment is kept in sync.
+ * The whole fragment is replaced (no merge/append),
+ * mirroring the 'replace' body path: the new title fully supersedes the old.
+ *
+ * DELIBERATE TRADE-OFF: because this does a FULL clear+replace of the 'title'
+ * fragment, a REST/MCP rename arriving while a user is actively editing the
+ * title in an open editor WILL overwrite that in-progress edit. This is
+ * acceptable — the title is a short, rarely-concurrently-edited field — and is
+ * preferable to leaving a stale Yjs title that onStoreDocument would revert the
+ * DB column to on the next save.
+ */
+export function writeTitleFragment(doc: Y.Doc, title: string): void {
+  const titleFragment = doc.getXmlFragment('title');
+
+  if (titleFragment.length > 0) {
+    titleFragment.delete(0, titleFragment.length);
+  }
+
+  const newTitleDoc = buildTitleSeedYdoc(title);
+  Y.applyUpdate(doc, Y.encodeStateAsUpdate(newTitleDoc));
+}
+
 @Injectable()
 export class CollaborationHandler {
   private readonly logger = new Logger(CollaborationHandler.name);

apps/server/src/collaboration/collaboration.util.spec.ts

@@ -1,9 +1,12 @@
 import * as Y from 'yjs';
+import { TiptapTransformer } from '@hocuspocus/transformer';
 import {
   getPageId,
   isEmptyParagraphDoc,
   jsonToNode,
   prosemirrorNodeToYElement,
+  buildTitleSeedYdoc,
+  jsonToText,
 } from './collaboration.util';
 import { Node } from '@tiptap/pm/model';
 
@@ -241,3 +244,43 @@ describe('prosemirrorNodeToYElement', () => {
     expect(element.get(1).get(0).toString()).toBe('two');
   });
 });
+
+describe('buildTitleSeedYdoc', () => {
+  it('builds a level-1 heading carrying the title text', () => {
+    const doc = buildTitleSeedYdoc('Hello World');
+    const json: any = TiptapTransformer.fromYdoc(doc, 'title');
+
+    const first = json.content?.[0];
+    expect(first.type).toBe('heading');
+    expect(first.attrs.level).toBe(1);
+    expect(jsonToText(json).trim()).toBe('Hello World');
+  });
+
+  it('produces a non-empty title fragment for a non-empty title', () => {
+    const doc = buildTitleSeedYdoc('Some Title');
+    expect(doc.get('title', Y.XmlFragment).length).toBeGreaterThan(0);
+  });
+
+  it('produces a heading with no text child for an empty title', () => {
+    const doc = buildTitleSeedYdoc('');
+    const json: any = TiptapTransformer.fromYdoc(doc, 'title');
+
+    const first = json.content?.[0];
+    expect(first.type).toBe('heading');
+    // No text content for an empty title.
+    expect(first.content ?? []).toHaveLength(0);
+    expect(jsonToText(json).trim()).toBe('');
+  });
+
+  it('round-trips a title through build -> extract -> build -> extract', () => {
+    const title = 'Round Trip Title';
+    const doc1 = buildTitleSeedYdoc(title);
+    const text1 = jsonToText(TiptapTransformer.fromYdoc(doc1, 'title')).trim();
+
+    const doc2 = buildTitleSeedYdoc(text1);
+    const text2 = jsonToText(TiptapTransformer.fromYdoc(doc2, 'title')).trim();
+
+    expect(text1).toBe(title);
+    expect(text2).toBe(text1);
+  });
+});

apps/server/src/collaboration/collaboration.util.ts

@@ -59,6 +59,7 @@ import { generateHTML, generateJSON } from '../common/helpers/prosemirror/html';
 import { Node, Schema } from '@tiptap/pm/model';
 import * as Y from 'yjs';
 import { Logger } from '@nestjs/common';
+import { TiptapTransformer } from '@hocuspocus/transformer';
 
 export const tiptapExtensions = [
   StarterKit.configure({
@@ -143,6 +144,34 @@ export function jsonToText(tiptapJson: JSONContent) {
   return generateText(tiptapJson, tiptapExtensions);
 }
 
+/**
+ * Build a standalone Y.Doc that holds ONLY the page title, in a dedicated Yjs
+ * fragment named exactly 'title' (the collaborative title-editor contract with
+ * the client). The ProseMirror shape is a doc with a single level-1 heading
+ * whose text is the title (empty title => heading with no text child).
+ *
+ * The encoded state of the returned doc can be merged into a body doc via
+ * `Y.applyUpdate(doc, Y.encodeStateAsUpdate(titleSeed))` to seed the title
+ * fragment for legacy pages. Seeding MUST be guarded by an emptiness check on
+ * the existing 'title' fragment to avoid the Yjs duplication trap.
+ */
+export function buildTitleSeedYdoc(title: string): Y.Doc {
+  return TiptapTransformer.toYdoc(
+    {
+      type: 'doc',
+      content: [
+        {
+          type: 'heading',
+          attrs: { level: 1 },
+          content: title ? [{ type: 'text', text: title }] : [],
+        },
+      ],
+    },
+    'title',
+    tiptapExtensions,
+  );
+}
+
 export function jsonToNode(tiptapJson: JSONContent) {
   const schema = getSchema(tiptapExtensions);
   try {

apps/server/src/collaboration/constants.ts

@@ -1,3 +1,9 @@
 export const HISTORY_INTERVAL = 5 * 60 * 1000;
 export const HISTORY_FAST_INTERVAL = 60 * 1000;
 export const HISTORY_FAST_THRESHOLD = 5 * 60 * 1000;
+
+// Redis pub/sub channel that bridges a PAGE_UPDATED tree snapshot (a title/icon
+// rename) from the standalone collab process to the API process, which is the
+// single broadcast authority. Imported by both halves of the bridge:
+// PageTreeBridgePublisher (collab process) and PageTreeBridgeSubscriber (API process).
+export const COLLAB_TREE_UPDATE_CHANNEL = 'collab:tree-update';

apps/server/src/collaboration/extensions/persistence.extension.spec.ts

@@ -0,0 +1,483 @@
+import * as Y from 'yjs';
+import { TiptapTransformer } from '@hocuspocus/transformer';
+import { PersistenceExtension } from './persistence.extension';
+import { buildTitleSeedYdoc, tiptapExtensions } from '../collaboration.util';
+
+// Direct instantiation with stub deps, mirroring the auth/env unit specs.
+const bodyJson = {
+  type: 'doc',
+  content: [{ type: 'paragraph', content: [{ type: 'text', text: 'hello' }] }],
+};
+
+// Build a body Y.Doc with a known JSON, plus a monkey-patched broadcastStateless
+// (the real Hocuspocus Document supplies it; a bare Y.Doc does not).
+const buildDoc = () => {
+  const d: any = TiptapTransformer.toYdoc(bodyJson, 'default', tiptapExtensions);
+  d.broadcastStateless = jest.fn();
+  return d;
+};
+
+const cloneOut = (doc: any) =>
+  JSON.parse(JSON.stringify(TiptapTransformer.fromYdoc(doc, 'default')));
+
+const addTitleFragment = (doc: any, title: string) =>
+  Y.applyUpdate(doc, Y.encodeStateAsUpdate(buildTitleSeedYdoc(title)));
+
+describe('PersistenceExtension', () => {
+  let pageRepo: any;
+  let pageHistoryRepo: any;
+  let trx: any;
+  let db: any;
+  let aiQueue: any;
+  let historyQueue: any;
+  let notificationQueue: any;
+  let collabHistory: any;
+  let transclusionService: any;
+  let ext: PersistenceExtension;
+
+  beforeEach(() => {
+    pageRepo = {
+      findById: jest.fn(),
+      updatePage: jest.fn().mockResolvedValue(undefined),
+    };
+    pageHistoryRepo = {
+      findPageLastHistory: jest.fn(),
+      saveHistory: jest.fn(),
+    };
+    trx = {};
+    db = { transaction: () => ({ execute: (fn: any) => fn(trx) }) };
+    aiQueue = { add: jest.fn().mockResolvedValue(undefined) };
+    historyQueue = { add: jest.fn().mockResolvedValue(undefined) };
+    notificationQueue = { add: jest.fn().mockResolvedValue(undefined) };
+    collabHistory = { addContributors: jest.fn().mockResolvedValue(undefined) };
+    transclusionService = {
+      syncPageTransclusions: jest.fn().mockResolvedValue(undefined),
+      syncPageReferences: jest.fn().mockResolvedValue(undefined),
+      syncPageTemplateReferences: jest.fn().mockResolvedValue(undefined),
+    };
+
+    ext = new PersistenceExtension(
+      pageRepo as any,
+      pageHistoryRepo as any,
+      db as any,
+      aiQueue as any,
+      historyQueue as any,
+      notificationQueue as any,
+      collabHistory as any,
+      transclusionService as any,
+    );
+  });
+
+  describe('seedTitleFragment', () => {
+    it('returns false for empty/whitespace/null titles', () => {
+      const doc = new Y.Doc();
+      expect((ext as any).seedTitleFragment(doc, '')).toBe(false);
+      expect((ext as any).seedTitleFragment(doc, '   ')).toBe(false);
+      expect((ext as any).seedTitleFragment(doc, null)).toBe(false);
+    });
+
+    it('does NOT re-seed an existing non-empty title fragment', () => {
+      const doc = new Y.Doc();
+      addTitleFragment(doc, 'Existing');
+
+      expect((ext as any).seedTitleFragment(doc, 'Other')).toBe(false);
+
+      const text = TiptapTransformer.fromYdoc(doc, 'title');
+      expect(JSON.stringify(text)).toContain('Existing');
+      expect(JSON.stringify(text)).not.toContain('Other');
+    });
+
+    it('seeds an empty fragment from a non-empty title and returns true', () => {
+      const doc = new Y.Doc();
+      expect((ext as any).seedTitleFragment(doc, 'Hello')).toBe(true);
+
+      const json: any = TiptapTransformer.fromYdoc(doc, 'title');
+      expect(JSON.stringify(json)).toContain('Hello');
+    });
+
+    it('returns false (defensive) when reading the fragment throws', () => {
+      const fakeDoc = {
+        get: () => {
+          throw new Error('boom');
+        },
+      };
+      expect((ext as any).seedTitleFragment(fakeDoc as any, 'X')).toBe(false);
+    });
+  });
+
+  describe('onStoreDocument', () => {
+    const basePage = (overrides: any) => ({
+      id: 'PAGE_ID',
+      slugId: 'slug',
+      spaceId: 'space',
+      parentPageId: null,
+      creatorId: 'creator',
+      contributorIds: ['creator'],
+      workspaceId: 'ws',
+      title: 'whatever',
+      content: null,
+      lastUpdatedSource: 'user',
+      createdAt: new Date().toISOString(),
+      ...overrides,
+    });
+
+    const context = { user: { id: 'u1', name: 'U', avatarUrl: null } };
+
+    it('no-op when neither body nor title changed', async () => {
+      const document = buildDoc();
+      const page = basePage({
+        content: cloneOut(document),
+        title: 'hello title',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      expect(pageRepo.updatePage).not.toHaveBeenCalled();
+      expect(document.broadcastStateless).not.toHaveBeenCalled();
+      expect(collabHistory.addContributors).not.toHaveBeenCalled();
+      expect(transclusionService.syncPageTransclusions).not.toHaveBeenCalled();
+      expect(aiQueue.add).not.toHaveBeenCalled();
+      expect(historyQueue.add).not.toHaveBeenCalled();
+    });
+
+    it('title-only change persists the title without body side-effects', async () => {
+      const document = buildDoc();
+      addTitleFragment(document, 'New Title');
+      const page = basePage({
+        content: cloneOut(document),
+        title: 'Old Title',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      expect(call[0].title).toBe('New Title');
+      expect(call[0].ydoc).toBeDefined();
+      expect(call[0].contributorIds).toBeDefined();
+      expect('content' in call[0]).toBe(false);
+      // Title-only must not touch the body-authorship provenance.
+      expect('lastUpdatedSource' in call[0]).toBe(false);
+      expect(call[1]).toBe('PAGE_ID');
+      expect(call[3].treeUpdate.title).toBe('New Title');
+
+      expect(collabHistory.addContributors).toHaveBeenCalledTimes(1);
+      expect(collabHistory.addContributors).toHaveBeenCalledWith(
+        'PAGE_ID',
+        expect.any(Array),
+      );
+      expect(document.broadcastStateless).toHaveBeenCalledTimes(1);
+      expect(transclusionService.syncPageTransclusions).not.toHaveBeenCalled();
+      expect(aiQueue.add).not.toHaveBeenCalled();
+      expect(historyQueue.add).not.toHaveBeenCalled();
+    });
+
+    it('an EMPTY title fragment does NOT overwrite a non-empty page.title (anti-corruption guard, Bug 2)', async () => {
+      // The client can momentarily seed the 'title' fragment as an EMPTY heading
+      // (hasTitleFragment true, extracted text '') before the real title syncs.
+      // Body is unchanged here, so the only candidate write is the title -> the
+      // guard must turn this into a full no-op (no updatePage, no broadcast).
+      const document = buildDoc();
+      addTitleFragment(document, ''); // empty heading: length > 0 but text ''
+      const page = basePage({
+        content: cloneOut(document),
+        title: 'Real Title',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      // No write at all: the empty title is not authoritative and the body is
+      // unchanged, so onStoreDocument must take the no-op fast path.
+      expect(pageRepo.updatePage).not.toHaveBeenCalled();
+      expect(document.broadcastStateless).not.toHaveBeenCalled();
+    });
+
+    it('an EMPTY title fragment alongside a body change persists the body but NOT an empty title (anti-corruption guard, Bug 2)', async () => {
+      const document = buildDoc();
+      addTitleFragment(document, ''); // empty title fragment
+      const page = basePage({
+        content: { type: 'doc', content: [] }, // different body -> bodyChanged
+        title: 'Real Title',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      // Body is persisted, but the title is NOT included (empty == not
+      // authoritative) and no tree update is broadcast for the title.
+      expect(call[0].content).toBeTruthy();
+      expect('title' in call[0]).toBe(false);
+      expect(call[3]).toBeUndefined();
+    });
+
+    it('body + title change persists both with full body side-effects', async () => {
+      const document = buildDoc();
+      addTitleFragment(document, 'New Title');
+      const page = basePage({
+        content: { type: 'doc', content: [] },
+        title: 'Old Title',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      expect(call[0].content).toBeTruthy();
+      expect(call[0].title).toBe('New Title');
+      expect(call[0].ydoc).toBeDefined();
+      expect(call[0].lastUpdatedSource).toBe('user');
+      expect(call[3].treeUpdate).toBeDefined();
+
+      expect(transclusionService.syncPageTransclusions).toHaveBeenCalled();
+      expect(aiQueue.add).toHaveBeenCalled();
+      expect(historyQueue.add).toHaveBeenCalled();
+      expect(collabHistory.addContributors).toHaveBeenCalled();
+      expect(document.broadcastStateless).toHaveBeenCalled();
+    });
+
+    it('body-only change persists the body without a tree update', async () => {
+      const document = buildDoc();
+      const page = basePage({
+        content: { type: 'doc', content: [] },
+        title: 'whatever',
+      });
+      pageRepo.findById.mockResolvedValue(page);
+
+      await ext.onStoreDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+        context,
+      } as any);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      expect(call[0].content).toBeTruthy();
+      expect('title' in call[0]).toBe(false);
+      // No treeUpdate for a body-only save.
+      expect(call[3]).toBeUndefined();
+
+      expect(transclusionService.syncPageTransclusions).toHaveBeenCalled();
+      expect(aiQueue.add).toHaveBeenCalled();
+      expect(historyQueue.add).toHaveBeenCalled();
+      expect(document.broadcastStateless).toHaveBeenCalled();
+    });
+  });
+
+  describe('onLoadDocument', () => {
+    it('returns early (no DB read) when the document is not empty', async () => {
+      const document = { isEmpty: () => false };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      expect(result).toBeUndefined();
+      expect(pageRepo.findById).not.toHaveBeenCalled();
+    });
+
+    it('returns undefined and does not persist when the page is null', async () => {
+      const document = { isEmpty: () => true };
+      pageRepo.findById.mockResolvedValue(null);
+
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      expect(result).toBeUndefined();
+      expect(pageRepo.updatePage).not.toHaveBeenCalled();
+    });
+
+    it('seeds + persists under a lock when the persisted ydoc lacks a title fragment', async () => {
+      const src = TiptapTransformer.toYdoc(bodyJson, 'default', tiptapExtensions);
+      const page = {
+        id: 'PAGE_ID',
+        title: 'Legacy Title',
+        ydoc: Buffer.from(Y.encodeStateAsUpdate(src)),
+        content: null,
+      };
+      // Both the cheap pre-check and the locked re-read return the same row.
+      pageRepo.findById.mockResolvedValue(page);
+
+      const document = { isEmpty: () => true };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      // The locked re-read must take the row lock inside the tx.
+      const lockedReadCall = pageRepo.findById.mock.calls.find(
+        (c: any[]) => c[1]?.withLock,
+      );
+      expect(lockedReadCall).toBeDefined();
+      expect(lockedReadCall[1].trx).toBe(trx);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      expect(Buffer.isBuffer(call[0].ydoc)).toBe(true);
+      expect(call[1]).toBe('PAGE_ID');
+      // Persist must run inside the transaction.
+      expect(call[2]).toBe(trx);
+      expect(result).toBeTruthy();
+    });
+
+    it('does NOT lock or persist when the ydoc already has a title fragment', async () => {
+      const src = TiptapTransformer.toYdoc(bodyJson, 'default', tiptapExtensions);
+      Y.applyUpdate(src, Y.encodeStateAsUpdate(buildTitleSeedYdoc('Has Title')));
+      const page = {
+        id: 'PAGE_ID',
+        title: 'Has Title',
+        ydoc: Buffer.from(Y.encodeStateAsUpdate(src)),
+        content: null,
+      };
+      pageRepo.findById.mockResolvedValue(page);
+
+      const document = { isEmpty: () => true };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      // Hot path: only the cheap lock-free read, no locked re-read, no write.
+      expect(pageRepo.findById).toHaveBeenCalledTimes(1);
+      expect(pageRepo.findById.mock.calls[0][1]?.withLock).toBeFalsy();
+      expect(pageRepo.updatePage).not.toHaveBeenCalled();
+      expect(result).toBeTruthy();
+    });
+
+    it('converts legacy content -> ydoc inside a tx and persists a {ydoc} Buffer', async () => {
+      const page = {
+        id: 'PAGE_ID',
+        title: 'T',
+        ydoc: null,
+        content: bodyJson,
+      };
+      pageRepo.findById.mockResolvedValue(page);
+
+      const document = { isEmpty: () => true };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      const lockedReadCall = pageRepo.findById.mock.calls.find(
+        (c: any[]) => c[1]?.withLock,
+      );
+      expect(lockedReadCall).toBeDefined();
+      expect(lockedReadCall[1].trx).toBe(trx);
+
+      expect(pageRepo.updatePage).toHaveBeenCalledTimes(1);
+      const call = pageRepo.updatePage.mock.calls[0];
+      expect(Buffer.isBuffer(call[0].ydoc)).toBe(true);
+      expect(call[2]).toBe(trx);
+      // The rebuilt doc carries the body.
+      expect(JSON.stringify(cloneOut(result))).toContain('hello');
+    });
+
+    it('SKIPS rebuild when the locked re-read shows the ydoc was already healed', async () => {
+      // Simulate a concurrent process: the cheap pre-check sees ydoc=null (legacy
+      // rebuild path), but by the time we hold the lock another process has
+      // already persisted a healthy ydoc. We must adopt it, not rebuild/clobber.
+      const healed = TiptapTransformer.toYdoc(
+        { type: 'doc', content: [{ type: 'paragraph', content: [{ type: 'text', text: 'healed' }] }] },
+        'default',
+        tiptapExtensions,
+      );
+      Y.applyUpdate(healed, Y.encodeStateAsUpdate(buildTitleSeedYdoc('Healed Title')));
+      const healedYdoc = Buffer.from(Y.encodeStateAsUpdate(healed));
+
+      const preCheck = { id: 'PAGE_ID', title: 'T', ydoc: null, content: bodyJson };
+      const lockedRow = {
+        id: 'PAGE_ID',
+        title: 'Healed Title',
+        ydoc: healedYdoc,
+        content: bodyJson,
+      };
+      pageRepo.findById
+        .mockResolvedValueOnce(preCheck) // cheap pre-check
+        .mockResolvedValueOnce(lockedRow); // locked re-read
+
+      const document = { isEmpty: () => true };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      // The healthy ydoc had a title fragment already, so nothing was rebuilt or
+      // seeded -> no clobbering write.
+      expect(pageRepo.updatePage).not.toHaveBeenCalled();
+      // The returned doc is the healed body, NOT a fresh rebuild of bodyJson.
+      expect(JSON.stringify(cloneOut(result))).toContain('healed');
+    });
+
+    it('REJECTS the load when the rebuild persist fails (does not return an unpersisted doc)', async () => {
+      const page = { id: 'PAGE_ID', title: 'T', ydoc: null, content: bodyJson };
+      pageRepo.findById.mockResolvedValue(page);
+      pageRepo.updatePage.mockRejectedValue(new Error('db down'));
+      const errSpy = jest
+        .spyOn((ext as any).logger, 'error')
+        .mockImplementation(() => undefined);
+
+      const document = { isEmpty: () => true };
+      await expect(
+        ext.onLoadDocument({
+          documentName: 'page.PAGE_ID',
+          document,
+        } as any),
+      ).rejects.toThrow('db down');
+      expect(errSpy).toHaveBeenCalled();
+    });
+
+    it('seed-only persist FAILURE returns the doc from the existing ydoc (no throw)', async () => {
+      const src = TiptapTransformer.toYdoc(bodyJson, 'default', tiptapExtensions);
+      const page = {
+        id: 'PAGE_ID',
+        title: 'Legacy Title',
+        ydoc: Buffer.from(Y.encodeStateAsUpdate(src)),
+        content: null,
+      };
+      pageRepo.findById.mockResolvedValue(page);
+      pageRepo.updatePage.mockRejectedValue(new Error('db down'));
+      const errSpy = jest
+        .spyOn((ext as any).logger, 'error')
+        .mockImplementation(() => undefined);
+
+      const document = { isEmpty: () => true };
+      const result = await ext.onLoadDocument({
+        documentName: 'page.PAGE_ID',
+        document,
+      } as any);
+
+      // Non-fatal: we fall back to the doc loaded from the existing page.ydoc.
+      expect(result).toBeTruthy();
+      expect(JSON.stringify(cloneOut(result))).toContain('hello');
+      expect(errSpy).toHaveBeenCalled();
+    });
+  });
+});

apps/server/src/collaboration/extensions/persistence.extension.ts

@@ -9,6 +9,7 @@ import * as Y from 'yjs';
 import { Injectable, Logger } from '@nestjs/common';
 import { TiptapTransformer } from '@hocuspocus/transformer';
 import {
+  buildTitleSeedYdoc,
   getPageId,
   isEmptyParagraphDoc,
   jsonToText,
@@ -116,6 +117,10 @@ export class PersistenceExtension implements Extension {
       return;
     }
 
+    // Cheap, lock-free pre-check (hot path stays lock-free). It tells us whether
+    // any heal (legacy rebuild and/or title seed) is needed; the heal itself
+    // re-reads the row FOR UPDATE and re-validates inside a transaction so it
+    // runs exactly once (see healUnderLock).
     const page = await this.pageRepo.findById(pageId, {
       includeContent: true,
       includeYdoc: true,
@@ -127,33 +132,193 @@ export class PersistenceExtension implements Extension {
     }
 
     if (page.ydoc) {
-      this.logger.debug(`ydoc loaded from db: ${pageId}`);
-
       const doc = new Y.Doc();
-      const dbState = new Uint8Array(page.ydoc);
+      Y.applyUpdate(doc, new Uint8Array(page.ydoc));
 
-      Y.applyUpdate(doc, dbState);
-      return doc;
+      // Legacy pages persisted their title only in the `page.title` column; the
+      // ydoc has no 'title' fragment. Decide cheaply (no lock) whether a seed is
+      // needed by inspecting the loaded doc's 'title' fragment. A seed is needed
+      // only when that fragment is empty AND there is a non-empty column title.
+      let titleSeedNeeded = false;
+      try {
+        const titleFrag = doc.get('title', Y.XmlFragment);
+        titleSeedNeeded = titleFrag.length === 0 && !!page.title?.trim();
+      } catch (err) {
+        // A malformed title fragment must not break loading; skip the seed.
+        this.logger.warn(`failed to inspect title fragment: ${err?.['message']}`);
+        titleSeedNeeded = false;
+      }
+
+      if (!titleSeedNeeded) {
+        // Fully healthy: a ydoc with a title fragment (or nothing to seed).
+        this.logger.debug(`ydoc loaded from db: ${pageId}`);
+        return doc;
+      }
+
+      // SEED-ONLY heal: a valid page.ydoc already exists; we only need to add the
+      // title fragment. If the persist fails we must NOT hand out an unpersisted
+      // fresh-client-id seed (it could later duplicate the title), so we fall
+      // back to the healthy doc loaded from the EXISTING page.ydoc, without the
+      // seed. The title just won't render until a later successful heal —
+      // non-fatal, non-corrupting.
+      try {
+        return await this.healUnderLock(pageId);
+      } catch (err) {
+        this.logger.error(
+          `Failed to persist seeded ydoc for page ${pageId}; serving existing ydoc without title seed`,
+          err,
+        );
+        return doc;
+      }
     }
 
-    // if no ydoc state in db convert json in page.content to Ydoc.
+    // NOTE (offline-sync M1, Goal 2): this per-load self-heal converts +
+    // title-seeds + persists every legacy page (content set, ydoc null) on its
+    // first open, which neutralizes the duplication trap incrementally. A
+    // proactive one-shot BATCH migration over all such pages could be added
+    // later, but it requires the tiptap schema + TiptapTransformer (Node/Yjs),
+    // which a Kysely SQL migration cannot run; no runnable-task/CLI convention
+    // exists in this repo yet, so we deliberately avoid a fragile migration.
+    //
+    // If no ydoc state in db, REBUILD a Y.Doc from the JSON in page.content under
+    // a row lock (see healUnderLock).
     if (page.content) {
-      this.logger.debug(`converting json to ydoc: ${pageId}`);
-
-      const ydoc = TiptapTransformer.toYdoc(
-        page.content,
-        'default',
-        tiptapExtensions,
-      );
-
-      Y.encodeStateAsUpdate(ydoc);
-      return ydoc;
+      // REBUILD heal: surface failures. If the persist fails we REFUSE the load
+      // (re-throw) rather than hand out an unpersisted fresh-client-id rebuild —
+      // returning it would re-arm the duplication trap. A transient DB failure
+      // means the client reconnects and retries: correctness over availability.
+      try {
+        return await this.healUnderLock(pageId);
+      } catch (err) {
+        this.logger.error(
+          `Failed to persist rebuilt ydoc for page ${pageId}; refusing load`,
+          err,
+        );
+        throw err;
+      }
     }
 
     this.logger.debug(`creating fresh ydoc: ${pageId}`);
     return new Y.Doc();
   }
 
+  /**
+   * Serialize the legacy self-heal (rebuild from page.content and/or seed the
+   * title fragment, then persist) so it runs exactly ONCE per page, closing the
+   * Yjs duplication trap. Both TiptapTransformer.toYdoc and buildTitleSeedYdoc
+   * mint FRESH Yjs client-ids every call, so two concurrent rebuilds (the API
+   * process via openDirectConnection AND the standalone collab process both
+   * seeing `ydoc IS NULL`) could each persist a different-client-id state and let
+   * a long-offline client merge-and-duplicate. We prevent that by re-reading the
+   * row FOR UPDATE inside a transaction and re-validating state under the lock:
+   * whoever wins the lock heals; the loser observes the healthy `ydoc` and adopts
+   * it instead of rebuilding. The persist happens IN THE SAME TX, so a failed
+   * write rolls back and propagates out (the caller then decides refuse vs.
+   * fall-back).
+   */
+  private async healUnderLock(pageId: string): Promise<Y.Doc> {
+    return executeTx(this.db, async (trx) => {
+      const locked = await this.pageRepo.findById(pageId, {
+        withLock: true,
+        includeContent: true,
+        includeYdoc: true,
+        trx,
+      });
+
+      const doc = new Y.Doc();
+      let rebuilt = false;
+
+      if (locked?.ydoc) {
+        // Another process already healed (or the page always had a ydoc): adopt
+        // the healthy persisted state, do NOT rebuild.
+        Y.applyUpdate(doc, new Uint8Array(locked.ydoc));
+      } else if (locked?.content) {
+        this.logger.debug(`converting json to ydoc: ${pageId}`);
+        const built = TiptapTransformer.toYdoc(
+          locked.content,
+          'default',
+          tiptapExtensions,
+        );
+        Y.applyUpdate(doc, Y.encodeStateAsUpdate(built));
+        rebuilt = true;
+      }
+      // else: no ydoc and no content -> a fresh empty doc.
+
+      // Idempotent, emptiness-guarded title seed (safe to call always).
+      const seeded = this.seedTitleFragment(doc, locked?.title ?? null);
+
+      if (rebuilt || seeded) {
+        // Persist IN THE SAME TX. If this throws, the tx rolls back and the
+        // error propagates out of executeTx to the caller.
+        await this.pageRepo.updatePage(
+          { ydoc: Buffer.from(Y.encodeStateAsUpdate(doc)) },
+          pageId,
+          trx,
+        );
+        this.logger.debug(`persisted rebuilt/seeded ydoc: ${pageId}`);
+      }
+
+      return doc;
+    });
+  }
+
+  /**
+   * Seed the 'title' fragment of `doc` from the `page.title` column for legacy
+   * pages whose persisted ydoc has no title fragment yet.
+   *
+   * Guarded STRICTLY by emptiness: we only seed when the existing 'title'
+   * fragment is empty AND there is a non-empty column title. Seeding a non-empty
+   * fragment would re-introduce the Yjs duplication trap, so we never do it.
+   * Returns true when a seed was applied (so the caller can persist).
+   * Defensive: a malformed title must not break document loading.
+   */
+  private seedTitleFragment(doc: Y.Doc, title: string | null): boolean {
+    const trimmed = (title ?? '').trim();
+    if (!trimmed) return false;
+
+    try {
+      const titleFrag = doc.get('title', Y.XmlFragment);
+      if (titleFrag.length !== 0) return false;
+
+      const titleSeed = buildTitleSeedYdoc(title);
+      Y.applyUpdate(doc, Y.encodeStateAsUpdate(titleSeed));
+      this.logger.debug('seeded title fragment from page.title column');
+      return true;
+    } catch (err) {
+      this.logger.warn(`failed to seed title fragment: ${err?.['message']}`);
+      return false;
+    }
+  }
+
+  /**
+   * Persist an already-encoded Y.Doc state directly to `page.ydoc`, mirroring the
+   * `pageRepo.updatePage({ ydoc })` write that onStoreDocument uses.
+   *
+   * Used by the gateway's writePageTitle (F1, variant C). A REST/MCP/agent rename
+   * with no live editor writes the new title into the in-memory 'title' fragment,
+   * but onStoreDocument's no-op fast-path (page.title column already equals the
+   * new title) does NOT persist that in-memory fragment, so the stored `page.ydoc`
+   * keeps the OLD title — and a later body edit then reverts the rename (loads the
+   * OLD fragment, sees it differs from the column, overwrites the column back to
+   * OLD). Writing the ydoc here keeps the persisted fragment consistent with the
+   * column so the rename survives.
+   *
+   * Broadcast-safe / no double broadcast: this carries no `treeUpdate`, so the
+   * tree WS + redis listeners (which gate on `treeUpdate`) do NOT re-broadcast the
+   * rename — only PageService.update's own PAGE_UPDATED does. The only extra
+   * side-effect is an idempotent search reindex.
+   *
+   * Idempotent and lock-free, so it is safe whether or not a live editor is
+   * connected: Yjs state is cumulative, so a concurrent onStoreDocument simply
+   * persists a superset of this state later.
+   */
+  async persistTitleFragmentYdoc(
+    pageId: string,
+    ydocState: Buffer,
+  ): Promise<void> {
+    await this.pageRepo.updatePage({ ydoc: ydocState }, pageId);
+  }
+
   async onStoreDocument(data: onStoreDocumentPayload) {
     const { documentName, document, context } = data;
 
@@ -171,7 +336,34 @@ export class PersistenceExtension implements Extension {
       this.logger.warn('jsonToText' + err?.['message']);
     }
 
+    // Title lives in the SAME Y.Doc as the body, in a dedicated 'title' fragment
+    // (the collaborative title-editor contract with the client). Extract it
+    // defensively: a malformed title fragment must NOT crash the document store.
+    // `hasTitleFragment` distinguishes "the doc actually carries a title
+    // fragment" from "legacy doc with no title fragment" — only the former may
+    // write page.title, so a legacy doc never clobbers the column with ''.
+    let titleText = '';
+    let hasTitleFragment = false;
+    try {
+      const titleFrag = document.get('title', Y.XmlFragment);
+      hasTitleFragment = !!titleFrag && titleFrag.length > 0;
+      if (hasTitleFragment) {
+        const titleJson = TiptapTransformer.fromYdoc(document, 'title');
+        titleText = titleJson ? jsonToText(titleJson).trim() : '';
+      }
+    } catch (err) {
+      this.logger.warn('title extraction: ' + err?.['message']);
+      hasTitleFragment = false;
+    }
+
     let page: Page = null;
+    // Tracks whether the BODY ('default') changed in this store. The heavy
+    // body-only side-effects (transclusion sync, mentions, RAG, history) stay
+    // gated on this so a title-only change does not trigger them.
+    let bodyChanged = false;
+    // Tracks a successful title-only persist so the post-tx contributor folding
+    // (collabHistory.addContributors) runs for the title-only case too.
+    let titleOnlyPersisted = false;
     const editingUserIds = this.consumeContributors(documentName);
     // Sticky agent marker: 'agent' if any agent edit landed in this window, OR
     // if the current writer is the agent (covers a store with no prior onChange
@@ -205,11 +397,80 @@ export class PersistenceExtension implements Extension {
             return;
           }
 
-          if (isDeepStrictEqual(tiptapJson, page.content)) {
+          bodyChanged = !isDeepStrictEqual(tiptapJson, page.content);
+          // Only a populated 'title' fragment may update page.title; compare
+          // against the current column value (treat null as '').
+          //
+          // ANTI-CORRUPTION GUARD (Bug 2): the client's collaborative title-editor
+          // can momentarily initialize the 'title' fragment as an EMPTY heading
+          // (so `hasTitleFragment` is true, but the extracted `titleText` is '')
+          // BEFORE the server's real-title seed has synced. Writing that '' would
+          // silently wipe a non-empty page.title to "untitled". A wiki page is
+          // never legitimately retitled to empty via this path, so we treat an
+          // empty extracted title as "not authoritative" and never persist it.
+          // The `titleText.length > 0` clause makes this guard apply to BOTH the
+          // title-only branch and the body+title branch below.
+          //
+          // DELIBERATE: this intentionally makes it impossible to retitle a page
+          // to EMPTY via the collab path — a wiki page is never legitimately
+          // empty-titled. If a non-empty-title rule ever needs relaxing or
+          // enforcing differently, the REST UpdatePageDto is the place to validate
+          // the title, not this collab guard.
+          const titleChanged =
+            hasTitleFragment &&
+            titleText.length > 0 &&
+            titleText !== (page.title ?? '');
+
+          // No-op fast path: neither body nor title changed.
+          if (!bodyChanged && !titleChanged) {
             page = null;
             return;
           }
 
+          // Title-only change: the body is unchanged, so skip the heavy body
+          // history/contributor logic and persist just the new title and the
+          // ydoc (the title fragment edit lives in the same ydoc). The early-skip
+          // used to drop this case entirely, losing the title change.
+          if (!bodyChanged) {
+            // Fold the window's editing users into contributors the same way the
+            // body branch does, so a user who edited ONLY the title is not dropped
+            // from page.contributorIds.
+            const contributorIds = Array.from(
+              new Set([
+                ...(page.contributorIds || []),
+                ...editingUserIds,
+                page.creatorId,
+              ]),
+            );
+            await this.pageRepo.updatePage(
+              {
+                title: titleText,
+                ydoc: ydocState,
+                lastUpdatedById: context.user.id,
+                contributorIds,
+                // A title-only change is not a body-authorship transition; leave
+                // lastUpdatedSource/aiChatId untouched so the user->agent history
+                // boundary in the body branch is not bypassed.
+              },
+              pageId,
+              trx,
+              // Mirror PageService.update's tree snapshot so a collaborative rename
+              // propagates to other users' sidebar/breadcrumbs like the REST rename.
+              {
+                treeUpdate: {
+                  id: pageId,
+                  slugId: page.slugId,
+                  spaceId: page.spaceId,
+                  parentPageId: page.parentPageId ?? null,
+                  title: titleText,
+                },
+              },
+            );
+            this.logger.debug(`Page title updated: ${pageId} - SlugId: ${page.slugId}`);
+            titleOnlyPersisted = true;
+            return;
+          }
+
           let contributorIds = undefined;
           try {
             const existingContributors = page.contributorIds || [];
@@ -227,29 +488,22 @@ export class PersistenceExtension implements Extension {
           // Approach A — boundary snapshot before the agent's first edit.
           // When this store is the agent's and the page's currently persisted
           // state was authored by a human, pin that human state as its own
-          // history version BEFORE the agent overwrites it. `page` still holds
-          // the OLD content/provenance here, so saveHistory(page) captures the
-          // pre-agent state tagged 'user'. The agent's new content is
-          // snapshotted later by the debounced PAGE_HISTORY job ('agent'). Skip
-          // if the prior state is already agent-authored (boundary already
-          // pinned on the user->agent transition), if the page is effectively
-          // empty, or if the latest existing snapshot already equals this human
-          // state (avoid duplicates).
-          if (
-            lastUpdatedSource === 'agent' &&
-            page.lastUpdatedSource !== 'agent'
-          ) {
+          // history version BEFORE the agent overwrites it. `page` still holds the
+          // OLD content/provenance here, so saveHistory(page) captures the
+          // pre-agent state tagged 'user'. The agent's new content is snapshotted
+          // later by the debounced PAGE_HISTORY job ('agent'). Skip if the prior
+          // state is already agent-authored (boundary already pinned on the
+          // user->agent transition), if the page is effectively empty, or if the
+          // latest existing snapshot already equals this human state (avoid
+          // duplicates).
+          if (lastUpdatedSource === 'agent' && page.lastUpdatedSource !== 'agent') {
             const lastHistory = await this.pageHistoryRepo.findPageLastHistory(
               pageId,
               { includeContent: true, trx },
             );
             const humanBaselineMissing =
-              !lastHistory ||
-              !isDeepStrictEqual(lastHistory.content, page.content);
-            if (
-              !isEmptyParagraphDoc(page.content as any) &&
-              humanBaselineMissing
-            ) {
+              !lastHistory || !isDeepStrictEqual(lastHistory.content, page.content);
+            if (!isEmptyParagraphDoc(page.content as any) && humanBaselineMissing) {
               await this.pageHistoryRepo.saveHistory(page, {
                 contributorIds: page.contributorIds ?? undefined,
                 trx,
@@ -267,9 +521,27 @@ export class PersistenceExtension implements Extension {
               lastUpdatedSource,
               lastUpdatedAiChatId: context?.aiChatId ?? null,
               contributorIds: contributorIds,
+              // Persist the title in the SAME transaction when the title fragment
+              // changed alongside the body.
+              ...(titleChanged ? { title: titleText } : {}),
             },
             pageId,
             trx,
+            // Mirror PageService.update's tree snapshot so a collaborative rename
+            // propagates to other users' sidebar/breadcrumbs like the REST rename.
+            // Only attach when the title actually changed; a body-only save must
+            // not trigger a tree broadcast.
+            titleChanged
+              ? {
+                  treeUpdate: {
+                    id: pageId,
+                    slugId: page.slugId,
+                    spaceId: page.spaceId,
+                    parentPageId: page.parentPageId ?? null,
+                    title: titleText,
+                  },
+                }
+              : undefined,
           );
 
           this.logger.debug(`Page updated: ${pageId} - SlugId: ${page.slugId}`);
@@ -290,6 +562,8 @@ export class PersistenceExtension implements Extension {
       }
     }
 
+    // `page` is truthy whenever anything was persisted (body OR title-only), so
+    // the page.updated broadcast fires for a title-only change too.
     if (page) {
       document.broadcastStateless(
         JSON.stringify({
@@ -307,11 +581,20 @@ export class PersistenceExtension implements Extension {
             : undefined,
         }),
       );
+    }
 
+    // Record the window's editing users in collab history for a title-only
+    // change too (the body branch does this below, gated on bodyChanged).
+    if (page && titleOnlyPersisted) {
+      await this.collabHistory.addContributors(pageId, editingUserIds);
+    }
+
+    // Body-only side-effects: skip them for a title-only change (body unchanged).
+    if (page && bodyChanged) {
       await this.syncTransclusion(pageId, page.workspaceId, tiptapJson);
     }
 
-    if (page) {
+    if (page && bodyChanged) {
       await this.collabHistory.addContributors(pageId, editingUserIds);
 
       const mentions = extractMentions(tiptapJson);

apps/server/src/collaboration/listeners/page-tree-bridge.publisher.spec.ts

@@ -0,0 +1,81 @@
+import { Test, TestingModule } from '@nestjs/testing';
+import { RedisService } from '@nestjs-labs/nestjs-ioredis';
+import { PageTreeBridgePublisher } from './page-tree-bridge.publisher';
+import { COLLAB_TREE_UPDATE_CHANNEL } from '../constants';
+import {
+  PageEvent,
+  TreeUpdateSnapshot,
+} from '../../database/listeners/page.listener';
+
+const treeUpdate: TreeUpdateSnapshot = {
+  id: 'page-1',
+  slugId: 'slug-1',
+  spaceId: 'space-1',
+  parentPageId: null,
+  title: 'Renamed',
+  icon: '🚀',
+};
+
+describe('PageTreeBridgePublisher', () => {
+  let publisher: PageTreeBridgePublisher;
+  let redis: { publish: jest.Mock };
+
+  beforeEach(async () => {
+    redis = { publish: jest.fn().mockResolvedValue(1) };
+    const redisService = { getOrThrow: () => redis } as unknown as RedisService;
+
+    const module: TestingModule = await Test.createTestingModule({
+      providers: [
+        PageTreeBridgePublisher,
+        { provide: RedisService, useValue: redisService },
+      ],
+    }).compile();
+
+    publisher = module.get<PageTreeBridgePublisher>(PageTreeBridgePublisher);
+  });
+
+  it('WITH a `treeUpdate`: publishes the JSON snapshot on the channel', async () => {
+    const event: PageEvent = {
+      pageIds: ['page-1'],
+      workspaceId: 'ws-1',
+      treeUpdate,
+    };
+
+    await publisher.onPageUpdated(event);
+
+    expect(redis.publish).toHaveBeenCalledTimes(1);
+    expect(redis.publish).toHaveBeenCalledWith(
+      COLLAB_TREE_UPDATE_CHANNEL,
+      JSON.stringify(treeUpdate),
+    );
+  });
+
+  it('content-only save (NO `treeUpdate`): does NOT publish', async () => {
+    const event: PageEvent = {
+      pageIds: ['page-1'],
+      workspaceId: 'ws-1',
+    };
+
+    await publisher.onPageUpdated(event);
+
+    expect(redis.publish).not.toHaveBeenCalled();
+  });
+
+  it('a publish rejection is caught (no throw)', async () => {
+    redis.publish.mockRejectedValueOnce(new Error('redis down'));
+    const errorSpy = jest
+      .spyOn(publisher['logger'], 'error')
+      .mockImplementation(() => undefined);
+
+    const event: PageEvent = {
+      pageIds: ['page-1'],
+      workspaceId: 'ws-1',
+      treeUpdate,
+    };
+
+    await expect(publisher.onPageUpdated(event)).resolves.toBeUndefined();
+    expect(errorSpy).toHaveBeenCalledTimes(1);
+
+    errorSpy.mockRestore();
+  });
+});

apps/server/src/collaboration/listeners/page-tree-bridge.publisher.ts

@@ -0,0 +1,55 @@
+import { Injectable, Logger } from '@nestjs/common';
+import { OnEvent } from '@nestjs/event-emitter';
+import { RedisService } from '@nestjs-labs/nestjs-ioredis';
+import type { Redis } from 'ioredis';
+import { EventName } from '../../common/events/event.contants';
+import { PageEvent } from '../../database/listeners/page.listener';
+import { COLLAB_TREE_UPDATE_CHANNEL } from '../constants';
+
+/**
+ * Collab-process half of the cross-process tree-update bridge.
+ *
+ * The standalone collab process bootstraps `CollabAppModule`, which does NOT
+ * import `WsModule`/`PageWsListener`. So when a collaborative title/icon rename
+ * persists and emits `EventName.PAGE_UPDATED` with a `treeUpdate` snapshot, there
+ * is no listener in this process to broadcast it — the live tree update would be
+ * lost for 2-process (COLLAB_URL set) deployments.
+ *
+ * This publisher fills that gap: it forwards the `treeUpdate` snapshot over a
+ * Redis pub/sub channel to the API process, which re-broadcasts it via
+ * `WsTreeService` (the single broadcast authority).
+ *
+ * It is registered ONLY in `CollabAppModule.providers`, so it never runs in the
+ * API process (where `PageWsListener` already broadcasts the same event locally).
+ * That module placement is what prevents a double broadcast. In single-process
+ * mode `CollabAppModule` is not loaded at all, so this publisher never runs.
+ */
+@Injectable()
+export class PageTreeBridgePublisher {
+  private readonly logger = new Logger(PageTreeBridgePublisher.name);
+  private readonly redis: Redis;
+
+  constructor(private readonly redisService: RedisService) {
+    this.redis = this.redisService.getOrThrow();
+  }
+
+  @OnEvent(EventName.PAGE_UPDATED)
+  async onPageUpdated(event: PageEvent): Promise<void> {
+    // Mirror PageWsListener's gating: only title/icon changes carry a snapshot.
+    // Content-only saves leave `treeUpdate` undefined and are ignored.
+    if (!event.treeUpdate) return;
+
+    try {
+      await this.redis.publish(
+        COLLAB_TREE_UPDATE_CHANNEL,
+        JSON.stringify(event.treeUpdate),
+      );
+    } catch (err) {
+      // A Redis publish failure must not break the store path.
+      this.logger.error(
+        `Failed to publish tree update to ${COLLAB_TREE_UPDATE_CHANNEL}`,
+        err instanceof Error ? err.stack : String(err),
+      );
+    }
+  }
+}

apps/server/src/collaboration/server/collab-app.module.ts

@@ -20,6 +20,7 @@ import { CaslModule } from '../../core/casl/casl.module';
 import { ThrottleModule } from '../../integrations/throttle/throttle.module';
 import { CacheModule } from '@nestjs/cache-manager';
 import KeyvRedis from '@keyv/redis';
+import { PageTreeBridgePublisher } from '../listeners/page-tree-bridge.publisher';
 
 @Module({
   imports: [
@@ -54,6 +55,6 @@ import KeyvRedis from '@keyv/redis';
       ? [CollaborationController]
       : []),
   ],
-  providers: [AppService],
+  providers: [AppService, PageTreeBridgePublisher],
 })
 export class CollabAppModule {}

apps/server/src/collaboration/title-rename-durability.spec.ts

@@ -0,0 +1,187 @@
+import * as Y from 'yjs';
+import { TiptapTransformer } from '@hocuspocus/transformer';
+import { CollaborationGateway } from './collaboration.gateway';
+import { PersistenceExtension } from './extensions/persistence.extension';
+import {
+  buildTitleSeedYdoc,
+  jsonToText,
+  tiptapExtensions,
+} from './collaboration.util';
+
+/**
+ * F1 (variant C) — rename durability for a page with an already-persisted Yjs
+ * 'title' fragment and NO live editor (the REST/MCP/agent rename path).
+ *
+ * The bug: PageService.update writes the NEW title to the `page.title` COLUMN,
+ * then calls gateway.writePageTitle, which loads the page's ydoc (fragment =
+ * OLD) and overwrites it to NEW in memory. On disconnect, onStoreDocument sees
+ * titleText(NEW) === column(NEW) → no-op fast-path → it does NOT persist the
+ * in-memory fragment. So `page.ydoc` keeps the OLD title, and a LATER body edit
+ * loads the OLD fragment, sees it differs from the column, and silently reverts
+ * the column back to OLD.
+ *
+ * The fix: writePageTitle persists the 'title' fragment to `page.ydoc` DIRECTLY
+ * (via PersistenceExtension.persistTitleFragmentYdoc) after the transact, so the
+ * persisted fragment and the column stay consistent.
+ *
+ * This test drives the REAL writePageTitle + the REAL onStoreDocument against an
+ * in-memory page row, so it FAILS on the pre-fix no-op behaviour and PASSES after.
+ */
+
+const PAGE_ID = '550e8400-e29b-41d4-a716-446655440000';
+const USER_ID = 'user-1';
+const OLD_TITLE = 'Old Title';
+const NEW_TITLE = 'Renamed Title';
+
+const bodyJson = (text: string) => ({
+  type: 'doc',
+  content: [{ type: 'paragraph', content: [{ type: 'text', text }] }],
+});
+
+// Build the initial persisted ydoc carrying BOTH a 'title' fragment and a body.
+const makeInitialYdoc = (title: string, body: any): Buffer => {
+  const doc = new Y.Doc();
+  Y.applyUpdate(doc, Y.encodeStateAsUpdate(buildTitleSeedYdoc(title)));
+  Y.applyUpdate(
+    doc,
+    Y.encodeStateAsUpdate(TiptapTransformer.toYdoc(body, 'default', tiptapExtensions)),
+  );
+  return Buffer.from(Y.encodeStateAsUpdate(doc));
+};
+
+// Load a doc from a persisted buffer (mirrors openDirectConnection loading from
+// persistence when no editor is connected). hocuspocus augments the live doc
+// with broadcastStateless(); a bare Y.Doc lacks it, so stub it.
+const loadDoc = (buf: Buffer): Y.Doc => {
+  const doc = new Y.Doc();
+  if (buf) Y.applyUpdate(doc, new Uint8Array(buf));
+  (doc as any).broadcastStateless = jest.fn();
+  return doc;
+};
+
+// Read the 'title' fragment text from a persisted buffer.
+const readTitle = (buf: Buffer): string => {
+  const doc = loadDoc(buf);
+  const titleJson = TiptapTransformer.fromYdoc(doc, 'title');
+  return titleJson ? jsonToText(titleJson).trim() : '';
+};
+
+describe('rename durability (F1 variant C): persisted title fragment survives a body edit', () => {
+  it('persists the renamed title into page.ydoc so a later body edit does not revert it', async () => {
+    // In-memory page row = the DB.
+    const row: any = {
+      id: PAGE_ID,
+      slugId: 'slug-1',
+      spaceId: 'space-1',
+      workspaceId: 'ws-1',
+      creatorId: 'creator-1',
+      contributorIds: ['creator-1'],
+      createdAt: new Date('2020-01-01T00:00:00Z'),
+      lastUpdatedSource: 'user',
+      title: OLD_TITLE,
+      // content column mirrors the normalized body in the ydoc.
+      content: TiptapTransformer.fromYdoc(
+        loadDoc(makeInitialYdoc(OLD_TITLE, bodyJson('BODY V1'))),
+        'default',
+      ),
+      ydoc: makeInitialYdoc(OLD_TITLE, bodyJson('BODY V1')),
+    };
+
+    const pageRepo = {
+      findById: jest.fn(async () => ({ ...row })),
+      updatePage: jest.fn(async (data: any, _pageId?: string) => {
+        Object.assign(row, data, { updatedAt: new Date() });
+      }),
+    };
+    const pageHistoryRepo = {
+      saveHistory: jest.fn().mockResolvedValue(undefined),
+      findPageLastHistory: jest.fn().mockResolvedValue(null),
+    };
+    const noopQueue = { add: jest.fn().mockResolvedValue(undefined) };
+    const collabHistory = { addContributors: jest.fn().mockResolvedValue(undefined) };
+    const transclusionService = {
+      syncPageTransclusions: jest.fn().mockResolvedValue(undefined),
+      syncPageReferences: jest.fn().mockResolvedValue(undefined),
+      syncPageTemplateReferences: jest.fn().mockResolvedValue(undefined),
+    };
+    // db whose transaction().execute(fn) runs fn with a trx stub (drives the
+    // real executeTx helper without a database).
+    const db = {
+      transaction: () => ({
+        execute: (fn: (trx: any) => Promise<any>) => fn({ __trx: true }),
+      }),
+    };
+
+    const ext = new PersistenceExtension(
+      pageRepo as any,
+      pageHistoryRepo as any,
+      db as any,
+      noopQueue as any,
+      noopQueue as any,
+      noopQueue as any,
+      collabHistory as any,
+      transclusionService as any,
+    );
+    jest.spyOn(ext['logger'], 'debug').mockImplementation(() => undefined);
+    jest.spyOn(ext['logger'], 'warn').mockImplementation(() => undefined);
+    jest.spyOn(ext['logger'], 'error').mockImplementation(() => undefined);
+
+    const documentName = `page.${PAGE_ID}`;
+    // Fake hocuspocus: openDirectConnection loads a doc from the CURRENT persisted
+    // ydoc (no live editor) and, on disconnect, runs the real onStoreDocument —
+    // exactly the no-live-editor unload path.
+    const fakeHocuspocus = {
+      openDirectConnection: jest.fn(async (name: string, context: any) => {
+        const liveDoc = loadDoc(row.ydoc);
+        return {
+          transact: async (fn: (doc: Y.Doc) => void) => fn(liveDoc),
+          disconnect: async () => {
+            await ext.onStoreDocument({
+              documentName: name,
+              document: liveDoc,
+              context,
+            } as any);
+          },
+        };
+      }),
+    };
+
+    const gateway: CollaborationGateway = Object.create(
+      CollaborationGateway.prototype,
+    );
+    (gateway as any).hocuspocus = fakeHocuspocus;
+    (gateway as any).persistenceExtension = ext;
+
+    // --- REST/service rename (no live editor) ---
+    // 1) PageService.update writes the NEW title to the column.
+    await pageRepo.updatePage({ title: NEW_TITLE }, PAGE_ID);
+    // 2) PageService.update syncs the Yjs 'title' fragment.
+    await gateway.writePageTitle(PAGE_ID, NEW_TITLE, {
+      user: { id: USER_ID } as any,
+    });
+
+    // Reload the persisted ydoc: the 'title' fragment must now be NEW.
+    // (Pre-fix this is still OLD — writePageTitle did not persist the fragment.)
+    expect(readTitle(row.ydoc)).toBe(NEW_TITLE);
+
+    // --- a later body edit must NOT revert the title ---
+    const editDoc = loadDoc(row.ydoc);
+    const frag = editDoc.getXmlFragment('default');
+    const p = new Y.XmlElement('paragraph');
+    const t = new Y.XmlText();
+    t.insert(0, 'appended');
+    p.insert(0, [t]);
+    frag.insert(frag.length, [p]);
+
+    await ext.onStoreDocument({
+      documentName,
+      document: editDoc,
+      context: { user: { id: USER_ID } },
+    } as any);
+
+    // The body edit was persisted, and the title stayed NEW in BOTH the column
+    // and the persisted ydoc fragment (pre-fix the column reverts to OLD).
+    expect(row.title).toBe(NEW_TITLE);
+    expect(readTitle(row.ydoc)).toBe(NEW_TITLE);
+  });
+});

apps/server/src/core/ai-chat/ai-chat-run.service.spec.ts

@@ -1,492 +0,0 @@
-import { Logger } from '@nestjs/common';
-import {
-  AiChatRunService,
-  RunAlreadyActiveError,
-  ONE_ACTIVE_RUN_PER_CHAT_INDEX,
-  mapTurnStatusToRun,
-} from './ai-chat-run.service';
-
-/** Shape a Postgres unique-violation the way the postgres.js driver surfaces it:
- *  SQLSTATE 23505 + the offending index in `constraint_name`. */
-function uniqueViolation(constraintName: string): Error & {
-  code: string;
-  constraint_name: string;
-} {
-  return Object.assign(
-    new Error('duplicate key value violates unique constraint'),
-    {
-      code: '23505',
-      constraint_name: constraintName,
-    },
-  );
-}
-
-/**
- * Unit coverage for the #184 phase-1 run lifecycle (AiChatRunService) with a
- * hand-rolled mock repo — no Nest graph, no DB. The invariant under test is the
- * one that makes a run "autonomous": a run keeps going when its SUBSCRIBER (the
- * browser) detaches, and ONLY an explicit stop aborts it. We assert that at the
- * abort-signal level (the signal the agent loop actually consumes).
- */
-
-/** Minimal EnvironmentService stub. Single-instance (CLOUD unset) by default. */
-function makeEnv(isCloud = false) {
-  return { isCloud: () => isCloud };
-}
-
-function makeRepo(overrides: Record<string, jest.Mock> = {}) {
-  return {
-    insert: jest.fn(async (v: any) => ({
-      id: 'run-1',
-      status: v.status ?? 'running',
-      chatId: v.chatId,
-      workspaceId: v.workspaceId,
-    })),
-    update: jest.fn(async () => ({ id: 'run-1' })),
-    markStopRequested: jest.fn(async () => ({ id: 'run-1' })),
-    findActiveByChat: jest.fn(async () => undefined),
-    findLatestByChat: jest.fn(async () => undefined),
-    findById: jest.fn(async () => undefined),
-    sweepRunning: jest.fn(async () => 0),
-    ...overrides,
-  };
-}
-
-describe('mapTurnStatusToRun', () => {
-  it('maps the turn terminal status to the run terminal status', () => {
-    expect(mapTurnStatusToRun('completed')).toBe('succeeded');
-    expect(mapTurnStatusToRun('error')).toBe('failed');
-    expect(mapTurnStatusToRun('aborted')).toBe('aborted');
-  });
-});
-
-describe('AiChatRunService.onModuleInit (startup sweep)', () => {
-  afterEach(() => jest.restoreAllMocks());
-
-  it('calls sweepRunning and resolves; logs when > 0', async () => {
-    const repo = makeRepo({ sweepRunning: jest.fn(async () => 2) });
-    const logSpy = jest
-      .spyOn(Logger.prototype, 'log')
-      .mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(svc.onModuleInit()).resolves.toBeUndefined();
-    expect(repo.sweepRunning).toHaveBeenCalledTimes(1);
-    expect(logSpy).toHaveBeenCalledTimes(1);
-    expect(String(logSpy.mock.calls[0][0])).toContain('2');
-  });
-
-  it('a sweep failure is swallowed (never blocks startup)', async () => {
-    const repo = makeRepo({
-      sweepRunning: jest.fn(async () => {
-        throw new Error('db down');
-      }),
-    });
-    const warnSpy = jest
-      .spyOn(Logger.prototype, 'warn')
-      .mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(svc.onModuleInit()).resolves.toBeUndefined();
-    // The first warn is the sweep failure (the multi-instance warn never fires
-    // single-instance), so the message is the db error.
-    expect(String(warnSpy.mock.calls[0][0])).toContain('db down');
-  });
-
-  it('F1 (DECISION C): the boot sweep is UNCONDITIONAL — sweepRunning is called with NO staleness window, so a fresh running run (updatedAt = now) is settled, not skipped', async () => {
-    // The bug: a fast restart (deploy/OOM within minutes of the last step) left a
-    // run stuck 'running' under the old 10-min window, 409ing every later turn in
-    // the chat. The fix settles ALL pending|running on boot. We assert the service
-    // invokes sweepRunning with no `staleMs` (the unconditional path); the repo's
-    // own spec proves no-window => no updatedAt filter.
-    const repo = makeRepo({ sweepRunning: jest.fn(async () => 1) });
-    jest.spyOn(Logger.prototype, 'log').mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.onModuleInit();
-    expect(repo.sweepRunning).toHaveBeenCalledTimes(1);
-    const callArgs = repo.sweepRunning.mock.calls[0] as unknown[];
-    const firstArg = callArgs[0] as { staleMs?: number } | undefined;
-    // Either no opts at all, or opts without a staleMs window => unconditional.
-    expect(firstArg?.staleMs).toBeUndefined();
-  });
-
-  it('F2 (DECISION A): warns at startup that autonomousRuns is single-instance-only when a horizontally-scaled deployment (CLOUD) is detected', async () => {
-    const repo = makeRepo();
-    const warnSpy = jest
-      .spyOn(Logger.prototype, 'warn')
-      .mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv(true) as never);
-    await svc.onModuleInit();
-    const warned = warnSpy.mock.calls.some((c) =>
-      /single-instance-only/i.test(String(c[0])),
-    );
-    expect(warned).toBe(true);
-  });
-
-  it('F2: does NOT warn about multi-instance on a single-instance (CLOUD unset) deployment', async () => {
-    const repo = makeRepo();
-    const warnSpy = jest
-      .spyOn(Logger.prototype, 'warn')
-      .mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv(false) as never);
-    await svc.onModuleInit();
-    const warned = warnSpy.mock.calls.some((c) =>
-      /single-instance-only/i.test(String(c[0])),
-    );
-    expect(warned).toBe(false);
-  });
-});
-
-describe('AiChatRunService run lifecycle', () => {
-  it('beginRun inserts a running row and registers a live abort controller', async () => {
-    const repo = makeRepo();
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const handle = await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-    expect(repo.insert).toHaveBeenCalledWith(
-      expect.objectContaining({
-        chatId: 'chat-1',
-        workspaceId: 'ws-1',
-        createdBy: 'user-1',
-        status: 'running',
-        trigger: 'user',
-      }),
-    );
-    expect(handle.runId).toBe('run-1');
-    expect(handle.signal.aborted).toBe(false);
-    expect(svc.isLocallyActive('run-1')).toBe(true);
-  });
-
-  it('beginRun REJECTS the racer: a 23505 on the one-active-per-chat index throws RunAlreadyActiveError (not swallowed) and registers no controller', async () => {
-    // The race: the controller's cheap pre-check passed for BOTH concurrent
-    // turns, so the loser's INSERT hits the partial unique index. That rejection
-    // is the authoritative gate — it must surface, not be swallowed into an
-    // untracked turn.
-    const repo = makeRepo({
-      insert: jest.fn(async () => {
-        throw uniqueViolation(ONE_ACTIVE_RUN_PER_CHAT_INDEX);
-      }),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(
-      svc.beginRun({ chatId: 'chat-1', workspaceId: 'ws-1', userId: 'user-1' }),
-    ).rejects.toBeInstanceOf(RunAlreadyActiveError);
-    // No controller leaked for a rejected start.
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-  });
-
-  it('beginRun does NOT mask an unrelated unique violation as already-active', async () => {
-    // A 23505 on some OTHER constraint is a real bug, not the race — it must
-    // propagate unchanged so it is never silently treated as "already active".
-    const other = uniqueViolation('ai_chat_runs_pkey');
-    const repo = makeRepo({
-      insert: jest.fn(async () => {
-        throw other;
-      }),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(
-      svc.beginRun({ chatId: 'chat-1', workspaceId: 'ws-1', userId: 'user-1' }),
-    ).rejects.toBe(other);
-  });
-
-  it('beginRun propagates a non-unique insert failure unchanged', async () => {
-    const boom = new Error('connection reset');
-    const repo = makeRepo({
-      insert: jest.fn(async () => {
-        throw boom;
-      }),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(
-      svc.beginRun({ chatId: 'chat-1', workspaceId: 'ws-1', userId: 'user-1' }),
-    ).rejects.toBe(boom);
-  });
-
-  it('two concurrent begins on one chat: exactly one wins, the other is rejected as already-active', async () => {
-    // Integration-style: model the DB partial unique index with a one-shot slot.
-    // The first insert claims it; the second hits a 23505 on the active index.
-    let slotTaken = false;
-    const repo = makeRepo({
-      insert: jest.fn(async (v: any) => {
-        if (slotTaken) throw uniqueViolation(ONE_ACTIVE_RUN_PER_CHAT_INDEX);
-        slotTaken = true;
-        return { id: 'run-win', status: v.status, chatId: v.chatId };
-      }),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const results = await Promise.allSettled([
-      svc.beginRun({ chatId: 'chat-1', workspaceId: 'ws-1', userId: 'user-1' }),
-      svc.beginRun({ chatId: 'chat-1', workspaceId: 'ws-1', userId: 'user-1' }),
-    ]);
-    const fulfilled = results.filter((r) => r.status === 'fulfilled');
-    const rejected = results.filter((r) => r.status === 'rejected');
-    expect(fulfilled).toHaveLength(1);
-    expect(rejected).toHaveLength(1);
-    expect((rejected[0] as PromiseRejectedResult).reason).toBeInstanceOf(
-      RunAlreadyActiveError,
-    );
-    // Exactly the winner is locally active.
-    expect(svc.isLocallyActive('run-win')).toBe(true);
-  });
-
-  it('a SUBSCRIBER detaching does NOT abort the run (only an explicit stop does)', async () => {
-    const repo = makeRepo();
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const handle = await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-    // Model a browser disconnect: nothing in the run service is told to stop.
-    // The signal the agent loop consumes must stay un-aborted and the run stays
-    // locally active — i.e. it keeps running server-side.
-    expect(handle.signal.aborted).toBe(false);
-    expect(svc.isLocallyActive('run-1')).toBe(true);
-    // markStopRequested was never called by a mere detach.
-    expect(repo.markStopRequested).not.toHaveBeenCalled();
-  });
-
-  it('requestStop aborts the live controller, marks the row, and reports true', async () => {
-    const repo = makeRepo();
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const handle = await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-    const aborted = jest.fn();
-    handle.signal.addEventListener('abort', aborted);
-
-    const result = await svc.requestStop('run-1', 'ws-1');
-
-    expect(result).toBe(true);
-    expect(handle.signal.aborted).toBe(true);
-    expect(aborted).toHaveBeenCalledTimes(1);
-    expect(repo.markStopRequested).toHaveBeenCalledWith('run-1', 'ws-1');
-  });
-
-  it('requestStop on a run this replica does NOT hold still marks the row (true)', async () => {
-    // e.g. after a restart, or a sibling replica owns the controller. The row is
-    // marked so the owning replica/sweep settles it; we report a stop took effect.
-    const repo = makeRepo({
-      markStopRequested: jest.fn(async () => ({ id: 'run-9' })),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const result = await svc.requestStop('run-9', 'ws-1');
-    expect(result).toBe(true);
-    expect(svc.isLocallyActive('run-9')).toBe(false);
-  });
-
-  it('requestStop on an already-settled run (nothing active) reports false', async () => {
-    const repo = makeRepo({
-      markStopRequested: jest.fn(async () => undefined),
-    });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    const result = await svc.requestStop('run-done', 'ws-1');
-    expect(result).toBe(false);
-  });
-
-  it('finalizeRun settles the row to the mapped status with finishedAt and drops the in-memory entry', async () => {
-    const repo = makeRepo();
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-    expect(svc.isLocallyActive('run-1')).toBe(true);
-
-    await svc.finalizeRun('run-1', 'ws-1', 'error', 'provider blew up');
-
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-    expect(repo.update).toHaveBeenCalledWith(
-      'run-1',
-      'ws-1',
-      expect.objectContaining({
-        status: 'failed',
-        error: 'provider blew up',
-        finishedAt: expect.any(Date),
-      }),
-    );
-  });
-
-  it('finalizeRun is IDEMPOTENT: a second settle no-ops (single terminal write)', async () => {
-    // The #184 review fix: AiChatService.stream wraps the turn in a safety-net
-    // catch that settles a failed turn AND streamText's terminal callback may
-    // also settle — both routes call finalizeRun. Only the FIRST may write the
-    // terminal row; the second must no-op so a late settle can never clobber the
-    // real terminal status or double-write the row.
-    const repo = makeRepo();
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-
-    await svc.finalizeRun('run-1', 'ws-1', 'error', 'first');
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-    // A second settle (e.g. a streamText callback firing after the catch) no-ops.
-    await svc.finalizeRun('run-1', 'ws-1', 'completed', undefined);
-
-    expect(repo.update).toHaveBeenCalledTimes(1);
-    expect(repo.update).toHaveBeenCalledWith(
-      'run-1',
-      'ws-1',
-      expect.objectContaining({ status: 'failed', error: 'first' }),
-    );
-  });
-
-  it('CONCURRENCY: two simultaneous finalizeRun on the same run write the terminal row EXACTLY ONCE (the 2nd caller exits synchronously at the atomic claim)', async () => {
-    // The CRITICAL race: AiChatService.stream's safety-net catch settles the turn
-    // to 'error' while a streamText terminal callback also settles it — both call
-    // finalizeRun for the SAME runId. The once-gate must close ATOMICALLY: a
-    // `settled.has` check alone is read BEFORE the awaited UPDATE, so both callers
-    // would pass it and BOTH write the row (last-write-wins clobber + double
-    // write). The fix claims the run with a SYNCHRONOUS `active.delete` before any
-    // await, so the second caller returns in the same tick, before the UPDATE.
-    //
-    // We force the two calls to overlap by making `update` return a promise we
-    // resolve only AFTER both finalizeRun calls have run their synchronous bodies.
-    let resolveUpdate!: (v: unknown) => void;
-    const updateGate = new Promise((res) => {
-      resolveUpdate = res;
-    });
-    const update = jest.fn(() => updateGate);
-    const repo = makeRepo({ update });
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-
-    // Fire both before the (pending) update resolves. The first synchronously
-    // claims the entry (active.delete) and awaits update; the second, started in
-    // the same macrotask, finds the entry already gone and returns at the claim
-    // WITHOUT ever calling update.
-    const p1 = svc.finalizeRun('run-1', 'ws-1', 'completed');
-    const p2 = svc.finalizeRun('run-1', 'ws-1', 'error', 'safety-net');
-
-    // The decisive assertion: exactly one caller reached the terminal UPDATE.
-    expect(update).toHaveBeenCalledTimes(1);
-
-    // Let the single in-flight update land; both calls resolve cleanly.
-    resolveUpdate({ id: 'run-1' });
-    await Promise.all([p1, p2]);
-
-    expect(update).toHaveBeenCalledTimes(1);
-    // The winner is the FIRST caller ('completed' -> 'succeeded'); the late
-    // 'error' settle never wrote, so it could not clobber the real status.
-    expect(update).toHaveBeenCalledWith(
-      'run-1',
-      'ws-1',
-      expect.objectContaining({ status: 'succeeded' }),
-    );
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-  });
-
-  it('F6: a TRANSIENT terminal-write failure is ridden out by the bounded retry — the run is settled, not stranded', async () => {
-    // The bug: finalizeRun used to DROP the in-memory entry BEFORE the terminal
-    // UPDATE, then only warn-log a failure. A single transient blip (pool
-    // exhaustion / deadlock / connection hiccup) on that PK UPDATE left the row
-    // 'running' with nothing left to recover it -> every later turn in that chat
-    // 409s until a restart. The fix updates FIRST and retries.
-    let calls = 0;
-    const repo = makeRepo({
-      update: jest.fn(async () => {
-        calls += 1;
-        if (calls === 1) throw new Error('deadlock detected');
-        return { id: 'run-1' };
-      }),
-    });
-    jest.spyOn(Logger.prototype, 'warn').mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-
-    await svc.finalizeRun('run-1', 'ws-1', 'completed');
-
-    // The retry landed the terminal write: the entry is dropped (slot freed) and
-    // the row carries the real terminal status — NOT stranded at 'running'.
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-    expect(repo.update).toHaveBeenCalledTimes(2);
-    expect(repo.update).toHaveBeenLastCalledWith(
-      'run-1',
-      'ws-1',
-      expect.objectContaining({ status: 'succeeded' }),
-    );
-  });
-
-  it('F6: if the terminal write keeps failing, the entry is RETAINED and a LATER settle completes it (chat not permanently 409d)', async () => {
-    // Worst case: the DB is down for the whole first finalize (all attempts fail).
-    // The run must NOT be silently lost — the entry stays so a subsequent settle
-    // (a streamText callback, requestStop -> onAbort, or a future sweep) can retry.
-    let healthy = false;
-    const repo = makeRepo({
-      update: jest.fn(async () => {
-        if (!healthy) throw new Error('pool exhausted');
-        return { id: 'run-1' };
-      }),
-    });
-    jest.spyOn(Logger.prototype, 'warn').mockImplementation(() => undefined);
-    const errorSpy = jest
-      .spyOn(Logger.prototype, 'error')
-      .mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await svc.beginRun({
-      chatId: 'chat-1',
-      workspaceId: 'ws-1',
-      userId: 'user-1',
-    });
-
-    // First settle: every bounded attempt fails -> entry retained, NOT settled.
-    await svc.finalizeRun('run-1', 'ws-1', 'completed');
-    expect(svc.isLocallyActive('run-1')).toBe(true);
-    // F12: the give-up emits ONE explicit, greppable ERROR (run + chat context)
-    // so an operator can tell "gave up, run held in memory" from a per-attempt
-    // blip — distinct from the per-attempt warns.
-    const gaveUp = errorSpy.mock.calls.some(
-      (c) =>
-        /NON-TERMINAL/.test(String(c[0])) &&
-        /run-1/.test(String(c[0])) &&
-        /chat-1/.test(String(c[0])),
-    );
-    expect(gaveUp).toBe(true);
-
-    // The DB recovers; a later settle now succeeds and frees the slot.
-    healthy = true;
-    await svc.finalizeRun('run-1', 'ws-1', 'completed');
-    expect(svc.isLocallyActive('run-1')).toBe(false);
-    expect(repo.update).toHaveBeenLastCalledWith(
-      'run-1',
-      'ws-1',
-      expect.objectContaining({ status: 'succeeded' }),
-    );
-
-    // And it is now idempotent: a further settle no-ops (terminal row already
-    // written), so a double-settle can never clobber the real status.
-    const callsBefore = repo.update.mock.calls.length;
-    await svc.finalizeRun('run-1', 'ws-1', 'error', 'late');
-    expect(repo.update).toHaveBeenCalledTimes(callsBefore);
-  });
-
-  it('recordStep / linkAssistantMessage are best-effort: a repo failure is swallowed', async () => {
-    const repo = makeRepo({
-      update: jest.fn(async () => {
-        throw new Error('transient');
-      }),
-    });
-    jest.spyOn(Logger.prototype, 'warn').mockImplementation(() => undefined);
-    const svc = new AiChatRunService(repo as never, makeEnv() as never);
-    await expect(svc.recordStep('run-1', 'ws-1', 3)).resolves.toBeUndefined();
-    await expect(
-      svc.linkAssistantMessage('run-1', 'ws-1', 'msg-1'),
-    ).resolves.toBeUndefined();
-  });
-});

apps/server/src/core/ai-chat/ai-chat-run.service.ts

@@ -1,426 +0,0 @@
-import { Injectable, Logger, OnModuleInit } from '@nestjs/common';
-import { AiChatRunRepo } from '@docmost/db/repos/ai-chat/ai-chat-run.repo';
-import { AiChatRun } from '@docmost/db/types/entity.types';
-import { isUniqueViolation, violatedConstraint } from '@docmost/db/utils';
-import { EnvironmentService } from '../../integrations/environment/environment.service';
-
-/** Name of the partial unique index enforcing "one active run per chat" (see the
- *  ai_chat_runs migration). A 23505 on THIS constraint is the race-safe signal
- *  that a concurrent turn already owns the chat — distinct from any other unique
- *  collision, which must NOT be silently treated as "already active". */
-export const ONE_ACTIVE_RUN_PER_CHAT_INDEX = 'ai_chat_runs_one_active_per_chat';
-
-/**
- * Thrown by {@link AiChatRunService.beginRun} when the run-row INSERT loses the
- * race for a chat's single active slot (the partial unique index rejects it with
- * a 23505). This is the AUTHORITATIVE concurrency gate: the controller's cheap
- * pre-check is only a fast-path, and a request that slips past it must NOT run
- * untracked. The caller (AiChatService.stream) translates this into a 409 and
- * aborts the turn BEFORE any AI/provider call.
- */
-export class RunAlreadyActiveError extends Error {
-  constructor(public readonly chatId: string) {
-    super(`An agent run is already in progress for chat ${chatId}`);
-    this.name = 'RunAlreadyActiveError';
-  }
-}
-
-/**
- * The terminal status of a TURN (the #183 assistant-row lifecycle) maps onto the
- * terminal status of a RUN (#184). A turn that completed -> the run succeeded; a
- * turn that errored -> the run failed; a turn aborted (explicit user stop) -> the
- * run aborted. Pure + unit-testable.
- */
-export type TurnTerminalStatus = 'completed' | 'error' | 'aborted';
-export type RunTerminalStatus = 'succeeded' | 'failed' | 'aborted';
-
-export function mapTurnStatusToRun(
-  status: TurnTerminalStatus,
-): RunTerminalStatus {
-  switch (status) {
-    case 'completed':
-      return 'succeeded';
-    case 'error':
-      return 'failed';
-    case 'aborted':
-      return 'aborted';
-  }
-}
-
-/** An in-flight run held in process memory: its AbortController is the ONLY thing
- *  that can stop the turn (an explicit user stop), independent of the browser
- *  socket. A mere disconnect never touches it, so the run keeps going. */
-interface ActiveRun {
-  controller: AbortController;
-  chatId: string;
-  workspaceId: string;
-}
-
-/** The live handle the streaming path drives a run through (returned by
- *  {@link AiChatRunService.beginRun}). The `signal` governs the agent loop's
- *  abort — wired to the run, NOT to the HTTP socket. */
-export interface RunHandle {
-  runId: string;
-  signal: AbortSignal;
-}
-
-/**
- * AiChatRunService (#184 phase 1) — owns the agent RUN as a first-class,
- * server-side lifecycle object detached from the HTTP request / browser window.
- *
- * Responsibilities:
- *  - create a run row when a turn starts (pending -> running) and register an
- *    in-memory AbortController for it (the explicit-stop lever);
- *  - finalize the run row (succeeded / failed / aborted) and unregister it;
- *  - service an EXPLICIT user stop (`requestStop`) — the ONLY thing that aborts a
- *    run; a browser disconnect deliberately does NOT;
- *  - crash-recovery sweep of dangling runs on startup.
- *
- * The agent loop itself still runs in AiChatService.stream (reusing #183's
- * step-granular durable write path, `consumeStream` already drains it independent
- * of the socket); this service only wraps it in a durable lifecycle and an
- * abort handle that outlives the subscriber.
- */
-@Injectable()
-export class AiChatRunService implements OnModuleInit {
-  private readonly logger = new Logger(AiChatRunService.name);
-
-  // runId -> ActiveRun. Process-local on purpose (phase 1 is single-process /
-  // in-memory transport; a cross-process BullMQ runner + Redis stop-signal is
-  // deferred to phase 2). A stop for a runId not in this map (e.g. after a
-  // restart) still records `stop_requested_at` on the row.
-  private readonly active = new Map<string, ActiveRun>();
-
-  // runIds whose TERMINAL row write has SUCCEEDED — the idempotency once-gate
-  // (F6). A finalize must short-circuit only AFTER the terminal write has landed,
-  // NOT merely after the in-memory entry was dropped: a transient UPDATE failure
-  // has to stay retryable, so "already settled" means "row already terminal", not
-  // "entry already gone". Grows by one short UUID per finished run over process
-  // uptime — negligible in phase 1's single process.
-  private readonly settled = new Set<string>();
-
-  // Bounded retry for the terminal write (F6): a single PK UPDATE can fail
-  // transiently under many fire-and-forget writes (pool exhaustion, deadlock, a
-  // brief connection blip). Riding out that blip in-place matters because the
-  // dominant success path (streamText onFinish) settles exactly ONCE — if that
-  // write is dropped and never retried, the row is stranded 'running' and the
-  // one-active-run gate 409s every future turn in the chat until a restart (no
-  // periodic sweep in phase 1).
-  private static readonly FINALIZE_MAX_ATTEMPTS = 3;
-  private static readonly FINALIZE_RETRY_BASE_MS = 50;
-
-  constructor(
-    private readonly runRepo: AiChatRunRepo,
-    private readonly environment: EnvironmentService,
-  ) {}
-
-  /**
-   * Crash-recovery sweep on server start: settle EVERY run still left
-   * pending/running to 'aborted' (F1 / DECISION C). The boot sweep is
-   * UNCONDITIONAL — no staleness window — because phase 1 is single-process: on a
-   * fresh boot any pending|running run is definitionally hung (no live runner owns
-   * it), so even a fast restart (deploy/OOM within minutes of the last step) can
-   * no longer leave a run stuck 'running' forever (which would make the
-   * one-active-run gate 409 every future turn in that chat). The staleness window
-   * is reintroduced only for the phase-2 multi-instance timer sweep, where a
-   * booting replica must not abort a run another replica is actively executing.
-   * Best-effort — a sweep failure is logged but MUST NOT block startup (mirrors
-   * AiChatService.onModuleInit for #183).
-   */
-  async onModuleInit(): Promise<void> {
-    this.warnIfMultiInstance();
-    try {
-      // No `staleMs`: unconditional boot sweep (F1). See AiChatRunRepo.sweepRunning.
-      const swept = await this.runRepo.sweepRunning();
-      if (swept > 0) {
-        this.logger.log(
-          `Startup sweep: marked ${swept} dangling agent run(s) as 'aborted'.`,
-        );
-      }
-    } catch (err) {
-      this.logger.warn(
-        `Startup sweep of dangling runs failed: ${
-          err instanceof Error ? err.message : 'unknown error'
-        }`,
-      );
-    }
-  }
-
-  /**
-   * F2 (DECISION A): autonomous runs are SINGLE-INSTANCE-ONLY in phase 1. An
-   * explicit Stop, and the in-memory AbortController that backs it, are
-   * process-local: a Stop only aborts the live turn if it lands on the SAME
-   * replica that owns the run (it still stamps `stop_requested_at` cross-instance,
-   * but nothing reads that flag during an active run yet). Cross-instance pub/sub
-   * stop is phase 2. So if the deployment is horizontally scaled, warn loudly at
-   * startup that a Stop may not reach a run executing on another replica.
-   *
-   * DETECTION: this codebase always wires the socket.io Redis adapter (REDIS_URL
-   * is mandatory), so the adapter alone is NOT a horizontal-scaling signal. The
-   * authoritative signal the codebase has is `CLOUD=true` (EnvironmentService
-   * .isCloud()), the Docmost-cloud multi-replica deployment. We warn whenever that
-   * is set, because any workspace could enable settings.ai.autonomousRuns. A
-   * self-hosted operator running multiple replicas behind a load balancer is also
-   * multi-instance; the deploy docs (.env.example / AGENTS.md) spell out the
-   * single-instance constraint for that case.
-   */
-  private warnIfMultiInstance(): void {
-    if (this.environment.isCloud()) {
-      this.logger.warn(
-        'Autonomous agent runs (settings.ai.autonomousRuns) are SINGLE-INSTANCE-ONLY ' +
-          'in phase 1: a horizontally-scaled deployment was detected (CLOUD=true). ' +
-          'An explicit Stop only aborts a run executing on the same replica that owns ' +
-          'it (cross-instance Stop is not yet reliable — phase 2). Run a single ' +
-          'instance if you enable autonomousRuns, or keep the flag off.',
-      );
-    }
-  }
-
-  /**
-   * Start a run for a turn: insert the run row (status 'running', startedAt now),
-   * register a fresh AbortController for it, and return a {@link RunHandle} whose
-   * `signal` the agent loop uses. The DB partial unique index guarantees at most
-   * one active run per chat — a second concurrent start on the same chat REJECTS
-   * at the insert (a 23505 on {@link ONE_ACTIVE_RUN_PER_CHAT_INDEX}). That
-   * rejection is the AUTHORITATIVE race gate: it is surfaced as a distinct
-   * {@link RunAlreadyActiveError} (NOT swallowed), so the caller turns it into a
-   * 409 and never streams an untracked turn. The controller is registered AFTER a
-   * successful insert so a rejected start leaks nothing.
-   */
-  async beginRun(args: {
-    chatId: string;
-    workspaceId: string;
-    userId: string;
-    trigger?: string;
-  }): Promise<RunHandle> {
-    let run: AiChatRun;
-    try {
-      run = await this.runRepo.insert({
-        chatId: args.chatId,
-        workspaceId: args.workspaceId,
-        createdBy: args.userId,
-        trigger: args.trigger ?? 'user',
-        status: 'running',
-        startedAt: new Date(),
-      });
-    } catch (err) {
-      // The race backstop: a concurrent turn already holds this chat's single
-      // active slot, so the partial unique index rejected our insert. Surface a
-      // distinct signal — the caller MUST reject this turn (409), not run it
-      // untracked. Any OTHER error propagates unchanged.
-      if (
-        isUniqueViolation(err) &&
-        violatedConstraint(err) === ONE_ACTIVE_RUN_PER_CHAT_INDEX
-      ) {
-        throw new RunAlreadyActiveError(args.chatId);
-      }
-      throw err;
-    }
-    const controller = new AbortController();
-    this.active.set(run.id, {
-      controller,
-      chatId: args.chatId,
-      workspaceId: args.workspaceId,
-    });
-    return { runId: run.id, signal: controller.signal };
-  }
-
-  /** Link the assistant message (the #183 projection) to its run. Best-effort. */
-  async linkAssistantMessage(
-    runId: string,
-    workspaceId: string,
-    assistantMessageId: string,
-  ): Promise<void> {
-    try {
-      await this.runRepo.update(runId, workspaceId, { assistantMessageId });
-    } catch (err) {
-      this.logger.warn(
-        `Failed to link assistant message to run ${runId}: ${
-          err instanceof Error ? err.message : 'unknown error'
-        }`,
-      );
-    }
-  }
-
-  /** Persist progress: bump the run's finished-step count. Best-effort (never
-   *  blocks or breaks the stream). */
-  async recordStep(
-    runId: string,
-    workspaceId: string,
-    stepCount: number,
-  ): Promise<void> {
-    try {
-      await this.runRepo.update(runId, workspaceId, { stepCount });
-    } catch (err) {
-      this.logger.warn(
-        `Failed to record step for run ${runId}: ${
-          err instanceof Error ? err.message : 'unknown error'
-        }`,
-      );
-    }
-  }
-
-  /**
-   * Finalize a run to its terminal status (succeeded / failed / aborted),
-   * stamping finishedAt + any error. Best-effort, but ROBUST against a transient
-   * terminal-write failure (F6) AND atomically safe against a concurrent settle.
-   *
-   * ATOMIC ONCE-CLAIM (the gate must close in ONE synchronous tick): two
-   * finalizeRun calls for the SAME run can race — the documented real path is
-   * AiChatService.stream's safety-net catch settling the turn to 'error' while a
-   * streamText terminal callback (onFinish/onAbort/onError) ALSO settles it. The
-   * `settled.has` check alone is NOT a gate: it is read BEFORE the awaited UPDATE,
-   * so two callers can both see `false` and both write the row (last-write-wins
-   * clobbers the real terminal status, and the bounded retry only widens that
-   * window). The claim therefore happens via `active.delete`, a SYNCHRONOUS
-   * check-and-clear with NO await between the gate and the entry removal: the
-   * second concurrent caller finds the entry already gone and returns in the same
-   * tick, before any UPDATE. The transition "nobody is finalizing" -> "I am
-   * finalizing" is thus a single atomic step.
-   *
-   * ORDER MATTERS (F6): once we own the claim, the terminal UPDATE happens FIRST;
-   * only once it SUCCEEDS do we record the run as settled. If the UPDATE fails on
-   * every bounded attempt we RESTORE the in-memory entry, leave the run UNsettled,
-   * and emit an ERROR signal that the row is left non-terminal 'running' (which
-   * would 409 every future turn in the chat until recovery). An in-process retry
-   * by a LATER settle is only POSSIBLE, never guaranteed: it needs (a) the entry
-   * to have been restored at the give-up path AND (b) a fresh settler to arrive
-   * AFTER that restore. A concurrent settler that arrives DURING the retry window
-   * — while the entry is deleted for backoff and not yet restored — is consumed at
-   * the synchronous `active.delete` claim (it finds nothing to delete and returns
-   * a no-op), so it does NOT become an in-process retrier. The NO-streamText path
-   * (the turn threw before streamText was wired, so ONLY the safety-net ever
-   * settles) likewise has no second in-process settler at all. The UNCONDITIONAL
-   * backstop in every case is the boot sweep on the next restart (phase 1 has no
-   * periodic in-process sweep); the retained entry is bounded (cleared on restart)
-   * and harmless meanwhile.
-   *
-   * IDEMPOTENT on SUCCESS (#184 review): the terminal write happens AT MOST ONCE
-   * per run. After a successful write the once-gate keys off {@link settled} (the
-   * terminal row already written) so a settle arriving AFTER the entry was already
-   * dropped-and-settled returns early; a settle racing the in-flight write is
-   * stopped earlier still, by the `active.delete` claim. Either way a genuine
-   * double-settle collapses to a single write and a late settle can never clobber
-   * the real terminal status or double-write the row.
-   */
-  async finalizeRun(
-    runId: string,
-    workspaceId: string,
-    turnStatus: TurnTerminalStatus,
-    error?: string,
-  ): Promise<void> {
-    // ---- Atomic once-claim (synchronous; NO await before the gate closes) ----
-    // Already terminally written -> idempotent no-op.
-    if (this.settled.has(runId)) return;
-    // Capture the entry BEFORE the delete so a total-failure path can restore it.
-    const entry = this.active.get(runId);
-    // SYNCHRONOUS check-and-clear: the FIRST caller deletes (claims) the entry;
-    // any concurrent SECOND caller finds nothing to delete and returns HERE, in
-    // the same tick, before any await — so it can never reach the UPDATE.
-    if (!this.active.delete(runId)) return;
-
-    let lastError: unknown;
-    for (
-      let attempt = 1;
-      attempt <= AiChatRunService.FINALIZE_MAX_ATTEMPTS;
-      attempt++
-    ) {
-      try {
-        await this.runRepo.update(runId, workspaceId, {
-          status: mapTurnStatusToRun(turnStatus),
-          finishedAt: new Date(),
-          error: error ?? null,
-        });
-        // Terminal write landed: arm the once-gate. The entry is already gone
-        // (claimed above); we do NOT restore it. The slot is now free.
-        this.settled.add(runId);
-        return;
-      } catch (err) {
-        lastError = err;
-        this.logger.warn(
-          `Failed to finalize run ${runId} (attempt ${attempt}/${
-            AiChatRunService.FINALIZE_MAX_ATTEMPTS
-          }): ${err instanceof Error ? err.message : 'unknown error'}`,
-        );
-        if (attempt < AiChatRunService.FINALIZE_MAX_ATTEMPTS) {
-          await this.delay(AiChatRunService.FINALIZE_RETRY_BASE_MS * attempt);
-        }
-      }
-    }
-    // Every attempt failed: this is a give-up, materially worse than a per-attempt
-    // blip — the row is left NON-TERMINAL ('running'), so emit ONE explicit,
-    // greppable ERROR so an operator can tell "survived a blip" from "gave up, run
-    // held in memory until recovery" (the last warn alone says only "attempt 3/3").
-    this.logger.error(
-      `Run ${runId} (chat ${entry?.chatId ?? 'unknown'}) left NON-TERMINAL ` +
-        `('running'): terminal write failed after ${
-          AiChatRunService.FINALIZE_MAX_ATTEMPTS
-        } attempts; entry retained in memory, recovery deferred to next settle / ` +
-        `boot sweep`,
-      lastError,
-    );
-    // RESTORE the claimed entry (and leave the run UNsettled) so a LATER settle
-    // that arrives AFTER this restore MAY retry the terminal write — but that
-    // in-process retry is NOT guaranteed (a concurrent settler caught in the retry
-    // window above is consumed at the `active.delete` claim, and the no-streamText
-    // path has no second settler at all). The UNCONDITIONAL backstop in every case
-    // is the boot sweep on the next restart; the restored entry is bounded and
-    // cleared on restart.
-    if (entry) this.active.set(runId, entry);
-  }
-
-  /** Small async backoff between terminal-write retries (F6). Isolated so it is
-   *  trivial to stub/fake-time in tests. */
-  private delay(ms: number): Promise<void> {
-    return new Promise((resolve) => setTimeout(resolve, ms));
-  }
-
-  /**
-   * Request an EXPLICIT stop of a run (the user pressed Stop). This is the ONLY
-   * thing that aborts a run — distinct from a browser disconnect, which leaves
-   * the run going. Records `stop_requested_at` on the row (only while active) and
-   * aborts the in-process controller if this replica owns the run. Returns true
-   * when a stop took effect (row marked and/or controller aborted), false when
-   * there was nothing active to stop.
-   */
-  async requestStop(runId: string, workspaceId: string): Promise<boolean> {
-    const marked = await this.runRepo.markStopRequested(runId, workspaceId);
-    const entry = this.active.get(runId);
-    if (entry) {
-      // Abort the live turn -> streamText onAbort fires -> the partial is
-      // persisted (#183) and finalizeRun settles the row as 'aborted'.
-      entry.controller.abort();
-    }
-    return Boolean(marked) || Boolean(entry);
-  }
-
-  /** Latest persisted run for a chat — the reconnect target (an in-flight or
-   *  finished run). Pure read-through to the repo. */
-  getLatestForChat(
-    chatId: string,
-    workspaceId: string,
-  ): Promise<AiChatRun | undefined> {
-    return this.runRepo.findLatestByChat(chatId, workspaceId);
-  }
-
-  /** Fetch a run by id (workspace-scoped). Used to resolve + ownership-check an
-   *  explicit stop targeting a runId. */
-  getRun(runId: string, workspaceId: string): Promise<AiChatRun | undefined> {
-    return this.runRepo.findById(runId, workspaceId);
-  }
-
-  /** The active run on a chat, if any (used to reject a concurrent start with a
-   *  clean 409 before committing to the stream). */
-  getActiveForChat(
-    chatId: string,
-    workspaceId: string,
-  ): Promise<AiChatRun | undefined> {
-    return this.runRepo.findActiveByChat(chatId, workspaceId);
-  }
-
-  /** Test/diagnostic seam: whether this replica is holding a live controller for
-   *  the run. */
-  isLocallyActive(runId: string): boolean {
-    return this.active.has(runId);
-  }
-}

apps/server/src/core/ai-chat/ai-chat.controller.bound-chat.spec.ts

@@ -19,7 +19,6 @@ describe('AiChatController.boundChat', () => {
     };
     const controller = new AiChatController(
       {} as never,
-      {} as never, // aiChatRunService
       aiChatRepo as never,
       {} as never,
       {} as never,

apps/server/src/core/ai-chat/ai-chat.controller.export.spec.ts

@@ -53,7 +53,6 @@ describe('AiChatController.export', () => {
     };
     const controller = new AiChatController(
       {} as never,
-      {} as never, // aiChatRunService
       aiChatRepo as never,
       aiChatMessageRepo as never,
       {} as never,

apps/server/src/core/ai-chat/ai-chat.controller.run.spec.ts

@@ -1,163 +0,0 @@
-import { BadRequestException, ForbiddenException } from '@nestjs/common';
-import { AiChatController } from './ai-chat.controller';
-import type { User, Workspace } from '@docmost/db/types/entity.types';
-
-/**
- * Wiring spec for the #184 run-reconnect / run-stop endpoints
- * (`POST /ai-chat/run` and `POST /ai-chat/stop`). Both are OWNER-gated via
- * assertOwnedChat (the requesting user must own the chat) and NOT flag-gated.
- * Exercised with hand-rolled mocks — no Nest graph, no DB. The controller's
- * constructor order is (aiChatService, aiChatRunService, aiChatRepo,
- * aiChatMessageRepo, aiTranscription).
- */
-describe('AiChatController run endpoints (#184)', () => {
-  const user = { id: 'u1' } as User;
-  const workspace = { id: 'ws1' } as Workspace;
-
-  function makeController(opts: {
-    chat?: unknown; // what aiChatRepo.findById returns (owner-gate)
-    run?: unknown; // getLatestForChat / getRun result
-    activeRun?: unknown; // getActiveForChat result
-    message?: unknown; // aiChatMessageRepo.findById result
-    stopped?: boolean; // requestStop result
-  }) {
-    const aiChatRunService = {
-      getLatestForChat: jest.fn().mockResolvedValue(opts.run),
-      getRun: jest.fn().mockResolvedValue(opts.run),
-      getActiveForChat: jest.fn().mockResolvedValue(opts.activeRun),
-      requestStop: jest.fn().mockResolvedValue(opts.stopped ?? false),
-    };
-    const aiChatRepo = {
-      findById: jest.fn().mockResolvedValue(opts.chat),
-    };
-    const aiChatMessageRepo = {
-      findById: jest.fn().mockResolvedValue(opts.message),
-    };
-    const controller = new AiChatController(
-      {} as never, // aiChatService
-      aiChatRunService as never,
-      aiChatRepo as never,
-      aiChatMessageRepo as never,
-      {} as never, // aiTranscription
-    );
-    return { controller, aiChatRunService, aiChatRepo, aiChatMessageRepo };
-  }
-
-  describe('POST /ai-chat/run (getRun)', () => {
-    it('owner-gates: a chat the user does not own throws ForbiddenException', async () => {
-      const { controller, aiChatRunService } = makeController({
-        chat: { id: 'c1', creatorId: 'someone-else' },
-      });
-      await expect(
-        controller.getRun({ chatId: 'c1' }, user, workspace),
-      ).rejects.toBeInstanceOf(ForbiddenException);
-      // It must NOT reach the run lookup once the owner-gate fails.
-      expect(aiChatRunService.getLatestForChat).not.toHaveBeenCalled();
-    });
-
-    it('returns { run: null, message: null } when the chat has never had a run', async () => {
-      const { controller, aiChatRunService } = makeController({
-        chat: { id: 'c1', creatorId: 'u1' },
-        run: undefined,
-      });
-      const res = await controller.getRun({ chatId: 'c1' }, user, workspace);
-      expect(res).toEqual({ run: null, message: null });
-      expect(aiChatRunService.getLatestForChat).toHaveBeenCalledWith(
-        'c1',
-        'ws1',
-      );
-    });
-
-    it('returns the run and its projected assistant message', async () => {
-      const run = { id: 'run-1', chatId: 'c1', assistantMessageId: 'm1' };
-      const message = { id: 'm1', role: 'assistant' };
-      const { controller, aiChatMessageRepo } = makeController({
-        chat: { id: 'c1', creatorId: 'u1' },
-        run,
-        message,
-      });
-      const res = await controller.getRun({ chatId: 'c1' }, user, workspace);
-      expect(res).toEqual({ run, message });
-      expect(aiChatMessageRepo.findById).toHaveBeenCalledWith('m1', 'ws1');
-    });
-
-    it('returns message: null when the run has no linked assistant message', async () => {
-      const run = { id: 'run-1', chatId: 'c1', assistantMessageId: null };
-      const { controller, aiChatMessageRepo } = makeController({
-        chat: { id: 'c1', creatorId: 'u1' },
-        run,
-      });
-      const res = await controller.getRun({ chatId: 'c1' }, user, workspace);
-      expect(res).toEqual({ run, message: null });
-      expect(aiChatMessageRepo.findById).not.toHaveBeenCalled();
-    });
-  });
-
-  describe('POST /ai-chat/stop (stopRun)', () => {
-    it('throws BadRequestException when neither runId nor chatId is given', async () => {
-      const { controller } = makeController({});
-      await expect(
-        controller.stopRun({}, user, workspace),
-      ).rejects.toBeInstanceOf(BadRequestException);
-    });
-
-    it('stops by runId: owner-gates via the run’s chat, then requests the stop', async () => {
-      const { controller, aiChatRunService, aiChatRepo } = makeController({
-        run: { id: 'run-1', chatId: 'c1' },
-        chat: { id: 'c1', creatorId: 'u1' },
-        stopped: true,
-      });
-      const res = await controller.stopRun({ runId: 'run-1' }, user, workspace);
-      expect(res).toEqual({ stopped: true });
-      expect(aiChatRunService.getRun).toHaveBeenCalledWith('run-1', 'ws1');
-      expect(aiChatRepo.findById).toHaveBeenCalledWith('c1', 'ws1');
-      expect(aiChatRunService.requestStop).toHaveBeenCalledWith('run-1', 'ws1');
-    });
-
-    it('stops by runId: a foreign run’s chat throws ForbiddenException (no stop)', async () => {
-      const { controller, aiChatRunService } = makeController({
-        run: { id: 'run-1', chatId: 'c1' },
-        chat: { id: 'c1', creatorId: 'someone-else' },
-      });
-      await expect(
-        controller.stopRun({ runId: 'run-1' }, user, workspace),
-      ).rejects.toBeInstanceOf(ForbiddenException);
-      expect(aiChatRunService.requestStop).not.toHaveBeenCalled();
-    });
-
-    it('stops by runId: an unknown run reports { stopped: false }', async () => {
-      const { controller, aiChatRunService } = makeController({
-        run: undefined,
-      });
-      const res = await controller.stopRun({ runId: 'gone' }, user, workspace);
-      expect(res).toEqual({ stopped: false });
-      expect(aiChatRunService.requestStop).not.toHaveBeenCalled();
-    });
-
-    it('stops by chatId: owner-gates, resolves the active run, requests the stop', async () => {
-      const { controller, aiChatRunService, aiChatRepo } = makeController({
-        chat: { id: 'c1', creatorId: 'u1' },
-        activeRun: { id: 'run-9' },
-        stopped: true,
-      });
-      const res = await controller.stopRun({ chatId: 'c1' }, user, workspace);
-      expect(res).toEqual({ stopped: true });
-      expect(aiChatRepo.findById).toHaveBeenCalledWith('c1', 'ws1');
-      expect(aiChatRunService.getActiveForChat).toHaveBeenCalledWith(
-        'c1',
-        'ws1',
-      );
-      expect(aiChatRunService.requestStop).toHaveBeenCalledWith('run-9', 'ws1');
-    });
-
-    it('stops by chatId: reports { stopped: false } when no run is active', async () => {
-      const { controller, aiChatRunService } = makeController({
-        chat: { id: 'c1', creatorId: 'u1' },
-        activeRun: undefined,
-      });
-      const res = await controller.stopRun({ chatId: 'c1' }, user, workspace);
-      expect(res).toEqual({ stopped: false });
-      expect(aiChatRunService.requestStop).not.toHaveBeenCalled();
-    });
-  });
-});

apps/server/src/core/ai-chat/ai-chat.controller.ts

@@ -1,7 +1,6 @@
 import {
   BadRequestException,
   Body,
-  ConflictException,
   Controller,
   ForbiddenException,
   HttpCode,
@@ -21,25 +20,14 @@ import { JwtAuthGuard } from '../../common/guards/jwt-auth.guard';
 import { AuthUser } from '../../common/decorators/auth-user.decorator';
 import { AuthWorkspace } from '../../common/decorators/auth-workspace.decorator';
 import { SkipTransform } from '../../common/decorators/skip-transform.decorator';
-import {
-  AiChat,
-  AiChatMessage,
-  AiChatRun,
-  User,
-  Workspace,
-} from '@docmost/db/types/entity.types';
+import { AiChat, User, Workspace } from '@docmost/db/types/entity.types';
 import { PaginationOptions } from '@docmost/db/pagination/pagination-options';
 import { AiChatRepo } from '@docmost/db/repos/ai-chat/ai-chat.repo';
 import { AiChatMessageRepo } from '@docmost/db/repos/ai-chat/ai-chat-message.repo';
 import { UserThrottlerGuard } from '../../integrations/throttle/user-throttler.guard';
 import { AI_CHAT_THROTTLER } from '../../integrations/throttle/throttler-names';
 import { FileInterceptor } from '../../common/interceptors/file.interceptor';
-import {
-  AiChatRunHooks,
-  AiChatService,
-  AiChatStreamBody,
-} from './ai-chat.service';
-import { AiChatRunService } from './ai-chat-run.service';
+import { AiChatService, AiChatStreamBody } from './ai-chat.service';
 import { AiTranscriptionService } from './ai-transcription.service';
 import {
   BoundChatDto,
@@ -47,9 +35,7 @@ import {
   ExportChatDto,
   GeneratePageTitleDto,
   GetChatMessagesDto,
-  GetRunDto,
   RenameChatDto,
-  StopRunDto,
 } from './dto/ai-chat.dto';
 import { describeProviderError } from '../../integrations/ai/ai-error.util';
 import { buildChatMarkdown } from './chat-markdown.util';
@@ -66,7 +52,6 @@ export class AiChatController {
 
   constructor(
     private readonly aiChatService: AiChatService,
-    private readonly aiChatRunService: AiChatRunService,
     private readonly aiChatRepo: AiChatRepo,
     private readonly aiChatMessageRepo: AiChatMessageRepo,
     private readonly aiTranscription: AiTranscriptionService,
@@ -152,75 +137,6 @@ export class AiChatController {
     return { markdown };
   }
 
-  /**
-   * Reconnect to the latest run of a chat (#184 phase 1). Returns the run's
-   * persisted lifecycle state ({ status, error, stepCount, timings, ... }) plus
-   * the assistant message it projects (the partial/final output) — the DB is the
-   * source of truth, so this works for an in-flight run (the browser dropped, the
-   * run kept going) and a finished one alike. Owner-gated via assertOwnedChat.
-   * `{ run: null }` when the chat has never had a run.
-   */
-  @HttpCode(HttpStatus.OK)
-  @Post('run')
-  async getRun(
-    @Body() dto: GetRunDto,
-    @AuthUser() user: User,
-    @AuthWorkspace() workspace: Workspace,
-  ): Promise<{ run: AiChatRun | null; message: AiChatMessage | null }> {
-    await this.assertOwnedChat(dto.chatId, user, workspace);
-    const run = await this.aiChatRunService.getLatestForChat(
-      dto.chatId,
-      workspace.id,
-    );
-    if (!run) return { run: null, message: null };
-    const message = run.assistantMessageId
-      ? await this.aiChatMessageRepo.findById(
-          run.assistantMessageId,
-          workspace.id,
-        )
-      : undefined;
-    return { run, message: message ?? null };
-  }
-
-  /**
-   * Explicitly STOP an agent run (#184 phase 1) — the user pressed Stop. This is
-   * the ONLY thing that ends a detached run; a browser disconnect deliberately
-   * does not. Target by `runId` (from the streamed start metadata) or by `chatId`
-   * (stop whatever run is active on it). Owner-gated. Returns
-   * `{ stopped }` — false when there was nothing active to stop.
-   */
-  @HttpCode(HttpStatus.OK)
-  @Post('stop')
-  async stopRun(
-    @Body() dto: StopRunDto,
-    @AuthUser() user: User,
-    @AuthWorkspace() workspace: Workspace,
-  ): Promise<{ stopped: boolean }> {
-    let runId = dto.runId;
-    if (!runId && !dto.chatId) {
-      throw new BadRequestException('runId or chatId is required');
-    }
-    if (runId) {
-      // Resolve the run to its chat and owner-gate via that chat.
-      const run = await this.aiChatRunService.getRun(runId, workspace.id);
-      if (!run) return { stopped: false };
-      await this.assertOwnedChat(run.chatId, user, workspace);
-    } else {
-      await this.assertOwnedChat(dto.chatId!, user, workspace);
-      const active = await this.aiChatRunService.getActiveForChat(
-        dto.chatId!,
-        workspace.id,
-      );
-      if (!active) return { stopped: false };
-      runId = active.id;
-    }
-    const stopped = await this.aiChatRunService.requestStop(
-      runId,
-      workspace.id,
-    );
-    return { stopped };
-  }
-
   /** Rename a chat. */
   @HttpCode(HttpStatus.OK)
   @Post('rename')
@@ -272,20 +188,11 @@ export class AiChatController {
     @AuthWorkspace() workspace: Workspace,
   ): Promise<void> {
     // A7 gate: the workspace must have AI chat explicitly enabled.
-    const settings = (workspace.settings ?? {}) as {
-      ai?: { chat?: boolean; autonomousRuns?: boolean };
-    };
+    const settings = (workspace.settings ?? {}) as { ai?: { chat?: boolean } };
     if (settings.ai?.chat !== true) {
       throw new ForbiddenException('AI chat is disabled');
     }
 
-    // #184 phase 1 flag: when ON, the turn becomes a detached, durable RUN — its
-    // lifecycle is tracked in ai_chat_runs, a browser disconnect no longer aborts
-    // it, and only an explicit /ai-chat/stop ends it. When OFF (the default) the
-    // turn is socket-bound exactly as before, so existing deployments are
-    // unaffected.
-    const autonomousRuns = settings.ai?.autonomousRuns === true;
-
     const sessionId = (req.raw as { sessionId?: string }).sessionId;
     if (!sessionId) {
       // The chat requires an interactive session to mint loopback tokens
@@ -309,58 +216,6 @@ export class AiChatController {
     // HttpException) instead of breaking mid-stream.
     const model = await this.aiChatService.getChatModel(workspace.id, role);
 
-    // #184: one active run per chat. For an EXISTING chat reject a concurrent
-    // start with a clean 409 BEFORE hijack (the common double-submit / second-tab
-    // case), so the user gets JSON, not a mid-stream error. A brand-new chat
-    // (no chatId) cannot have a prior run, and the DB partial unique index is the
-    // backstop against any race that slips past this check.
-    if (autonomousRuns && body.chatId) {
-      const active = await this.aiChatRunService.getActiveForChat(
-        body.chatId,
-        workspace.id,
-      );
-      if (active) {
-        throw new ConflictException({
-          message: 'An agent run is already in progress for this chat',
-          code: 'A_RUN_ALREADY_ACTIVE',
-        });
-      }
-    }
-
-    // Run-lifecycle hooks (#184), only when the flag is on. They wrap the turn in
-    // a durable run whose abort is governed by the run (explicit stop), persist
-    // its progress, and settle its terminal status — see AiChatRunService.
-    const runHooks: AiChatRunHooks | undefined = autonomousRuns
-      ? {
-          begin: (chatId) =>
-            this.aiChatRunService.beginRun({
-              chatId,
-              workspaceId: workspace.id,
-              userId: user.id,
-              trigger: 'user',
-            }),
-          onAssistantSeeded: (runId, messageId) =>
-            this.aiChatRunService.linkAssistantMessage(
-              runId,
-              workspace.id,
-              messageId,
-            ),
-          onStep: (runId, stepCount) =>
-            void this.aiChatRunService.recordStep(
-              runId,
-              workspace.id,
-              stepCount,
-            ),
-          onSettled: (runId, status, error) =>
-            this.aiChatRunService.finalizeRun(
-              runId,
-              workspace.id,
-              status,
-              error,
-            ),
-        }
-      : undefined;
-
     // Abort the agent loop when the client disconnects. `close` also fires on
     // normal completion, so only abort when the response has not finished
     // writing (a genuine disconnect). `once` fires at most once and self-removes;
@@ -375,44 +230,18 @@ export class AiChatController {
       // A genuine disconnect leaves the response unfinished (unlike a normal
       // completion, which also fires `close`). Such a drop — e.g. a reverse
       // proxy cutting the SSE mid-answer — is otherwise invisible server-side,
-      // so log it here.
+      // so log it here before aborting the agent loop.
       if (!res.raw.writableEnded) {
-        if (autonomousRuns) {
-          // #184: the turn is a DETACHED run. A disconnect must NOT abort it —
-          // the run keeps executing and persisting server-side; the client
-          // reconnects via /ai-chat/run (or re-stops via /ai-chat/stop). Log only.
-          this.logger.log(
-            `AI chat stream: client disconnected; run continues server-side ` +
-              `(elapsed=${Date.now() - reqStartedAt}ms since request received)`,
-          );
-        } else {
-          this.logger.warn(
-            `AI chat stream: client disconnected before completion; aborting turn ` +
-              `(elapsed=${Date.now() - reqStartedAt}ms since request received)`,
-          );
-          controller.abort();
-        }
+        this.logger.warn(
+          `AI chat stream: client disconnected before completion; aborting turn ` +
+            `(elapsed=${Date.now() - reqStartedAt}ms since request received)`,
+        );
+        controller.abort();
       }
     };
     req.raw.once('close', onClose);
     res.raw.once('finish', () => req.raw.off('close', onClose));
 
-    // #184: in detached mode the turn is NOT aborted on disconnect, so the SDK's
-    // pipe keeps writing to a socket the client may have dropped — for the rest of
-    // the (continuing) run. A write to the dead socket can emit an 'error' on the
-    // raw response; without a listener that surfaces as an unhandled error event.
-    // Swallow it (the run continues server-side regardless). Legacy mode aborts on
-    // disconnect, so it does not need this and keeps its exact prior behavior.
-    if (autonomousRuns) {
-      res.raw.on('error', (err) => {
-        this.logger.debug(
-          `AI chat detached stream: post-disconnect socket error swallowed: ${
-            err instanceof Error ? err.message : String(err)
-          }`,
-        );
-      });
-    }
-
     // Commit to streaming: hijack so Fastify stops managing the response and
     // the AI SDK can write the UI-message stream directly to the Node socket.
     res.hijack();
@@ -427,32 +256,15 @@ export class AiChatController {
         signal: controller.signal,
         model,
         role,
-        // #184: present only when the flag is on; wraps the turn in a durable run.
-        runHooks,
       });
     } catch (err) {
-      // Any failure AFTER hijack can no longer go through Nest's exception
-      // filter, so emit the error on the raw socket if nothing has been written
-      // yet. The lost-the-race 409 (RunAlreadyActiveError -> ConflictException)
-      // is raised by stream() BEFORE it writes a byte, so headers are still
-      // unsent here: honor the HttpException's real status + body (a clean 409),
-      // not a blanket 500. Everything else stays a 500.
-      const isHttp = err instanceof HttpException;
-      if (!isHttp) {
-        this.logger.error('AI chat stream failed', err as Error);
-      }
+      // Any failure AFTER hijack can no longer send a clean JSON error, so emit
+      // a minimal error on the raw socket if nothing has been written yet.
+      this.logger.error('AI chat stream failed', err as Error);
       if (!res.raw.headersSent) {
-        const status = isHttp ? err.getStatus() : 500;
-        const payload = isHttp
-          ? err.getResponse()
-          : { error: 'Internal server error' };
-        res.raw.statusCode = status;
+        res.raw.statusCode = 500;
         res.raw.setHeader('Content-Type', 'application/json');
-        res.raw.end(
-          JSON.stringify(
-            typeof payload === 'string' ? { message: payload } : payload,
-          ),
-        );
+        res.raw.end(JSON.stringify({ error: 'Internal server error' }));
       } else if (!res.raw.writableEnded) {
         res.raw.end();
       }

apps/server/src/core/ai-chat/ai-chat.generate-page-title.spec.ts

@@ -57,7 +57,6 @@ describe('AiChatController.generatePageTitle', () => {
     const aiChatService = { generatePageTitle: generate };
     const controller = new AiChatController(
       aiChatService as never,
-      {} as never, // aiChatRunService
       {} as never,
       {} as never,
       {} as never,

apps/server/src/core/ai-chat/ai-chat.module.ts

@@ -3,7 +3,6 @@ import { AiModule } from '../../integrations/ai/ai.module';
 import { TokenModule } from '../auth/token.module';
 import { AiChatController } from './ai-chat.controller';
 import { AiChatService } from './ai-chat.service';
-import { AiChatRunService } from './ai-chat-run.service';
 import { AiTranscriptionService } from './ai-transcription.service';
 import { AiChatToolsService } from './tools/ai-chat-tools.service';
 import { EmbeddingModule } from './embedding/embedding.module';
@@ -43,7 +42,6 @@ import { PublicShareChatToolsService } from './tools/public-share-chat-tools.ser
   controllers: [AiChatController, PublicShareChatController],
   providers: [
     AiChatService,
-    AiChatRunService,
     AiTranscriptionService,
     AiChatToolsService,
     PublicShareChatService,

apps/server/src/core/ai-chat/ai-chat.service.lifecycle.spec.ts

@@ -1,7 +1,5 @@
 import { Logger } from '@nestjs/common';
-import { AiChatService, AiChatRunHooks } from './ai-chat.service';
-import { AiChatRunService } from './ai-chat-run.service';
-import type { User, Workspace } from '@docmost/db/types/entity.types';
+import { AiChatService } from './ai-chat.service';
 
 /**
  * Lifecycle unit tests for AiChatService.onModuleInit (#183 crash-recovery
@@ -61,97 +59,3 @@ describe('AiChatService.onModuleInit (startup sweep)', () => {
     expect(String(warnSpy.mock.calls[0][0])).toContain('db unavailable');
   });
 });
-
-/**
- * #184 CRITICAL run-lifecycle safety net (review fix). A transient failure
- * AFTER a successful beginRun but BEFORE streamText's terminal callbacks own the
- * lifecycle must STILL settle the run — otherwise the run row is stuck 'running'
- * forever (sweepRunning only runs at startup) and the partial unique index + the
- * controller pre-check 409 every future turn in that chat until a restart. Here
- * we model the very first bare await after beginRun (the user-message insert)
- * throwing, wiring the run hooks to a REAL AiChatRunService (mock repo) exactly
- * as the controller does, and assert the run is settled to 'error' and its
- * in-memory entry dropped (so a follow-up turn would NOT be 409'd).
- */
-describe('AiChatService.stream run-lifecycle safety net (#184)', () => {
-  const user = { id: 'u1' } as User;
-  const workspace = { id: 'ws1' } as Workspace;
-
-  afterEach(() => jest.restoreAllMocks());
-
-  it('an exception after beginRun settles the run to error and drops the in-memory entry', async () => {
-    jest.spyOn(Logger.prototype, 'error').mockImplementation(() => undefined);
-
-    // Real run service over a mock repo, so finalizeRun's in-memory bookkeeping
-    // (active.delete) is exercised for real.
-    const runRepo = {
-      insert: jest.fn().mockResolvedValue({ id: 'run-1', status: 'running' }),
-      update: jest.fn().mockResolvedValue({ id: 'run-1' }),
-    };
-    const runService = new AiChatRunService(runRepo as never, { isCloud: () => false } as never);
-
-    // The user-message insert (the first bare await after beginRun) throws.
-    const aiChatMessageRepo = {
-      insert: jest.fn().mockRejectedValue(new Error('insert boom')),
-    };
-    const aiChatRepo = {
-      // Existing chat -> chatId stays, no new-chat insert path.
-      findById: jest.fn().mockResolvedValue({ id: 'chat-1', creatorId: 'u1' }),
-    };
-
-    const service = new AiChatService(
-      {} as never, // ai
-      aiChatRepo as never,
-      aiChatMessageRepo as never,
-      {} as never, // aiSettings
-      {} as never, // tools
-      {} as never, // mcpClients
-      {} as never, // aiAgentRoleRepo
-      {} as never, // pageRepo
-      {} as never, // pageAccess
-    );
-
-    const runHooks: AiChatRunHooks = {
-      begin: (chatId) =>
-        runService.beginRun({
-          chatId,
-          workspaceId: workspace.id,
-          userId: user.id,
-          trigger: 'user',
-        }),
-      onSettled: (runId, status, error) =>
-        runService.finalizeRun(runId, workspace.id, status, error),
-    };
-
-    await expect(
-      service.stream({
-        user,
-        workspace,
-        sessionId: 'sess',
-        body: {
-          chatId: 'chat-1',
-          messages: [
-            { id: 'm', role: 'user', parts: [{ type: 'text', text: 'hi' }] },
-          ],
-        },
-        res: {} as never,
-        signal: new AbortController().signal,
-        model: {} as never,
-        role: null,
-        runHooks,
-      }),
-    ).rejects.toThrow('insert boom');
-
-    // The run was begun...
-    expect(runRepo.insert).toHaveBeenCalledTimes(1);
-    // ...then settled to a terminal FAILED status by the safety net...
-    expect(runRepo.update).toHaveBeenCalledTimes(1);
-    expect(runRepo.update).toHaveBeenCalledWith(
-      'run-1',
-      'ws1',
-      expect.objectContaining({ status: 'failed' }),
-    );
-    // ...and the in-memory entry is gone, so a follow-up turn is NOT 409'd.
-    expect(runService.isLocallyActive('run-1')).toBe(false);
-  });
-});

apps/server/src/core/ai-chat/ai-chat.service.run-race.spec.ts

@@ -1,337 +0,0 @@
-import { ConflictException, Logger } from '@nestjs/common';
-
-// Mock the AI SDK so we can PROVE no provider call is made for the turn we are
-// about to reject. The race rejection happens at runHooks.begin(), long before
-// any streamText/generateText, so these never resolve a real model.
-jest.mock('ai', () => ({
-  streamText: jest.fn(),
-  generateText: jest.fn(),
-  convertToModelMessages: jest.fn(() => []),
-  stepCountIs: jest.fn(() => () => false),
-}));
-
-import { streamText, generateText } from 'ai';
-import { AiChatService } from './ai-chat.service';
-import { RunAlreadyActiveError } from './ai-chat-run.service';
-
-/**
- * Race-closure coverage for the "one active run per chat" guard (#184).
- *
- * THE BUG: two simultaneous POST /ai-chat/stream on the same chat both pass the
- * controller's cheap pre-check (TOCTOU), so the loser's run-row INSERT hits the
- * partial unique index. Previously that 23505 was SWALLOWED and the second turn
- * streamed UNTRACKED (no runId, not stoppable). THE FIX: beginRun surfaces a
- * RunAlreadyActiveError and stream() turns it into a 409 BEFORE any AI call —
- * the second turn never runs.
- */
-describe('AiChatService.stream — concurrent-run race rejection (#184)', () => {
-  const streamTextMock = streamText as unknown as jest.Mock;
-  const generateTextMock = generateText as unknown as jest.Mock;
-
-  beforeEach(() => {
-    streamTextMock.mockReset();
-    generateTextMock.mockReset();
-  });
-
-  // Minimal service whose only reachable deps before begin() are aiChatRepo
-  // (resolve the existing chat) — everything past begin must remain untouched.
-  function makeService(beginImpl: () => Promise<unknown>) {
-    const aiChatMessageRepo = { insert: jest.fn() };
-    const aiChatRepo = {
-      // An existing chat: stream keeps the supplied chatId and skips creation.
-      findById: jest.fn(async () => ({ id: 'chat-1', workspaceId: 'ws-1' })),
-      insert: jest.fn(),
-    };
-    const svc = new AiChatService(
-      {} as never, // ai
-      aiChatRepo as never,
-      aiChatMessageRepo as never,
-      {} as never, // aiSettings
-      {} as never, // tools
-      {} as never, // mcpClients
-      {} as never, // aiAgentRoleRepo
-      {} as never, // pageRepo
-      {} as never, // pageAccess
-    );
-    const begin = jest.fn(beginImpl);
-    return { svc, begin, aiChatRepo, aiChatMessageRepo };
-  }
-
-  const baseArgs = (begin: jest.Mock) => ({
-    user: { id: 'user-1' } as never,
-    workspace: { id: 'ws-1' } as never,
-    sessionId: 'sess-1',
-    body: { chatId: 'chat-1', messages: [] } as never,
-    res: { raw: {} } as never,
-    signal: new AbortController().signal,
-    model: {} as never,
-    role: null,
-    runHooks: {
-      begin,
-      onAssistantSeeded: jest.fn(),
-      onStep: jest.fn(),
-      onSettled: jest.fn(),
-    } as never,
-  });
-
-  it('rejects the racer with a 409 ConflictException BEFORE any AI call, and never persists an untracked turn', async () => {
-    // begin loses the unique-index race -> RunAlreadyActiveError.
-    const { svc, begin, aiChatMessageRepo } = makeService(() => {
-      throw new RunAlreadyActiveError('chat-1');
-    });
-
-    const promise = svc.stream(baseArgs(begin));
-
-    await expect(promise).rejects.toBeInstanceOf(ConflictException);
-    await promise.catch((err: ConflictException) => {
-      expect(err.getStatus()).toBe(409);
-      expect((err.getResponse() as { code?: string }).code).toBe(
-        'A_RUN_ALREADY_ACTIVE',
-      );
-    });
-
-    // The decisive assertions: the rejected racer spent NO tokens and left NO
-    // untracked turn behind.
-    expect(begin).toHaveBeenCalledTimes(1);
-    expect(streamTextMock).not.toHaveBeenCalled();
-    expect(generateTextMock).not.toHaveBeenCalled();
-    expect(aiChatMessageRepo.insert).not.toHaveBeenCalled();
-  });
-});
-
-/**
- * F3 — the LOAD-BEARING run-detach wiring: `effectiveSignal = handle.signal`
- * after runHooks.begin, then `abortSignal: effectiveSignal` passed to streamText.
- * That single line is what makes a run survive a browser disconnect (the agent
- * loop's abort is governed by the RUN's signal, not the socket): a regression to
- * the socket-bound signal would still pass every other test green while silently
- * breaking Stop + durability. These two tests pin the exact signal streamText
- * consumes on both paths.
- */
-describe('AiChatService.stream — abortSignal wiring (#184 F3)', () => {
-  const streamTextMock = streamText as unknown as jest.Mock;
-
-  // A streamText result stub: the post-call drain + pipe are no-ops here; we only
-  // care WHICH abortSignal streamText was handed.
-  function makeStreamResult() {
-    return {
-      consumeStream: jest.fn(),
-      pipeUIMessageStreamToResponse: jest.fn(),
-    };
-  }
-
-  // A raw-response stub sufficient for the post-streamText wiring
-  // (stripStreamingHopByHopHeaders binds writeHead; startSseHeartbeat registers
-  // close/finish listeners; flushHeaders is belt-and-braces).
-  function makeRes() {
-    return {
-      raw: {
-        writeHead: jest.fn(),
-        write: jest.fn(),
-        once: jest.fn(),
-        on: jest.fn(),
-        flushHeaders: jest.fn(),
-        writableEnded: false,
-        destroyed: false,
-      },
-    };
-  }
-
-  // Wire only the deps reached on the way to streamText: resolve the existing
-  // chat, persist the user + seed the assistant row, load (empty) history, the
-  // admin settings, an empty external toolset + Docmost toolset.
-  function makeService() {
-    const aiChatRepo = {
-      findById: jest.fn(async () => ({ id: 'chat-1', workspaceId: 'ws-1' })),
-      insert: jest.fn(),
-    };
-    const aiChatMessageRepo = {
-      insert: jest.fn(async () => ({ id: 'msg-1' })),
-      findAllByChat: jest.fn(async () => []),
-      update: jest.fn(async () => ({ id: 'msg-1' })),
-    };
-    const aiSettings = { resolve: jest.fn(async () => ({})) };
-    const tools = { forUser: jest.fn(async () => ({})) };
-    const mcpClients = {
-      toolsFor: jest.fn(async () => ({
-        tools: {},
-        clients: [],
-        outcomes: [],
-        instructions: [],
-      })),
-    };
-    const svc = new AiChatService(
-      {} as never, // ai
-      aiChatRepo as never,
-      aiChatMessageRepo as never,
-      aiSettings as never,
-      tools as never,
-      mcpClients as never,
-      {} as never, // aiAgentRoleRepo
-      {} as never, // pageRepo (openPage undefined -> never touched)
-      {} as never, // pageAccess
-    );
-    return { svc };
-  }
-
-  const body = {
-    chatId: 'chat-1',
-    messages: [
-      { id: 'm1', role: 'user', parts: [{ type: 'text', text: 'hi' }] },
-    ],
-  };
-
-  beforeEach(() => {
-    streamTextMock.mockReset();
-    streamTextMock.mockImplementation(() => makeStreamResult());
-    jest
-      .spyOn(Logger.prototype, 'log')
-      .mockImplementation(() => undefined as never);
-  });
-
-  afterEach(() => jest.restoreAllMocks());
-
-  it('happy path (run-wrapped): streamText is driven with abortSignal === handle.signal (the RUN signal, NOT the socket)', async () => {
-    const { svc } = makeService();
-    const runController = new AbortController();
-    const runSignal = runController.signal;
-    const socketSignal = new AbortController().signal;
-
-    const begin = jest.fn(async () => ({ runId: 'run-1', signal: runSignal }));
-    await svc.stream({
-      user: { id: 'user-1' } as never,
-      workspace: { id: 'ws-1' } as never,
-      sessionId: 'sess-1',
-      body: body as never,
-      res: makeRes() as never,
-      signal: socketSignal,
-      model: {} as never,
-      role: null,
-      runHooks: {
-        begin,
-        onAssistantSeeded: jest.fn(),
-        onStep: jest.fn(),
-        onSettled: jest.fn(),
-      } as never,
-    });
-
-    expect(begin).toHaveBeenCalledTimes(1);
-    expect(streamTextMock).toHaveBeenCalledTimes(1);
-    // THE assertion: the agent loop's abort is wired to the RUN, so a browser
-    // disconnect (which aborts only `socketSignal`) cannot end the turn.
-    expect(streamTextMock.mock.calls[0][0].abortSignal).toBe(runSignal);
-    expect(streamTextMock.mock.calls[0][0].abortSignal).not.toBe(socketSignal);
-  });
-
-  it('legacy path (no runHooks): streamText is driven with the SOCKET signal', async () => {
-    const { svc } = makeService();
-    const socketSignal = new AbortController().signal;
-
-    await svc.stream({
-      user: { id: 'user-1' } as never,
-      workspace: { id: 'ws-1' } as never,
-      sessionId: 'sess-1',
-      body: body as never,
-      res: makeRes() as never,
-      signal: socketSignal,
-      model: {} as never,
-      role: null,
-      // No runHooks -> the turn stays socket-bound (flag off / default).
-    });
-
-    expect(streamTextMock).toHaveBeenCalledTimes(1);
-    expect(streamTextMock.mock.calls[0][0].abortSignal).toBe(socketSignal);
-  });
-
-  /**
-   * F9 — streamText's TERMINAL callbacks carry the #184 run lifecycle:
-   *   onStepFinish -> runHooks.onStep(runId, stepCount)
-   *   onFinish     -> runHooks.onSettled(runId, 'completed')   (dominant path)
-   *   onAbort      -> runHooks.onSettled(runId, 'aborted')
-   *   onError      -> runHooks.onSettled(runId, 'error', cause)
-   * makeStreamResult() ignores the streamText options, so these callbacks never
-   * fire on their own — a regression in this wiring (esp. the success path) would
-   * strand the run with NO test catching it. Here we CAPTURE the options streamText
-   * was handed and invoke each callback with the real wiring, asserting the run
-   * hooks fire with the right args.
-   */
-  // Drive stream() to the point streamText is called, capturing the options object
-  // (which carries onStepFinish/onFinish/onError/onAbort) and the run hooks.
-  async function captureStreamCallbacks() {
-    const { svc } = makeService();
-    let capturedOpts: any;
-    streamTextMock.mockImplementation((opts: any) => {
-      capturedOpts = opts;
-      return makeStreamResult();
-    });
-    const runHooks = {
-      begin: jest.fn(async () => ({
-        runId: 'run-1',
-        signal: new AbortController().signal,
-      })),
-      onAssistantSeeded: jest.fn(),
-      onStep: jest.fn(),
-      onSettled: jest.fn(),
-    };
-    await svc.stream({
-      user: { id: 'user-1' } as never,
-      workspace: { id: 'ws-1' } as never,
-      sessionId: 'sess-1',
-      body: body as never,
-      res: makeRes() as never,
-      signal: new AbortController().signal,
-      model: {} as never,
-      role: null,
-      runHooks: runHooks as never,
-    });
-    expect(capturedOpts).toBeDefined();
-    return { capturedOpts, runHooks };
-  }
-
-  it('F9: onStepFinish bumps the run step count, onFinish settles the run "completed" (the dominant autonomous-run path)', async () => {
-    const { capturedOpts, runHooks } = await captureStreamCallbacks();
-
-    // A finished step -> onStep(runId, finishedStepCount).
-    capturedOpts.onStepFinish({ text: 'step one', toolCalls: [], content: [] });
-    expect(runHooks.onStep).toHaveBeenCalledWith('run-1', 1);
-    capturedOpts.onStepFinish({ text: 'step two', toolCalls: [], content: [] });
-    expect(runHooks.onStep).toHaveBeenLastCalledWith('run-1', 2);
-
-    // The success terminal callback settles the run.
-    await capturedOpts.onFinish({
-      text: 'done',
-      finishReason: 'stop',
-      totalUsage: {},
-      usage: {},
-      steps: [],
-    });
-    expect(runHooks.onSettled).toHaveBeenCalledWith('run-1', 'completed');
-  });
-
-  it('F9: onAbort settles the run "aborted"', async () => {
-    jest
-      .spyOn(Logger.prototype, 'warn')
-      .mockImplementation(() => undefined as never);
-    const { capturedOpts, runHooks } = await captureStreamCallbacks();
-
-    await capturedOpts.onAbort({ steps: [] });
-    expect(runHooks.onSettled).toHaveBeenCalledWith('run-1', 'aborted');
-  });
-
-  it('F9: onError settles the run "error" carrying the provider cause', async () => {
-    jest
-      .spyOn(Logger.prototype, 'error')
-      .mockImplementation(() => undefined as never);
-    jest
-      .spyOn(Logger.prototype, 'warn')
-      .mockImplementation(() => undefined as never);
-    const { capturedOpts, runHooks } = await captureStreamCallbacks();
-
-    await capturedOpts.onError({ error: new Error('provider exploded') });
-    expect(runHooks.onSettled).toHaveBeenCalledWith(
-      'run-1',
-      'error',
-      expect.stringContaining('provider exploded'),
-    );
-  });
-});

apps/server/src/core/ai-chat/ai-chat.service.spec.ts

@@ -371,12 +371,6 @@ describe('chatStreamMetadata', () => {
     });
   });
 
-  it('attaches the runId on the start part when a run wraps the turn (#184)', () => {
-    expect(
-      chatStreamMetadata({ type: 'start' }, 'chat-1', undefined, 'run-1'),
-    ).toEqual({ chatId: 'chat-1', runId: 'run-1' });
-  });
-
   it('returns the CUMULATIVE step usage passed in for the finish-step part', () => {
     // finish-step usage is per-step in v6; the caller accumulates and passes the
     // running sum, which this just wraps.

apps/server/src/core/ai-chat/dto/ai-chat.dto.ts

@@ -43,30 +43,6 @@ export class BoundChatDto {
   pageId: string;
 }
 
-/**
- * Reconnect to the latest run of a chat (#184): fetch its persisted lifecycle
- * state (and the assistant message it projects) for an in-flight or finished run.
- */
-export class GetRunDto {
-  @IsString()
-  chatId: string;
-}
-
-/**
- * Explicitly STOP an agent run (#184): the user pressed Stop — distinct from a
- * browser disconnect, which never stops a run. Either the run id (preferred, from
- * the streamed start metadata) or the chat id (stop whatever run is active on it).
- */
-export class StopRunDto {
-  @IsOptional()
-  @IsString()
-  runId?: string;
-
-  @IsOptional()
-  @IsString()
-  chatId?: string;
-}
-
 /** Export a chat to Markdown (#183). `lang` localizes the few fixed
  *  role/tool-action labels; defaults to English server-side. */
 export class ExportChatDto {

apps/server/src/core/auth/auth.controller.spec.ts

@@ -19,4 +19,87 @@ describe('AuthController', () => {
   it('should be defined', () => {
     expect(controller).toBeDefined();
   });
+
+  // The EE MFA module is absent in this repo, so require() throws and is caught;
+  // login falls through to authService.login -> setAuthCookie -> returnToken.
+  describe('login returnToken branch', () => {
+    const workspace = { id: 'ws1', enforceSso: false };
+
+    const makeController = () => {
+      const authService = {
+        login: jest.fn().mockResolvedValue('jwt-token-123'),
+      };
+      const environmentService = {
+        getCookieExpiresIn: jest.fn().mockReturnValue(new Date()),
+        isHttps: jest.fn().mockReturnValue(false),
+      };
+      const ctrl = new AuthController(
+        authService as any,
+        {} as any,
+        environmentService as any,
+        {} as any,
+        {} as any,
+      );
+      const res = { setCookie: jest.fn() };
+      return { ctrl, authService, res };
+    };
+
+    it('returns the body token and sets the cookie when returnToken is true', async () => {
+      const { ctrl, authService, res } = makeController();
+      const loginInput = {
+        email: 'a@b.com',
+        password: 'pw',
+        returnToken: true,
+      };
+
+      const result = await ctrl.login(
+        workspace as any,
+        res as any,
+        loginInput as any,
+      );
+
+      expect(result).toEqual({ authToken: 'jwt-token-123' });
+      expect(res.setCookie).toHaveBeenCalledTimes(1);
+      expect(res.setCookie).toHaveBeenCalledWith(
+        'authToken',
+        'jwt-token-123',
+        expect.objectContaining({ httpOnly: true }),
+      );
+      expect(authService.login).toHaveBeenCalled();
+    });
+
+    it('returns no body token but still sets the cookie when returnToken is omitted', async () => {
+      const { ctrl, res } = makeController();
+      const loginInput = { email: 'a@b.com', password: 'pw' };
+
+      const result = await ctrl.login(
+        workspace as any,
+        res as any,
+        loginInput as any,
+      );
+
+      expect(result).toBeUndefined();
+      expect(res.setCookie).toHaveBeenCalledTimes(1);
+    });
+
+    // Guards against an `!== undefined`-style bug: an explicit `false` must
+    // behave exactly like the omitted case (cookie set, no token in the body).
+    it('returns no body token but still sets the cookie when returnToken is false', async () => {
+      const { ctrl, res } = makeController();
+      const loginInput = {
+        email: 'a@b.com',
+        password: 'pw',
+        returnToken: false,
+      };
+
+      const result = await ctrl.login(
+        workspace as any,
+        res as any,
+        loginInput as any,
+      );
+
+      expect(result).toBeUndefined();
+      expect(res.setCookie).toHaveBeenCalledTimes(1);
+    });
+  });
 });

apps/server/src/core/auth/auth.controller.ts

@@ -97,6 +97,12 @@ export class AuthController {
         } else if (mfaResult.authToken) {
           // User doesn't have MFA and workspace doesn't require it
           this.setAuthCookie(res, mfaResult.authToken);
+          // Opt-in body token for native clients (Bearer auth). The response is
+          // wrapped by TransformHttpResponseInterceptor, so clients read it at
+          // `data.authToken`. Web clients omit returnToken and keep the cookie.
+          if (loginInput.returnToken) {
+            return { authToken: mfaResult.authToken };
+          }
           return;
         }
       }
@@ -104,6 +110,12 @@ export class AuthController {
 
     const authToken = await this.authService.login(loginInput, workspace.id);
     this.setAuthCookie(res, authToken);
+    // Opt-in body token for native clients (Bearer auth). The response is wrapped
+    // by TransformHttpResponseInterceptor, so clients read it at `data.authToken`.
+    // Web clients omit returnToken and keep using the httpOnly cookie only.
+    if (loginInput.returnToken) {
+      return { authToken };
+    }
   }
 
   @UseGuards(SetupGuard)

apps/server/src/core/auth/dto/login.dto.ts

@@ -1,4 +1,10 @@
-import { IsEmail, IsNotEmpty, IsString } from 'class-validator';
+import {
+  IsBoolean,
+  IsEmail,
+  IsNotEmpty,
+  IsOptional,
+  IsString,
+} from 'class-validator';
 
 export class LoginDto {
   @IsNotEmpty()
@@ -8,4 +14,13 @@ export class LoginDto {
   @IsNotEmpty()
   @IsString()
   password: string;
+
+  // When true, the access token is returned in the response body (in addition
+  // to the httpOnly cookie) so native/mobile clients can store it in
+  // Keychain/Keystore and send it as 'Authorization: Bearer'. Web clients omit
+  // this flag and keep using the cookie. Opt-in only: the token is never put in
+  // the body otherwise.
+  @IsOptional()
+  @IsBoolean()
+  returnToken?: boolean;
 }

apps/server/src/core/page/services/page.service.spec.ts

@@ -31,6 +31,102 @@ describe('PageService', () => {
     expect(service).toBeDefined();
   });
 
+  describe('update — title sync into collab doc (Bug 1)', () => {
+    const makeUpdateService = () => {
+      const pageRepo = {
+        updatePage: jest.fn().mockResolvedValue(undefined),
+        findById: jest.fn().mockResolvedValue({ id: 'page-1' }),
+      };
+      const generalQueue = { add: jest.fn().mockResolvedValue(undefined) };
+      const collaborationGateway = {
+        writePageTitle: jest.fn().mockResolvedValue(undefined),
+      };
+
+      const svc = new PageService(
+        pageRepo as any, // pageRepo
+        {} as any, // pagePermissionRepo
+        {} as any, // attachmentRepo
+        {} as any, // db
+        {} as any, // storageService
+        {} as any, // attachmentQueue
+        {} as any, // aiQueue
+        generalQueue as any, // generalQueue
+        {} as any, // eventEmitter
+        collaborationGateway as any, // collaborationGateway
+        {} as any, // watcherService
+        {} as any, // transclusionService
+      );
+
+      return { svc, pageRepo, collaborationGateway };
+    };
+
+    const basePage = (): Page =>
+      ({
+        id: 'page-1',
+        slugId: 'slug-1',
+        spaceId: 'space-1',
+        workspaceId: 'ws-1',
+        parentPageId: null,
+        title: 'Old Title',
+        icon: null,
+        contributorIds: [],
+      }) as any;
+
+    const user = { id: 'u1' } as any;
+
+    it('writes the new title into the collab doc when the title actually changed', async () => {
+      const { svc, collaborationGateway } = makeUpdateService();
+
+      await svc.update(basePage(), { title: 'New Title' } as any, user);
+
+      // Must use the Redis-independent writePageTitle (direct
+      // openDirectConnection), NOT handleYjsEvent which no-ops without Redis.
+      expect(collaborationGateway.writePageTitle).toHaveBeenCalledTimes(1);
+      expect(collaborationGateway.writePageTitle).toHaveBeenCalledWith(
+        'page-1',
+        'New Title',
+        expect.objectContaining({ user }),
+      );
+    });
+
+    it('threads agent provenance into the collab title write', async () => {
+      const { svc, collaborationGateway } = makeUpdateService();
+
+      await svc.update(basePage(), { title: 'New Title' } as any, user, {
+        actor: 'agent',
+        aiChatId: 'chat-1',
+      } as any);
+
+      expect(collaborationGateway.writePageTitle).toHaveBeenCalledWith(
+        'page-1',
+        'New Title',
+        expect.objectContaining({ actor: 'agent', aiChatId: 'chat-1' }),
+      );
+    });
+
+    it('does NOT write into the collab doc when the title is unchanged', async () => {
+      const { svc, collaborationGateway } = makeUpdateService();
+
+      // Same title -> titleChanged is false; an icon-only change must not fire
+      // the title sync.
+      await svc.update(
+        basePage(),
+        { title: 'Old Title', icon: '📄' } as any,
+        user,
+      );
+
+      expect(collaborationGateway.writePageTitle).not.toHaveBeenCalled();
+    });
+
+    it('does NOT write into the collab doc when the DTO omits the title', async () => {
+      const { svc, collaborationGateway } = makeUpdateService();
+
+      await svc.update(basePage(), { icon: '📄' } as any, user);
+
+      expect(collaborationGateway.writePageTitle).not.toHaveBeenCalled();
+    });
+  });
+
   describe('movePage cycle guard (#67)', () => {
     // A valid fractional-indexing key — movePage validates `position` by feeding
     // it to generateJitteredKeyBetween(position, null) before anything else.

apps/server/src/core/page/services/page.service.ts

@@ -265,6 +265,8 @@ export class PageService {
     contributors.add(user.id);
     const contributorIds = Array.from(contributors);
 
+    const isAgent = provenance?.actor === 'agent';
+
     // Detect a real title/icon change so the WS tree listener can broadcast an
     // `updateOne` to the space (rename / icon swap) WITHOUT re-broadcasting on a
     // content-only save. Only treat a field as changed when the DTO actually
@@ -307,6 +309,43 @@ export class PageService {
         : undefined,
     );
 
+    // Bug 1: a REST/MCP rename wrote the new title ONLY to the page.title DB
+    // column above. The title's source of truth is the Yjs 'title' fragment in
+    // the page's collab doc, which onStoreDocument re-extracts on every save —
+    // so leaving the fragment stale would REVERT this rename on the page's next
+    // collaborative save (and re-broadcast the old title). Push the new title
+    // into the Yjs 'title' fragment so Yjs stays in sync and never reverts.
+    //
+    // Use the gateway's writePageTitle (direct openDirectConnection) rather than
+    // a Redis-routed handleYjsEvent path: handleYjsEvent routes through
+    // redisSync and SILENTLY no-ops when Redis is disabled
+    // (COLLAB_DISABLE_REDIS=true), which would let the rename revert in a
+    // single-process deployment. writePageTitle is Redis-independent and
+    // openDirectConnection loads the doc from persistence when no editor is
+    // connected, so this also works for an offline page. Thread agent provenance
+    // into the context so onStoreDocument tags the title store 'agent' too.
+    if (titleChanged) {
+      try {
+        await this.collaborationGateway.writePageTitle(
+          page.id,
+          updatePageDto.title,
+          {
+            user,
+            ...(isAgent
+              ? { actor: 'agent', aiChatId: provenance.aiChatId }
+              : {}),
+          },
+        );
+      } catch (err) {
+        // The DB column write already succeeded (fast-read source stays
+        // correct); a failure to sync Yjs here must not fail the rename. Log so
+        // a persistent desync is visible.
+        this.logger.warn(
+          `Failed to sync renamed title into collab doc for page ${page.id}: ${err?.['message']}`,
+        );
+      }
+    }
+
     this.generalQueue
       .add(QueueJob.ADD_PAGE_WATCHERS, {
         userIds: [user.id],

apps/server/src/database/database.module.ts

@@ -31,7 +31,6 @@ import { FavoriteRepo } from '@docmost/db/repos/favorite/favorite.repo';
 import { TemplateRepo } from '@docmost/db/repos/template/template.repo';
 import { AiChatRepo } from '@docmost/db/repos/ai-chat/ai-chat.repo';
 import { AiChatMessageRepo } from '@docmost/db/repos/ai-chat/ai-chat-message.repo';
-import { AiChatRunRepo } from '@docmost/db/repos/ai-chat/ai-chat-run.repo';
 import { AiProviderCredentialsRepo } from '@docmost/db/repos/ai-chat/ai-provider-credentials.repo';
 import { AiMcpServerRepo } from '@docmost/db/repos/ai-chat/ai-mcp-server.repo';
 import { AiAgentRoleRepo } from '@docmost/db/repos/ai-agent-roles/ai-agent-roles.repo';
@@ -105,7 +104,6 @@ import { normalizePostgresUrl } from '../common/helpers';
     TemplateRepo,
     AiChatRepo,
     AiChatMessageRepo,
-    AiChatRunRepo,
     AiProviderCredentialsRepo,
     AiMcpServerRepo,
     AiAgentRoleRepo,
@@ -139,7 +137,6 @@ import { normalizePostgresUrl } from '../common/helpers';
     TemplateRepo,
     AiChatRepo,
     AiChatMessageRepo,
-    AiChatRunRepo,
     AiProviderCredentialsRepo,
     AiMcpServerRepo,
     AiAgentRoleRepo,

apps/server/src/database/migrations/20260627T130000-ai-chat-runs.ts

@@ -1,104 +0,0 @@
-import { type Kysely, sql } from 'kysely';
-
-/**
- * `ai_chat_runs` — the agent RUN as a first-class, server-side lifecycle object
- * (#184 phase 1: autonomous agent runs detached from the browser window).
- *
- * Until now an agent turn lived ONLY as long as the HTTP request was open
- * (`res.hijack()` in ai-chat.controller.ts); a browser disconnect aborted it.
- * This table makes a turn a persistent object the server owns: it is created
- * when a run starts, transitions pending -> running -> succeeded|failed|aborted,
- * and survives the subscriber (browser) going away. The DB is the source of
- * truth — a later client reconnects/sees the result by reading this row plus the
- * assistant message it projects (`assistant_message_id`).
- *
- * The assistant message row (#183 step-granular durability) is the PROJECTION of
- * a run's output; this row is the run's LIFECYCLE. They are linked by
- * `assistant_message_id` (SET NULL if the message is later pruned).
- *
- * `status`  : 'pending' | 'running' | 'succeeded' | 'failed' | 'aborted'.
- * `trigger` : 'user' | 'autostart' | 'schedule' | 'api' | 'continue' — only
- *             'user' is produced in phase 1; the others are reserved for the
- *             autonomy triggers deferred to phase 2 so they need no later
- *             migration.
- *
- * ONE ACTIVE RUN PER CHAT is enforced by a partial unique index on `chat_id`
- * WHERE status IN ('pending','running'): an autonomous run and a user run can
- * never trample each other on the same chat. Settled runs (succeeded/failed/
- * aborted) are excluded from the index so a chat can accumulate any number of
- * historical runs.
- */
-export async function up(db: Kysely<any>): Promise<void> {
-  await db.schema
-    .createTable('ai_chat_runs')
-    .ifNotExists()
-    .addColumn('id', 'uuid', (col) =>
-      col.primaryKey().defaultTo(sql`gen_uuid_v7()`),
-    )
-    .addColumn('chat_id', 'uuid', (col) =>
-      col.references('ai_chats.id').onDelete('cascade').notNull(),
-    )
-    .addColumn('workspace_id', 'uuid', (col) =>
-      col.references('workspaces.id').onDelete('cascade').notNull(),
-    )
-    // The human who triggered the run (audit). SET NULL on user deletion so the
-    // run history outlives its author; NULL is also the natural value for a
-    // future system/cron/api trigger with no human actor.
-    .addColumn('created_by', 'uuid', (col) =>
-      col.references('users.id').onDelete('set null'),
-    )
-    // The assistant message this run materializes (the #183 projection). SET NULL
-    // if that message row is later deleted; nullable because the run row is
-    // created a moment BEFORE the assistant row is seeded.
-    .addColumn('assistant_message_id', 'uuid', (col) =>
-      col.references('ai_chat_messages.id').onDelete('set null'),
-    )
-    .addColumn('trigger', 'varchar(20)', (col) =>
-      col.notNull().defaultTo('user'),
-    )
-    .addColumn('status', 'varchar(20)', (col) =>
-      col.notNull().defaultTo('pending'),
-    )
-    // Terminal error message for a failed run (provider/transport cause),
-    // mirroring the assistant message's metadata.error.
-    .addColumn('error', 'text', (col) => col)
-    // Number of agent steps finished so far (kept monotonic with the projection).
-    .addColumn('step_count', 'integer', (col) => col.notNull().defaultTo(0))
-    // Set when an EXPLICIT user stop is requested (distinct from a mere browser
-    // disconnect, which never stops a run). The runner aborts the turn and the
-    // run settles as 'aborted'.
-    .addColumn('stop_requested_at', 'timestamptz', (col) => col)
-    .addColumn('started_at', 'timestamptz', (col) => col)
-    .addColumn('finished_at', 'timestamptz', (col) => col)
-    .addColumn('created_at', 'timestamptz', (col) =>
-      col.notNull().defaultTo(sql`now()`),
-    )
-    .addColumn('updated_at', 'timestamptz', (col) =>
-      col.notNull().defaultTo(sql`now()`),
-    )
-    .execute();
-
-  // Reconnect / "latest run for this chat" reads hit chat_id first.
-  await db.schema
-    .createIndex('ai_chat_runs_chat_id_idx')
-    .ifNotExists()
-    .on('ai_chat_runs')
-    .column('chat_id')
-    .execute();
-
-  // One ACTIVE run per chat (advisory at the DB level): a second pending/running
-  // run on the same chat is rejected, so a user turn and an autonomous turn can
-  // never race on the same chat. Partial so settled runs do not collide.
-  await db.schema
-    .createIndex('ai_chat_runs_one_active_per_chat')
-    .ifNotExists()
-    .on('ai_chat_runs')
-    .column('chat_id')
-    .unique()
-    .where(sql.ref('status'), 'in', sql`('pending','running')`)
-    .execute();
-}
-
-export async function down(db: Kysely<any>): Promise<void> {
-  await db.schema.dropTable('ai_chat_runs').execute();
-}