Zurich Patch 10

The Zurich Patch 10 release contains important problem fixes.

Zurich Patch 10 was released on June 16, 2026.

• Build date: 06-12-2026_2311
• Build tag: glide-zurich-07-01-2025__patch10-05-22-2026

Important:

For more information about how to upgrade an instance, see ServiceNow upgrades.

For more information about the release cycle, see the ServiceNow Release Cycle.

Note:

This ServiceNow AI Platform® major family release is now available in ServiceNow's Regulated Market environments. For more information about services available in isolated environments, see KB0743854.

For a downloadable, sortable version of the fixed problems in this release, click here.

Overview

Zurich Patch 10 includes 356 problem fixes in various categories. The chart previous shows the top 10 problem categories included in this patch.

Figure 1. Top 10 problem categories

Security-related fixes

Zurich Patch 10 includes fixes for security-related problems that affected certain ServiceNow® applications and the ServiceNow AI Platform®. We recommend that customers upgrade to this release for the most secure and up-to-date features. For more details on security problems fixed in Zurich Patch 10, refer to KB3046592.

Changes in Zurich Patch 10

Authentication factors

Authentication factors help identify and verify callers, allowing only authorized users to access AI voice agents on the ServiceNow AI Platform.

Explore authentication factors for AI voice agents

Email OTP delivers a temporary numeric code to the caller’s registered email address. It is easy to deploy and familiar to most users. Callers can enter the code via keypad or by speaking the digits. Email OTP is susceptible to email account compromise and phishing, and should not be used as a standalone factor for sensitive operations.

Configure authentication factors for AI voice agents

To secure voice agent environments, configure authentication factors that first identify the caller, then authenticate them before granting access.

Theming for AI Search in Service Portal

The default value for the background color for title highlights in search results is transparent.

Notable fixes

The following problems and their fixes are ordered by potential impact to customers, starting with the most significant fixes.

Problem	Short description	Description	Steps to reproduce
Authentication PRB1969882 KB2718536	Login screen performance issues on iOS 26.2 RC and in Zurich instances	After upgrading to iOS 26.2, users are unable to log in to any instance using the Now Mobile app. The login page is extremely slow or unresponsive, and it can fail to load or accept credentials. The issue affects multiple users and device models. It's reproducible across different instances, and it also impacts the Agent app.	On iOS, navigate to the Now Mobile app. Connect to a Zurich or later family release instance. Observe that the screen loads slowly and is non-responsive.
Database Persistence - Data Access PRB2007256 KB2925734	The text search doesn't return results in non-English languages	After downloading language plugins and switching the system language to a non-English language, results aren't returned when the property 'glide.db_query.replace _distinct_with_groupby' is set to 'false.' When the language is set to English, results are returned.	Refer to the listed KB article for details.
Instance Scan PRB1992382 KB2901125	Instance scan jobs get stuck in a sleep loop for days, which causes the subsequent scans to fail	Instance scan findings are written to the database asynchronously and are tracked using a global counter. If any write fails, the counter doesn't decrement properly; it goes up but never comes back down. This causes all future scans to hang indefinitely, waiting for a counter that will never reach zero. There's no timeout or logging to flag this, so scans can get stuck silently.	Refer to the listed KB article for details.
Password2 Encryption PRB2015257 KB2978056	There's excessive growth on the sys_rollback_incremental table and higher binlog generation due to 'Mass Encryption Job'	This causes a large increase in the disk space on the database server.	Refer to the listed KB article for details.
Related Lists PRB2025356 KB3032264	User can't add records in a related list because of the strict ACL check and missing ACLs for required roles	Users with missing ACLs aren't able to add new records in the related list due to a strict ACL check.	Refer to the listed KB article for details.
Request Management PRB1973824 KB2815909	Added approvers names don't appear on the Approver list of Request and RITM	When adding approvers to a Request (REQ) or Requested Item (RITM) via the Multi-Row Add (MRA) component in the 'Related Records' section of Service Operations Workspace (SOW), a success message is displayed confirming that the approvers have been added. However, the Approver list doesn't display the newly added approver names. Similarly, when attempting to add approvers from the UI16 related list, the approvers aren't added.	Refer to the listed KB article for details.
Virtual Agent PRB2007255 KB3045151	There's memory pressure on nodes due to high memory for the cache 'com.glide.cs.qlue.module. coma.MessageBatchingSession'	Users with 2GB nodes may encounter memory issues that can cause the events process jobs to yield.	Run a heap dump. Observe that MacMessageBatchingSession or MessageBatchingSession uses over 50 MB of memory.

All other fixes

Problem	Short description	Description	Steps to reproduce
Access Control PRB1892720	'Applies to' in the Deny Unless Read ACL doesn't work for users with elevated privileges	It always returns true for any record for users with elevated privileges.
Activity Stream PRB2005727	Navigation fails for sources beyond sliding windows in an activity stream	When selecting a citation source in an incident summary response, navigation doesn't work properly if the cited activity (comment or email) is more than three pages away or before the current activity stream view. The user receives a message stating 'The source may be hidden in your current view', even though the source exists and isn't filtered.	Navigate to any instance on track/anowassist. Open Service Operations Workspace. Open a record with 200+ activity stream entries. Scroll to a recent AI-generated response containing citation references. Select Show sources. The cited source (comment/email) should be an older entry — at least 3+ pages before the current viewport. Select the citation link with worknotes/comments.
Activity Stream PRB2016617	Unguarded Jelly expression in form.xml causes RhinoEcmaError warnings on every classic form load	On any classic UI16 form load, the platform Jelly template evaluator throws a RhinoEcmaError warning in the system log. The warning is present on every classic form, regardless of application scope or record type. There is no functional outage; forms load and save correctly, and all transactions return HTTP 200 with normal response times. The main impact is high-volume system log noise.	Log in to a Zurich instance. Open any classic UI16 form record without a sysparam_citation query parameter in the URL. Example without activity stream: Navigate to /sys_dictionary.do and open any record. Example with activity stream: Navigate to /incident.do and open any record. After the form loads, open the system log. Expected behavior: No RhinoEcmaError warnings are generated during a standard form load. Actual behavior: On forms without an activity stream, one warning entry is generated per form load. On forms with an activity stream, two warning entries are generated per form load.
Activity Stream PRB2018487	A citation URL is retained and fields are highlighted when a user logs in to an instance again		Navigate to any instance with the 'Record Summarization' skill activated. Open any incident in Workspace. Ensure that the activity stream has several entries for the 'Record Summarization' skill to pick up. Run the 'Record Summarization' skill. Select a citation source that points to an activity stream entry. Log out of the instance. Log back in to the instance and return to the same record. Expected behavior: The citation URL shouldn't be preserved after triggering the citation source flow. Actual behavior: The URL that is created after selecting a citation source is preserved, and causes the citation scroll and highlight to trigger.
Activity Stream PRB2023624	For the Core UI only, update the Activity Stream icon for AI specialist vs Agentic Workflow	This is a product update.
Activity Stream PRB2033885	A new activity comment has the type as 'Field changes' instead of 'Comments'		Open any incident in UI16. Post a comment. Note that the comment type (top right of tile) is 'Field changes [dot timestamp]' instead of 'Comments [dot timestamp]'.
Agent Chat PRB1943203	When hovering through the response templates, the preview is obstructed/hidden behind the case form	The response templates are obstructed in the header.	Open an instance on a Safari browser. Open the same instance in the incognito window of Safari or in a new window on a Chrome browser. Request a live agent chat from a virtual agent (for example, through the ESC portal). Accept the chat as an agent in the SOW/HR Agent Workspace (keep the plugins on the latest version for workspace). Resize the browser window to be small. Type /r to open the response template. Hover through the response templates. Observe that the preview is obstructed/hidden behind the case form.
Agent Chat PRB2000390	The date format is incorrect in the Active Chat window	The date/time format is displayed as 'MMM DD, YYYY HH:MM a.m.' instead of 'DD-MM-YYYY HH:MM:SS' in the Active Chat window.	Enable the conversation history by navigating to Conversational Interface > Agent chat settings. Set the value for 'Number of conversations to display'. Navigate to an instance where Advanced Work Assignment (AWA) is configured in two browsers. Log in as Abel Tuter. Set the date/time preference to 'DD-MM-YYYY HH:MM:SS'. Initiate a chat from the portal. Connect to a live agent. Log in as an agent. Accept the chat. Close the conversation. Initiate a new chat. Connect to a live agent. Accept the chat as an agent. Observe that the previous chat conversation is visible in the Active Chat window. Expected behavior: The date/time format should display as 'DD-MM-YYYY HH:MM:SS' at the top of the Active Chat window. Actual behavior: The date/time format is displayed as 'MMM DD, YYYY HH:MM a.m.' in the Active Chat window.
Agent Chat PRB2012130	There's a misleading VA Message while agent connection is still in progress	Before the agent accepts the workItem, there's a VA Message that says 'Live Agent Engaged'.	Connect to LA. Observe that the VA Message says 'Live Agent Engaged' before the agent accepts the workItem.
Agent Chat PRB2013475	LA conversation ends when sys_prop is set to empty	When sys_prop is deleted, the off-Glide LA topic is still called for LA transfer.	Scenario 1: Create the sys_prop, but leave the value as empty. Connect to LA. Observe that the conversation ends. Scenario 2: Delete the sys_prop. Start a LA topic. Observe that the off-Glide LA topic is still called for LA transfer, even when sys_prop is deleted.
Agent Chat PRB2013947	Inbox Advanced Work Assignment (AWA) interactions aren't presented according to the Configuration field and order for the walkup experience	When an agent receives multiple chats in Service Operations Workspace (SOW), they are initially displayed in descending order, with the most recent chat at the top. However, after a refresh, the chat order appears to change and seemingly displays in a different order.	Open an instance. Impersonate an agent with the role 'system_administrator'. Open SOW. Set the agent status as 'available'. Open the instance again in new incognito tab, or in any another browser. Initiate 4-5 live agent chats by impersonating the end users from service portal. Check the order of the chats the agent received. Notice that it would be in descending order, with the most recent chat appearing first. Refresh the page as an agent. Notice that the chat order is jumbled in a different order.
Agent Chat PRB2014178	Optimize the request to fetch previous interactions for a requester to show conversation history in Agent chat	If there are no custom filters, there is no need to create the list of interaction records by going through all the returned records from the DB query.	Enable the 'Conversation history' feature in Agent chat settings. Set the limit to 10. Ensure that there's many closed interactions (more than 100) for Abel Tuter. Check the server memory in /stats.do. Start a new conversation as Abel Tuter. Accept the work item as Beth Anglin. Ensure that the history shows in the agent chat. Check the server memory again in /stats.do. Observe that the server memory has gone up significantly.
Agent Chat PRB2021490	Enhance the chat desktop notification	When the user selects the new chat message desktop notification, they should get focused back on the workspace tab. The workspace tab that the new chat message belongs to should also be focused and opened.
Agent Chat PRB2022900	AWA workItem responder changes	Update WorkItemResponder's onEnter and OnExit to add and remove notifications in the ui_notification_inbox table. Also, create a system property (glide.awa.work_ item.notifications.enabled) to determine whether to show or hide user preference at the agent level.
Agent Chat PRB2022906	Add user preference for app shell notifications	Add support to display a new user preference 'Inbox Workspace Notifications' based on the system property glide.awa.work_ item.notifications.enabled.
Agent Chat PRB2022911	Polaris app shell changes	This is a product update.
Agent Chat PRB2023682	Add the interaction number to desktop notifications to differentiate between interactions	The desktop notification should contain the IMS number to indicate which chat the notification is for.	Enable desktop notifications in the agent workspace. Start a live agent chat so that a new desktop notification is rendered. Expected behavior: The desktop notification contains the IMS number to indicate which chat the notification is for. Actual behavior: The desktop notification only says that a new chat arrived.
Agile Development PRB2022589	Add 'None' choice to the rm_epic.status field and make it the default for new epics	The Epic Status field (rm_epic.status) currently doesn't include a 'No status' option, which prevents epics from having an unset/neutral status. This causes ambiguity in reporting and makes it difficult to distinguish between epics that have not yet been assigned a status vs. those intentionally marked as green, yellow, or red.	Navigate to Agile Development > Epics (or open an Epic from a board/backlog). Open any active Epic record Select the Status field. Expected behavior: 'No status' is available as a selectable option to represent epics with no status assigned, enabling accurate status transition tracking for usage analytics. Actual behavior: 'No status' isn't available as an option in the Status field list. Available options are limited to green, yellow, and red.
AI Agents (Glide Family) PRB1997916	There's an extra 'response' layer in the JSON input passed by the agent to a scripted tool, causing performance issues	The final tool of creating the regulatory change tasks expects the input to be in the format - { 'finalPlan_json': {...}, 'finalPlan_html': '...' }. But with the latest code, one 'response' layer is added, which causes performance issues.	Scenario 1: Log in to an instance. Navigate to /now/risk/compliance/record/sn_grc_reg_change_regulatory_feed. Open a record. Navigate to AskNowAssist > Get action oplan. Select Yes after the action plan is displayed. Notice that number of regulatory tasks in the related list doesn't change. Tasks aren't being created. Scenario 2: Create a skill tool that outputs a JSON object. Create a script tool that takes the skill output JSON as input. Return that input as is for displaying in VA. Expected behavior: The JSON object returned by the skill tool should display in VA. Actual behavior: The JSON object will be wrapped/contained in a format like this: { response: { model_output: 'actual JSON object' }}.
AI Agents (Glide Family) PRB2011068	Data to Glide from off-Glide isn't getting logged with the actual user	Any record update or creation doesn't have the created_by or updated_by set as the actual user.	Make a set cache call to Glide from off-Glide. Observe that any record update or creation doesn't have the created_by or updated_by set as the actual user.
AI Agents (Glide Family) PRB2018916	AI Agent execution fails with 'Tool access denied' for triggering user after the update set migration	After migrating AI Agents (built in Agent Studio) from a development environment to a client test environment via update set, the agents fail to execute. The triggering user receives an error indicating that they don't have access to the agent's tools, and the agent never starts.	In a source instance, build an AI Agent in Agent Studio with one or more tools (for example, a subflow tool). Configure the agent's ACL to allow any authenticated user to trigger. Configure the agent's dynamic user with the snc_internal role. Verify the agent runs successfully end-to-end in the source instance. Capture the agent, its tools, ACLs, and dynamic-user configuration in an update set. Move the update set to the target instance. Commit it. As any authenticated user on the target instance (admin or otherwise), trigger the agent. Observe that the agent fails to start. An error states that the triggering user doesn't have access to the agent's tools.
AI Agents (Glide Family) PRB2022336	Caching improvements for NextWave architecture	This is a product update.
AI Agents (Glide Family) PRB2022344	Java changes for conversation history support for a Knowledge Graph tool in AI Agent Studio	This is a product update.
AI Agents (Glide Family) PRB2022649	AI Agents are stuck indefinitely and the facts response is empty	When running AutoEval for the Enterprise Architecture Explorer – Query Agent, the evaluation run gets stuck indefinitely while processing certain questions and datasets. During execution, AutoEval doesn't progress to completion, progress keeps retrying without advancing, and the run never completes successfully. GenAI logs repeatedly show: 'JSON{ 'facts': []}Show more lines'. Despite retries, no new facts are generated and the evaluation pipeline makes no forward progress.	Navigate to Now Assist Skill Kit > Evaluation Results Dashboard. Start an AutoEval run for: Agent: Enterprise Architecture Explorer – Query Agent Dataset containing multiple evaluation questions Allow the run to process. Observe that the evaluation stalls on certain questions, AutoEval keeps retrying internally, and the run never completes.
AI Agents (Glide Family) PRB2025221	Cache service intermittently fails and gets calls across different keys
AI Agents (Glide Family) PRB2025859	The OffGlideScriptObject. generateAuthorizationInfo API creates JSON Web Tokens (JWT) with current session users	The API sn_cs_offglide.OffGlideScriptObject .generateAuthorizationInfo() creates JWT with current user sessions, even though the userID value is passed in the request.
AI Agents (Glide Family) PRB2025974	User session is logged out when opening AI-generated interaction records	The user session is logged out when opening AI-generated interaction records that were created/updated by the incident_intelligence_agent. The interaction record remains in the WIP state, even though the associated conversation has been marked as faulted. The issue is specific to the in Service Operations Workspace view, as opening the same record in platform view works without issue.
AI Agents (Glide Family) PRB2030410	An instance isn't invoking DARE calls due to new properties not being allow listed in the cache configuration's invalidation script		Turn on the DARE sysprop. Open Now Assist Portal. Run the utterance 'List my incidents'. Expected behavior: The response should be received from DARE. Actual behavior: The response is coming from NextWave.
AI Agents (Glide Family) PRB2030948	Tools using data stream actions aren't able to retrieve data in NAVA		Log in to an instance with user credentials. In Service Portal, give the utterance, such as 'using smartsheet agents, look up groups stream'. Observe that the tool isn't able to retrieve data.
AI Experience Framework - Glide PRB2015131	Turn on UI validation for 'Record Update' and 'Create Agent'	This is a product update.
AI Experience Framework - Glide PRB2018366	/api/now/aiux_service/sysprops doesn't enforce an allowlist
AI Experience Framework - Glide PRB2024769	Chat gate service uses a GlideRecord query with ACL evaluation, blocking non-admin users from reading the 'Deployment channel' table	In chat-gate-service.js (line 13), the chatEnabled check uses glideRecord_query to query the 'Deployment channel' table. This evaluates all ACLs for the current user, so only admin users have the necessary read access to that table. Non-admin users are blocked from fetching chatEnabled, which prevents the chat from functioning for them.	Log in as a non-admin user. Navigate to an AI Control Tower (AICT) page where the chat is expected to load. Observe that the chat doesn't initialize because chatEnabled can't be fetched. Log in as an admin user and repeat. Observe that the chat works correctly. Expected behavior: chatEnabled is fetched successfully regardless of the user role. Actual behavior: glideRecord_query evaluates ACLs and denies non-admin users read access to the deployment channel table.
AI Experience Framework - Glide PRB2027126	AIX Webserver Glide integration	This is a product update.
AI Experience Framework - Glide PRB2027128	Migration in Glide	This is a product update.
AI Gateway - Security PRB2024062	AI Gateway security Q2 complete feature update	This is a product update.
AI Search (Glide) PRB2001677	A Knowledge Graph (KG) NLQ call should pass the caller ID in the Caller field, not meta		In an instance with KG, navigate to 'Make a kgnlq query'. Observe one_api_service_ plan_feature invocation. It doesn't match the design with respect to the Caller field.
AI Search (Glide) PRB2002346	When a KB only has an embedded KBB with embedded_match = false, the KB should send itself instead of the KBB	The KBB content is sent to the LLM instead of KB.	Create a KB titled 'abc' with some text content and an embedded KBB 'defhi'. Perform a search for 'abc' in the search preview with the 'Now Assist in VA' profile in sys_generative_ai_log. Observe the KB sent to LLM. Expected behavior: The content of the KB is sent to LLM. Actual behavior: Only the KBB content (defhi) is sent to the LLM.
AI Search for Service Portal PRB2023882	The 'View all results' option isn't available in the chat response shared by assistant		Open an instance. Impersonate the system admin. Navigate to /csm portal. Input 'How to repair fan fuse'. Expected behavior: The 'View all results' option is visible. Actual behavior: There is no 'View all results' option in the side panel bottom.
AI Search UX PRB1989246	When records are searched in global search, it isn't opening the form directly	When a user tries to search the incident number in the global search when the zoom size is 150% and left navigator is pinned, then the incident record isn't opening directly. It instead displays a search result page. However, when the zoom size is 100%, then the incident record is opening directly in a form page.	Navigate to an instance where AI Search is turned on. Update the browser zoom to 150%. Pin the 'All' option in the filter navigator. Copy any incident number. Search with the incident number in the global search and press enter. See the search results page. Expected behavior: Users should see the incident form record directly. Actual behavior: Users are seeing the search results page.
AI Search UX PRB1991340	Search input intermittently skips characters	The search input intermittently drops characters when users type quickly, especially under a heavy event load.	Open any workspace/heavy loaded page. In the global search, type quickly. Observe intermittent character loss.
AI Search UX PRB1994212	The hidden description in the source code is visible in the chatbot for 'Order guide' and not for 'Catalog item' in Now Assist Virtual Agent (NAVA)	On the catalog item, it's possible to hide the description using the following in the code source: '<div style=''display: none;''> TEXT TO HIDE.' When attempting to do the same in the order guide, it's not working properly in the chatbot, and both texts are visible. This issue occurs only on the chatbot preview of the order guide, and not when consulting it directly from portal.
AI Search UX PRB1998976	Search event records don't indicate the search mode (keyword, hybrid, or semantic)	When a search is executed in an AIS-enabled portal with Hybrid Search enabled on the corresponding search application, the resulting sys_search_event and sys_search_signal_event records don't capture the search mode (keyword, hybrid, or semantic) that was used to execute the query. This makes it impossible to distinguish query types in search analytics and signal data.	Turn on Hybrid Search on a Search auto-complete Configuration (SAC). Navigate to a Dynamic Window-enabled portal that uses the configured SAC. Perform a search in the portal. Open the corresponding sys_search_event and sys_search_signal_event records generated by the query. Observe that neither record contains any field or indicator specifying the search mode used (keyword, hybrid, or semantic).
AI Search UX PRB2005193	Exact Match, when configured to honor conditions of search sources, should be configurable for a single application and not the entire instance	A system property called glide.search.exact_match _use_search_source_filter enables all applications on the instance with exact match limited to those tables included in a search source and honors the conditions set within them. Users don't expect this feature to be enabled at the instance level, but require it to be enabled at the application level.	Set the system property glide.search.exact_match_use_search_source_filter to true. Notice how setting this property to true applies to all search applications. Instead, it should be enabled/inactive for a single application.
AI Search UX PRB2011697	RAG popular search suggestion reader doesn't consider sys_rag_search _suggestion.active	The sys_rag_search_suggestion records with active=false are still returned.	Generate sys_rag_search_suggestion entries. Query the Suggestion API. Observe that the sys_rag_search_suggestion records with active=false are still returned.
AI Search UX PRB2014138	$sp-search-result-title- highlight--background-color in sp-variables.scss	There is no default assigned for variabl '$sp-search-result-title- highlight--background-color'.	Log in to Portal with AI search enabled. Search for 'laptop' and navigate to the AI search page. Notice that the variable '--search-result-title- highlight--background-color: $sp-search-result-title- highlight--background-color;' is not resolved.
AI Search UX PRB2016024	Synthesized responses on a mobile sized screen don't show 'Show full answer' after truncation in Zurich on Safari	The answer is truncated without the option to expand: 'Show full answer'.	Find a Zurich instance with Multi Content Synthesized GRs working in Portal that can work on Safari mobile browser. Search for a term that would yield a Synthesized GR. Observe that the answer is truncated without the option to expand: 'Show full answer'.
AI Search UX PRB2021873	Facets aren't rendered as the sensitivity filter fails quietly	Facets aren't returned because of a failure to fetch hasMatch.	Make sure the Now Assist setup isn't done. Make sure the sensitivity filter is null. Observe that facets aren't returned as a result of before line failing to fetch hasMatch.
AI Search UX PRB2024551	New/different experience in the search when AI Search is turned on in the portal level		Navigate to an instance. Impersonate as system admin. Navigate to /csm portal. In search bar, search 'Controllers and applications'. Observe that there's no follow up and articles displayed are shown previously in the enhanced chat.
AI Search UX PRB2024704	RAG search signals aren't getting logged	After certain searches, the sys_search_event and sys_search_signal_event records should be created. The response from the REST API call should have a signalsCorrelationData object on it. Instead, no records get created and no signalsCorrelationData is returned.
AI Search UX PRB2026614	Auto-Enable Hybrid Search on Now Assist for Search Installation	When a new users installs the 'Now Assist for Search' plugin, the system should automatically enable Hybrid Search for all search applications configured in the instance. This ensures new users benefit from Hybrid Search by default without requiring manual configuration. The behavior applies to all new user installations going forward and shouldn't retroactively affect existing users who have already installed Now Assist for Search.
AI Search UX PRB2031624	Enhanced Chat's search bar appears on portals with NextWave	Note that the NextWave omni-bar displays on the homepage as expected. However, the search bar from Enhanced Chat is also appearing on other pages unexpectedly.
Analytics Data API PRB1844011	Using 'Trend by week' leads to an incorrect record count when the new year starts	When using 'Trend by' with the options 'Opened', 'Calendar', 'Standard Calendar' and 'per-week' and running the report, the record count is 3 but the List view record count is 5.
Analytics Data API PRB1919741	Calendar API doesn't accept requests from external users	The following message appears: 'message': 'External User Not Authorized'.	Provision an instance with the plugin 'com.glide.explicit_roles' installed. Open a record for the user 'Abel.tuter'. Replace Abel.tuter's current role with 'snc_external'. Set a new password for Abel.tuter. Using Postman or any other tool, make a request using the credentials from step 4. Expected behavior: The user is able to make the request. Actual behavior: The following message appears: 'message': 'External User Not Authorized'.
Analytics Data API PRB1928857	The maximum number of groups on the heat map visualization in Platform Analytics resets to 10 when filtering	After saving and refreshing, the column count resets to 10. Instead, all selected groups should remain visible.	Open an instance. Create a heatmap data visualization. Add a 'Group By' under Columns with more than 10 columns. Select Apply. Observe that all columns display correctly at this point. Save and refresh the visualization or browser. Expected behavior: All selected groups remain visible after saving or refreshing. Actual behavior: The column count resets to 10 every time.
Analytics Data API PRB2003111	Pivot table using multiple data sources shows 'No data' on the widget when one of the data sources returns no data	Even though data is available, the chart shows 'No data available'.	Provision an instance with the com.snc.itsm_pa.demo plugin installed to have more test data. Open the list with records of the problem table (problem.LIST). Filter records by Assigned to = Abel Tuter and State = Assess so that two records appear. Remove one of the records. Move the other record to the Fix in Progress state by selecting the Fix button. Navigate to All > Library > Data visualizations. Select New. Select Visualization type: Pivot table. Specify the following data sources: Incident table with the conditions Assigned to = Abel Tuter and State in (New, On Hold, Resolved). Problem table with the conditions Assigned to = Abel Tuter and State in (New, Fix in Progress, Assess). Specify group by's: Columns: Assign to; Show all; Individual metric. Rows: State. Set 'Show row total' and 'Show column total' to false. Check the chart. Observe that the chart renders and shows expected data. Open the list with records of the problem table (problem.LIST). Filter records by Assigned to = Abel Tuter and State = Fix in Progress so that one record appears. Open that record. Change the status to Re-Analyze. Refresh the chart. Expected behavior: The chart is rendered and shows the expected data with zeros. Actual behavior: The chart shows 'No data available', despite data being available.
Analytics Data API PRB2005078	When using dynamic date ranges, appliedFilters returns incorrect dates		Open UI Builder. Create a page. Add a data visualization component. Use the type, 'Line' Check the Define Data Manually toggle under 'Under data sources'. Copy and paste the 'defect response.txt' in the 'Data' section. Notice that the chart is rendered, though no information from March is present. Edit the 'Data' section. Update the 'end' information in appliedFilters, from ''end':'2026-02-28' to ''end':'2026-03-31'. Notice that the data for March is now visible.
Analytics Data API PRB2006926	Geomap doesn't fetch locationData correctly when the incorrect query gets applied	Locations which have valid data in cmn_location table aren't getting plotted. Every valid location should get plotted correctly.	Navigate to Platform Analytics > Library > Dashboards. Create a dashboard. Add a Geomap widget based on the [sys_user] table with the following configuration: Metric: Count Group by: 'Location' Save the dashboard. Exit 'Edit' mode. Observe that the created Geomap widget doesn't render all the data points and shows an error: 'Some of the data you selected can't be visualized on the map due to latitude/longitude mapping errors'.
Analytics Export API PRB2007025	Dashboard PDF export doesn't apply dashboard filters on the list	Dashboard filters aren't applied to the list in the dashboard PDF exports. Only the filters configured in the list are honored.	Export a dashboard to PDF with list visualization. Make sure the dashboard has dashboard filters and the list is following the filters. Observe that the filters are applied correctly on the dashboard to the list, but they aren't applied in the exported PDF.
Analytics Export API PRB2019202	Show the total row count in the analytics list	The list-simple property 'hideTitleRowCount' should migrate to showRecordCount in the new analytics list.	In Yokohama, create a list-simple visualization. Change the Hide total number of records toggle switch in the configuration. Save the visualization. Upgrade the instance. Open the same visualization. Observe that it doesn't migrate to the new list visualization and have the corresponding property set for showing count.
Analytics Export API PRB2023200	Hide the export option when it's inactive or the roles list is empty	The system provides export on demand functionality at dashboard and data visualization level (server-side) for the Next Experience. There is a need to control the availability of the export option based on system properties and user roles.	As the system administrator, configure a system property to enable or disable the export feature. Make sure the export feature is enabled. Configure a second system property with a comma-separated list of roles that are allowed to use the export feature when it is enabled. Observe the system checks if the user's role is included in the configured list of roles (if the list is not empty). If the second system property (roles list) is empty, the export feature is available to all users when enabled. If the second system property (roles list) is populated, only users with roles in the list can access the export feature.
Application Manager PRB1995859	KMF key isn't copied to the target instance during the clone, causing encrypted app attachment decryption failure and store re-download fallback	After cloning an instance, the target instance fails to decrypt the encrypted app ZIP attachments. This happens because the KMF key used on the source instance isn't copied to the target during the clone. The target instance falls back to re-downloading and re-encrypting every installed app from the ServiceNow App Store, causing extended startup times, node timeouts, and degraded instance availability.	Provision a source instance (sub-prod) with one or more apps installed, so that the encrypted app ZIP attachments exist in sys_attachment. Clone the source instance to a target instance. Start the target instance after the clone completes. Check the instance startup logs. Observe that there are repeated re-download attempts for apps already installed on the source. Also, the decryption of app ZIP attachments fails on the target (KMF key from source is not present on target). Finally, enroll_module_for _resource_exchange is false on the com_glide_snc_app_client KMF crypto spec.
Application Rationalization PRB1993865	Base instance business rule fails to exclude the current record during duplicate name validation due to incorrect property reference	The business rule on the cmdb_ci_business_app table incorrectly blocks updates when changing the case of a Business Application name (for example, 'abc' to 'ABC'). This happens because it uses current.sysId instead of current.sys_id, resulting in an undefined sys_id comparison.	Create a Business Application (cmdb_ci_business_app) with the name 'abc'. Edit the same Business Application and change the name to 'ABC' (only change the case). Attempt to save the record. Expected behavior: The save is allowed because it's the same record, just with a different case. Actual behavior: The save is blocked with the error: 'A Business Application exists with the same name. Please verify.'
Appointment Booking PRB1961234	Future maximum bookable days are inconsistent when the time is 8:58AM in JST and 9:00AM in JST	The issue is caused by the internal time zone conversion logic. JST 19th Aug 08:58 a.m. converts to GMT 18th Aug 11:58 p.m.. Since GMT is behind JST, slots for September 18th aren't shown (the future maximum only reaches September 17th). However, JST 19th Aug 08:00 a.m. converts to GMT 19th Aug 12:00 a.m.. Future maximum days are added on GMT, and slots for September 18th appear.
Asset Management PRB2018791	The navigation filter isn't applied correctly	The relevant filter should be automatically applied after navigation.	Navigate to Software Asset Workspace. Open the 'Software Asset Overview' page. In the 'Activity Center' section, select the Asset Request link. Observe the navigation to the target page. Verify whether the expected navigation filter is applied. Expected behavior: The relevant filter is automatically applied after navigation. Actual behavior: The filter isn't applied.
Authentication Factors PRB2026011	Introduce a new MFA factor, Email OTP	This is a product update.
Authentication Factors PRB2026012	KBA improvements of Platform capabilities	This is a product update.
Authentication Factors PRB2026014	KBA session context persistence	This is a product update.
Authentication Factors PRB2026015	NextWave AI authentication migration to a new authentication service	This is a product update.
Authentication Factors PRB2026060	The kba_session_context value is not in JSON format	The kba_session_context value is logged in this format: '{q1_keyword=q1_user_input, q2_keyword=q2_user_input}'.	Create three scriptable authentication type answers. Attach them to three different questions. Log kba_session_context in these three answers, along with kba_auth_result=true. Navigate to a voice deployment. Select the three questions as a KBA auth factor. Call the voice agent. Authenticate a user. Navigate to the syslog table. Observe that the kba_session_context value for the third question is logged in this format: '{q1_keyword=q1_user_input, q2_keyword=q2_user_input}'. It should be logged in a standard json format.
Authentication Factors PRB2030597	Fall back to secondary identification when the primary resolves a non-sys_user	In cases when the identification questions are configured to a non-sys user, then it should ask fallback questions. This isn't happening in v2 and the call is ended since the userID isn't resolved. It should fallback to the secondary identification to get the userID. In cases where both the questions lead to a non-sys_user, then it should fail.
Authentication PRB2033124	Case-sensitive comparison is applied during knowledge-based authentication (KBA) answer matching	Case sensitiveness of KBA answer matching is controlled by the system property 'glide.auth_factors. kba.case_insensitive _validation'. It is by default case insensitive in previous patches, but seems to have the incorrect default value in recent tracks.	Set up knowledge_based_answer for a KBA Service. As an end user, try to pass answers and validate identification/authentication. Expected behavior: The user-given input should be matched case insensitively. Actual behavior: Answers are matched case sensitively.
Automated Test Framework (ATF) PRB2022891	MVP - UI test script (Testing Library) test step in ATF for UI testing	This is a product update.
Case and Knowledge Management for HR Service Delivery PRB1983139	Approval records aren't created when creating a HR case		Log in to the instance. Assign some manager (for example, system admin) to Abel Tuter. If the HR service is inactive, activate it. Create a HR case for the 'Request Personal Information Report' service. Observe that the approval record isn't created.
Change Management PRB1994240	Translation issues in the Conflict Checker Progress modal window	Translation issues were observed after upgrading from Yokohama to Zurich. The 'conflict_detection' header appears to be missing the message completely when it previously read 'Conflict Detection'. The message 'The conflict check is complete' is translated into Finnish, even when the user's language is in English. This issue also occurs when the instance language is set to French, and the label is translated in English.	Open a Zurich instance. Ensure the instance is set to the English language. Open a Change record. Select the Check Conflicts button in the 'Conflicts' tab. Notice that the progress bar title displays as 'conflict_detection' and the value is shown rather than the label, 'The conflict check is complete'. The label is translated in Finnish even though the user's instance language is set to English.
Code Signing PRB2014768	Increase the code signing validity window maximum from 60 minutes to 4 hours	In the 'Scan Signatures' module, the Data source field is incorrectly set to false for records that have valid, successfully created signatures. These signatures return true when validated via a background script. The issue occurs because the effective maximum of com.snc.kmf.signature .validity_window is 60 minutes, which prevents users from configuring longer validity windows, especially when the consecutive signature generation takes time.	Open an instance. Navigate to 'Update Set: Test Code Signing Updates'. Validate that the signature is created for the Data source field. Select Scan Signature. Observe that the signature state for the Data source field still reflects as false.
Code Signing PRB2014831	Scheduled scripts incorrectly appear in 'Update Set Signature States'	Currently, users are required to hold the security_admin role to perform code signing operations on update sets. This creates several operational and security challenges. Instead, there should be a separate role that grants the ability to perform code signing operations on update sets and related artifacts within the Code Signing framework. It shouldn't grant access to security policies, ACLs, encryption contexts, or other security administration functions.	Open an instance. Navigate to 'Update Set: Code Signing Scheduled Jobs'. Navigate to the tab 'Update Set Signature States'. Observe that there's no option to run signing jobs unless elevated to the 'security_admin' role.
Code Signing PRB2014833	Enable digest creation for users with the code signing role	For users with the code_signing_user role, the 'Create Digest' functionality should be enabled for Flow Designer flows and actions. The current dependency on the security_admin role should be removed.	Open an instance. Navigate to 'Update Set: Code Signing Scheduled Jobs'. Navigate to the tab 'Update Set Signature States'. Observe that the user can't generate digest for flows unless they have the 'security_admin' role.
Condition Builder in Workspace PRB2001225	The messaging for deleting saved filters is misleading	The confirmation dialog says 'Active will no longer be available in your saved filters. This action can't be undone.' Instead, it should be more explicit that the filter may be deleted for everyone.	Navigate to a list in Service Operations Workspace. Open the condition builder. Add some filters to the condition builder. In the 'Saved Filters' list, select Save Filter. Select Save. Open the 'Saved Filters' list. Select the Trash Can icon next to the saved filter that was just created. Expected behavior: The confirmation explicitly says that when a filter is deleted—especially one that has been shared to a group or globally—it is deleted for everyone. Actual behavior: The confirmation dialog opens and shows: 'Active will no longer be available in your saved filters. This action can't be undone'.
Configuration Management Database (CMDB) PRB1923632	CMDB Health Correctness group score doesn't consider inclusion rules when computing health score	The CMDB Health correctness group score is incorrect because the parent metric doesn't consider the inclusion rules. The code in the correctness manager that would return the health inclusion rule for the class is missing. It works correctly for other metrics (for example, completeness and conformance).
Connections and Credentials PRB2007222	AuthMetadataProvider .getAliasInfoList() performs unfiltered full-table scan on sys_alias (29,000+ rows)	This causes QueryWarning and excessive DB load. The same behavior occurs via any code path that calls PersonalAuthClient .getAuthCredentialInfo (credentialId).	Open an instance with many aliases in sys_alias (for example, 29,000+ records). Trigger a flow that invokes sn_cc.PersonalAuthAPI() .getInitiatorURL(aliasId) with a valid alias ID. Observe that the following warning appears in the system logs: 'QueryWarning *** WARNING *** Large Table: Table handling an extremely large result set: 29180'. Additionally, observe that the info-level log for orphaned credential aliases contains 'can't find a credential with the given alias sys_id...'
Connections and Credentials PRB2015199	An initiator URL isn't returned when the AT has expired and the record is present in oauth_credential_list
Content Library Portal PRB2005753	Content lookup historical data triggers unintended deduplication for pa_manual_breakdowns records	The fix script runs with the 'isHistorical' parameter set to true, then it calls the deduplicateByValueField function. The duplication logic doesn't query for the breakdown; it only looks for duplicated value to identify duplicate records. In theory, there could be duplicated value across different breakdowns, causing unwanted data loss.
Content Library Portal PRB2010143	Multiple ais_index events are triggered for sam_sw_product_lifecycle during content updates	During content updates in the ITAM Content Library, many ais_index events are generated for the sam_sw_product_lifecycle (SAM software lifecycle) and sn_hamp_lifecycle_definition (HAM hardware lifecycle) tables. These events appear in the event log. The excessive events are triggered because system metadata columns on the AI Search datasources for these two tables are missing the no_text_index field attribute. Without this attribute, the AI Search indexing engine includes these columns in the vector index.
Database Persistence - Data Access PRB1919872	There's duplicate edges in the output of jsFunction_getEdgeList for custom created graphs and when includeEdgesForTables OutsideGraph is true
Database Persistence - Data Access PRB1930279	Users can create a cyclic relation through graph hierarchy		Create a graph. Name it test1. Create another graph with the parent as test1 (with extension model) and name it test2. Edit test1 to have a parent as test2 (with extension model). Notice that it creates a cyclic relation in graphs test2-parent-test1 and test1-parent-test2.
Database Persistence - Data Access PRB1946286	Workflow Data Fabric (WDF) queries fail when trying to execute on connection for trino_primary	When trying to get a specific column instead of the entire record, the error occurs in the response: 'FAILED TRYING TO EXECUTE ON CONNECTION trino_primary'.
Database Persistence - Data Access PRB1962536	The prefix is missing in some places for the Knowledge Graph UI		Navigate to All > Knowledge Graph Designer. Select the Open global graph button. Select Start with user table > Any node. Note that the prefix is missing for some of the columns. Expend any node on canvas. Note that the prefix is missing for expended nodes. Open the contribution graph. Note that the prefix is missing for all the nodes, columns, and relations.
Database Persistence - Data Access PRB1969193	SELECT isn't generated correctly for a script when a business rule is active with ORDER BY for Postgres	There's an error message: 'FAILED TRYING TO EXECUTE ON CONNECTION glide.1 (connpid=3758451): SELECT...'
Database Persistence - Data Access PRB1984070 KB2928335	Trend reports and dashboard show inaccurate data by week across the end of a year	On instances using a PostgreSQL database, an incorrect year reference (yearref) is returned for the non-ISO week functions sunday_week and monday_week. As a result, the column data displays trends for future dates where no actual record data exists. This issue affects reports and dashboards that group or summarize data by week across the end of a year. Reports may display an unexplained empty week in one year, a duplicated week in another, or records appearing in the incorrect annual bucket.	Refer to the listed KB article for details.
Database Persistence - Graph PRB1923336	The isTableExcluded API should exclude text search tables
Database Persistence - Graph PRB1938573	A 'WHERE' clause appends all query node conditions with an 'AND' operator, even when there are 'OPTIONAL MATCH 'statements	When trying to a build cypher query for the following use case, 'Identify all CIs which are connected to Database catalog or Application', in the final built cypher query the 'WHERE' clause is using all 'AND' conditions to append all node filters to the query. When using 'OPTIONAL MATCH' statements, an 'OR' logic is expected.
Database Persistence - Graph PRB1944327	Global graph should be the default and all other graphs should be non-default when the instance is upgraded from Yokohama to Zurich	Default graphs and Global graphs are all set to non-default.	Create two default graphs on a Yokohama instance. Upgrade the instance to Zurich. Check the graphs on sys_meta_graph. Expected behavior: Global graph should have default = true and the other two graphs that were created in step one should have default = false. Actual behavior: Both of the created default graphs and the Global graph are default = false.
Database Persistence - Graph PRB1967193	CMDB Query Builder doesn't work when multiple OR conditions are used on location attribute	When multiple OR conditions are used on the location filter, the CMDB query builder doesn't return any results.	Scenario 1: Navigate to the query builder. Open the configuration item query block. Select Filter. Run the builder with multiple OR conditions on the location filter. Observe that no results are returned. Remove all the OR conditions for location (or use one location filter). Run it. Observe that once a second condition is added, the query builder fails. Scenario 2: Navigate to the CMDB Query Builder. Add a server node to the canvas. Apply filters to the node with the following conditions: Location contains 'bock' Location contains 'via' Run the query. Observe that no results are returned. Remove one of the conditions (for example, Location Contains 'via'). Rerun the query. Observe that results are now returned.
Database Persistence - Graph PRB1975732	There's a repeated query execution in GraphUtils #getGraphByQualifiedName for a metadata child table [sys_meta_graph] for every iteration/execution
Database Persistence - Graph PRB1988516	Weeks, milliseconds, and microseconds aren't supported in a DURATION map	There's an error: 'com.glide.db. DBGraphApiException: Error executing cypher: FAILED TRYING TO EXECUTE ON CONNECTION...Syntax Error or Access Rule Violation detected by database (ERROR: function duration_between(date, date) does not exist tooltip: No function matches the given name and argument types. You might need to add explicit type casts. Position: 294)'.
Database Persistence - Graph PRB1988520	TimeKeyExpression isn't supported as a DURATION function parameter in DBSqlParser	There's an error: 'Error: com.glide.db. DBGraphApiException: Error executing cypher: FAILED TRYING TO EXECUTE ON CONNECTION...Syntax Error or Access Rule Violation detected by database (ERROR: function duration_between(timestamp with time zone, date) does not exist Hint: No function matches the given name and argument types. You might need to add explicit type casts. Position: 307)'.
Database Persistence - Graph PRB1990632	Node/edge types should allow special characters for GraphMetadataAPI	Predefined filters fail in the Graph API version of the CMDB query builder. After a platform upgrade, CMDB query builder throws the 'Problem Running Query' error.
Database Persistence - Graph PRB1995712	Weeks, milliseconds, and microseconds aren't supported in the Duration map
Database Persistence - Graph PRB1996879	Guardrail for maximum number of edges returned from getForTables	The number of nodes returned from getForTables is capped to 1000, but there's no such limitation for the number of edges. When inbound edges are specified and the number of nodes is at maximum, the number of edges can be quite large.	Run getForTables for sys_user with inbound edge flag = true.
Database Persistence - Graph PRB1997409	L2 query displays non-empty edge values		Navigate to Query Builder. Drag the 'Database and Computer' class into the canvas. Connect them with the L2 relationship edge. Expected behavior: With the data, there should be two L1 relationships and three L2 relationships with the edge shown as '(empty)'. Actual behavior: The three L2 relationships have unexpected edge values.
Database Persistence - Graph PRB2007566	com.glide.db. DBGraphApiException has an error executing cypher	There's an error: 'This user (admin) is not allowed access to table: sys_user_has_role'.
Database Persistence - Graph PRB2008809	Rename the ScopedGraphQueryBuilder .createNode method	To avoid potential confusion with createNode as a write operation, rename the 'createNode()' method to 'node()'.
Database Persistence - Graph PRB2020140	Union queries aren't supported
Database Persistence - Graph PRB2021889	GlideRecord.setCategory() is used in a way that prevents routing to read-replica databases	The query is categorized as 'graphqueryexecutor' instead of 'cmdb_queries', and therefore doesn't get routed to the read replica because the category isn't a second database category.	Ensure the instance is set up with a read replica database. Ensure the CMDB Queries (or all Secondary Database Categories) are enabled and are mapped to the read replica. Execute a KG query from the CMDB Query Builder app. Expected behavior: The query is categorized as 'cmdb_queries' and is routed to the read replica. Actual behavior: The query is categorized as 'graphqueryexecutor', and since that category isn't a secondary database category, is not routed to the read replica.
Database Persistence PRB1866305	DictionaryXMLParser rejects large table alters for Postgresql db based on MariaDB logic		Create an instance on Postgresql with any table that has over 150M records (for example, a table like sys_journal_field). Update the sys_journal_field.xml to have a new string column. Upgrade the instance. Observe that the alter is rejected.
Data Management Console PRB1974031	A repeated query with the hash -423529530 comes up when running stats gatherer
Data Snapshots PRB1999238	There's a node outage due to an out-of-memory error as a result of a memory bloat in ScriptableCDCDataCollector .jsFunction_mineAllChanges()	During the investigation, it was noticed via heapdumps that the com.snc.db.cdc. implementation. ReplicationTableReader object held an array list of size 500. Each entry in this array list held up to a max of 800 Glide record objects. The size of each of these 500 entries was ~21 MB. The Glide record objects were from the cmdb_ci_vm_instance table. These were particularly heavy in terms of row size (reaching ~400 KB). The existing sizing must be revisited to take into account this outlier. In particular, in addition to a limit on number of rows to be polled from cdc_queue_par, it should also include a limit to cap the memory size.	Create an 'All activity data snapshots' source on the em_alert table. Run the 'Incremental all activity' job. Run the first snapshot job and let the first mining complete. Run the 'Incremental all activity' job. Notice the memory bloat in the worker thread running for the 'All changes' job.
Developer Sandboxes PRB1995018	NowMQ messages can be claimed by an unexpected recipient	If the user sets the App Operation Queue Health Monitor job to inactive, it should prevent the sandbox from claiming any application operation queue messages in sys_nowmq_message. On an instance without sandboxes, any messages that were left unclaimed or any newly added messages should still be claimable when the job is re-enabled.	Make sure one controller and one sandbox are running. On the sandbox, set the App Operation Queue Health Monitor job to inactive. Enqueue an operation on the sandbox by initiating an app install from app manager or via /sn_cicd/update_set/commit. Navigate to sys_nowmq_message. Observe that the corresponding message from step 3 is unexpectedly gone. It should still be present but unclaimed as a result of step 2.
Developer Sandboxes PRB2011753	Sys_flow_trigger_plan_chunk is currently shared and causing updated flows to error out	If the user updates the base flow, the sandbox flow errors out with 'Compiled flow not found for snapshotId'.	On a base instance, create a flow that is scheduled to run every minute. Navigate to Flow Designer > Operations > Flow > Today's execution. Observe that the flow is running successfully. Create a sandbox. In the sandbox, navigate to Flow Designer > Operations > Flow > Today's execution. Observe that the flow is running successfully. In the sandbox, update the flow. Select Activate. Navigate to the base instance. Observe that, under 'Today's execution', the flow starts erroring out with 'Compiled flow not found for snapshotId'.
Developer Sandboxes PRB2015080	The scheduled flow inherited from the main instance does not run automatically on the sandbox instance	The flow runs in main instance at the scheduled interval, but the flow doesn't run in the sandbox instance even though it exists.	Open the main instance. Navigate to All > Flow Designer. Create a flow. Schedule that flow to run at a certain interval, for example, 'Repeat every minute'. Activate the flow. Navigate to All > Flow Designer > Operations > Flows > Today's executions. Notice that the flow is running every minute. Create a sandbox instance. Navigate to All > Flow Designer. Notice that the flow exists. Navigate to All > Flow Designer > Operations > Flows > Today's executions. Notice that the flow is not running.
Developer Sandboxes PRB2017438	DSB update sets aren't exported before clone	Because sandboxes are retired, update sets from a sandbox should be saved as remote update sets on the base instance on upgrade. However, due to this problem, they may not be saved correctly.	Create a sandbox. Make changes in the sandbox. Upgrade the instance. Expected behavior: There is a way to recover the work. Actual behavior. The work is gone. Logs show jsFunction_exportUpdateSets is called.
Developer Sandboxes PRB2018545	BA conversations are shared between sandboxes/base instance	The base instance conversation contains messages from the sandbox, and vice versa.	Provision an instance with BA installed. Create a sandbox. From the base instance, open IDE or Studio. Start a conversation (for example, 'What's 1+1'). Visit the sandbox. Open IDE or Studio again. Observe that the conversation already contains the messages from the base instance, even though it should be empty. Add another prompt in the sandbox. Refresh the base instance. Expected behavior: The conversation doesn't contain what was typed in the sandbox. Actual behavior: The conversation contains messages from the sandbox.
Discovery PRB1986211	Discovery.complete and discovery.canceled are triggered incorrectly	Discovery may trigger both discovery.complete and discovery.canceled events for the same status, and it may trigger discovery.canceled twice.	Create or pick a cloud-resource discovery schedule configured to use a specific MID Server (midSelMethod = 'specific_mid'). Ensure that the MID Server is in a state where validateMIDCondition() returns false (typically because the MID is down / not validated). Run the schedule (or trigger it via the Discover Now action). Observe that discovery.canceled fires twice for the same status record. The StartDiscovery error log for 'Discovery canceled on previously started schedule ... due to mid not available' appears twice. Instead, there should be a single discovery.canceled event and a single error log entry per cancelled run.
Document Intelligence Unified Backend PRB2018937	Local naively checks for empty text rather than 'empty' when classifying PDF pages for DocReader	The PDF page is classified as 'digital' instead of 'scanned' because it checks for empty text instead of a empty PDF page.	Call DocReader Local with a scanned PDF containing empty text, for example, '\n\n\n'. Notice that the page is classified as 'digital' rather than 'scanned'.
Document Intelligence Unified Backend PRB2029867	Users can configure more than 50 fields per use case, which does not match the product and pricing based limitations
Edge Encryption PRB1945424	There's high memory utilization on proxy servers after the Yokohama upgrade	After upgrading and running Yokohama, the memory consumption is over the limit set in wrapper.conf. The user set the memory limit for 10G in the wrapper.conf, but the edge proxy doesn't respect this setting in the build. The memory goes to the physical limit and causes the proxy server to restart.
Email Accounts PRB2022999	Add the capability to capture mailbox as part of headers	This is a product update.
Email Notifications PRB2006670	Push Notifications aren't sending the badge count	This issue occurs on mobile instances.	Log in to a mobile instance. Fire a notification. Observe that the badge count isn't sent in the push payload.
Email Notifications PRB2022602	Client script include should be created to make client scripts modular and reusable	In Zurich, the wiring's not implemented for mini composer in CSM when activity stream compose isn't present.	Update the 'glide.email_client. configurable_send_enabled' value to 'sn_customerservice_case'. Make sure configuration is done. Log in as a user with elevated privileges. Navigate to CSM workspace. Open a case record. Ensure activity stream isn't present. Select Compose Email to open a modeless dialog. Add [private] in the Subject field. Add 'unknown user' in the To field. Select the Send button. Expected behavior: The pop-up is displayed based on the configuration. Actual behavior: The pop-up isn't displayed when the user selects the Send button.
Event Management PRB1980951 KB2738069	When text-based grouping is turned on, tag-based grouping doesn't work consistently	When text-based grouping is turned on, the related tag-based alerts aren't always grouped. This behavior isn't consistently reproducible and occurs at random. If text-based grouping is turned off, the related alerts are grouped by tags.	Refer to the listed KB article for details.
Event Management PRB1982324	Group type appears automated, but the group was created according to the tag-based definition	The issue occurs when there is a pattern (same metric, different node), and tag-based is enabled but with the wrong one turned on (same metric, same node). The user gets an automated group, but worknotes are received from tag-based.	Define the pattern attribute node and metric. Set the following properties: sa_analytics.enable_no_ci_grouping=true sa_analytics.agg.query. group_logic_order= MIXED,PATTERN,GENERALIZED _PATTERNS,NETWORK_ TRAFFIC,TEXTBASE (default) sa_analytics.agg.mi_graph_min_freq_threshold=2 Deactivate all tag-based definitions except the base instance definition (group by exact match of node). Create two events with the same node N1, empty CI, and different metrics M1 and M2. Wait at least 10 minutes. Create the same pair of events again. Run the job 'Service Analytics Alert Aggregation Learner'. Verify that the pattern with features N1&&&M1 and N1&&&M3 is created. Close all alerts. Create one event with N1 and M1. Wait for the grouping job to handle it. Observe that the relevant alert appears in the staging table as an isolated alert with no parent. Create one event with N1 and M2. Observe that an automated group with two alerts is created in the next cycle of the grouping job. The group alert should have the proper worknotes.
Event Management PRB1999707	An app node crash occurs due to OutOfMemory (OOM), and GroupingDebugLogger accumulates log messages even if logging is disabled	After stimulating alerts for Metric rule/Event rule and checking the tables for node restarts and active transactions, the queue in the sys_trigger table is huge.	Provision an instance with the following installed: sn-dex-4.0.0-SNAPSHOT agent-client-collector-framework-6.0.1 com.snc.samp com.sn_sam_saas_int Load 400k agents with E2E data simulation using SIM Version: 3.2.45. Simulate alerts for Metric rule/Event rule at 300k-600k per 5 minute rate, so that it runs at 1-2k per second. Check the node restarts from the v_cluster_nodes table. Check the active transactions from the v_cluster_transaction table. Notice that there are slow 'Event Management – process events' jobs. Check queue in the sys_trigger table. Notice that there is a huge queue found in the sys_trigger table.
Explicit Roles PRB2011506	The NullPointerException in ExplicitRoleCollisions .handleSysUserGrmember invalidates ATF rollback context during CreateUserStepRunner execution	During the scheduled nightly ATF test execution, a null pointer exception occurs in the explicit roles collision check logic, which propagates up and invalidates the ATF rollback context. This prevents automatic rollback of test-created records (for example, Incidents, Users, Business Services), leaving orphaned data in the environment.	Activate the explicit roles plugin. Create a test user. Assign the user the snc_internal role only. Create a group with any one role. Add the test user to the created group. Expected behavior: There is no null pointer exception in the node log. Actual behavior: There is the following null pointer exception in the node log: 'Error: java.lang.NullPointerException: can't invoke 'com.glide.explicit_roles. MutuallyExclusiveRoleCheckResult $MutualExclusionResultType .ordinal()' because the return value of 'com.glide.explicit_roles. MutuallyExclusiveRoleCheckResult .getResultType()' is null'.
Field Service Capacity and Reservations Management (Glide Family Channel) PRB2017642	The hours/tasks capacity definition doesn't consider the assignment time zone when computing demand metrics and calculating capacity usage	The metrics table records (from_date, to_date) are based on the system time zone, not the assignment time zone.	Create a capacity assignment with the capacity definition as hours/tasks. Give any time zone other than system time zone. Observe that the metrics table records (from_date, to_date) are based on the system time zone, not the assignment time zone.
Field Service Capacity and Reservations Management (Glide Family Channel) PRB2020554	All capacity console event values should be displayed based on the capacity assignment time zone	The events should be displayed according to the assignment time zone, not the system time zone.	Navigate to the Capacity Console. Create a capacity assignment for any territory. Open CSP. Check the dates of the event. Observe that the events aren't displayed according to the assignment time zone.
Flow Engine PRB1992819	Impact handling of Scripting Governance disable switch	A disable switch for the Scripting Governance feature is being added. Any ACL or Java code dependent on the scripting role/group will break if a user flips the disable switch.	Check the ACL and Java code for usage of the snc_required_script_writer_permission role and conditional script writer group. verify that the feature doesn't break when the switch is turned off/on.
Flow Engine PRB2014479	Add the roles 'taas writer' and 'taas reader' with API access only, and not with table ownership	Adding the role trigger_designer directly to policy_manager was evaluated and ruled out, as the 'policy manager' role should not have access to trigger the 'designer' role.
Form Templates PRB1963747 KB2962231	A template sometimes can't populate both the 'Assigned to' and 'Assignment' group	Following the upgrade to Zurich, the user can't update 'Assigned to' through templates, but they can perform the update manually. In workspace, if a template sets both 'Assigned To' and 'Assignment Group', the template application may fail. Assignment group validation determines if the 'Assigned To' user belongs to the assignment group when a template is applied. However, there's an issue where the validation only uses the record value's assignment group and the assignment group in the template is ignored. This behavior doesn't occur when templates are applied using UI16.	Refer to the listed KB article for details.
Generative AI Controller PRB2026284	Agents aren't working in Zurich nightly with the latest snapshot		Navigate to agents playground. Trigger any agent. See that the agent is stuck in starting and no context is passed to it.
GlideAggregate API PRB1951351	GlideAggregate returns incorrect results for the business calendar with a condition on Date field	The issue occurs on both MariaDB and RaptorDB.
GlideRecord PRB1922298	Assessment Metrics (asmt_metric) in a question bank disappear after they're used in a published survey	On an instance running on RaptorDB (postresql), the assessment metric in the question bank can disappear. The issue can't be reproduced on MariaDB.
GlideRecord PRB1958518 KB2942052	A Null Pointer Exception is thrown if a table has a GlideElement WikiText type field and queries at start up before extension points are loaded	Adding a GlideElement Wikitext type field to the user table can cause an outage. The page is empty and errors are seen. On a local instance, the instance throws servlet exceptions and doesn't start.	Refer to the listed KB article for details.
Hardware Asset Management PRB2023284	The stockrooms missing distribution channel yields empty records in list view		Provision a Zurich instance with HAM 12.1.2 installed. Navigate to Hardware Asset Workspace > Inventory. Select the View button on stockrooms missing distribution channel (important action). Observe that no records are visible.
Horizon Avatar Component PRB2010763	The gradient border is missing for AI Specialist avatars in the activity stream for Seismic interfaces
Horizon Component Library PRB1910095	The VA Chat icon disappears in responsive view		Log in to the ESC Portal. Configure NAVA. Enable pinning behavior. Pin the chat window. Change the view to the responsive view. Unpin the chat window. Observe that the VA Chat icon disappears.
Horizontal Portal Capabilities for Customer Service PRB2010425	Additional comments added via the CSM Portal are duplicated	This happens if the state is set to Awaiting Info.	Impersonate a CSM agent. Open a case. Add an additional comment and request info. Observe that the state of the case is set to Awaiting Info. Impersonate an external customer account. Open the case. Add a comment. Observe that the comment appears twice in the case activity log. With the state of the case set to open, add a second comment. Observe that the second comment is not duplicated.
HR Service Delivery PRB2016516	RCA is generated for 'Populate Manager Reportee Count Using Eligible Users' and 'EmployeeHubOrgChart ReporteeUtilSNC'	The following two RCA's are generated for the new scripts, causing test failures: sys_script_include_ a302f8807873f250f877079523d275e1 and sysauto_script_ 33a5e72453f7b210f2ebffc230e5e69d.
Instance Clone (Family) PRB2023467	Incorrect values are shown on the clone summary page
Interactive Filters PRB2007031	Applying the filter in the Dashboard shows 'Some applied options are currently unavailable' when the filter options have a comma ',' in the value	While trying to apply an indicator filter with the 'Filter' option in the breakdown, the messages 'Some applied options are currently unavailable' and 'Unavailable option' appear.	Create a filter with a data source as an indicator. Create a breakdown on any table. Ensure the field has a comma in the field value. Add the filter to a dashboard. Apply the filter. Notice the message, 'Some applied options are currently unavailable.'
Internationalization Features PRB1892286 KB2277705	A non-admin user can't change a dashboard name by specific steps when the system language is set to Japanese	The dashboard name should be updated correctly and reflected in both the primary record and its translated fields, as it is in the Washington and Xanadu versions. In the Yokohama version, the update to the dashboard name fails silently when the Japanese language is enabled and the sys_translated record exists.	Refer to the listed KB article for details.
Key Management Framework (KMF) PRB1964424	WebServices Mutual Auth connections fail if the instances are in FIPS mode	If an instance is using FIPS mode, REST Web Services connections fail if the authentication type is 'Mutual Auth'. The following error is thrown: 'org.bouncycastle.tls.TlsFatalAlert: unsupported_extension(110)'.	Turn on FIPS mode on an instance. Set up an outbound REST call to a third-party API. Make sure the authentication type is 'Mutual Auth'. Set up a protocol profile record to associate with the REST connection. Apply the API's endpoint .bsfks keystore certificate to the Protocol Profile Keystore record. Run a test connection to the endpoint. See that it fails with an error: 'org.bouncycastle.tls.TlsFatalAlert: unsupported_extension(110)'.
Knowledge Graph (Family) PRB2021133	Knowledge Graph (KG) description deactivator triggers long-running queries, causing HLL spikes	Post Version Upgrade Job : KG Description Deactivator triggers Long running queries and causes HLL spike
Knowledge Management PRB2000521	Support for Knowledge blocks in ECE	Knowledge blocks aren't supported in ECE. Ideally, blocks should be supported in ECE.
Knowledge Management PRB2005195	KM Import Article conversion utility should be exposed as a callable API for external consumption by the GRC/IRM policy flow	The KM Import Article feature uses Apache Tika and Apache POI libraries for DOCX-to-HTML conversion and produces flexible, translation-safe HTML output. On the other hand, the GRC policy flow's current GroupDocs-based conversion applies static absolute positioning. To resolve the translation formatting issue in GRC's policy flow, the KM conversion utility needs to be exposed as a standalone API that GRC/IRM can consume directly—without requiring an existing Knowledge Article upfront and without deleting the source attachment document during processing.	Navigate to Knowledge Management > Import Article. Import a Word document. Open the created Knowledge Article. View the source HTML. Observe that the output contains flexible/relative styling and there's no static absolute positioning. Compare against a Knowledge Article created via the GRC/IRM policy flow and note the static positioning difference.
Knowledge Management PRB2010029	Knowledge blocks aren't honoring KC redirection properties		Turn on redirection properties for KC: sn_km_center.glide.knowman.redirect.enable. Navigate to 'Knowledge Blocks' in Navigator in UI16. Select Create. Expected behavior: Knowledge blocks should honor KC redirection properties. Actual behavior: Creating or editing a from UI16 doesn't redirect to KC.
Knowledge Management PRB2014078	Turn on the Apriel 2.0 model and set the new 3P model as the default	This is a product update.
Knowledge Management PRB2027236	True-up for KC and NAKM, ECE, and KC in UI Builder	This is a product update.
List Administration PRB1933683	List column widths aren't persistent when columns are personalized	The user can adjust the list column width. However, when the list columns are personalized (either by adding or removing columns), the adjusted width doesn't persist and resets to the default width.	Navigate to CSM/FSM Configurable Workspace. Open any list from the List Page. Adjust the width of any column. Select the Personalize action. Add or remove columns. Select Apply. Notice that the adjusted width is reset to some default width.
List Administration PRB1970036	List type data visualization doesn't show new lines (multi-lines)	The data visualization doesn't add new lines (multi-lines) in Zurich, but it does in Yokohama.	Open an incident record. Enter the following strings in the Description field: 改行1 改行1 改行1 改行1 改行1 改行1 Create a dashboard. Add a list data visualization based on the incident table. Show the description column on the data visualization. Observe that the data visualization doesn't add new lines (multi-lines).
List Administration PRB2012758	List fails to load due to a DataFetchingException from NowAssistSkillConfig	The list page in Service Operations Workspace doesn't load; it looks like the page is processing. NAA code is called and throws an exception, which results in list load error.
List Administration PRB2015114	The user is unable to load the results table frame on the same page	This issue was found in Query Builder for Yokohama and Zurich.	Navigate to the CMDB Query Builder. Create a simple one node query. Select the Settings icon on top right. Turn off 'Display Results in New Tab' so that the results are displayed on the current page. Run the query without saving. Expected behavior: Even when there are no results for this query, the result table frame is shown with headers and columns. Actual behavior: The result table frame does not show up, and only a blank popover appears.
List Administration PRB2023617	Grouping and summarization for telemetry for Multi-Record Actions (MRA) - List AI	This is a product update.
List Administration PRB2023634	Implement add/remove an AI indicator for the rows created/modified by an AI agent and clear an AI indicator when inline edits are made	This is a product update.
List Controller PRB1773052	The check box in presentational list can't be both inactive and selected	When the user sets the row as selected and inactive, the check box is only inactive.	Add a presentational list on UIB. Use a script to set a row as selected and inactive. Expected behavior: The check box in presentational list is both inactive and selected. Actual behavior: The check box in presentational list is inactive, but not selected.
MetricBase PRB2008126	The rollup scheduling script has a race condition that allows duplicate event queuing in MetricRaptor	When no classic shards are configured and the mb_shard_mapping table is either empty or has no rows with stores_classic_data = true, there is excessive logging for ClothoClientManager and ClothoRrdWriter. The error '*** ERROR *** No endpoints available' occurs for ClothoClientManager, and '*** WARNING *** No Classic Clotho Shard Configured' occurs for ClothoRrdWriter.	Have an active mb_raptor_table with the rollup spec configured. Wait for the scheduled script to queue an mb_rollup event. While the event is in 'processing' state, trigger or wait for the next scheduled run. Wait for a one minute interval. Observe that a duplicate mb_rollup event is queued because the state = 'ready' check misses the processing event.
Mobile Platform PRB1999865	Support onSubmit UI Rule	Configure and execute UI Rules at the time when the input form screen is submitted both online and offline.
Multimodal Service (Family Channel) PRB2022350	Vision Agent MMS Glide changes	This is a product update
Multimodal Service (Family Channel) PRB2022351	Implement improvements to the MMService plugin	This is a product update.
Multimodal Service (Family Channel) PRB2029156	Move MMS to GA	The MMS plugin name displays 'MAINT ONLY', and should be updated.
Next Experience Unified Navigation PRB2003211	/api/now/ui/polaris/menu is slow to build the cache on login	1000s of each of these queries are run to retrieve user preferences and screen accessibilities. It should be able to retrieve all of most of these items in 1 query.
Next Experience Unified Navigation PRB2022500	If a work item is assigned to the same agent repeatedly, the Desktop Notification stops appearing entirely	A new desktop notification should appear on every work item assignment.
Now Assist in Document Intelligence PRB2030125	Support attachments where the file names doesn't contain '.type' at the end	There's an error: '{'attachment':{}, 'task':{'708a54b23b810f50141 a0e0f23e45a0b': {'error_msg':'Prediction server is not able to process the attachments: begin 0, end -1, length 9'}}}'.	Have a png image file with no errors. Rename it to have a file name with '.png'. Upload it to an instance. Test the attachment summarization. Observe an error.
Now Assist Nextwave Experience PRB2010718	Portal URL resolver mapping is broken in NextWave conversations
Now Assist Panel PRB2009288	Feedback isn't updated in the Gen AI log table		Type 'what is spam?'. Select any feedback Thumbs Up/Down icon immediately. Observe that feedback is successfully stored in the Gen AI log table. Unselect submitted feedback or submit different feedback. Expected behavior: Feedback should be updated in the Gen AI log table. Actual behavior: Feedback isn't updated in Gen AI log table.
Now Assist Panel PRB2025041	A mic enabled for NAVA Premium Chat is incorrectly surfaced on Enhanced Chat		Use the latest version of NAVA. Enable a mic from NAVA assistant designer. Change the experience to Enhanced Chat. Expected behavior: The Send button appears in the input bar. Actual behavior: On-Glide mic appears in the input bar.
OneExtend PRB2001785	Accuracy choice should be added for the Group field in the eval suggestion table		Navigate to the sys_one_extend_eval_suggestion table. Open any record. Select the Group choice list. Observe that accuracy choice isn't present.
OneExtend PRB2013928	JSON Web Token (JWT) and metadata are missing domain visibility information	The JWT API only contains a claim for the domain, and is missing the list of domains visible to the user.
OneExtend PRB2017386	Now Assist jobs are running for extended periods and slowing processing
OneExtend PRB2025105	Responses fix for Java flow	Validate all the Java flow from NAVA, NAP and Skill Kit.
OneExtend PRB2025250	Correlation Insights skill is stuck on the progress bar when off-Glide is enabled	Insights should be generated and displayed. Instead, the progress bar spins indefinitely and no generative AI logs are produced. The issue occurs because the app prepares a bulk request with three payloads and invokes it in async mode with a callback handler. When off-Glide is enabled for the capability, the async callback flow ('NowAssistforSIR CorrelationInsightsCallback') appears to break silently. No generative AI logs are generated, indicating the LLM call never completes or the response isn't handled correctly.
OneExtend PRB2025268	Refactor the logic to update the URL and status in GeoRoutingDaoImpl	On repairing the GAIC plugin, if the Geo-Licensed record is not available, the http_connection records are randomly updating.	Log in to an instance as an admin user. Navigate to Connection & Credential aliases. Filter Now LLM. Open the record. Verify that the Now LLM and Geo-Licensed record exists in http_connection. Delete only the Geo-Licensed record. Repair the GAIC plugin. Once the repair is complete, navigate to the http_connection record. Filter by latest updated. Observe that it's randomly updating any of the http_connection.
OneExtend PRB2025367	A cloning instance should preserve mosaic functionality in the target instance
OneExtend PRB2025540	Off-glide Mosaic pipeline ignores the per-request timeout and every call falls back to the platform default, regardless of caller intent or service-plan configuration	All Mosaic calls fall back to the platform default; caller / control-setting / service-plan values are ignored. There's no phase-boundary timeout enforcement and requests run well past their stated SLA. DT translation always uses 0L.	Enable off-Glide execution. Ensure the Mosaic endpoint is reachable. Submit an off-Glide execution request with an explicit short timeout (for example, 5 seconds) that targets a capability with a slow model/heavy prompt so that the Mosaic call takes around 10–20 seconds. Observe that the request doesn't abort at 5 seconds. It runs until the platform default (around 60 seconds) or the upstream gateway times out. In the Mosaic-client logs, confirm timeoutMs=0 (or the default) on the outgoing HTTP call, regardless of the caller-supplied value. Update the sys_generative_ai control setting (Mosaic timeout) to a small value. Send a request with no explicit timeout. Observe that the control setting is also ignored. Update one_api_service_plan.timeout_ms for the plan backing the capability. Send a request. Observe that the service-plan value is not applied either. Pick a capability whose postprocessor emits a _dtRequest (DT reverse translation). Send a request with timeout=5 seconds where the primary Mosaic call returns at around 4 seconds. Confirm the subsequent DT-translation call still fires with timeoutMs=0 (platform default), so the total elapsed time exceeds 5 seconds. Repeat steps 9-11, driving the budget to exhaustion before postprocess. Confirm the DT call still fires instead of being skipped, and there is no 400001 / timeout response. Send a request via the routed-from-Mosaic > OneExtend callback path (validateOffGlideConfig) that triggers DT translation. Confirm that no timeout is honored there either. Expected behavior: The caller-supplied timeout is the strongest signal and is honored by every Mosaic HTTP call in the pipeline (preprocess gate, primary execute, postprocess DT translation). In its absence, the control-setting, service-plan, and 60 second default apply (in that order). Exhausting the budget at any phase returns a timeout response (status=error, errorCode=400001). Override values are persisted to one_api_service_plan.timeout_ms and cascade. Actual behavior: All Mosaic calls fall back to the platform default; caller / control-setting / service-plan values are ignored. There's no phase-boundary timeout enforcement and requests run well past their stated SLA. DT translation always uses 0L.
OneExtend PRB2025975	Do not log prompts or responses to Splunk for in-country routing users	Prompt and response information is removed from logs for users with in-country routing SKU.
OneExtend PRB2026081	There's a 100501 error with an async call
OneExtend PRB2026173	Glide-cs unit test fails in Zurich mainline
OneExtend PRB2027496	The metric batch_result.status is stuck at in_progress and isSkillEvaluationRun() doesn't recognize the conversational_ai evaluation type	For conversational_ai evaluation runs, metric-level batch_result records are never updated to status=completed. This causes the evaluation dashboard to display zero scores for all the metrics, even though valid evaluation data exists in the sys_one_extend _eval_metric_result and sys_generative_ai_metric tables.	Create a Conversational AI evaluation run (AutoChat) with metrics like conversation success, skill selection accuracy, and turn count. Wait for the evaluation to complete (batch_run_task status = processed). Query the metric-level batch_result records. Observe that all the metric batch_result records have status=in_progress instead of completed. Navigate to the evaluation dashboard. Observe that all the metric scores show as 0 because the Java API getCustomMetric ScoresPerPrompt filters batch_result.status! =in_progress.
OneExtend PRB2028533	Java scriptable methods aren't working in some instance nodes
Performance Analytics Dashboards PRB1993682	'This report has been migrated to a visualization' banner is shown when the unified property is false	The banner message appears: 'This report has been migrated to a visualization. Migrated visualization is here.' The link redirects to a random visualization.	Open an instance with sys_ui_context menus. Navigate to Migration Center (/now/platform-analytics-workspace/migration-center). Perform a bulk migration. Activate all content. Navigate to System Properties. Set the property 'com.glide.par.unified_analytics.enabled' to false. Navigate to any record list (for example, /incident_list.do). Create a bar or pie chart report from any column to open the Report Designer. Expected behavior: The banner isn't shown, as no report was migrated. Actual behavior: The banner message appears: 'This report has been migrated to a visualization. Migrated visualization is here.' The link redirects to a random visualization.
Performance Analytics Dashboards PRB2005720	The CoreUI dashboard selector triggers excessive ACL checks	The ACL is evaluated with a different access path and context for each check, so it is re‑executed repeatedly. When there are thousands of dashboards, this results in excessive ACL evaluations and causes the instance to hang.	Open any CoreUI dashboard. Open the dashboard selector. Select All Dashboards. Expected behavior: The dashboard selector avoids excessive ACL checks and runs one check per dashboard. Actual behavior: It triggers permission evaluation for every dashboard the user has access to, including an ACL check per dashboard and multiple field‑level ACL checks.
Performance Analytics PRB1990900	The system property com.snc.pa.dm.enable.mlb is set to true on non-Raptor DB Pro instances	Users can install a plugin with Raptor DB (pro or standard). With a schedule job, they can disable the feature by disabling the global property com.snc.pa.dm.enable.mlb.	Open an instance with raptor standard. Validate the system property: com.snc.pa.dm.enable.mlb. Expected behavior: The property is set to false. The user shouldn't be able to control it and set its values. Actual behavior: The property is set to true. The user can control it and set its values.
Performance Analytics PRB1995131	There are Query Builder (QB) and Workspace UI error pop-up clips in the surrounding UI if a query fails for CMDB NLQ	Turning off the NLQ Prediction server results in an error message in both QB and in the CMDB Workspace, and the UI doesn't render as expected in CMDB Workspace.	Turn off the NLQ Prediction server from sys_properties by deleting the URL value 'glide.prediction_service.url'. Navigate to Query Builder. Create an NLQ query, which adds nodes to the canvas. Notice that this results in the 'Oops...' error. Navigate to CMDB Workspace. Create an NLQ query, which should open a new tab. Notice that this results in the 'Oops...' error, and the UI is incorrect.
Performance Analytics PRB1999245	Discrepancies with Core UI on change and change percentage	The change and change percentage may be displayed on the widget but not on the data visualization.	Create an indicator (an indicator source isn't necessary). Enable Render continuous lines under the 'Other' tab. Open the scoresheet to edit the scores of the before indicator: Feb 28th: 200 Jan 31st: 100 Create a single score visualization using this indicator. Enable change and change percentage. Notice that the change and change percentage aren't displayed. Disable the system property 'com.glide.par.unified_analytics.enabled'. Create a single score PA widget on the before indicator, with Visualization: Latest Score. Leave all other fields as the default value. Add the before widget to a Core UI dashboard. Notice that the widget displays the change (100) and change percentage (100%).
Platform Analytics Dashboard API PRB1988004	Platform analytics displays an incorrect view for the most recently viewed dashboards	The Recent Dashboards list includes dashboards that were not previously accessed. While actively using the instance, the list appears to update as the user opens certain dashboards. However, after logging out and logging back in, the Recent Dashboards list resets and once again displays dashboards that haven't been used.	Open a base instance. Navigate to All > Platform Analytics > Library > Dashboards. Open any dashboard. Select the Arrow to view the Recent Dashboards list. If this is the first login of the day, observe that the list contains dashboards that may not have been previously accessed. Switch between several dashboards. Observe that the Recent Dashboards list updates to reflect the dashboards opened during the session. Log out of the instance. Log back in using a new browser tab. Observe that the Recent Dashboards list resets and displays a different, seemingly random list of dashboards instead of the ones previously accessed.
Platform Analytics Dashboard API PRB1992984	Can't share a dashboard to all internal users (users with at least one role)	When the itil user opens the dashboard, the dashboard isn't found.	Create a dashboard. Share the dashboard to the role 'dashboard_user' with view mode. Impersonate an itil user. Open the dashboard. Expected behavior: The dashboard is visible to the itil user. Actual behavior: The dashboard isn't found.
Platform Analytics Dashboard API PRB1998865	Sys ID is automatically populated for the value in sys_translated	This happens when the user changes the value of a tab when using a non-English language.	Navigate to Dashboards. Create a dashboard. Add a tab. Switch to a non-English language. Change the value of the tab. Save. Notice that the value in sys_translated contains sys ID.
Platform Analytics Dashboard API PRB2002552	Due to VCS update set collision, a dashboard save fails with a generic error: 'Your dashboard could not be saved'	This problem only happens in scopes with VCS-controlled apps, no matter where users create the dashboard. The important factor is whether the scope has a repo configuration set.
Platform Analytics Dashboard API PRB2007784	DomainService uses DomainSupport.hasAccess instead of DomainHierarchy for validation	When the user tries to access the dashboard, an error appears: 'Error while getting domain for dashboard'.
Platform Analytics Dashboard API PRB2025067	The sys_translated record for par_dashboard_tab is overwritten	This can cause translations to be lost.	In English, create a sys_translated record as follows: Label: あいう Table: par_dashboard_tab Element: name Language: ja Value: TabName Create a PAE Dashboard. Add a tab with the same name as the sys_translated value (TabName). Save the dashboard. Create another dashboard. Add a tab with the same name as the sys_translated value again. Save the dashboard. Switch the language to Japanese. Return to one of the dashboards. Rename the tab あいう to さしす. Check the other dashboard tab. Observe that the translation is lost because the sys_translated record is overwritten.
Platform Analytics Filters PRB2013316	Additional controls for single and multi select	For both radio buttons and check boxes, there should be a way to clear the selection instead of using the 'Clear all filters' capability on the dashboard level.
Platform Analytics Filters PRB2014365	JavaScript-based sys_choice labels are rendered as raw text in dashboard filters	A sys_choice record uses JavaScript to dynamically populate the country label when the value is null. However, when this field is used in Platform Analytics dashboards (via indicator breakdown with sys_choice as the facts table), the JavaScript isn't evaluated and is instead displayed as raw text in the filter.
Platform Analytics Migration API PRB1998821	The Platform Analytics Migration Center summary window isn't counting all domain dashboards when doing migration in the global domain
Platform Analytics PRB1933496 KB2599826	The main Refresh icon in the dashboard doesn't work with the list component added to the dashboard	In Zurich, the main Refresh icon in dashboards doesn't refresh list components when they are added to a dashboard. While other widgets (such as single score visualizations) refresh correctly, the list component remains static. This issue occurs because the dashboard refresh event isn't mapped to the list component.	Refer to the listed KB article for details.
Platform Runtime PRB2014434	The Granular Delegation API isn't returning any results
Playbooks (Family Channel) PRB2009677	The complex object's format is different for playbook output/input	The flow gives an 'com.snc.process_ flow.exception. ProcessAutomationException: Could not serialize value' error when running playbook.
Playbooks (Family Channel) PRB2023731	Child agent changes in playbook snapshot for hybrid agent configuration	This is a product update.
Process Mining Workspace PRB1980574	The Process Mining app should be free on the store	Process Mining is shown as a licensed app on the store. However, it's auto-installed on many instances, which causes confusion when users aren't able to upgrade.	Scenario 1: Navigate to the store. Search for the Process Mining app. Check the pricing. Scenario 2: Navigate to the plugin list. Search for the Process Mining (sn_po) app. Make sure there are some app updates available. Expected behavior: Users can update the app to the latest version. Actual behavior: Users get a license error, so they can't proceed with the app update.
Process Mining Workspace PRB2011383	Update the Appsee event
Project Management PRB1908085	Error on 'Manage View'
Project Management PRB2009176	Opening the Project Status Report in the Project Workspace triggers an update to the Overall Health field on the corresponding project record	Opening the Project Status Report within the Project Workspace updates the Overall Health field on the corresponding project record to match the value from the status report.	Open a project in the Project Workspace (PW). Navigate to the 'Status Report' tab. Create a status report. Navigate to the 'Details' tab. Locate the Status field. Ensure the Status field is different from the Overall Health field in the status report. If both values are the same, manually update the Status field on the project. Open the 'Status Report' tab from the left panel. Navigate back to the 'Details' tab. Notice that the Status field is automatically reverted to match the Overall Health value from the status report.
Project Management PRB2021806	Add a 'No status' option to the Project Status field	The Project Status field (pm_project.status) currently doesn't include a 'No status' option, which prevents projects from having an unset/neutral status state. This causes ambiguity in reporting and makes it difficult to distinguish between projects that have not yet been assigned a status vs. those intentionally marked as 'On Track' or 'At Risk'. A 'No status' option should be added to the Status field list. This verifies that status transitions — including from an unset state to a defined status — can be accurately captured and reflected in usage analytics.	Navigate to Project Management > Projects > All Projects. Open any active project. Select the Status field. Observe that a 'No status' option isn't available.
Record Hierarchy PRB2013585	Reconciliation of record hierarchy with missing path field skipped	If the user deletes the rh.hierarchy.create event, there is no 'Reconcile Hierarchy' link under Related Links. If the user executes SNC.RecordHierarchyAPI.get ().reconcilePaths(sysId), no event is posted to resume creation of the hierarchy.	Create a sys_record_hierarchy. Immediately delete the associated rh.hierarchy.create event that's generated in sysevent. Observe that the sys_record_hierarchy is in a PENDING_CREATE state with no event to progress it to the next step. Visit the sys_record_hierarchy record. Notice that there is no 'Reconcile Hierarchy' link under Related Links. Copy the sysId of the sys_record_hierarchy record. Navigate to Scripts - background. Execute SNC.RecordHierarchyAPI. get().reconcilePaths(sysId). Observe that nothing happens. No event is posted to resume creation of the hierarchy.
Record Hierarchy PRB2014979	Allow RecordHierarchies to always be usable optionally unless it is inactive	The current behavior shows that after a record hierarchy is defined, any query operations that use the hierarchy will explicitly return 'false' until every record in the hierarchy has been assigned a path. After paths have been assigned, the hierarchy is said to be 'usable' as its fully initialized and ready for queries. A GP was added to optionally allow the hierarchy to be usable while records are being assigned paths. The results for queries may be incomplete or partially stale depending on how long the initialization takes place, but should otherwise be accurate. Users are beginning to use IN_HIERARCHY conditions for all sorts of features ranging in criticality, up to and including ACLs. If the record hierarchy runs into some sort of issue, such as it can render the record hierarchy unusable and all queries would begin to fail. However, the paths assigned to all records are still valid as long as changes are minimal, and can continue to return mostly accurate results while any issues with the record hierarchy feature are addressed. This fix is to address usability; as long as the record hierarchy isn't explicitly deactivated, it can be made 'usable' and queries will return results based on whatever paths may be assigned at the time.
Related Lists PRB2000158 KB2823064	Related lists don't show row data	Related lists don't show row data in list cells [Core UI].	Refer to the listed KB article for details.
ReleaseOps - Family PRB2009639	Instance scan failure doesn't return the correct message	The scan failure doesn't mention a reason for the failure in the work notes.	Open a Zurich instance. In development, start a full instance scan. In production, mark a DR as ready for assessment. Expected behavior: The scan failure shows the reason for the failure in the work notes. Actual behavior: The scan failure doesn't mention a reason in the work notes.
Resource Management PRB2003945	Cost plan breakdowns are duplicated	When the user changes the start date and end date from a resource plan, the business rule (BR) UpdateCostPlan AssociatedToResourcePlan launches an event. The BR 'Recreate Requested Allocations' is also launched, updating the resource plan again and changing the planned cost. In summary, when the user updates the dates, the system updates the resource plan twice, one time for the dates and another for the planned cost, practically at the same time. Both changes trigger the BR UpdateCostPlan AssociatedToResourcePlan and two events are created. Since there are two events in the system, they will try to perform the same action, and this could lead to problem concurrency. The two events are processed at the same time and the cost plan breakdowns are created twice.	Navigate to pm_project.list. Open the project where the duplicated cost plan breakdowns were created.
Resource Management PRB2020809	Updating dates in the assignment should update the resource availability	Dependent choice config should also be supported in the project workspace bottom grid.	Open the project workspace. Navigate to the project. Create an unassigned assignment from the bottom grid. Check any resource. Notice its availability. Update the assignment dates. Check the resource availability again. Observe that it's still the same, even though the availability should be updated as soon as the dates are updated. Create any field on the resource assignment table which is dependent on any other reference field. Observe that the dependent config isn't supported in bottom grid choices. It should be supported.
Resource Management PRB2024520	Fetching choices with their dependent field values to prepare criteria	The field doesn't show the choices according to the dependent config defined.	Create any field on the resource assignment table which is dependent on any other reference field. Add that field in the resource assignment bottom grid in PWS. Open the project workspace. Navigate to the project. Create an unassigned assignment from the bottom grid. Observe that the field doesn't show the choices according to the dependent config defined.
Server-side scripts PRB1994381	Discovery has issues on some node after upgrading in Australia	After upgrading to Australia, JavaScript running in app nodes fails to call Java functions. The following warning appears: '*** WARNING *** Evaluator: com.glide.script.RhinoEcmaError: undefined is not a function.' This impacts various features, including Discovery and Event Management.
Server-side scripts PRB2022176	The script name should be logged when a page title is unavailable in the 'Guarded Scripts' list entry	The page title can be empty in some cases, such as a scheduled job and probably more. It should become the script name if there isn't a page name that makes sense.
Server-side scripts PRB2022324	Logging for all sandbox scripts is broken		Execute a script in the sandbox. Observe that it should be logged, but isn't.
Service Catalog PRB1976031	Lookup select box variables with 'Choices depend on' store a choice value instead of a label	When a lookup select box variable on a record producer is configured to depend on another variable, the system saves the underlying choice value rather than the human‑readable label in the created record. This behavior occurs only for lookup select box variables with the 'Choices depend on' setting; standard select box variables work correctly. As a result, the Subcategory field of the incident shows the internal value (for example, '1023') instead of the expected label (for example, 'Internal Application'), which can cause confusion for users reviewing incident data. This is observed in portal pages like the ESC standard ticket page, where the value is shown in the variable summarizer.
Service Catalog PRB2014228	Update the copy in the alert banner	Catalog item forms should be automatically pre-filled using the context from the conversation. The user shouldn't have to re-enter information they've already shared in the chat.
Service Catalog PRB2020118	Ensure that a search term is a minimum of 3 words before triggering the slot fill	.
ServiceNow SDK (Glide) PRB2016794	Skip delete issue for 'Coalsce' tables	When a Fluent app defines a role, the SDK generates a new sys_id. The existing ScopeConflictDetector keys off sys_update_name (which encodes the sys_id), and finds nothing in sys_metadata, so no conflict is flagged. The installer then coalesces onto the real global admin role by matching the Name field and silently overwrites its sys_scope, stealing a system-owned record into the app's scope.	Navigate to IDE. Create a fscoped Fluent app. Create a role with the name as 'admin'. Build and install. Expected behavior: The Global role shouldn't be overridden. Actual behavior: The Global role is overridden.
ServiceNow SDK (Glide) PRB2021468	Dictionary updates aren't propagated when installed via FluentXMLUploader	When the user deploys changes, the dictionary updates get skipped.	Create a fluent configuration project. Update the table column maxLength value on the instance in the UI. Update it again in the fluent configuration project. Deploy the changes. Expected behavior: The maxLength of the column is updated. Actual behavior: The dictionary updates get skipped.
Service Portal PRB2014571	Alignment of the Now Assist search results gets overflow	This happens on the portal while loading on browser resolution size 360 x 640.	Make sure the Now Assist genius result is enabled on the portal. Navigate to a base instance. Change the session language to Deutsch. Navigate to the /esc portal. Open the developer tool of the browser. Set the resolution as 360 x 640. Search for any keyword (for example, people manager). Observe that the filter text 'Am relevantesten' gets overflow.
Session Management PRB2024286	Make application nodes more resilient to DDoS attacks from unauthenticated traffic	This is a product update.
Sidebar (Family Release) PRB2019090	The cached RecordCard is returned without checking if the user has access to record	This is a caching issue, and occurs for both with incident and HR records. When the code checks if the Record Card is in the cache, it does not re-check if the user can access the record or not. When clearing the cache, the record access check is calculated correctly.	Create an incident (INC1) as the user Abel Tuter. Create a discussion on it. Paste INC1 in the chat. Notice that it renders a record card, and the user Abraham Lincoln can't access INC1 in the sidebar chat. As the user Abraham Lincoln, attempt to paste the INC1 number in the chat. Notice that Abraham Lincoln can see the INC1 record card and its details, when Abraham Lincoln should not be able to because they have no access. Create another incident (INC2) as the user Abel Tuter. Create a discussion on it. Paste INC2 in the chat. Notice that it renders a record card, and the user Abraham Lincoln can't access INC2 in the sidebar chat. The INC2 record card is not in the cache.
Software Asset Management PRB1959756	The 'Connection Setup' section is missing on the Custom Integration view for the Workday Integration Profile after upgrading sn_sam_saas_int	When trying to integrate with Workday (SaaS License Management), some tabs are missing (for example, the 'Connection Setup' section and related list). This makes the user unable to validate the connection and publish the profile. The 'Connection Setup' section should be visible and accessible on the custom integration view after the upgrade.
Software Asset Management PRB2018383 KB2986694	A Microsoft per‑core license metric isn't visible after a Zurich or Australia upgrade	The MS per core metric was moved from the apply_once folder to the update folder. The fix script to set the metric group was overwritten, since the update folder file insert ran after. Thus, the MS per core uploaded with no metric group.	Refer to the listed KB article for details.
Software Asset Management PRB2021450	The Add Lifecycle Data table UI action for lifecycle playbook is incorrectly available in Zurich	The Add Lifecycle Data button is available for Zurich users, despite it being an Australia feature. When the user selects the button, nothing happens.	Download version 10.0.1 or 10.0.2 of the app-sam-workspace store app. Navigate to Workspaces > Software Asset Workspace > Software asset analytics. Select the 'Lifecycle Management' tab. Select a report result to drill down into any report. Select the Add Lifecycle Data button. Observe that nothing happens.
Software Asset Reclamation PRB2008233	Potential savings on reclamation candidates should be set regardless of licensing status	It should be set for both licensed and unlicensed subscriptions/installs.
Software Asset Reconciliation PRB2021614	Potential savings on reclamation candidates should be set regardless of licensing status	It should be set for both licensed and unlicensed subscriptions/installs. A software's license status doesn't impact the potential savings for removing that software. The reclamation candidate should reflect the cost of having that software installed.
Software Entitlements PRB2023511	Add an asset tag, agreement number, and location to entitlement duplicate check attributes	This is a product update.
Standard Ticket Page PRB2023619	Family changes to fetch Summary fields from the m2m table	This is a product update.
Suite Engine PRB1966133	Assigned software should handle suite of suite	The reconciliation logic for licensing subscriptions should license corresponding installations for the same user.
System Archiving PRB1938103	Reparenting of sys_attachment doesn't work when the table name is longer than 37 characters	.
System Archiving PRB1988156	Archive reparenting doesn't work with peripherals and large table names	The records for sys_audit and sys_journal_field don't reparent, and an error is found in the logs.	Create a table with sys_audit and sys_journal records with a length that is over 60 characters. For example, 'u_supercalifragilisticexpiextralongname_restore_destroy_test'. Attempt to archive it. Notice that the sys_audit and sys_journal_field records don't reparent. Instead, the log shows this error, 'Found in current data management track'.
System Events PRB1939688	Inputs aren't getting processed for subflows with inputs	Implement thread pool with the configurations/design. The scope excludes any changes related to autoscaling or processing framework (which means it co-exists with the jobs that are already there for flow_engine).	Create a subflow with inputs. Trigger it. Expected behavior: All flows are completed without any errors. Actual behavior: A couple flows move to the completed state with an error log for not processing inputs.
Table Administration and Data Management PRB2025009	Ordering a query by a Function field that wraps a translated_text column returns empty values for that list		Create a table. Add a translated_text column. Add Function field on that column. Navigate to v_plugin.LIST. Activate the plugin 'com.snc.i18n.swedish'. Insert records (New Name=Alice and New Name=Bob). Submit. Add the Swedish translations (Älice_sv for Alice and Bob_sv for Bob). Submit. Switch the session to Swedish. Open u_repro_translated_function.LIST. Personalize the columns to show 'Name' and 'Name Function'. Select the Name Function header to sort by ascending. Expected behavior: The 'Name Function' column shows Älice_sv and Bob_sv (or the English fallback). Actual behavior: The 'Name Function' column is empty for both rows.
Territory Planning PRB2016459	Unused form field mappings in the 'Suggested Agents' and 'Relocate Agents' pages cause performance issues	This happens in the CSM/FSM Configurable Workspace. There are two UX screen conditions where not all form field mappings are used.
Trace Collector - Family Release PRB2023730	Trace collector Epic	This is a product update.
Trace Collector - Family Release PRB2030648	The Azure Application Insight throws an error	During testing of the Application Insight feature, users are intermittently seeing an error: 'Unable to find app_insights_config.json'.
UI Builder (Family Channel) PRB2023628	Configuration table, APIs and ACLs	This is a product update.
UI Field Administration PRB1917016	TinyMCE Editor is not dark theme compatible in the builder screen	The Description field and the 'Preview' and 'Default Value' sections don't honor the dark theme.	Log in to an instance as an admin. Navigate to Catalog Builder. Select the Admin Profile icon. Set the theme as dark. Select Build From Scratch (the create catalog item flow). Select Standard. Continue the flow from the pop-up. Scroll a bit. Observe that the Description field doesn't honor the dark theme. The same issue is also seen in the 'Preview' and 'Default Value' sections.
UI Field Administration PRB1919959	An exception displays in a syslog when adding/removing a property in full-screen	TableName is returned as an empty string in a function, which causes tabledescriptor to throw an error in the log during initialization.	Log in to a Zurich or Australia instance as an admin. Navigate to kb_knowledge_base.list or sys_user.list. Select any record or user. Open the related list. Edit any properties. Select Save. View syslog.list. Order it by 'Created by most recent' first. Notice the following exception: 'Table name can't be null: java.lang.IllegalStateException: Table name can't be null: com.glide.db.TableDescriptor.init(TableDescriptor.java:98)...'
UI Field Administration PRB1969665	Currency values change automatically in workspaces	The currency value changes automatically when selecting in and out of the Currency field, without actually changing the value.	Open any record from incident.list. Select View - Service Operations Workspace. Create a field type 'Price'. Give it any name and any value. Save the record. Open sys_user table for the name 'Beth Anglin' who has 'ITIL' role. Add a County Code field with the value set to 'system(US)'. Impersonate the user Beth Anglin. Open Service Operations Workspace. Select into the Price field under the 'Details' tab. Without changing anything, select out of the field. Notice that the value doesn't change. End the impersonation. Update the County Code field to 'Germany' or any other country for the user Beth Anglin. Impersonate the user Beth Anglin again. Open Service Operations Workspace. Select into the Price field under the 'Details' tab. Without changing anything, select out of the field again. Observe that the value changes.
UI Field Administration PRB1982868	The pop-up related to the Comment/Work Note web link only displays in the boundary of the 'Compose' tab	The message, 'The URL you entered seems to be an external link. Do you want to add the required http:// prefix' is confined to the 'Compose' section and not overlaying the page. This occurs in the CSM workspace, Service Operations Worspace (SOW), and the HR workspace. This issue occurs in Safari browsers, but has been resolved in Chrome browsers.	Open a Safari browser. Log in to a Xanadu instance. Open the CSM workspace. Open any non-closed incident. Navigate to the Activity Stream. Navigate to the 'Compose' section. Under 'Subject', select the 3 dots (...) if formatting options aren't available. Select the Link button. In the URL field, enter 'www.servicenow.com'. Select the Save button. Observe the message, 'The URL you entered seems to be an external link. Do you want to add the required http:// prefix'. The message is confined to the 'Compose' section, and not overlaying the entire page.
UI Field Administration PRB1985085	Quick action visibility script is executed in global scope, preventing access to scoped script includes	The quick action runs into an error: 'Evaluator.evaluateString() problem: java.lang.SecurityException: Illegal access to package_private script include...'	Create a quick action in a scoped app. From the visibility script, call a script include that is accessible only from the same scope. Notice the quick action runs into an error: 'Evaluator.evaluateString() problem: java.lang.SecurityException: Illegal access to package_private script include CMDBGenAICIFormContextualHelp: caller not in scope sn_cmdb_gen_ai: com.glide.script. RhinoEnvironment. checkScriptableAccess (RhinoEnvironment.java:723) com.glide.script. ARhinoScope. checkScriptableAccess (ARhinoScope.java:134)'. This happens because the visibility script is evaluated in the global scope.
UI Field Administration PRB1998819	WWNAglobal utils should be removed
UI Field Administration PRB2025633	AI Agent modified fields API returns aiCreated instead of isAiCreated in JSON response	The Boolean field is named incorrectly.	Provision an instance with Gen AI installed. Open any existing record page (for example, an incident in Ui16). Observe the AI Agent modified field API response. Inspect the JSON response body. Expected behavior: The Boolean field is named 'isAiCreated'. Actual behavior: The Boolean field is named 'aiCreated'.
UI Form Administration PRB1999196	The citation URL is retained and fields are highlighted when the user logs in to an instance again	The highlight should happen only if the user selects links in a summary or the email outputs.
UI Form Administration PRB2006958	Introduce system property to hide form level TI banner	The alert appears when the form is loaded. There's no way to avoid seeing the alert if not required.	Open a SOW record with Task Intelligence field level recommendations. Notice that the form level banner appears with deep links to fields that have recommendations. Cancel the alert. Load the form. Observe that the alert appears again.
Upgrade Center PRB1889947	Instance nodes don't appear and an error appears: 'javax.crypto.AEADBadTagException: Error finalizing cipher data: mac check in GCM failed'.	Instance JVMs won't start after the error, 'java.io.IOException: javax.crypto.AEADBadTagException: Error finalizing cipher data: mac check in GCM failed'.
Upgrade Center PRB1992027	'Add node' automation times out too quickly	The 'Add node' automation times out before the node is online. PluginManager takes a long time to complete initialization.
Upgrade Center PRB2023239 KB3015307	There is a mismatch in the Glide version between the app node and the database following an upgrade	A new code path introduced the MariaDBI18NSQLFormatter class. When the sys_properties record of the 'com.glide.db.session _language_collation_feature' property is set to true, it takes a code path on upgrade or restart where an instance will not come up. When 'com.glide.db.session _language_collation_feature' is false, the code path exits early and doesn't cause this issue.	Refer to the listed KB article for details.
Usage Analytics PRB2021289	Long session durations are observed for few Usage Insights Sessions, causing inaccurate session duration metrics	A fix for long session duration in some Usage Insights sessions.
UX Framework PRB1969085 KB2677276	GraphQL schema changes for client-interaction.graphl in ux-metrics	The user sees error logs from the ClientMetricsRestService.java file. The error logs are logged to the syslog table.	Refer to the listed KB article for details.
UX Framework PRB1986291	List dynamic routing shouldn't be applied when creating a record from a related list	Dynamic routing is handled within the recordRoutesMapping Client Script Include. Kb_knowledge mapping uses source_component: 'list' to route article clicks to kb_view. When the New button is selected on a related list, the kb_view route is incorrectly applied. Since kb_view can't handle new records, the page breaks.
UX Framework PRB1987200	GLOBAL_NAVIGATION_REQUESTED isn't routing to the feature on select after a hard refresh of the page	When the user selects the desktop notification, it doesn't navigate back to the workspace.	Enable desktop notifications in the Agent Chat. Make an agent self available. When an incoming work item appears for the agent, select the desktop notification. Observe that it doesn't navigate back to the workspace.
UX Framework PRB2002772	Preset for the presentational list (nested inside the record list bundle) wasn't discovered during pre-population	The list is completely broken.	On a base instance, build a custom UIB page where the record list component is nested deeply under a tabs viewport. Perform cache.do. Visit the UIB page. Observe that the list component is empty and broken. No data is rendered. Visit the SOW list page: /now/sow/list. Observe that the SOW list page is also broken and empty.
UX Framework PRB2018001	An blank page issue on some nodes during upgrade	This issue occurs in Australia instances.	Upgrade the instance. Navigate to the node. Observe that an empty page is displayed with a spinning wheel.
UX Framework PRB2023614	In-product surveys MVP	This is a product update.
UX Framework PRB2030213	When a user navigates between two pages that both have IPS surveys configured, the survey triggered on the first page isn't dismissed on navigation
Virtual Agent Designer Legacy PRB1752802	The virtual agent designer page isn't accessible on Safari	All seismic and Core UI pages should work on Safari 17.1+ for Washington and 17.3+ for Xanadu.	Open a Safari browser. Navigate to All > Virtual Agent > Designer. Observe the following error: 'Safari and older versions of internet Explorer are unsupported browsers, please switch to a different or newer browser.'
Virtual Agent PRB1977665	The 'Pagination' API doesn't reset totalSearchResultsCount if a search term is removed		Execute a topic with dynamic input. Scroll through the input list. Note that the totalSearchResultsCount is 0 in the response for /options, during scrolling. Enter a search text in the input. Note that totalSearchResultsCount has some valid integer values in the response for /options. Remove the search text in the input. Note that totalSearchResultsCount still has the same value from step three in the response for /options and isn't reset to 0.
Virtual Agent PRB1985280	Duplicate attachment is displayed in Virtual Agent after portal refresh		Open an instance in two browsers. In browser one, impersonate the agent 'Fred Luddy'. Set the agent as available in Service Operations Workspace. Switch to browser two. Impersonate an admin. Initiate a chat from the portal or web-client. Connect to a live agent. In browser one, accept the chat as the agent. Send any image as end-user to agent. Refresh the portal at end-user. Expected behavior: Nothing happens. Actual behavior: The duplicate attachment/image is displayed only in Virtual Agent.
Virtual Agent PRB1999511	Message preview and unread badge count don't work on page refresh	When the user refreshes the page, the message preview doesn't show up. On standard chat, there's also no unread badge count.	Navigate to /sp. Query 'What is spam'. While standard or enhanced chat is processing, close the VA. Observe that there is an unread badge count (1) and the message preview pops up. Refresh the page. Expected behavior: There is still an unread badge count and the message preview shows up. Actual behavior: The message preview doesn't show up. On standard chat, there's also no unread badge count.
Virtual Agent PRB2003424	External users are unable to delete closed chats from the Virtual Agent chat history	External users are unable to delete closed chats from the Virtual Agent chat history and encounter the error message: 'Failed to delete conversation. Please try again'.	Impersonate an external user. Navigate to the ESC portal. Open the Virtual Agent chat history. Select the Bin icon to delete a closed chat. Notice that the error message appears: 'Failed to delete conversation. Please try again'.
Virtual Agent PRB2004944	Teams conversations can pull in context from NAP if not using a NAP channel	When the user executes a skill, it should prompt for more context if needed instead of using NAP context.	Set up an instance using NAP with the standard chat widget (not enhanced chat). Make sure the instance has Teams integration installed for NAVA. In SOW, navigate to an incident as a user. In Teams, as the same user, start a conversation. Execute the 'Summarize Record' skill. Expected behavior: The skill prompts for the record to summarize. Actual behavior: The skill uses the incident from NAP context.
Virtual Agent PRB2006240	The Virtual Agent server doesn't process pre-chat surveys for bot 2 bot	Users aren't routed to a live agent in bot to bot conversations. This was due to a misconfiguration where pre-chat survey was enabled and the user wasn't able to provide a response. Interactions aren't assigned to the queue or an agent in Agent Workspace. It was working when the user first upgraded, but recently has stopped.	Start a conversation in crescendo using the ServiceNow Virtual Agen API. Send a request to transfer to a live agent using endpoint /api/sn_va_as_service/bot/integration. Notice that the conversation doesn't get routed to the live agent.
Virtual Agent PRB2012889	vaSystem.getSearchText isn't working with off-Glide
Virtual Agent PRB2013847	Create a scriptable API for endChatSummarization
Virtual Agent PRB2013993	The deployment channel for a conversation isn't populated in context variables
Virtual Agent PRB2015140	Multiple interactions are created during live agent execution	This only happens when a conversation is reset or abandoned.
Virtual Agent PRB2015460	Now Assist Portal's unread conversation count disappears after the execution is started via AI Agent triggers or Runtime APIs
Virtual Agent PRB2015866	Handshake fails when it fails to load an in-progress conversation	The same error impacts async tool execution intermittently.	Impersonate any user who has existing open conversations. Open Now Assist Portal. Expected behavior: It should load the chat client successfully with a new or older conversation. Actual behavior: It keeps loading and throws an error.
Virtual Agent PRB2017392	Add topic, agent execution to the sys_cs_now_ assist_execution table for NextWave	When a topic/agent is run, the sys_cs_now_assist_execution doesn't populate.
Virtual Agent PRB2018098	AI agents are incorrectly displayed on the Virtual Agent's topics list	On Virtual Agent, users see AI agents in the 'View All Topics' section under the 'Others' category. But none of those agents are marked to be visible on that section.	Open Virtual Agent. Navigate to the 'View All Topics' section. Scroll to the 'Others' section. Observe that multiple AI Agents are visible there.
Virtual Agent PRB2018878	Employeeslate gets a 404 error for promoted-skills		Navigate to Employeeslate's home. Observe the 404 error.
Virtual Agent PRB2021412	Send flag to AIEL to identify new vs. existing catalog pages	When a catalog item is opened in the NextWave interactive view, AIEL needs to be able to distinguish whether the catalog page being rendered is a new (modern/interactive) catalog page or an existing (legacy) catalog page. Currently, no flag is sent to AIEL to make this distinction, which can cause incorrect rendering behavior or routing decisions. A flag should be passed to AIEL at the time of catalog page load so it can apply the appropriate logic based on the page type.	Open Now Assist NextWave chat experience. Trigger a catalog item via conversational flow. Observe that AIEL doesn't receive any flag to differentiate between a new catalog page and an existing/legacy catalog page. This can result in incorrect behavior when AIEL applies logic without awareness of the page type.
Virtual Agent PRB2022716	NextWave Premium Chat doesn't work for external users (snc_external) on CSM portal	The console error '403 FORBIDDEN' appears on ais_auth_token_refresh API. The external user can't use the chat experience.
Virtual Agent PRB2024076	Pickers aren't honoring pre-selected options	This works in the web version and is considered a feature gap.	Set up a topic that has pre-selected options. Execute a topic in MS Teams. Expected behavior: The pre-selected options defined in the topic should be presented to the user as selected. Actual behavior: An empty picker is presented with nothing selected.
Virtual Agent PRB2024849	Simplified handshake for NextWave needs fixes for session creation, channels, and context	The existing session needs to be looked up by conversation before creating a new one. Also, if the consumer account context name is provided, it needs to be stamped and associated with the consumer account when created. Finally, the channel user profile needs to be properly resolved for channels.
Virtual Agent PRB2025027	Create a custom fallback message for AI Control Tower (AICT) Assistant
Virtual Agent PRB2025311	Add an ai_greeting type mapping in getContextProfile MessagesNoDomain to support a Premium Chat greeting message feature	The Now Assist Virtual Agent (VA) admin UI saves static AI greeting messages as sys_cs_context _profile_message records with type=ai_greeting. For these to load correctly, getContextProfile MessagesNoDomain in NowAssistIn VAadmin ConsoleUtil needs an explicit branch mapping ai_greeting → aiGreetingMessage. Without this Glide-services change, the greeting message isn't populated for multi-domain instances when the feature ships.	On a multi-domain instance, navigate to Now Assist VA admin UI > Chat Experience > Premium Messages tab. Configure a static greeting message. Save. Refresh the page. Navigate back to the chat. Observe that the custom greeting message isn't populated even though the message was saved earlier.
Virtual Agent PRB2025956	Resolved issues after introducing four tables for Virtual Agent (VA) analytics	The search-related analytics can be written to the four tables instead of sys_ci_analytics as event entries.	Make sure sn_nowassist_va.analytics .persistence_strategy = 'table'. Launch VA client. Type anything in the search. Observe that all the expected entries are in the new tables and view.
Virtual Agent PRB2026053	sys_cs_context_profile_topics should be sent to the off-Glide cache for Authorizing Official (AO) to read
Virtual Agent PRB2026723	In OGChannels, there should be a picker search update
Virtual Agent PRB2026804	getBrandingConfig() doesn't provide the correct response in NextWave	api/sn_nowassist_va/ og_branding_configs/ {deploymentId} doesn't provide the correct response for branding configuration. It's missing a Value field and the type is null for the first item.
Virtual Agent PRB2028137	Glide-cs-test failures in Australia
Virtual Agent PRB2029447	Knowledge article links are broken in the sources citation when the article has an attachment	There is an issue with the link formation in the source citations for the KB article if there is an attachment. The link generated by Now Assist uses the sys_id of the PDF attachment instead of the KB article sys_id in the URL.
Virtual Agent PRB2030245	External users can't upload files	There's an HTTP 422 error and the file fails to upload.
Virtual Agent PRB2031950	AI-user fetch issue	The conversation user for the ZTSD flow is AI L1 Service Desk Specialist, which is the identity type ai_agent. The cache configuration get_user|table:{table} |field:{field}|value:{value} explicitly ignores users of type 'ai_agent'.	Perform a cache fetch call for the cache key 'get_user|table:{table} |field:{field}|value:{value}' for any user of the identity type 'ai_agent'. See that it doesn't return the User field value.
Virtual Agent PRB2032108	Glide-side changes for the NextWave release	This is a product update.
Virtual Agent Web Client PRB1993741	When live_agent_only is set to true, the agent chat triggers the greetings topic instead of going to the live agent	The /esc portal agent chat is configured to only trigger the live agent. However, when the user selects the Virtual Agent icon, it triggers the greetings topic and doesn't go to the live agent. When the user ends the first conversation and creates another conversation, it then goes straight to the live agent. The issue only occurs the first time the user selects the Virtual Agent icon in the session. It also only happens for Now Assist virtual agent sessions.	Enable Now Assist in Virtual Agent. Configure the agent chat config for the /esc portal. Disable all the other agent chat configs, only keeping one active. Navigate to /esc. Select the Virtual Agent icon. Expected behavior: It triggers the live agent support topic. Actual behavior: It triggers the greetings topic. However, if the user ends the conversation and creates a new conversation, it goes straight to the live agent.
Virtual Agent Web Client PRB2030686	The Stop Flow button is incorrectly displayed in the Virtual Agent input field during file upload	The Stop Flow button should only be visible when a dynamic loader control is active, and it should not appear for normal file upload controls.	Create a NLU topic with file upload. Open the topic in standard Virtual Agent chat. Observe the Stop Flow button in the bottom text input field when the file upload control appears.
Virtual Agent Web Client PRB2033755	SEARCH_FALLBACK_EVENT is missing in sys_ci_analytics	SEARCH_FALLBACK_EVENT records are no longer created in sys_ci_analytics in Australia.	Navigate to Dynamic Window. Create a conversation. Start a conversation. Search for 'iPhone'. Select See more of iPhone. Navigate through the conversation without starting a catalog request. End the iPhone search conversation with 'nothing else'.
Word Document APIs PRB1973625	Support unzipped font size up to 20 MB in Word Doc API	The property com.snc.word_doc_api .unzip_word_size_limit_mb supports only up to 10 MB file size. This is an enhancement request to support up to 20 MB.

Fixes included

Unless any exceptions are noted, you can safely upgrade to this release version from any of the versions listed below. These prior versions contain PRB fixes that are also included with this release. Be sure to upgrade to the latest listed patch that includes all of the PRB fixes you are interested in.

• Zurich Patch 9 Hotfix 2
• Zurich Patch 9 Hotfix 1
• Zurich Patch 9
• Zurich Patch 8 Hotifx 3
• Zurich Patch 8 Hotfix 2
• Zurich Patch 8
• Zurich Patch 7b Hotfix 1
• Zurich Patch 7b
• Zurich Patch 7a Hotfix 1
• Zurich Patch 7
• Zurich Patch 6
• Zurich Patch 5
• Zurich Patch 4
• Zurich Patch 3
• Zurich Patch 2
• Zurich Patch 1
• Zurich security and notable fixes
• All other Zurich fixes