fix(cloud): restore stateful Space account login

This reverts commit d50957fc4f.

Co-authored-by: dadachann <185672915+dadachann@users.noreply.github.com>

skills/skills.index.json

@@ -1349,7 +1349,8 @@
             "local-agent",
             "tools",
             "e2b",
-            "nsjail"
+            "nsjail",
+            "host"
           ],
           "automation": "",
           "setup_automation": [],

skills/skills/langbot-deploy/SKILL.md

@@ -63,7 +63,7 @@ Key settings:
 | `api.global_api_key` | **Global API key** for the HTTP API + MCP server. Non-empty = accepted with no login/DB record; no `lbk_` prefix required. Empty = disabled. Plaintext — trusted/internal only, serve over HTTPS. |
 | `plugin.runtime_ws_url` | Standalone plugin runtime WS URL (e.g. `ws://langbot_plugin_runtime:5400/control/ws`) |
 | `box.enabled` | Master switch for the Box sandbox runtime |
-| `box.backend` | `local` (Docker/nsjail autopick) / `docker` / `nsjail` / `e2b`; env override `BOX__BACKEND` |
+| `box.backend` | `local` (Docker/nsjail autopick) / `docker` / `nsjail` / `e2b` / explicit unsafe `host`; env override `BOX__BACKEND` |
 | `box.runtime.endpoint` | External Box runtime URL (e.g. `ws://127.0.0.1:5410`); empty = local auto-managed |
 
 Many keys have `ENV__SUBKEY` overrides (e.g. `BOX__BACKEND`, `BOX__ENABLED`).
@@ -75,6 +75,10 @@ Many keys have `ENV__SUBKEY` overrides (e.g. `BOX__BACKEND`, `BOX__ENABLED`).
   with `--standalone-runtime`.
 - Box has a parallel `--standalone-box` flag; the Docker box host is
   `langbot_box:5410`.
+- `box.backend: host` runs commands directly as the Box Runtime system user.
+  It is never auto-selected, provides no sandbox isolation, and is only for
+  trusted local development. A WebSocket-controlled host backend requires
+  `LANGBOT_BOX_CONTROL_TOKEN`; local stdio control is allowed.
 
 ## Global API key — enabling for agents/automation
 
@@ -93,5 +97,7 @@ login session. See `langbot-mcp-ops` for using it, and `docs/API_KEY_AUTH.md`.
 - "No supported sandbox backend (Docker / nsjail / E2B)" with Docker running
   usually means the user isn't in the `docker` group →
   `sudo usermod -aG docker <user>` and restart in a new shell.
+- Do not use `box.backend: host` as a production fallback. It cannot enforce
+  image, filesystem, network, PID, CPU, memory, or storage isolation.
 - Box root host/container path mismatch breaks sandbox container creation.
 - Don't commit a non-empty `api.global_api_key` to version control.

skills/skills/langbot-testing/cases/sandbox-skill-authoring-e2e.yaml

@@ -13,6 +13,7 @@ tags:
   - tools
   - e2b
   - nsjail
+  - host
 skills:
   - langbot-env-setup
   - langbot-testing
@@ -23,7 +24,7 @@ env:
   - LANGBOT_LOCAL_AGENT_PIPELINE_NAME
 preconditions:
   - "LANGBOT_LOCAL_AGENT_PIPELINE_URL or LANGBOT_LOCAL_AGENT_PIPELINE_NAME points to the local-agent pipeline under test."
-  - "LangBot is started with the sandbox backend intended for this run, such as e2b or nsjail."
+  - "LangBot is started with the Box backend intended for this run, such as e2b, nsjail, or explicit host development mode."
   - "The selected model route supports tool/function calling strongly enough to invoke sandbox tools."
 steps:
   - "Start LangBot with the target sandbox backend and confirm the Box status UI or LANGBOT_BACKEND_URL /api/v1/box/status reports the expected backend."
@@ -33,7 +34,7 @@ steps:
 checks:
   - "UI: Debug Chat final assistant response contains E2E_OK:<skill-name>."
   - "Logs: The model called exec, register_skill, activate, then exec again from the activated skill path."
-  - "Logs: The selected backend name is the expected one, such as e2b or nsjail."
+  - "Logs: The selected backend name is the expected one, such as e2b, nsjail, or host."
   - "Skill store: The registered package and activated writeback match references/sandbox-skill-authoring.md."
   - "Box status: recent_error_count is 0 after the run."
 evidence_required:

skills/skills/langbot-testing/references/sandbox-skill-authoring.md

@@ -4,7 +4,7 @@
 
 Verify that Local Agent can use sandbox tools to create, register, activate, and use a LangBot skill package through the same path a user would exercise in Debug Chat.
 
-This flow applies to Docker, nsjail, and E2B backends. API calls are useful diagnostics, but the primary pass/fail signal is the model-driven Debug Chat tool sequence.
+This flow applies to Docker, nsjail, E2B, and the explicit host development backend. Host runs commands directly as the Box Runtime user and must never be treated as sandbox-isolation coverage. API calls are useful diagnostics, but the primary pass/fail signal is the model-driven Debug Chat tool sequence.
 
 ## Preconditions
 
@@ -13,6 +13,7 @@ This flow applies to Docker, nsjail, and E2B backends. API calls are useful diag
    - `BOX_BACKEND=e2b` when validating E2B.
    - `BOX_BACKEND=nsjail` when validating nsjail.
    - `BOX_BACKEND=local` or `docker` when validating local container fallback.
+   - `BOX_BACKEND=host` only when validating explicit, trusted local direct execution.
 3. Confirm `/api/v1/box/status` reports `available: true` and the expected backend name.
 4. Confirm Debug Chat uses a model with function-calling ability.
 5. Confirm backend logs say native sandbox tools are available.
@@ -71,7 +72,7 @@ Backend logs should show:
 - `register_skill`
 - `activate`
 - a second `exec` whose workdir is `/workspace/.skills/<skill-name>`
-- `backend=e2b`, `backend=nsjail`, or the expected local backend
+- `backend=e2b`, `backend=nsjail`, `backend=host`, or the expected local backend
 
 After the run, verify the skill store through the UI or API:
 
@@ -125,6 +126,8 @@ For E2B raw HTTP diagnostics, include a valid template id such as `base`; a miss
 - Session metadata should keep LangBot logical paths such as `/workspace`; storing provider-internal paths can make later requests look incompatible.
 - nsjail versions differ. Some expose only `--disable_clone_new*` flags and use `--bindmount` instead of `--rw_bind`.
 - On WSL, cgroup v2 may exist but not be writable. The backend should warn and fall back to rlimits rather than fail the sandbox.
+- The host backend does not honor sandbox image, network, rootfs, process, or
+  resource isolation. Use a disposable workspace and low-privilege account.
 - If `ALL_PROXY` uses a SOCKS URL and `socksio` is not installed, some Python HTTP clients can fail during startup. Prefer consistent HTTP proxy variables unless SOCKS support is installed.
 
 ## Related Troubleshooting

skills/skills/langbot-testing/troubleshooting/sandbox-native-tools-unavailable.yaml

@@ -3,7 +3,7 @@ title: "Native sandbox tools are unavailable even though a backend is configured
 date: 2026-05-18
 symptoms:
   - "Backend logs show Native sandbox tools (exec/read/write/edit/glob/grep) are NOT available."
-  - "The Box runtime later reports that E2B, nsjail, or Docker is configured."
+  - "The Box runtime later reports that E2B, nsjail, Docker, or explicit host mode is configured."
   - "Debug Chat does not expose exec, register_skill, or activate as usable tools."
 patterns:
   - "Native sandbox tools ... are NOT available"
@@ -19,6 +19,7 @@ fix_steps:
   - "Ensure the Box runtime reselects a backend when get_backend_info is called and the cached backend is empty."
   - "For E2B, verify the key without printing it and confirm any required template setting."
   - "For nsjail, run nsjail --help and confirm the binary is on PATH for the LangBot process."
+  - "For trusted local development only, explicitly set box.backend=host; never use host as a production sandbox fallback."
 verification: "Run sandbox-skill-authoring-e2e. Logs should show Native sandbox tools are available and /api/v1/box/status should report available=true with the expected backend."
 related_cases:
   - sandbox-skill-authoring-e2e

src/langbot/libs/dingtalk_api/api.py

@@ -697,9 +697,10 @@ class DingTalkClient:
         if not await self.check_access_token():
             await self.get_access_token()
 
-        cardData: dict = {'cardParamMap': _stringify_card_param_map(card_param_map)}
+        template_params = dict(card_param_map or {})
         if card_data_config is not None:
-            cardData['config'] = json.dumps(card_data_config)
+            template_params['config'] = card_data_config
+        cardData: dict = {'cardParamMap': _stringify_card_param_map(template_params)}
 
         body: dict = {
             'cardTemplateId': card_template_id,

src/langbot/pkg/api/http/controller/groups/monitoring.py

@@ -218,6 +218,7 @@ class MonitoringRouterGroup(group.RouterGroup):
             pipeline_ids = quart.request.args.getlist('pipelineId')
             start_time_str = quart.request.args.get('startTime')
             end_time_str = quart.request.args.get('endTime')
+            user_query = quart.request.args.get('userQuery')
             is_active_str = quart.request.args.get('isActive')
             limit = int(quart.request.args.get('limit', 100))
             offset = int(quart.request.args.get('offset', 0))
@@ -237,6 +238,7 @@ class MonitoringRouterGroup(group.RouterGroup):
                 pipeline_ids=pipeline_ids if pipeline_ids else None,
                 start_time=start_time,
                 end_time=end_time,
+                user_query=user_query,
                 is_active=is_active,
                 limit=limit,
                 offset=offset,
@@ -396,7 +398,14 @@ class MonitoringRouterGroup(group.RouterGroup):
         @self.route('/sessions/<session_id>/analysis', methods=['GET'], permission=Permission.RESOURCE_VIEW)
         async def get_session_analysis(session_id: str, request_context: RequestContext) -> str:
             """Get detailed analysis for a specific session"""
-            analysis = await self.ap.monitoring_service.get_session_analysis(request_context, session_id)
+            start_time = parse_iso_datetime(quart.request.args.get('startTime'))
+            end_time = parse_iso_datetime(quart.request.args.get('endTime'))
+            analysis = await self.ap.monitoring_service.get_session_analysis(
+                request_context,
+                session_id,
+                start_time=start_time,
+                end_time=end_time,
+            )
 
             # Always return success with the analysis data
             # The frontend will handle the 'found: false' case

src/langbot/pkg/api/http/controller/groups/user.py

@@ -9,6 +9,7 @@ from .. import group
 from .....entity.errors import account as account_errors
 from ...context import RequestContext
 from .....cloud.launch import SpaceLaunchError
+from .....workspace.errors import WorkspaceNotFoundError
 from ...service.user import ControlPlaneDirectoryRequiredError, PublicRegistrationClosedError
 
 
@@ -429,13 +430,48 @@ class UserRouterGroup(group.RouterGroup):
                 )
 
             account = await self.ap.user_service.get_user_by_uuid(launch['account_uuid'])
+            projection_service = self.ap.directory_projection_service
+            access = None
+            # A first Cloud launch creates the personal Workspace immediately
+            # before redirecting here. Pull a bounded number of signed event
+            # pages until both the Account and its target Workspace membership
+            # are visible instead of rejecting during the background-sync window.
+            for attempt in range(4):
+                if account is not None:
+                    self.ap.user_service._require_active_account(account)
+                    try:
+                        access = await self.ap.workspace_collaboration_service.resolve_account_workspace(
+                            account.uuid,
+                            launch['workspace_uuid'],
+                        )
+                        break
+                    except WorkspaceNotFoundError:
+                        if projection_service is None:
+                            raise
+                elif projection_service is None:
+                    break
+
+                if attempt == 3:
+                    break
+                await projection_service.sync_once()
+                account = await self.ap.user_service.get_user_by_uuid(launch['account_uuid'])
+
+            if access is None and projection_service is not None:
+                # The target event may be deeper than the bounded incremental
+                # page budget. One authoritative signed snapshot catches this
+                # process up without turning the callback into unbounded polling.
+                await projection_service.refresh_snapshot()
+                account = await self.ap.user_service.get_user_by_uuid(launch['account_uuid'])
+                if account is not None:
+                    self.ap.user_service._require_active_account(account)
+                    access = await self.ap.workspace_collaboration_service.resolve_account_workspace(
+                        account.uuid,
+                        launch['workspace_uuid'],
+                    )
             if account is None:
                 raise SpaceLaunchError('Launch Account is not projected into Core')
-            self.ap.user_service._require_active_account(account)
-            access = await self.ap.workspace_collaboration_service.resolve_account_workspace(
-                account.uuid,
-                launch['workspace_uuid'],
-            )
+            if access is None:  # pragma: no cover - bounded loop resolves or raises.
+                raise SpaceLaunchError('Launch Workspace is not projected into Core')
             token = await self.ap.user_service.generate_jwt_token(account)
             return self.success(
                 data={

src/langbot/pkg/api/http/service/monitoring.py

@@ -1257,6 +1257,7 @@ class MonitoringService:
         pipeline_ids: list[str] | None = None,
         start_time: datetime.datetime | None = None,
         end_time: datetime.datetime | None = None,
+        user_query: str | None = None,
         is_active: bool | None = None,
         limit: int = 100,
         offset: int = 0,
@@ -1274,6 +1275,14 @@ class MonitoringService:
             conditions.append(persistence_monitoring.MonitoringSession.start_time >= start_time)
         if end_time:
             conditions.append(persistence_monitoring.MonitoringSession.start_time <= end_time)
+        if user_query and user_query.strip():
+            user_pattern = f'%{user_query.strip()}%'
+            conditions.append(
+                sqlalchemy.or_(
+                    persistence_monitoring.MonitoringSession.user_id.ilike(user_pattern),
+                    persistence_monitoring.MonitoringSession.user_name.ilike(user_pattern),
+                )
+            )
         if is_active is not None:
             conditions.append(persistence_monitoring.MonitoringSession.is_active == is_active)
 
@@ -1365,6 +1374,8 @@ class MonitoringService:
         self,
         context: TenantContext,
         session_id: str,
+        start_time: datetime.datetime | None = None,
+        end_time: datetime.datetime | None = None,
     ) -> dict:
         """Get bounded session details with full statistics computed in SQL."""
         workspace_uuid = require_workspace_uuid(context)
@@ -1478,12 +1489,17 @@ class MonitoringService:
             )
         )
         tool_stats = tool_stats_result.one()
+        tool_conditions = [
+            persistence_monitoring.MonitoringToolCall.workspace_uuid == workspace_uuid,
+            persistence_monitoring.MonitoringToolCall.session_id == session_id,
+        ]
+        if start_time is not None:
+            tool_conditions.append(persistence_monitoring.MonitoringToolCall.timestamp >= start_time)
+        if end_time is not None:
+            tool_conditions.append(persistence_monitoring.MonitoringToolCall.timestamp <= end_time)
         tool_query = (
             sqlalchemy.select(persistence_monitoring.MonitoringToolCall)
-            .where(
-                persistence_monitoring.MonitoringToolCall.workspace_uuid == workspace_uuid,
-                persistence_monitoring.MonitoringToolCall.session_id == session_id,
-            )
+            .where(*tool_conditions)
             .order_by(persistence_monitoring.MonitoringToolCall.timestamp.asc())
             .limit(detail_limit + 1)
         )

src/langbot/pkg/box/service.py

@@ -455,7 +455,9 @@ class BoxService:
 
     async def _require_validated_workspace_sandbox(self, execution_context: ExecutionContext) -> None:
         if not self._available:
-            raise BoxError('Box runtime is not available. Install and start Docker to use sandbox features.')
+            raise BoxError(
+                'Box runtime is not available. Configure an available Box backend before using Box features.'
+            )
         if self._cloud_managed:
             if self._admission is None:
                 raise BoxAdmissionError('Cloud Box sandbox admission is unavailable')
@@ -565,7 +567,9 @@ class BoxService:
         skip_host_mount_validation: bool = False,
     ) -> dict:
         if not self._available:
-            raise BoxError('Box runtime is not available. Install and start Docker to use sandbox features.')
+            raise BoxError(
+                'Box runtime is not available. Configure an available Box backend before using Box features.'
+            )
         execution_context = await self._validated_execution_context(self._query_execution_context(query))
         spec_payload = self._managed_policy_payload(execution_context, spec_payload)
         await self._require_validated_workspace_sandbox(execution_context)
@@ -2142,5 +2146,8 @@ class BoxService:
             if backend_name:
                 payload['connector_error'] = f'Configured sandbox backend "{backend_name}" is unavailable'
             else:
-                payload['connector_error'] = 'No supported sandbox backend (Docker / nsjail / E2B) is available'
+                payload['connector_error'] = (
+                    'No supported sandbox backend (Docker / nsjail / E2B) is available. '
+                    'Trusted local development may explicitly select the unsafe host backend.'
+                )
         return payload

src/langbot/pkg/cloud/directory_projection.py

@@ -125,10 +125,21 @@ class DirectoryProjectionService:
         # The database cursor remains the shared projection high-water mark,
         # while this cursor tracks what this process has actually observed.
         self._consumer_cursor: int | None = None
+        self._sync_lock = asyncio.Lock()
 
     async def initialize(self) -> None:
         """Block Cloud startup until one full signed snapshot is committed."""
 
+        async with self._sync_lock:
+            await self._refresh_snapshot()
+
+    async def refresh_snapshot(self) -> None:
+        """Refresh from one full signed snapshot within the sync single-flight."""
+
+        async with self._sync_lock:
+            await self._refresh_snapshot()
+
+    async def _refresh_snapshot(self) -> None:
         last_superseded: _DirectorySnapshotSuperseded | None = None
         for _attempt in range(5):
             snapshot = await self.provider.fetch_snapshot(self.instance_uuid)
@@ -159,9 +170,13 @@ class DirectoryProjectionService:
                 delay = min(max(delay * 2, self.sync_interval_seconds), self.max_staleness_seconds / 2)
 
     async def sync_once(self) -> None:
+        async with self._sync_lock:
+            await self._sync_once()
+
+    async def _sync_once(self) -> None:
         cursor = self._consumer_cursor
         if cursor is None:
-            await self.initialize()
+            await self._refresh_snapshot()
             return
         batch = await self.provider.fetch_events(
             self.instance_uuid,

src/langbot/pkg/platform/sources/lark.py

@@ -160,6 +160,29 @@ def _lark_should_update_stream_element(
     return not resume_from and not form_data and (msg_seq % 8 == 0 or is_final)
 
 
+def _lark_final_layout_texts(
+    *,
+    resume_from: bool,
+    text_message: str,
+    pre_pause_cached: str | None,
+    resume_cached: str,
+) -> tuple[str, str]:
+    """Return (main_text, resume_placeholder_text) for the final card update.
+
+    Non-resume round: the full reply belongs in the main streaming element
+    only — also rendering the resume placeholder duplicates the reply, since
+    both hold the same accumulated text. Resume round (Dify HITL): keep the
+    pre-pause text in the main element and the resumed text in the
+    placeholder, as they are distinct segments.
+    """
+    if resume_from:
+        # An empty pre-pause cache is valid (Dify paused before emitting any
+        # text); only a missing entry (None) falls back to the full text.
+        main_text = text_message if pre_pause_cached is None else pre_pause_cached
+        return main_text, resume_cached
+    return text_message, ''
+
+
 def _lark_display_input_value(field: dict, value: typing.Any) -> str:
     field_type = _dify_field_type(field)
     if field_type == 'file':
@@ -2358,16 +2381,21 @@ class LarkAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
                     self.card_form_input_defs[card_id] = _lark_form_input_defs(form_data)
                     self.card_form_inputs[card_id] = dict(form_data.get('inputs') or {})
             else:
-                # Normal finish: keep pre-pause + resume content visible,
-                # remove buttons/notice, drop the resume placeholder.
+                # Normal finish: remove buttons/notice and finalize the card.
+                main_text, resume_text = _lark_final_layout_texts(
+                    resume_from=resume_from,
+                    text_message=text_message,
+                    pre_pause_cached=self.card_pre_pause_text.get(card_id),
+                    resume_cached=resume_cached,
+                )
                 await self._update_card_layout(
                     card_id=card_id,
                     message_source=message_source,
-                    text_message=pre_pause,
+                    text_message=main_text,
                     sequence=final_seq,
                     form_data=None,
                     notice_text=selected_notice if resume_from else '',
-                    resume_placeholder_text=resume_cached,
+                    resume_placeholder_text=resume_text,
                 )
                 self._drop_card_state(card_id)
             self.card_id_dict.pop(message_id, None)

src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py

@@ -573,7 +573,7 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
             levels = ['provider_default', 'disabled', 'enabled']
         elif family == 'doubao':
             levels = ['provider_default', 'disabled', 'low', 'medium', 'high']
-        elif family == 'ollama':
+        elif family in ('ollama', 'ollama_chat'):
             levels = ['provider_default']
             levels.append('disabled')
             if normalized_name.startswith('gpt-oss') or '/gpt-oss' in normalized_name:
@@ -1345,7 +1345,14 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
         extra_args: dict[str, typing.Any] = {},
     ) -> tuple[list[list[float]], dict]:
         """Invoke embedding and return vectors with usage info."""
-        model_name = self._build_litellm_model_name(model.model_entity.name)
+        # litellm's embedding routing has no "ollama_chat" branch (that provider
+        # exists only for /api/chat completions) — embeddings still go through
+        # the plain "ollama" provider. Requesters configured for ollama_chat
+        # (to get native tool-calling on the chat path) must fall back to
+        # "ollama" here specifically, or embedding calls raise "Unmapped LLM
+        # provider for this endpoint".
+        embedding_provider = 'ollama' if self._get_custom_llm_provider() == 'ollama_chat' else None
+        model_name = self._build_litellm_model_name(model.model_entity.name, embedding_provider)
         api_key = model.provider.token_mgr.get_token()
 
         args = {
@@ -1541,6 +1548,12 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
                 event_hooks=httpclient.httpx_response_limit_hooks(),
             ) as client:
                 response = await client.get(models_url, headers=headers)
+                if response.status_code == 404 and not base_url.rstrip('/').endswith('/v1'):
+                    # Some OpenAI-compatible servers (notably a bare Ollama host,
+                    # e.g. http://host:11434) expose the model list under /v1/models
+                    # rather than /models. Providers whose configured base_url
+                    # already ends in /v1 keep their original (working) URL.
+                    response = await client.get(f'{base_url}/v1/models', headers=headers)
                 response.raise_for_status()
                 payload = await httpclient.parse_json_response(response)
 

src/langbot/pkg/provider/modelmgr/requesters/ollamachat.yaml

@@ -7,7 +7,7 @@ metadata:
     zh_Hans: Ollama
   icon: ollama.svg
 spec:
-  litellm_provider: ollama
+  litellm_provider: ollama_chat
   config:
   - name: base_url
     label:

src/langbot/pkg/provider/tools/loaders/native.py

@@ -222,6 +222,7 @@ class NativeToolLoader(loader.ToolLoader):
             self.ap.logger.warning(
                 'Native sandbox tools (exec/read/write/edit/glob/grep) are NOT available. '
                 'No sandbox backend (Docker/nsjail/E2B) is ready. '
+                'Trusted local development may explicitly select box.backend=host. '
                 'The LLM will not have access to code execution or file operation tools.'
             )
 

src/langbot/pkg/provider/tools/loaders/skill_authoring.py

@@ -42,7 +42,8 @@ class SkillToolLoader(loader.ToolLoader):
         else:
             self.ap.logger.info(
                 'Skill tools (activate/register_skill) are NOT available. '
-                'No sandbox backend (Docker/nsjail/E2B) is ready.'
+                'No sandbox backend (Docker/nsjail/E2B) is ready. '
+                'Trusted local development may explicitly select box.backend=host.'
             )
 
     async def _check_sandbox_available(self) -> bool:

src/langbot/templates/config.yaml

@@ -331,7 +331,10 @@ box:
     # skill tool, skill add/edit, and stdio-mode MCP servers. Skills can still
     # be listed read-only and http/sse MCP servers continue to work.
     enabled: true
-    backend: 'local'  # 'local' (Docker/nsjail), 'docker', 'nsjail', or 'e2b'. Can be written via BOX__BACKEND.
+    # 'host' runs commands directly as the Box Runtime user without sandbox
+    # isolation. It is never auto-selected and is only for trusted local
+    # development. Can be written via BOX__BACKEND.
+    backend: 'local'  # 'local' (Docker/nsjail), 'docker', 'nsjail', 'e2b', or explicit unsafe 'host'.
     runtime:
         # LANGBOT_BOX_CONTROL_TOKEN is optional for OSS external WebSocket
         # runtimes. To protect an exposed endpoint, set the same strong secret

src/langbot/templates/embed/widget.js

@@ -642,9 +642,10 @@
               .replace(/\s+/g, " ")
               .trim();
             if (
-              prevContent === content ||
-              prevContent.indexOf(content) >= 0 ||
-              content.indexOf(prevContent) >= 0
+              prevContent &&
+              (prevContent === content ||
+                prevContent.indexOf(content) >= 0 ||
+                content.indexOf(prevContent) >= 0)
             )
               return;
           }

tests/integration/api/test_monitoring.py

@@ -242,6 +242,22 @@ class TestMonitoringSessionsEndpoint:
 
         assert response.status_code == 200
 
+    @pytest.mark.asyncio
+    async def test_get_sessions_forwards_user_search_and_page_window(self, quart_test_client, fake_monitoring_app):
+        fake_monitoring_app.monitoring_service.get_sessions.reset_mock()
+
+        response = await quart_test_client.get(
+            '/api/v1/monitoring/sessions?botId=bot-1&userQuery=alice&limit=20&offset=40',
+            headers={'Authorization': 'Bearer test_token'},
+        )
+
+        assert response.status_code == 200
+        kwargs = fake_monitoring_app.monitoring_service.get_sessions.await_args.kwargs
+        assert kwargs['bot_ids'] == ['bot-1']
+        assert kwargs['user_query'] == 'alice'
+        assert kwargs['limit'] == 20
+        assert kwargs['offset'] == 40
+
 
 @pytest.mark.usefixtures('mock_circular_import_chain')
 class TestMonitoringErrorsEndpoint:
@@ -278,13 +294,19 @@ class TestMonitoringDetailsEndpoints:
     """Tests for detail endpoints."""
 
     @pytest.mark.asyncio
-    async def test_get_session_analysis(self, quart_test_client):
+    async def test_get_session_analysis(self, quart_test_client, fake_monitoring_app):
         """GET /api/v1/monitoring/sessions/{id}/analysis."""
         response = await quart_test_client.get(
-            '/api/v1/monitoring/sessions/sess-1/analysis', headers={'Authorization': 'Bearer test_token'}
+            '/api/v1/monitoring/sessions/sess-1/analysis'
+            '?startTime=2026-08-31T16%3A00%3A00.000Z'
+            '&endTime=2026-09-01T15%3A59%3A59.999Z',
+            headers={'Authorization': 'Bearer test_token'},
         )
 
         assert response.status_code == 200
+        kwargs = fake_monitoring_app.monitoring_service.get_session_analysis.await_args.kwargs
+        assert kwargs['start_time'].isoformat() == '2026-08-31T16:00:00'
+        assert kwargs['end_time'].isoformat() == '2026-09-01T15:59:59.999000'
 
     @pytest.mark.asyncio
     async def test_get_message_details(self, quart_test_client):

tests/integration/api/test_user_space_oauth.py

@@ -11,6 +11,7 @@ import pytest
 import quart
 
 from langbot.pkg.api.http.controller.groups.user import UserRouterGroup
+from langbot.pkg.workspace.errors import WorkspaceNotFoundError
 
 
 pytestmark = pytest.mark.integration
@@ -27,7 +28,8 @@ async def space_oauth_api():
         execution=SimpleNamespace(instance_uuid='instance-a', placement_generation=1),
     )
     application = Mock()
-    application.deployment = SimpleNamespace(multi_workspace_enabled=False)
+    application.deployment = SimpleNamespace(multi_workspace_enabled=False, mode='oss')
+    application.directory_projection_service = None
     application.persistence_mgr = None
     application.user_service.get_authenticated_account = AsyncMock(return_value=account)
     application.user_service.issue_space_oauth_state = AsyncMock(
@@ -69,6 +71,7 @@ async def space_oauth_api():
     application.space_service.get_oauth_authorize_url = Mock(
         side_effect=lambda redirect_uri, state: f'https://space.example/authorize?state={state}'
     )
+
     application.space_service.exchange_oauth_code = AsyncMock(
         return_value={
             'access_token': 'space-access-token',
@@ -125,6 +128,26 @@ async def test_cloud_launch_state_is_server_issued_and_workspace_bound(space_oau
     )
 
 
+@pytest.mark.asyncio
+async def test_cloud_login_entry_starts_stateful_space_oauth(space_oauth_api):
+    application, client = space_oauth_api
+    application.deployment.mode = 'cloud'
+
+    response = await client.get(
+        '/api/v1/user/space/authorize-url',
+        query_string={
+            'redirect_uri': 'http://localhost/auth/space/callback',
+            'cloud_entry': '1',
+        },
+        headers={'Origin': 'http://localhost'},
+    )
+
+    assert response.status_code == 200
+    authorize_url = (await response.get_json())['data']['authorize_url']
+    assert authorize_url.startswith('https://space.example/authorize?state=')
+    application.user_service.issue_space_oauth_state.assert_awaited_once_with('login')
+
+
 @pytest.mark.asyncio
 async def test_public_login_rejects_caller_supplied_state(space_oauth_api):
     application, client = space_oauth_api
@@ -414,3 +437,85 @@ async def test_direct_launch_assertion_does_not_consume_normal_oauth_state(space
     )
     application.user_service.consume_space_oauth_state.assert_not_awaited()
     application.space_service.exchange_oauth_code.assert_not_awaited()
+
+
+@pytest.mark.asyncio
+async def test_direct_launch_refreshes_new_workspace_projection_before_rejecting_account(space_oauth_api):
+    application, client = space_oauth_api
+    projected_account = SimpleNamespace(
+        uuid='account-a',
+        user='owner@example.com',
+        account_type='space',
+        status='active',
+    )
+    application.user_service.get_user_by_uuid = AsyncMock(side_effect=[None, None, projected_account])
+    application.directory_projection_service = SimpleNamespace(sync_once=AsyncMock())
+
+    response = await client.post(
+        '/api/v1/user/space/callback',
+        json={
+            'workspace_uuid': WORKSPACE_UUID,
+            'launch_assertion': 'signed-launch-token',
+        },
+    )
+
+    assert response.status_code == 200
+    assert (await response.get_json())['data']['workspace_uuid'] == WORKSPACE_UUID
+    assert application.directory_projection_service.sync_once.await_count == 2
+    assert application.user_service.get_user_by_uuid.await_count == 3
+
+
+@pytest.mark.asyncio
+async def test_direct_launch_refreshes_projection_when_account_exists_before_workspace(space_oauth_api):
+    application, client = space_oauth_api
+    projected_access = application.workspace_collaboration_service.resolve_account_workspace.return_value
+    application.workspace_collaboration_service.resolve_account_workspace = AsyncMock(
+        side_effect=[WorkspaceNotFoundError('Workspace not found'), projected_access]
+    )
+    application.directory_projection_service = SimpleNamespace(sync_once=AsyncMock())
+
+    response = await client.post(
+        '/api/v1/user/space/callback',
+        json={
+            'workspace_uuid': WORKSPACE_UUID,
+            'launch_assertion': 'signed-launch-token',
+        },
+    )
+
+    assert response.status_code == 200
+    assert (await response.get_json())['data']['workspace_uuid'] == WORKSPACE_UUID
+    application.directory_projection_service.sync_once.assert_awaited_once_with()
+    assert application.workspace_collaboration_service.resolve_account_workspace.await_count == 2
+
+
+@pytest.mark.asyncio
+async def test_direct_launch_falls_back_to_snapshot_when_event_backlog_exceeds_page_budget(space_oauth_api):
+    application, client = space_oauth_api
+    projected_access = application.workspace_collaboration_service.resolve_account_workspace.return_value
+    application.workspace_collaboration_service.resolve_account_workspace = AsyncMock(
+        side_effect=[
+            WorkspaceNotFoundError('Workspace not found'),
+            WorkspaceNotFoundError('Workspace not found'),
+            WorkspaceNotFoundError('Workspace not found'),
+            WorkspaceNotFoundError('Workspace not found'),
+            projected_access,
+        ]
+    )
+    application.directory_projection_service = SimpleNamespace(
+        sync_once=AsyncMock(),
+        refresh_snapshot=AsyncMock(),
+    )
+
+    response = await client.post(
+        '/api/v1/user/space/callback',
+        json={
+            'workspace_uuid': WORKSPACE_UUID,
+            'launch_assertion': 'signed-launch-token',
+        },
+    )
+
+    assert response.status_code == 200
+    assert (await response.get_json())['data']['workspace_uuid'] == WORKSPACE_UUID
+    assert application.directory_projection_service.sync_once.await_count == 3
+    application.directory_projection_service.refresh_snapshot.assert_awaited_once_with()
+    assert application.workspace_collaboration_service.resolve_account_workspace.await_count == 5

tests/unit_tests/api/service/test_monitoring_tenancy.py

@@ -138,6 +138,39 @@ async def test_same_session_and_resource_ids_do_not_collide(service):
     assert (await service.get_message_details(context_a, message_b))['found'] is False
 
 
+async def test_session_search_matches_user_id_or_name_within_workspace(service):
+    context_a = _context(WORKSPACE_A)
+    context_b = _context(WORKSPACE_B)
+    fixtures = [
+        (context_a, 'session-id-match', 'customer-42', 'Alice'),
+        (context_a, 'session-name-match', 'customer-99', 'Bob Alice Cooper'),
+        (context_a, 'session-no-match', 'customer-7', 'Bob'),
+        (context_b, 'session-other-workspace', 'customer-42', 'Alice'),
+    ]
+    for context, session_id, user_id, user_name in fixtures:
+        await service.record_session_start(
+            context,
+            session_id=session_id,
+            bot_id='same-bot',
+            bot_name='Same Bot',
+            pipeline_id='same-pipeline',
+            pipeline_name='Same Pipeline',
+            user_id=user_id,
+            user_name=user_name,
+        )
+
+    by_id, id_total = await service.get_sessions(context_a, user_query='customer-42')
+    by_name, name_total = await service.get_sessions(context_a, user_query='alice')
+
+    assert id_total == 1
+    assert [session['session_id'] for session in by_id] == ['session-id-match']
+    assert name_total == 2
+    assert {session['session_id'] for session in by_name} == {
+        'session-id-match',
+        'session-name-match',
+    }
+
+
 async def test_tool_call_inherits_context_from_connection_message_row(service):
     context = _context(WORKSPACE_A)
     message_id = await _record_message(service, context, 'tool context')

tests/unit_tests/cloud/test_directory_projection.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import asyncio
 import datetime
 import logging
 from types import SimpleNamespace
@@ -735,6 +736,72 @@ async def test_each_replica_consumes_events_with_its_own_cursor(projection_conte
     assert second_provider.after_cursors == [1, 2]
 
 
+async def test_concurrent_sync_once_calls_are_serialized_per_service(projection_context):
+    application, _session_factory = projection_context
+
+    class _ConcurrentProvider(_Provider):
+        def __init__(self) -> None:
+            super().__init__([_snapshot(1)])
+            self.first_fetch_started = asyncio.Event()
+            self.release_first_fetch = asyncio.Event()
+            self.active_fetches = 0
+            self.max_active_fetches = 0
+
+        async def fetch_events(
+            self,
+            instance_uuid: str,
+            after_cursor: int,
+            limit: int,
+        ) -> DirectoryEventBatch:
+            assert instance_uuid == INSTANCE_UUID
+            assert limit == 100
+            self.after_cursors.append(after_cursor)
+            self.active_fetches += 1
+            self.max_active_fetches = max(self.max_active_fetches, self.active_fetches)
+            try:
+                if len(self.after_cursors) == 1:
+                    self.first_fetch_started.set()
+                    await self.release_first_fetch.wait()
+                cursor = after_cursor + 1
+                return DirectoryEventBatch(
+                    instance_uuid=instance_uuid,
+                    after_cursor=after_cursor,
+                    cursor=cursor,
+                    high_water_cursor=cursor,
+                    events=(
+                        DirectoryEvent(
+                            cursor=cursor,
+                            uuid=f'40000000-0000-4000-8000-{cursor:012d}',
+                            aggregate_uuid=WORKSPACE_UUID,
+                            event_type='entitlement.changed',
+                            revision=cursor,
+                            payload={
+                                'workspace_uuid': WORKSPACE_UUID,
+                                'entitlement_revision': cursor,
+                            },
+                            created_at=datetime.datetime(2026, 7, 24, 12, cursor, tzinfo=datetime.UTC),
+                        ),
+                    ),
+                )
+            finally:
+                self.active_fetches -= 1
+
+    provider = _ConcurrentProvider()
+    service = DirectoryProjectionService(application, provider, INSTANCE_UUID)
+    await service.initialize()
+
+    first = asyncio.create_task(service.sync_once())
+    await provider.first_fetch_started.wait()
+    second = asyncio.create_task(service.sync_once())
+    await asyncio.sleep(0)
+    provider.release_first_fetch.set()
+    await asyncio.gather(first, second)
+
+    assert provider.max_active_fetches == 1
+    assert provider.after_cursors == [1, 2]
+    assert service._consumer_cursor == 3
+
+
 async def test_snapshot_coverage_allows_lagging_replica_to_replay_receipts(projection_context):
     application, session_factory = projection_context
     event_two = DirectoryEvent(

tests/unit_tests/platform/test_dingtalk_api.py

@@ -1,8 +1,11 @@
 """Tests for DingTalk API payload helpers."""
 
 import json
+from contextlib import asynccontextmanager
+from unittest.mock import AsyncMock
 
-from langbot.libs.dingtalk_api.api import _stringify_card_param_map
+from langbot.libs.dingtalk_api.api import DingTalkClient, _stringify_card_param_map
+from langbot.pkg.utils import httpclient
 
 
 def test_dingtalk_card_param_map_stringifies_select_component_arrays():
@@ -40,3 +43,41 @@ def test_dingtalk_card_param_map_stringifies_unregistered_structures():
 
     assert params['other'] == '["A"]'
     assert params['empty'] == ''
+
+
+async def test_create_card_embeds_layout_config_as_template_parameter(monkeypatch):
+    response = type('Response', (), {'status_code': 200})()
+    post = AsyncMock(return_value=response)
+
+    @asynccontextmanager
+    async def client_context():
+        yield type('HttpClient', (), {'post': post})()
+
+    client = object.__new__(DingTalkClient)
+    client.access_token = 'access-token'
+    client.robot_code = 'robot-code'
+    client.key = 'client-id'
+    client.logger = None
+    client.check_access_token = AsyncMock(return_value=True)
+    client._http_client_context = client_context
+    monkeypatch.setattr(httpclient, 'response_text', AsyncMock(return_value='{}'))
+
+    original_params = {'content': 'hello'}
+    delivered = await client.create_and_deliver_card(
+        card_template_id='template-id',
+        out_track_id='track-id',
+        open_space_id='dtv1.card//IM_ROBOT.user-id',
+        is_group=False,
+        card_param_map=original_params,
+        card_data_config={'autoLayout': True},
+    )
+
+    request_body = post.await_args.kwargs['json']
+    assert delivered is True
+    assert request_body['cardData'] == {
+        'cardParamMap': {
+            'content': 'hello',
+            'config': '{"autoLayout": true}',
+        }
+    }
+    assert original_params == {'content': 'hello'}

tests/unit_tests/platform/test_lark_adapter.py

@@ -1,7 +1,7 @@
 """Tests for Lark adapter helper behavior."""
 
 import threading
-from unittest.mock import MagicMock
+from unittest.mock import AsyncMock, MagicMock
 
 import pytest
 
@@ -12,6 +12,7 @@ from langbot.pkg.platform.sources.lark import (
     _lark_completed_input_lines,
     _lark_current_input_defs,
     _lark_extract_action_form_inputs,
+    _lark_final_layout_texts,
     _lark_should_update_stream_element,
     _lark_visible_form_content,
 )
@@ -221,3 +222,121 @@ def test_lark_completed_input_lines_display_select_value_from_object():
     )
 
     assert lines == ['✅ xiala：B']
+
+
+def test_lark_final_layout_texts_normal_round_drops_resume_placeholder():
+    """Non-resume final chunk: the reply must land in the main element only.
+
+    Regression: rendering the resume placeholder too duplicated the reply,
+    because the accumulated streaming text equals the final text on a normal
+    round (e.g. 'It is Sep 1, 2026.\nIt is Sep 1, 2026.' in the card).
+    """
+    main_text, resume_text = _lark_final_layout_texts(
+        resume_from=False,
+        text_message='It is Sep 1, 2026, 15:09:15.',
+        pre_pause_cached=None,
+        resume_cached='It is Sep 1, 2026, 15:09:15.',
+    )
+
+    assert main_text == 'It is Sep 1, 2026, 15:09:15.'
+    assert resume_text == ''
+
+
+def test_lark_final_layout_texts_resume_round_keeps_both_segments():
+    """Dify HITL resume final chunk: pre-pause text and resumed text differ,
+    both segments stay visible."""
+    main_text, resume_text = _lark_final_layout_texts(
+        resume_from=True,
+        text_message='resumed answer',
+        pre_pause_cached='partial answer before pause',
+        resume_cached='resumed answer',
+    )
+
+    assert main_text == 'partial answer before pause'
+    assert resume_text == 'resumed answer'
+
+
+def test_lark_final_layout_texts_resume_round_without_pre_pause_falls_back():
+    main_text, resume_text = _lark_final_layout_texts(
+        resume_from=True,
+        text_message='answer',
+        pre_pause_cached=None,
+        resume_cached='answer',
+    )
+
+    assert main_text == 'answer'
+    assert resume_text == 'answer'
+
+
+def test_lark_final_layout_texts_resume_round_empty_pre_pause_kept_empty():
+    """Dify paused before emitting any text: the pre-pause cache is a valid
+    empty string and must NOT be treated as a cache miss.
+
+    Regression: `pre_pause_cached or text_message` fell back to the full
+    text, so the final card rendered ('resumed answer', 'resumed answer')
+    and duplicated the reply.
+    """
+    main_text, resume_text = _lark_final_layout_texts(
+        resume_from=True,
+        text_message='resumed answer',
+        pre_pause_cached='',
+        resume_cached='resumed answer',
+    )
+
+    assert main_text == ''
+    assert resume_text == 'resumed answer'
+
+
+def _build_resume_final_chunk_adapter(message_text: str):
+    """Build a LarkAdapter whose card state mimics a Dify HITL round that
+    paused before emitting any text, then resumed and completed."""
+    adapter = LarkAdapter.model_construct(
+        api_client=MagicMock(),
+        message_converter=MagicMock(yiri2target=AsyncMock(return_value=([[{'tag': 'text', 'text': message_text}]], []))),
+    )
+    adapter.config = {'app_type': 'self'}
+    LarkAdapter.get_app_access_token = lambda self: None
+    LarkAdapter.get_tenant_access_token = lambda self, tenant_key: None
+    adapter.card_id_dict = {'msg-1': 'card-1'}
+    adapter.card_streaming_text = {'card-1': message_text}
+    adapter.card_pre_pause_text = {'card-1': ''}
+    adapter.card_resume_transitioned = {'card-1'}
+    adapter.card_sequence_dict = {}
+    adapter.card_last_accessed = {}
+    adapter.card_cleanup_at = 0.0
+    adapter.card_id_to_source_ids = {}
+    adapter.reply_message_card_ids = {}
+    adapter.card_form_content = {}
+    adapter.card_form_input_defs = {}
+    adapter.card_form_inputs = {}
+    adapter._update_card_layout = AsyncMock()
+    return adapter
+
+
+@pytest.mark.asyncio
+async def test_reply_message_chunk_resume_final_with_empty_pre_pause_keeps_main_empty():
+    """End-to-end regression via reply_message_chunk: Dify paused before any
+    text, so the pre-pause cache is ''. The final card update must render the
+    resumed answer only once (empty main text + resume placeholder), not
+    twice as ('resumed answer', 'resumed answer')."""
+    adapter = _build_resume_final_chunk_adapter('resumed answer')
+
+    bot_message = MagicMock(
+        resp_message_id='msg-1',
+        msg_sequence=1,
+        spec=['resp_message_id', 'msg_sequence', '_resume_from_form'],
+    )
+    bot_message._resume_from_form = True
+    message_source = MagicMock(source_platform_object=None)
+
+    await adapter.reply_message_chunk(
+        message_source,
+        bot_message,
+        MagicMock(),
+        is_final=True,
+    )
+
+    adapter._update_card_layout.assert_awaited_once()
+    layout_kwargs = adapter._update_card_layout.await_args.kwargs
+    assert layout_kwargs['text_message'] == ''
+    assert layout_kwargs['resume_placeholder_text'] == 'resumed answer'

web/src/app/home/bots/components/bot-session/BotSessionMonitor.tsx

@@ -17,6 +17,7 @@ import {
   Copy,
   Check,
   ChevronDown,
+  ChevronLeft,
   ChevronRight,
   Workflow,
   ThumbsUp,
@@ -117,16 +118,43 @@ interface BotSessionMonitorProps {
   botId: string;
 }
 
+const SESSION_PAGE_SIZE = 20;
+const MESSAGE_PAGE_SIZE = 50;
+
+const localDateBoundaryToISOString = (
+  dateValue: string,
+  endOfDay: boolean,
+): string => {
+  const [year, month, day] = dateValue.split('-').map(Number);
+  return new Date(
+    year,
+    month - 1,
+    day,
+    endOfDay ? 23 : 0,
+    endOfDay ? 59 : 0,
+    endOfDay ? 59 : 0,
+    endOfDay ? 999 : 0,
+  ).toISOString();
+};
+
 const BotSessionMonitor = forwardRef<
   BotSessionMonitorHandle,
   BotSessionMonitorProps
 >(function BotSessionMonitor({ botId }, ref) {
   const { t } = useTranslation();
   const [sessions, setSessions] = useState<SessionInfo[]>([]);
+  const [sessionTotal, setSessionTotal] = useState(0);
+  const [sessionPage, setSessionPage] = useState(0);
+  const [startDate, setStartDate] = useState('');
+  const [endDate, setEndDate] = useState('');
+  const [userQuery, setUserQuery] = useState('');
+  const [appliedUserQuery, setAppliedUserQuery] = useState('');
   const [selectedSessionId, setSelectedSessionId] = useState<string | null>(
     null,
   );
   const [messages, setMessages] = useState<SessionMessage[]>([]);
+  const [messageTotal, setMessageTotal] = useState(0);
+  const [messagePage, setMessagePage] = useState(0);
   const [loadingSessions, setLoadingSessions] = useState(false);
   const [loadingMessages, setLoadingMessages] = useState(false);
   const [copiedUserId, setCopiedUserId] = useState(false);
@@ -138,6 +166,8 @@ const BotSessionMonitor = forwardRef<
     Record<string, boolean>
   >({});
   const messagesContainerRef = useRef<HTMLDivElement>(null);
+  const sessionRequestIdRef = useRef(0);
+  const messageRequestIdRef = useRef(0);
   const { admins, reload: reloadAdmins } = useBotAdmins(botId);
   const [adminsDialogOpen, setAdminsDialogOpen] = useState(false);
   const [togglingAdmin, setTogglingAdmin] = useState<string | null>(null);
@@ -204,16 +234,33 @@ const BotSessionMonitor = forwardRef<
   };
 
   const loadSessions = useCallback(async () => {
+    const requestId = ++sessionRequestIdRef.current;
     setLoadingSessions(true);
     try {
-      const response = await httpClient.getBotSessions(botId);
+      const response = await httpClient.getBotSessions(botId, {
+        limit: SESSION_PAGE_SIZE,
+        offset: sessionPage * SESSION_PAGE_SIZE,
+        startTime: startDate
+          ? localDateBoundaryToISOString(startDate, false)
+          : undefined,
+        endTime: endDate
+          ? localDateBoundaryToISOString(endDate, true)
+          : undefined,
+        userQuery: appliedUserQuery || undefined,
+      });
+      if (requestId !== sessionRequestIdRef.current) return;
       setSessions(response.sessions ?? []);
+      setSessionTotal(response.total ?? 0);
     } catch (error) {
-      console.error('Failed to load sessions:', error);
+      if (requestId === sessionRequestIdRef.current) {
+        console.error('Failed to load sessions:', error);
+      }
     } finally {
-      setLoadingSessions(false);
+      if (requestId === sessionRequestIdRef.current) {
+        setLoadingSessions(false);
+      }
     }
-  }, [botId]);
+  }, [appliedUserQuery, botId, endDate, sessionPage, startDate]);
 
   useImperativeHandle(
     ref,
@@ -224,25 +271,39 @@ const BotSessionMonitor = forwardRef<
   );
 
   const loadMessages = useCallback(
-    async (sessionId: string) => {
+    async (sessionId: string, page: number) => {
+      const requestId = ++messageRequestIdRef.current;
       setLoadingMessages(true);
       setExpandedToolCallIds({});
       try {
-        const messagesRes = await httpClient.getSessionMessages(sessionId);
+        const messagesRes = await httpClient.getSessionMessages(
+          sessionId,
+          MESSAGE_PAGE_SIZE,
+          page * MESSAGE_PAGE_SIZE,
+        );
+        if (requestId !== messageRequestIdRef.current) return;
         const sorted = (messagesRes.messages ?? []).sort(
           (a, b) =>
             new Date(a.timestamp).getTime() - new Date(b.timestamp).getTime(),
         );
         setMessages(sorted);
+        setMessageTotal(messagesRes.total ?? 0);
 
         try {
+          const analysisParams = new URLSearchParams();
+          if (sorted.length > 0) {
+            analysisParams.set('startTime', sorted[0].timestamp);
+            analysisParams.set('endTime', sorted[sorted.length - 1].timestamp);
+          }
           const analysisRes = await httpClient.get<{
             tool_calls?: SessionToolCall[];
           }>(
-            `/api/v1/monitoring/sessions/${encodeURIComponent(sessionId)}/analysis`,
+            `/api/v1/monitoring/sessions/${encodeURIComponent(sessionId)}/analysis?${analysisParams.toString()}`,
           );
+          if (requestId !== messageRequestIdRef.current) return;
           setToolCalls(analysisRes?.tool_calls ?? []);
         } catch (analysisError) {
+          if (requestId !== messageRequestIdRef.current) return;
           console.error('Failed to load session tool calls:', analysisError);
           setToolCalls([]);
         }
@@ -259,6 +320,7 @@ const BotSessionMonitor = forwardRef<
           }>(
             `/api/v1/monitoring/feedback?botId=${encodeURIComponent(botId)}&limit=200`,
           );
+          if (requestId !== messageRequestIdRef.current) return;
 
           const map: Record<string, SessionFeedback> = {};
           if (feedbackRes?.feedback) {
@@ -273,9 +335,13 @@ const BotSessionMonitor = forwardRef<
           setFeedbackMap({});
         }
       } catch (error) {
-        console.error('Failed to load session messages:', error);
+        if (requestId === messageRequestIdRef.current) {
+          console.error('Failed to load session messages:', error);
+        }
       } finally {
-        setLoadingMessages(false);
+        if (requestId === messageRequestIdRef.current) {
+          setLoadingMessages(false);
+        }
       }
     },
     [botId],
@@ -285,16 +351,24 @@ const BotSessionMonitor = forwardRef<
     loadSessions();
   }, [loadSessions]);
 
+  useEffect(() => {
+    setSelectedSessionId(null);
+    setMessagePage(0);
+  }, [appliedUserQuery, botId, endDate, sessionPage, startDate]);
+
   useEffect(() => {
     if (selectedSessionId) {
-      loadMessages(selectedSessionId);
+      loadMessages(selectedSessionId, messagePage);
     } else {
+      messageRequestIdRef.current += 1;
+      setLoadingMessages(false);
       setMessages([]);
+      setMessageTotal(0);
       setToolCalls([]);
       setExpandedToolCallIds({});
       setFeedbackMap({});
     }
-  }, [selectedSessionId, loadMessages]);
+  }, [selectedSessionId, messagePage, loadMessages]);
 
   useEffect(() => {
     if (messages.length === 0 && toolCalls.length === 0) return;
@@ -552,6 +626,19 @@ const BotSessionMonitor = forwardRef<
   const selectedSession = sessions.find(
     (s) => s.session_id === selectedSessionId,
   );
+  const sessionPageCount = Math.max(
+    1,
+    Math.ceil(sessionTotal / SESSION_PAGE_SIZE),
+  );
+  const messagePageCount = Math.max(
+    1,
+    Math.ceil(messageTotal / MESSAGE_PAGE_SIZE),
+  );
+
+  const applyUserSearch = () => {
+    setSessionPage(0);
+    setAppliedUserQuery(userQuery.trim());
+  };
 
   return (
     <>
@@ -575,6 +662,65 @@ const BotSessionMonitor = forwardRef<
                 )}
               </span>
             </button>
+            <span className="text-[11px] text-muted-foreground tabular-nums">
+              {t('bots.sessionMonitor.totalSessions', {
+                defaultValue: '{{count}} sessions',
+                count: sessionTotal,
+              })}
+            </span>
+          </div>
+          <div className="p-1.5 border-b shrink-0 space-y-1.5">
+            <div className="flex gap-1">
+              <input
+                value={userQuery}
+                onChange={(event) => setUserQuery(event.target.value)}
+                onKeyDown={(event) =>
+                  event.key === 'Enter' && applyUserSearch()
+                }
+                aria-label={t('bots.sessionMonitor.userSearch', {
+                  defaultValue: 'User ID or name',
+                })}
+                placeholder={t('bots.sessionMonitor.userSearch', {
+                  defaultValue: 'User ID or name',
+                })}
+                className="h-7 min-w-0 flex-1 rounded border bg-background px-2 text-xs"
+              />
+              <button
+                type="button"
+                onClick={applyUserSearch}
+                className="h-7 rounded border px-2 text-[11px] hover:bg-accent"
+              >
+                {t('common.search', { defaultValue: 'Search' })}
+              </button>
+            </div>
+            <div className="grid grid-cols-2 gap-1">
+              <input
+                type="date"
+                value={startDate}
+                max={endDate || undefined}
+                onChange={(event) => {
+                  setSessionPage(0);
+                  setStartDate(event.target.value);
+                }}
+                aria-label={t('bots.sessionMonitor.startDate', {
+                  defaultValue: 'Start date',
+                })}
+                className="h-7 min-w-0 rounded border bg-background px-1 text-[10px]"
+              />
+              <input
+                type="date"
+                value={endDate}
+                min={startDate || undefined}
+                onChange={(event) => {
+                  setSessionPage(0);
+                  setEndDate(event.target.value);
+                }}
+                aria-label={t('bots.sessionMonitor.endDate', {
+                  defaultValue: 'End date',
+                })}
+                className="h-7 min-w-0 rounded border bg-background px-1 text-[10px]"
+              />
+            </div>
           </div>
           {/* Session List */}
           <ScrollArea className="flex-1 min-h-0">
@@ -601,7 +747,10 @@ const BotSessionMonitor = forwardRef<
                         'w-full text-left px-2.5 py-2 rounded-md transition-colors cursor-pointer',
                         isSelected ? 'bg-accent' : 'hover:bg-accent/50',
                       )}
-                      onClick={() => setSelectedSessionId(session.session_id)}
+                      onClick={() => {
+                        setSelectedSessionId(session.session_id);
+                        setMessagePage(0);
+                      }}
                     >
                       <div className="flex items-center justify-between mb-0.5">
                         <span className="text-sm font-medium truncate mr-2">
@@ -637,6 +786,29 @@ const BotSessionMonitor = forwardRef<
               </div>
             )}
           </ScrollArea>
+          <div className="h-8 border-t px-1.5 flex items-center justify-between shrink-0 text-[11px]">
+            <button
+              type="button"
+              aria-label={t('common.previous', { defaultValue: 'Previous' })}
+              disabled={sessionPage === 0 || loadingSessions}
+              onClick={() => setSessionPage((page) => Math.max(0, page - 1))}
+              className="p-1 rounded hover:bg-accent disabled:opacity-40"
+            >
+              <ChevronLeft className="size-3.5" />
+            </button>
+            <span className="tabular-nums text-muted-foreground">
+              {sessionPage + 1} / {sessionPageCount}
+            </span>
+            <button
+              type="button"
+              aria-label={t('common.next', { defaultValue: 'Next' })}
+              disabled={sessionPage + 1 >= sessionPageCount || loadingSessions}
+              onClick={() => setSessionPage((page) => page + 1)}
+              className="p-1 rounded hover:bg-accent disabled:opacity-40"
+            >
+              <ChevronRight className="size-3.5" />
+            </button>
+          </div>
         </div>
 
         {/* Right Panel: Messages */}
@@ -975,6 +1147,33 @@ const BotSessionMonitor = forwardRef<
                   )}
                 </div>
               </ScrollArea>
+              <div className="h-9 border-t px-3 flex items-center justify-center gap-3 shrink-0 text-xs">
+                <button
+                  type="button"
+                  disabled={messagePage === 0 || loadingMessages}
+                  onClick={() =>
+                    setMessagePage((page) => Math.max(0, page - 1))
+                  }
+                  className="inline-flex items-center gap-1 rounded px-2 py-1 hover:bg-accent disabled:opacity-40"
+                >
+                  <ChevronLeft className="size-3.5" />
+                  {t('common.previous', { defaultValue: 'Previous' })}
+                </button>
+                <span className="tabular-nums text-muted-foreground">
+                  {messagePage + 1} / {messagePageCount} · {messageTotal}
+                </span>
+                <button
+                  type="button"
+                  disabled={
+                    messagePage + 1 >= messagePageCount || loadingMessages
+                  }
+                  onClick={() => setMessagePage((page) => page + 1)}
+                  className="inline-flex items-center gap-1 rounded px-2 py-1 hover:bg-accent disabled:opacity-40"
+                >
+                  {t('common.next', { defaultValue: 'Next' })}
+                  <ChevronRight className="size-3.5" />
+                </button>
+              </div>
             </>
           )}
         </div>

web/src/app/infra/http/BackendClient.ts

@@ -474,8 +474,13 @@ export class BackendClient extends BaseHttpClient {
 
   public getBotSessions(
     botId: string,
-    limit: number = 100,
-    offset: number = 0,
+    options: {
+      limit: number;
+      offset: number;
+      startTime?: string;
+      endTime?: string;
+      userQuery?: string;
+    },
   ): Promise<{
     sessions: Array<{
       session_id: string;
@@ -495,8 +500,17 @@ export class BackendClient extends BaseHttpClient {
   }> {
     const queryParams = new URLSearchParams();
     queryParams.append('botId', botId);
-    queryParams.append('limit', limit.toString());
-    queryParams.append('offset', offset.toString());
+    queryParams.append('limit', options.limit.toString());
+    queryParams.append('offset', options.offset.toString());
+    if (options.startTime) {
+      queryParams.append('startTime', options.startTime);
+    }
+    if (options.endTime) {
+      queryParams.append('endTime', options.endTime);
+    }
+    if (options.userQuery) {
+      queryParams.append('userQuery', options.userQuery);
+    }
     return this.get(`/api/v1/monitoring/sessions?${queryParams.toString()}`);
   }
 
@@ -1376,7 +1390,9 @@ export class BackendClient extends BaseHttpClient {
   }> {
     return this.get(
       '/api/v1/user/space/authorize-url',
-      { redirect_uri: redirectUri },
+      {
+        redirect_uri: redirectUri,
+      },
       { skipWorkspace: true },
     );
   }

web/tests/unit/cloud-new-account-entry.test.mjs

@@ -0,0 +1,18 @@
+import assert from 'node:assert/strict';
+import fs from 'node:fs';
+import test from 'node:test';
+
+const source = fs.readFileSync(
+  new URL('../../src/app/login/page.tsx', import.meta.url),
+  'utf8',
+);
+
+test('normal Cloud login starts stateful Space OAuth', () => {
+  assert.match(source, /getSpaceAuthorizeUrl\(redirectUri\)/);
+  assert.doesNotMatch(source, /cloudEntry/);
+});
+
+test('invitation login remains on the OAuth callback path', () => {
+  assert.match(source, /getPendingInvitationToken\(\)/);
+  assert.match(source, /getSpaceAuthorizeUrl\(redirectUri\)/);
+});

web/tests/unit/embed-widget-assistant-dedupe.test.mjs

@@ -0,0 +1,201 @@
+import assert from 'node:assert/strict';
+import fs from 'node:fs';
+import path from 'node:path';
+import test from 'node:test';
+import vm from 'node:vm';
+import { fileURLToPath } from 'node:url';
+
+const testDirectory = path.dirname(fileURLToPath(import.meta.url));
+const widgetPath = path.resolve(
+  testDirectory,
+  '../../../src/langbot/templates/embed/widget.js',
+);
+const widgetSource = fs.readFileSync(widgetPath, 'utf8');
+
+class FakeElement {
+  constructor(tagName) {
+    this.tagName = tagName;
+    this.children = [];
+    this.className = '';
+    this.dataset = {};
+    this.style = {};
+    this.listeners = {};
+    this._innerHTML = '';
+  }
+
+  appendChild(child) {
+    this.children.push(child);
+    return child;
+  }
+
+  setAttribute(name, value) {
+    this[name] = String(value);
+  }
+
+  addEventListener(event, listener) {
+    this.listeners[event] = listener;
+  }
+
+  click() {
+    this.listeners.click?.({});
+  }
+
+  attachShadow() {
+    this.shadowRoot = new FakeElement('shadow-root');
+    return this.shadowRoot;
+  }
+
+  get classList() {
+    return {
+      add: (...names) => {
+        const classes = `${this.className} ${names.join(' ')}`
+          .trim()
+          .split(/\s+/);
+        this.className = [...new Set(classes)].join(' ');
+      },
+    };
+  }
+
+  set textContent(value) {
+    this._innerHTML = String(value ?? '');
+  }
+
+  set innerHTML(value) {
+    this._innerHTML = String(value ?? '');
+  }
+
+  get innerHTML() {
+    return this._innerHTML;
+  }
+
+  querySelectorAll(selector) {
+    const matches = [];
+    for (const child of this.children) {
+      if (
+        selector.startsWith('.') &&
+        child.className.split(/\s+/).includes(selector.slice(1))
+      ) {
+        matches.push(child);
+      }
+      matches.push(...child.querySelectorAll(selector));
+    }
+    return matches;
+  }
+
+  querySelector(selector) {
+    return this.querySelectorAll(selector)[0] ?? null;
+  }
+}
+
+class FakeDocument {
+  constructor() {
+    this.body = new FakeElement('body');
+    this.head = new FakeElement('head');
+    this.readyState = 'complete';
+    this.currentScript = { getAttribute: () => null };
+  }
+
+  createElement(tagName) {
+    return new FakeElement(tagName);
+  }
+
+  getElementById(id) {
+    const find = (element) => {
+      if (element.id === id) return element;
+      for (const child of element.children) {
+        const match = find(child);
+        if (match) return match;
+      }
+      return null;
+    };
+    return find(this.body) ?? find(this.head);
+  }
+}
+
+class FakeWebSocket {
+  static OPEN = 1;
+  static instances = [];
+
+  constructor(url) {
+    this.url = url;
+    this.readyState = FakeWebSocket.OPEN;
+    FakeWebSocket.instances.push(this);
+  }
+
+  send() {}
+}
+
+function launchWidget() {
+  FakeWebSocket.instances = [];
+  const document = new FakeDocument();
+  const window = {
+    crypto: { randomUUID: () => '00000000-0000-4000-8000-000000000000' },
+    sessionStorage: { getItem: () => null, setItem: () => {} },
+  };
+  const context = vm.createContext({
+    document,
+    window,
+    navigator: { clipboard: { writeText: () => Promise.resolve() } },
+    WebSocket: FakeWebSocket,
+    fetch: () => new Promise(() => {}),
+    requestAnimationFrame: () => 0,
+    setTimeout: () => 0,
+    clearTimeout: () => {},
+    setInterval: () => 0,
+    clearInterval: () => {},
+  });
+
+  vm.runInContext(widgetSource, context, { filename: widgetPath });
+  const root = document.getElementById('langbot-widget-root');
+  assert.ok(root, 'widget should initialize');
+  root.shadowRoot.querySelector('.lb-bubble').click();
+  const socket = FakeWebSocket.instances.at(-1);
+  assert.ok(socket, 'opening widget should connect its WebSocket');
+
+  return {
+    receive(message) {
+      socket.onmessage({
+        data: JSON.stringify({ type: 'response', data: message }),
+      });
+    },
+    assistantMessages() {
+      return root.shadowRoot.querySelectorAll('.lb-msg-assistant');
+    },
+  };
+}
+
+function assistant(id, content) {
+  return { id, role: 'assistant', content, is_final: true };
+}
+
+test('renders a non-empty reply after an empty assistant frame', () => {
+  const widget = launchWidget();
+
+  widget.receive(assistant('thought', ''));
+  widget.receive(assistant('answer', 'visible answer'));
+
+  const messages = widget.assistantMessages();
+  assert.equal(messages.length, 2);
+  assert.equal(
+    messages[1].querySelector('.lb-msg-bubble').innerHTML,
+    'visible answer',
+  );
+});
+
+test('still drops a duplicate non-empty assistant frame', () => {
+  const widget = launchWidget();
+
+  widget.receive(assistant('answer-1', 'same answer'));
+  widget.receive(assistant('answer-2', 'same answer'));
+
+  assert.equal(widget.assistantMessages().length, 1);
+});
+
+test('still keeps two distinct non-empty assistant frames', () => {
+  const widget = launchWidget();
+
+  widget.receive(assistant('answer-1', 'first answer'));
+  widget.receive(assistant('answer-2', 'second answer'));
+
+  assert.equal(widget.assistantMessages().length, 2);
+});

web/tests/unit/session-monitor-pagination.test.mjs

@@ -0,0 +1,139 @@
+import assert from 'node:assert/strict';
+import fs from 'node:fs';
+import path from 'node:path';
+import test from 'node:test';
+import { fileURLToPath } from 'node:url';
+
+const root = path.resolve(
+  path.dirname(fileURLToPath(import.meta.url)),
+  '../..',
+);
+const repoRoot = path.resolve(root, '..');
+const read = (file) => fs.readFileSync(path.join(root, file), 'utf8');
+const readRepo = (file) => fs.readFileSync(path.join(repoRoot, file), 'utf8');
+const includes = (source, token, message) =>
+  assert.ok(source.includes(token), message);
+
+test('session list request supports a server-side page and operator filters', () => {
+  const client = read('src/app/infra/http/BackendClient.ts');
+  includes(client, 'startTime?: string', 'client accepts a start date');
+  includes(client, 'endTime?: string', 'client accepts an end date');
+  includes(client, 'userQuery?: string', 'client accepts a user query');
+  includes(
+    client,
+    "queryParams.append('offset', options.offset.toString())",
+    'client sends the requested session offset',
+  );
+  includes(
+    client,
+    "queryParams.append('userQuery', options.userQuery)",
+    'client sends the user query',
+  );
+
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  for (const token of [
+    'SESSION_PAGE_SIZE',
+    'sessionTotal',
+    'sessionPage',
+    'startDate',
+    'endDate',
+    'userQuery',
+  ]) {
+    includes(monitor, token, `monitor includes ${token}`);
+  }
+});
+
+test('session detail requests and renders a bounded message page', () => {
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  for (const token of [
+    'MESSAGE_PAGE_SIZE',
+    'messageTotal',
+    'messagePage',
+    'page * MESSAGE_PAGE_SIZE',
+  ]) {
+    includes(monitor, token, `message pagination includes ${token}`);
+  }
+});
+
+test('backend filters sessions by user id or user name in the existing endpoint', () => {
+  const controller = readRepo(
+    'src/langbot/pkg/api/http/controller/groups/monitoring.py',
+  );
+  const service = readRepo('src/langbot/pkg/api/http/service/monitoring.py');
+  includes(
+    controller,
+    "quart.request.args.get('userQuery')",
+    'route accepts userQuery',
+  );
+  includes(controller, 'user_query=user_query', 'route forwards userQuery');
+  includes(
+    service,
+    'user_query: str | None = None',
+    'service accepts userQuery',
+  );
+  includes(
+    service,
+    'MonitoringSession.user_id.ilike',
+    'service searches user ids',
+  );
+  includes(
+    service,
+    'MonitoringSession.user_name.ilike',
+    'service searches user names',
+  );
+});
+
+test('stale session and message page responses cannot overwrite the latest page', () => {
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  for (const token of [
+    'sessionRequestIdRef',
+    'messageRequestIdRef',
+    'requestId !== sessionRequestIdRef.current',
+    'requestId !== messageRequestIdRef.current',
+    'messageRequestIdRef.current += 1',
+  ]) {
+    includes(monitor, token, `stale response guard includes ${token}`);
+  }
+});
+
+test('changing the session page or filters clears the selected detail', () => {
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  includes(monitor, 'setSelectedSessionId(null)', 'selection is cleared');
+  includes(
+    monitor,
+    '[appliedUserQuery, botId, endDate, sessionPage, startDate]',
+    'page and filters invalidate the selected session',
+  );
+});
+
+test('date filters use the operator local calendar day', () => {
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  includes(
+    monitor,
+    'localDateBoundaryToISOString(startDate, false)',
+    'local start-of-day conversion',
+  );
+  includes(
+    monitor,
+    'localDateBoundaryToISOString(endDate, true)',
+    'local end-of-day conversion',
+  );
+});
+
+test('session tool calls are bounded to the visible message page', () => {
+  const monitor = read(
+    'src/app/home/bots/components/bot-session/BotSessionMonitor.tsx',
+  );
+  includes(monitor, "analysisParams.set('startTime'", 'analysis page start');
+  includes(monitor, "analysisParams.set('endTime'", 'analysis page end');
+});