Yokohama Patch 3

  • Release version: Yokohama
  • Updated May 1, 2025
  • 79 minutes to read
  • The Yokohama Patch 3 release contains important problem fixes.

    Yokohama Patch 3 was released on May 01, 2025.
    • Build date: 04-30-2025_1345
    • Build tag: glide-yokohama-12-18-2024__patch3-04-17-2025
    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

    Yokohama Patch 3 includes 285 problem fixes in various categories. The chart below shows the top 10 problem categories included in this patch.

    Figure 1. Top 10 problem categories
    Fixed issues grouped by problem categories bar chart

    Changes in Yokohama Patch 3

    Activate indexing of catalog variable content on Catalog Item records
    Activate indexing of searchable content from variables on Catalog Item records. Configure the set of Catalog Items eligible for catalog variable indexing and the set of variables to index.
    AI Search system properties
    glide.ais.ingestion.catalog_variables_filter_query: Encoded query string for a query that Catalog Item records must satisfy to be eligible for catalog variable content indexing.
    glide.ais.ingestion.ignore_catalog_variables_read_roles: Option to allow indexing of searchable content from role-restricted catalog variables on Catalog Item records.
    glide.ais.ingestion.index_catalog_variables: Option to allow indexing of searchable content from globally accessible catalog variables on Catalog Item records.
    Field Service Management release notes

    :

    Variable types supported by AI Search indexing
    Service Catalog variable types indexed from Catalog Items.

    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

    Agent Chat

    PRB1862727

    A 'Show audio' modal appears when an agent presence state is changed to 'Available' for Safari users A user interaction is required on the page for the audio to work, especially in inactive tabs. This was enforced by a modal when the audioContext was detected to be suspended, which occurs when a new AudioContext is created in the page. Now, the modal appears for Safari whenever the presence changes in the inbox to 'Available' to avoid having users refresh the browser to see the modal.

    Form Builder

    PRB1820975

    A form isn't saved on a cross scope in ServiceNow studio There's an error in the networking calls.

    List Filters

    PRB1801767

    Japanese names cannot be filtered by a Japanese term but can be filtered by an English term under the Japanese setting Unable to filter lists by Translated Text fields in non-English languages when using the non-English translation

    MID Server

    PRB1870465

    KB2022624

    ECCSender does not use the correct character set when reading XML queue files ECCSender uses the default file encoding/character set when reading XML queue files from a disk. This results in incorrect characters in the response for some special characters. ECCSender should use UTF-8 when reading these files, as XML queue files on disk are always UTF-8.

    Refer to the listed KB article for details.

    All other fixes

    Problem Short description Description Steps to reproduce

    Access Control

    PRB1859273

    A query that consumes significant time is invoked every minute by the UserHasRole Patcher sys_trigger It's an issue with CSM performance.

    Prepare an instance with CSM with 5 times the data.

    Observe the time taken by the query executed as part of the UserHasRole Patcher scheduled job.

    Activity Stream

    PRB1830587

    Activity Stream doesn't limit the size of Journal fields The maximum size of journal entry appears to be either 50Kb or 100Kb by default. UI16 has a 10MB limit on the total Journal field size. Once this limit was reached it truncates every journal entry to 50K. The entry cap is 250 by default. The entry cap is 30 for attachments. The size cap + entry cap is what ensured no out of memory errors in UI16.
    1. Create a script that creates a 10MB work note. This can be a 100 character string that is appended to the work note 100,000 times in a loop.
    2. Add 2-3 of these to a single incident.
    3. Request the incident in Service Operations Workspace (SOW).

    Expected behavior: The incident should load with these massive strings.

    Actual behavior: Chrome and Edge crash as the browser runs out of memory.

    1. Add 100 of the 10MB work notes from #1.
    2. Request the incident.

    Expected behavior: The incident should load in SOW.

    Actual behavior: The Glide server crashes with an out of memory error.

    Activity Stream

    PRB1834000

    Reaper query needs an index to perform optimally When the reaper query runs, it does not use an index to find the expired records.

    Advanced Work Assignment

    PRB1867023

    When a blind transfer work item is created from a work item with the consult state as 'transferred', it retains the consult state The agent can accept only work items created in the 'requested' or empty state.
    1. Have Agent 1 get a call.
    2. Agent 1 consult transfers to Agent 2.
    3. Agent 1 leaves the call.
    4. Agent 2 blind transfers to Agent 1.
    5. Agent 1 gets the offer card in their inbox.
    6. Agent 1 accepts.

    Observe that the card stays, and the work item is also always in a 'pending accept' state.

    Agent Chat

    PRB1868186

    The conversation history in the Agent Chat only displays one previous interaction All previous closed conversations should be displayed, but only one past closed conversation is shown.
    1. Enable conversation history in Agent Chat in Conversational Settings > Agent Chat > Conversation History.
    2. Log in as an agent in Safari.
    3. Log in as a non-guest user in any other browser.
    4. Start a conversation and accept as an agent.
    5. Close the conversation.
    6. Have a few more conversations (about 2-3) and close them.
    7. Start a new conversation and scroll up to see the past closed conversations in the chat panel in Agent Chat.

    Expected behavior: All previous closed conversations (up to the amount configured in the conversation history settings) are shown.

    Actual behavior: Only one past closed conversation is shown.

    AI Search

    PRB1783862

    The wrong search events get deleted due to incorrect sort logic The method called via a scheduled job that runs weekly deletes signals that are both older than 90 days and are older than the 500K 'most recent' search events for the search application. When an end user closes a suggestion from their recently searched terms, it causes the suggest_to_user field on the sys_search_event record to be updated, which updates the sys_updated_on field in the process. This can lead to older search events that have suggest_to_user set to 'false' being preserved over newer ones that the user may still wants as a suggestion.

    AI Search

    PRB1824166

    A translated reference field is indexed in a different language in a single record update vs a full table index The peekahead is using the same value for different languages.
    1. Enable i18n Spanish.
    2. Insert a KB record kb_knowledge_base=knowledge, language=es.
    3. Wait for incremental index and dump the document.
    4. Verify that kb_knowledge_base is in Spanish.
    5. Re-index kb_knowledge.
    6. Wait for incremental index and dump the document.

    Expected behavior: kb_knowledge_base is in Spanish.

    Actual behavior: kb_knowledge_base is in English.

    AI Search

    PRB1841579

    sn-search-combobox-desktop's use of createGraphQLEffect dispatches errorActionType due to a 401 error for public GraphQL endpoints The batch request fails due to 401 even though the GraphQL request in the batch was actually successful.
    1. Set the Service Portal home page to public.
    2. Set the Typeahead Search widget to public.
    3. Open Service Portal without logging in.
    4. Open Developer Tools.
    5. Select the search box.

    Observe in Seismic Dev Tools that the batch request failed due to 401. Then, observe in the Network tab that the GraphQL request in the batch was actually successful.

    AI Search

    PRB1847810

    When multi-node ingestion is enabled on a production instance, all cloned instances from production enable multi-node ingestion

    AI Search

    PRB1856064

    Now Assist Panel (NAP) displays the error, 'Trouble processing your request' while promoted skills are working correctly The skill is not being identified and indexing is not occurring automatically on the 'Skill' table. Skill discovery is not happening, and the following error message is displayed: 'We're having trouble processing your request right now due to a temporary system issue. Please try again later.' However, when the 'Skill' table is manually indexed, the skill gets discovered.

    AI Search

    PRB1860558

    The indexing history of a quick test on an admin console does not show a link to the platform indexing history page The start time column should be populated with a link to an indexing history record, but instead it is blank.
    1. Using admin console V5, select any indexed source.

      Notice that it redirects to the indexing history page and the indexing history record is created after a couple seconds.

    2. Reload the list.

    Expected behavior: The start time column is populated with a link to an indexing history record on the platform.

    Actual behavior: The start time column is blank.

    AI Search

    PRB1863971

    RAGRertirval API throws an error if the searchSource given request is not present When the user submits a RAGRetrieval API request with a search source that is not part of the search profile or is not an indexed source, RAGRetrievalAPI fails with no search results.

    AI Search

    PRB1869616

    RAG capability isn't honoring return fields for child tables' columns

    Try invoking Uber Rag or Retriever capability with child tables columns as return fields.

    Expected behavior: Child table fields should be returned in the response.

    Actual behavior: Child table fields aren't returned in the response.

    AI Search

    PRB1870966

    When topics are searched, the term logged to the suggested utterance table is unrelated to the topic
    1. Ensure that utterance readers are mapped to the search application.
    2. Search a topic on the portal.

      It should return a synthesized Genius Results answer.

    3. Check that the utterance is logged to the sys_suggested_utterance table.

    Observe that the logged term is sometimes unrelated to the topic searched.

    AI Search

    PRB1871798

    Dynamic Window (DW) chat returns an error reading 'Sorry there was a problem on my side'
    1. Enable dynamic window on Service Portal
    2. Perform a search in chat.

    Notice an error that reads 'Sorry there was a problem on my side'.

    AI Search

    PRB1878130

    External content ingestion isn't sending a URL to AI Search's (AIS) backend
    1. Ingest a KB document.
    2. Observe AI Search (AIS) backend logs.
    3. Create external content setup.
    4. Start an external content ingestion, like web crawler.
    5. Observe AIS backend logs.

    AI Search

    PRB1878248

    Auto-correct is not turned off in search if it's triggered from the chat Auto-correct should be turned off if a search is triggered from chat, and a rewritten query should be used.
    1. Navigate to search toggle in Full Page Experience.
    2. Search for 'Abel Tuter'.

    Expected behavior: It should search for Abel Tuter.

    Actual behavior: The search term gets auto-corrected to Abel Tuner and does not return any result.

    AI Search

    PRB1881968

    The full page responsive mobile is unusable The pre-search box is cropped on both the left and right sides. The search results search box is cropped and buttons aren't accessible.

    AI Search

    PRB1882714

    A KG response in streaming mode might stream 'kg_information'
    1. Set up NA4S and NAVA on the instance.
    2. Set up KG.
    3. Ensure to enable streaming.
    4. Run a KG people search query.

    Expected behavior: 'kg_information' shouldn't be part of streaming text.

    Actual behavior: Sometimes a streaming answer might contain the string 'kg_information'.

    AI Search

    PRB1882832

    Fix tracer functionality end-to-end for search use cases
    1. Perform end-to-end searches.
    2. Collect search timings.

    AI Search

    PRB1883872

    A FPE new search isn't routed to the 'All' tab
    1. Open an instance with a May Store app.
    2. Ensure that FPE is enabled on the ESC portal.
    3. Search for the query 'what is cookie?'.
    4. Toggle the search.
    5. Perform the search again.
    6. Navigate to the 'Knowledge' tab.
    7. Perform a new search 'Apple iPhone' while still on the 'Knowledge' tab.

    Notice no results are found.

    AI Search

    PRB1883953

    Signals aren't being captured for a carousel on the Portal This isn't working for synthesized Genius results either.
    1. Open an instance with May store apps.
    2. Enable Now Assist QnA and Now Assist Actions on portal.
    3. Perform search for queries 'Apple Iphone' and 'what is spam?'.
    4. Perform actions like 'Open source'.
    5. Trigger the 'Process Queued Signals' job.

    Expected behavior: Signals should be logged in the sys_search_genius _result_event_action_list table.

    Actual behavior: Notice no entries are found in the sys_search_genius _result_event_action table.

    AI Search

    PRB1884610

    When a dynamic window is active, reloads on the search page result in double queries
    1. Navigate to the Portal.
    2. In the main search bar (not the NAVA chatbot), enter 'what is spam?'.

      Portal Genius results return.

    3. Navigate to sys_generative_ai_log and see there is one synthesized QnA response.
    4. Return to the page where the genius result is already displayed and reload the page.
    5. Navigate back to the sys_generative_ai_log and observe there are two synth QnA capability calls. One has a conversationID of DefaultAISearch and one has a NAVA conversation ID.

    Observe that it acts as if the query was simultaneously entered from both the NAVA chatbot interface and also the Portal search bar.

    AI Search

    PRB1887382

    For Global search, subsequent searches can't be submitted There is no issue on Service Portal.

    Analytics Data API

    PRB1847323

    The apply_to object isn't parsing properly for custom data sources in a multivisdata API
    1. Add a custom data source.
    2. Create a resolution API for the data source.
    3. Add a filter to the inline dashboard with the data source and a visualization.
    4. Save the dashboard.
    5. Select something from the filter.
    6. Check the multivisdata API and see the filter configuration where apply_to object has valid values.
    7. Once applied, check the data provider method defined in sys_viz_data_source table.
    8. Check the request body passed to this method.

    See the apply_to object is passed as an empty list.

    Application Install Engine

    PRB1846815

    Admins users aren't able to repair/install all dependencies when domain separation is enabled An issue was observed where a parent application (for example, Now Assist For CSM) was installed successfully, even though its dependencies (for example, sn_genai_platform) weren't installed during the process. This behavior is unexpected.
    1. Open an instance where domain separation is enabled.
    2. Create an admin user in a different domain.
    3. Log in or impersonate as that admin user.
    4. Change the domain to global.
    5. Attempt to install a parent application with known dependencies.

      New dependencies aren't installed and there's domain related errors in plugin logs.

    6. Try to repair the same plugin.
    7. Verify that the new dependencies are still not installed and the domain related errors in plugin logs are still present.

    All the dependencies should be installed with the parent app.

    Appointment Booking

    PRB1874123

    When a slot's end time is 'Arrive by', unavailable slots are displayed as available In the case of 'Arrive by', when an appointment is being booked, it is only considering the exact slot's worth of block while making the booking. However, the slot may have been shown as available since the work could have been done between a time. After post-performance changes, it isn't adding the additional 30 min that is normally done when booking of the slot, which is leading to an available slot not being booked.
    1. In the appointment booking configuration, select the use slot end time as 'Arrive by'.
    2. Configure the slots from 5AM, the appointment window as 2 hours, and the work duration as 1 hour.
    3. Create a work schedule for an agent from 8-5.

    Observe that the first slot 5-7 should be unavailable but it is displayed as available. However, when an appointment is booked for that slot, it throws an error.

    Authentication

    PRB1852202

    A 'Tracking' status isn't changing to self-enrollment in a tracking table
    1. Log in via an admin.
    2. Enable MFA, AA and MFA context.
    3. Navigate to sys_properties.list.
    4. Set SWP to 30 and MAX_SWP to 90.
    5. Log in via a newly created user or any existing user.
    6. Navigate to a profile and configure MFA.
    7. Log in via an admin.
    8. Navigate to the sys_user_mfa_enforcement_info.list.

    Expected behavior: The status should change from 'Tracking' to 'Self Enrolled'.

    Actual behavior: The status doesn't change; it remains as 'Tracking' only.

    Capacity and Reservations Management

    PRB1868046

    Event cells of capacity on a calendar aren't visible in the dark coral theme
    1. Enable the coral theme dark mode.
    2. Open capacity console using a user with elevated privileges.

    The user should be able to view the event cells without any issue.

    Case and Knowledge Management for HR Service Delivery

    PRB1852603

    A record handling agent is unable to fetch HR case details The record handling agent gives an error: 'The system is unable to get the record information'.

    Case Management

    PRB1872131

    Clustering isn't triggered because of a missing ACL
    1. Navigate to All > Now Assist Admin > Feature > Now Assist Skills.
    2. Select suggested steps generation skills.
    3. Select clustering progress.

    Observe that the clustering progress bar is stuck.

    Change Advisory Board (CAB) Workbench

    PRB1870269

    There's a 'Write operation against 'cab_definition' from scope 'sn_sow_chg' has been refused due to the table's cross-scope access policy' alert present after selecting Refresh CAB meetings on SOW CAB Definition Refresh CAB meetings actions in UI16 work without any issues, but returns an error in Service Operations Workspace.
    1. Log in.
    2. Navigate to SOW > Change Advisory Board > All CAB Definitions.
    3. Select New.
    4. Populate mandatory fields.
    5. Create scheduled entries from the 'Related records' tab via the New button.
    6. Populate mandatory fields.
    7. Select Submit.
    8. See a scheduled entry created for the CAB Definition.
    9. Select the 'Details' tab.
    10. Select the Refresh CAB Meetings button.

    Expected behavior: CAB Meetings are generated without errors/alerts.

    Actual behavior: CAB Meetings generated with an error/alert: 'Write operation against 'cab_definition' from scope 'sn_sow_chg' has been refused due to the table's cross-scope access policy" presents on the form'.

    Change Management

    PRB1849821

    The Closed Change Count field under the 'std_change_ record_producer' table is calculated incorrectly when multiple template records are created with the same template name This is caused because the 'group by' query defined in the script 'StdChangeUtilsSNC' inside the function 'calculateVersionAndTemplateStats' is attempting to group by on the name rather than the sysID, which is leading to the incorrect calculation of the closed change count.

    Change Management

    PRB1875123

    CAB AMB controller is too strict on the number of parameters passed on registration, preventing enhancements for Store applications

    Change Management

    PRB1880887

    com.snc.change_management .standard_change_catalog is activated on an upgrade com.snc.change_management. standard_change_flows is installed on an upgrade and zboot. This has a dependency on the standard change catalog. The dependency needs to be removed.

    Code Signing

    PRB1854759

    Frequent updates on the 'sn_kmf_record_signature' table leads to excessive generation of text index events Frequent updates on the 'sn_kmf_record_signature' table leads to excessive generation of text index events, delaying the text index events processing on the instances.

    Code Signing

    PRB1860810

    Generate signatures for sys_auto_script table with wildcard mode Generating signatures for sys_suto_script table records as they are invoking scripts writing to ecc_queue directly from sys_auto_Script.

    Code Signing

    PRB1869613

    Guardrails should respect filter conditions on the signature configuration Don't validate signatures for records not meeting filter conditions, and skip validating records that don't participate in MID.

    Run guardrails.

    Observe that it validates all signatures and doesn't care about filters.

    Column Level Encryption

    PRB1829680

    Key Migration should succeed when EFCs and tables are in different scopes When the EFC scope is different from the configured table scope, the key migration job will not modify it and fail the key migration. Key migration fails due to a scope mismatch between EFCs and the associated table during key migration.

    Condition Builder

    PRB1845617

    A duration editor is modifying the typed in value and adding to it
    1. Navigate to a Yokohama instance.
    2. Navigate to now/SOW/list.
    3. Select the Incidents > All.
    4. Open the PCB by selecting the Filter button.
    5. Select a duration value editor.
    6. Type the '1' character into the Days field.

    Expected behavior: It should set the value to what the user types in, but something is modifying the value before it goes to the editor.

    Actual behavior: The value for days jumps to 3079 with hours red and set to -1.

    Configuration Management Database (CMDB)

    PRB1868449

    Fix family compatibility issues for the 'Reject task' feature The Data Manager (DM) subflow does a double rejection process for rejected tasks in the Workspace, and the business rule 'Clean up TaskToCI records' fails to process records with a NullPointerException.
    1. Create a DM retire policy on the cmdb_ci table where the condition name starts with 'SAP'.
    2. Assign the policy to user.
    3. Publish the policy.
    4. Run the scheduled job 'SNC.CMDBDataManager ScriptableApis. executeRetirePolicies();'.
    5. Notice that a task is assigned to a user.
    6. Impersonate the user.
    7. Open the CMDB Workspace.
    8. Navigate to Open my task > Open the task > Review task > Reject.
    9. Give a reason for the rejection.
    10. Select Save.
    11. Observe that the task is rejected successfully, and CIs should be available for any other DM policy to pick up.
    12. Re-run the same scheduled job 'SNC.CMDBDataManager ScriptableApis. executeRetirePolicies();' or any other policy process.

    Expected behavior: A new task should be created with the CIs with the matching condition because the task was rejected.

    Actual behavior: No task is created, and the cmdb_data_management_task_to_ci table is not clearing the old CIs from the rejected task.

    Content Management System

    PRB1880403

    The CMS portal is corrupted after a Yokohama upgrade It looks like there are CSS files for the portal that aren't loaded as expected.

    Contextual Search

    PRB1872960

    An external link can no longer be accessed in the cxs_new_window UI page

    Core Platform

    PRB1881092

    In an activity stream, be able to determine actions performed by an AI Agent on behalf of a user This is a product update.

    Customer Service Management

    PRB1848565

    Creating a consumer user from a consumer form doesn't work Creating a consumer user from a consumer form creates a sys_user, not a csm_consumer_user. As a result, this user can't be linked with a consumer. This process should work according to documentation.
    1. Open any csm_consumer record.
    2. Navigate to the 'Login Details' related list.
    3. Select the User lookup icon.
    4. Select New to create a user.

    Expected behavior: The user created should be csm_consumer_user, so this user can be linked with csm_consumer.

    Actual behavior: The user created is sys_user, so users can't select this user to link it with csm_consumer.

    Database Persistence - Data Access

    PRB1833232

    There's an incomplete 'NULLS FIRST' clause in multi-column indexes on new (non-migrated) RaptorDB instances This means that all multicolumn indexes can't leverage their natural sort order.

    Database Persistence - Data Access

    PRB1881074

    Provide introspection methods to get back the table/prefixes for views in the Genius Results returned from cypher2Results This is a product update.

    Database Persistence - Data Management

    PRB1861048

    ServiceNow documentation is showing incorrect information for 'Clean peripheral' option ServiceNow documentation needs to be updated regarding the 'Clean peripheral' option in the Table cleaner form. The default behavior when a parent record is deleted is that the system will automatically delete records from sys_attachment table and not 'sys_journal_field' table.

    Database Persistence - Data Scale

    PRB1800841

    JavaScriptMembers, ScriptableObject, and JavaMembers use a HashTable that causes a concurrency bottleneck The 'get' method is synchronized. The HashTable is initialized and then not modified again. It's unnecessary to synchronize all 'get' access.

    Database Persistence - Data Scale

    PRB1846741

    Adjust tuple size thresholds to avoid failing legitimate queries Adjusting the threshold to a reasonable value to avoid failing some queries.
    1. Create a query pattern record with sizable executable SQL.
    2. Attempt to open the query pattern record.

    Database Persistence - Data Scale

    PRB1859279

    StringCache causes a concurrency bottleneck due to SynchronizedLRU Enhance StringCache to support a large volume of concurrent requests.

    Document Intelligence Unified Backend

    PRB1823462

    Document Q&A should skip .docx processing once it started using Document Reader Processing API Install the plugins 'Now Assist in Document Intelligence' and 'Generative AI Controller'.

    Document Intelligence Unified Backend

    PRB1862011

    Brief description shows a space in the 'General Details' page Fields are showing empty when referred, as the display name is empty.

    Document Intelligence Unified Backend

    PRB1862455

    An extraction flow throws an error for a use case that has only a field group Users get an error message: 'java.lang.IllegalArgumentException: objectToMap: invalid type class com.snc.process_ flow.engine.serialization. GRProxy$1 for object property targetRecord (sys_script_include. 4a035a2beb771110 f2549bf12a522841 .script; line 530'.
    1. Create a use case with a target table.
    2. Create fields that have a field group.
    3. Create integration flows and activate them.
    4. Create a record in the target table.
    5. Test the extraction flow with the created document task.

    Document Intelligence Unified Backend

    PRB1881095

    Add 'description' and 'autogenerated_summary' to a task definition This is a product update.

    Document Intelligence Unified Backend

    PRB1881127

    Turn off the embedded image extraction option for PDFs and Docx This is a product update.

    Document Intelligence Unified Backend

    PRB1881128

    Introduce the docx4j library to Glide env This is a product update.

    Document Intelligence Unified Backend

    PRB1881129

    Fix 'fileId' null issue in Glide Java This is a product update.

    Document Intelligence Unified Backend

    PRB1881130

    In 'Chat with doc', classify if it's a scanned or digital PDF This is a product update.

    Document Intelligence Unified Backend

    PRB1881131

    Move Docx4j to 8.3.13 in Xanadu and Yokohama Now Assist This is a product update.

    Document Intelligence Unified Backend

    PRB1881132

    In Docx4j, create a scriptable extractText Java function which takes a docx file and gives an attachment ID for a text file This is a product update.

    Document Intelligence Unified Backend

    PRB1881133

    Support the .txt file format This is a product update.

    Document Intelligence Unified Backend

    PRB1881134

    Adding a PDF > IMG null pointer exception to Xanadu and Yokohama Now Assist This is a product update.

    Document Intelligence Unified Backend

    PRB1881135

    Extract text and embedded images using Docx4j for Docx and pdfbox for PDFs This is a product update.

    Document Intelligence Unified Backend

    PRB1881136

    Glide API to convert a PDF to an image This is a product update.

    Document Intelligence Unified Backend

    PRB1881137

    Fix the Docx path for custom XML so extraction of the text is successful This is a product update.

    Dynamic Translation for Virtual Agent

    PRB1848023

    Messages sent for DT are always using the system language property (glide.sys.language), which could result in LLM messages not being translated This can cause LLM responses to go directly to the user without any translation taking place. This visible text is not hardcoded, but received from the LLM.
    1. Set glide.sys.language to a non-English language.
    2. Ensure the com.glide.cs.llm.default.Language is English/en.
    3. Ensure that native translations is turned off.
    4. Create a custom LLM topic in the same language as the system property language.
    5. Add some LLM components, such as an LLM enabled bot response.
    6. Publish the LLM topic.
    7. Set the session language to the same language as the glide.sys.language property.
    8. Start a Now Assist Virtual Agent (VA) conversation.
    9. Trigger the new topic.
    10. Observe the results.
    11. Set session language to the same language as the LLM default language.
    12. Start a Now Assist VA conversation.
    13. Trigger the new topic.
    14. Observe the results.
    15. Set session language to a language different than the LLM default language and the system property language.
    16. Start a Now Assist VA conversation.
    17. Trigger the new topic.
    18. Observe the results.

    Expected behavior: VA responses are in same language as the session language.

    Actual behavior: VA responses may be mixed, with LLM responses in the default LLM language.

    Dynamic Translation

    PRB1883401

    The synthesized QnA capability fails for P3 languages when invoked from a Portal interface
    1. Navigate to a Service Portal page.
    2. Enter 'Vad är Sverige?' as a Swedish session user in the main portal page.
    3. Navigate to the generative AI log and observe 'Error occurred during Generative AI execution' for a synth QnA invocation.

    In the one_api_service_plan _feature_invocation table, see that the error message is given: 'Cannot invoke 'com.glide.oneapi. entity.FeatureInput. getPayload()" because the return value of 'com.glide.cs.qlue. module.oneextend. proxy.trackers. GenAIExecutionTracker .getOriginalRequest()' is null'. If sn_vad_genai. com.glide.cs.one_ extend.auto_ proxy_enabled is set to false, the E2E works successfully.

    Embedded Help

    PRB1870178

    The embedded help link redirects to the page 'not_allowed.do' instead of the documentation site The issue occurs in the Yokohama release.
    1. Open a Yokohama instance.
    2. Open the 'cmdb_ci_business_app' or 'incident' table.
    3. In top right corner, select the ? or the 'Show help' icon.
    4. Scroll down to link.
    5. Select it.

    Expected behavior: Users are redirected to the documentation.

    Actual behavior: Users are redirected to the 'not_allowed.do' page.

    Employee Center

    PRB1818417

    JavaScript console errors on the Human Resource Management (HRM) ticket page Navigating to the 'to-do' page displays console errors.
    1. Create a HR case - Employee Travel Visa Request for an employee with the manager 'Abel Tuter'.
    2. Let the case move to 'Awaiting approval'.The approval record is created for a user with no roles (in this case - Abel).
    3. Open the approval task in the 'to-do's' page.
    4. Select HR Service, which opens the 'HRM ticket' page.
    5. Select the approval to-do in the 'HRM ticket' page.

    Expected behavior: The user should be able to approve or reject the task from the 'HRM ticket' page.

    Actual behavior: The 'to-do' page has console errors.

    Encryption

    PRB1854358

    Protected script decryption fails for a cloned instance which is upgraded before the 'Protected Script Values Migration' job is executed
    1. Create a Xanadu/Yokohama instance.
    2. Create a 3des data.
    3. Clone the instance into other instance.
    4. After the clone is successful, verify the key exchange won't happen and the cm_glide_protected _script_encrypter module key isn't present in the instance.
    5. Upgrade the instance before the 'Protected Script Values Migration' job runs and keys are created.

    After an upgrade is successful, as the static key is removed and the key isn't available, protected scripts decryptions fail.

    Field Service Quality Management

    PRB1860999

    The dictionary override on wm_task causes issues In a Washington DC instance, there's no override attribute for the State field at the wm_task. In the Xanadu, the override attribute is introduced for the State field at the wm_task.

    Flow Engine

    PRB1751544

    KB1710818

    There's a change in behavior from V1 to V2 in handling JS arrays This change in behavior causes the failure of a following REST step in a Surf action where the 'If_Matches' header who's value is 'etag' isn't matching.

    Refer to the listed KB article for details.

    Flow Engine

    PRB1854665

    Asynchronous business rules are not getting triggered on sys_flow_context when the condition has a 'change' operator The business rule 'Testing flow rules' is not triggered even if it is changed after the update, and does not work on async.

    Flow Engine

    PRB1867133

    The Summarize quick action gives a 'The current execution is in the waiting state' message The flow caller is present. Async HTTP is used and that causes flows to go to a waiting state while waiting for the Async HTTP response. The 'The current execution is in the waiting state' exception is expected when Async HTTP is used.
    1. Log in to the instance as an admin.
    2. Navigate to ESC portal.
    3. Initiate a chat with LA.
    4. Log in as a user and accept the work item.
    5. Trigger the Summarize quick action.

    Expected behavior: The Summarize action should show the chat summary.

    Actual behavior: The quick action gives a 'The current execution is in the waiting state' message and an error in the sys_log.

    Flow Engine

    PRB1874438

    There's a 'LLM Instruction Executor' error on flow execution
    1. Log in as a user with elevated privileges.
    2. Activate the plugin com.glide.utilities.capability_step.
    3. Create an action.
    4. Add a LLM step to the action.
    5. Create a flow.
    6. Add the action created in step 3.
    7. Run the flow.
    8. Users see a 'LLM Instruction Executor' error when running the flow with LLM step.

    Flow Engine

    PRB1874873

    KB2042735

    A deserialization issue occurs to flows when upgrading from Washington DC to Yokohama Flows containing the 'Make a decision' logic don't resume and error out after upgrading to Yokohama. This is caused by a deserialization issue in Flow Engine V2, which prevents the flow from progressing.

    Refer to the listed KB article for details.

    Flow Engine

    PRB1878005

    The success status for the Call Now Assist Skill action step is displayed differently
    1. Log in to an instance.
    2. Enable Now Assist.
    3. Enable flow summarization.
    4. Enable a flow generation skill.
    5. Create a flow.

      The install plugin ID: com.glide.utilities. now_assist_skill_step.

    6. Create an action to the above flow and add the 'Call Now Assist Skill' step.
    7. Run the flow.
    8. Once the flow succeeds, check the status.

    Notice that the success status for the 'Call Now Assist Skill' action step is empty.

    Flow Engine

    PRB1880998

    Unable to deserialize a 'Make a decision' instruction after upgrading to Yokohama When the flow is in a 'Waiting' state in the Washington DC release with Engine_V2, the event opening the flow execution fails after upgrading to the Yokohama release.
    1. Create a flow with a 'Make a decision' flow logic.
    2. Ensure the flow has a waiting scenario so that flow_context goes to a 'Waiting' state.
    3. Execute the flow.
    4. Ensure the flow is in a 'Waiting' state in the Washington DC release with Engine_V2.
    5. Upgrade the instance to Yokohama.
    6. Notice that the event opening the flow execution in Operation view fails, and resuming the flow by satisfying the waiting condition fails.

    Expected behavior: The flow should resume on Yokohama without any errors.

    Actual behavior: The flow errors out in Yokohama.

    Flow Engine

    PRB1881094

    Create the ability to pass the context of the AI agent that triggered the flow This is a product update.

    Flow Generation (Family)

    PRB1881140

    Flow generation Glide changes This is a product update.

    Flows (Family Channel)

    PRB1837338

    Seeing log errors and warnings when creating flows No functional breakdowns were seen, but the errors are consistent when creating flows.
    1. Impersonate a GenAI user with NAC, preferably an admin.
    2. Navigate to Workflow Studio.
    3. Create a flow.
    4. Use a build with Now Assist.
    5. Use any example and select Generate preview.
    6. While the flow is generated, select the 'Network' tab.
    7. Copy the transaction ID from the response payload.
    8. On the other tab, open All > System log > Utilities > Node Log File Browser.
    9. Paste the transaction ID in the Message field.

    Observe the logs.

    Flows (Family Channel)

    PRB1839860

    Trigger Designer role users face insufficient access errors and missing options in the 'Application' list for external triggers When attempting to open the external trigger created using Github Spoke or Jira Spoke as a user with the trigger_designer role, the error message appears, 'Insufficient level of access for eventsource'.
    1. Log in as a user with the admin role.
    2. Create an external trigger using GitHub Spoke or Jira Spoke.
    3. Log out.
    4. Create a new user with the trigger_designer role.
    5. Log in as the user with the trigger_designer role.
    6. Navigate to the Workflow Studio.
    7. Attempt to open the external trigger created.
    8. Observe the error banner that appears displaying the message, 'Insufficient level of access for eventsource'.
    9. Attempt to create a new external trigger.
    10. Select the trigger type as Event on the trigger creation page.
    11. Verify the Application list.

    Notice that the Application list no longer shows any options.

    Flows (Family Channel)

    PRB1865407

    Mismatching data on an action instance can cause issues with flow execution Actions don't perform as expected in various ways.

    Flows (Family Channel)

    PRB1869580

    Subflow creation through image issues Images are partially generated, or are returned with placeholders. Attempt to create a subflow with attached images.

    Notice that they are partially generated or placeholders are returned for whole image.

    Flows (Family Channel)

    PRB1873141

    CopyFlow doesn't set generationSource values as empty
    1. Create a flow with a trigger/action/subflow/flowlogic instances using Flow Generation.
    2. Copy the flow.
    3. View generationSource value on the flow and trigger/action/subflow/flowlogic instances on the copied flow.

    Expected behavior: All generationSource values are blank.

    Actual behavior: All generationSource values in trigger/action/subflow/flowlogic instances are 'text2Flow'.

    Flows (Family Channel)

    PRB1873639

    The icon for 'Call Now Assist Skill' isn't correct
    1. Create an action.
    2. Add a LLM step to the action.
    3. Create a flow.
    4. Add the action created in step 1.
    5. Run the flow.

    Observe a 'LLM Instruction Executor' error.

    Flows (Family Channel)

    PRB1874240

    Update the description of the flow family in Now Assist Features > Creator > Flow in 'Now Assist Features It should read: 'Reduce flow creation time by building a multi-step flow with generative AI. Make editing flows easier by showing users recommendations for the next step of their flow. Learn about a flow by generating a summary.'
    1. Navigate to Now Assist Features > Creator > Flow.
    2. Check the description.

    Flows (Family Channel)

    PRB1874414

    Changing the name of a step to 'Call Now Assist Skill' The plugin isn't automatically enabled. It needs to be enabled by an admin. The name of the step should be changed to 'Call Now Assist Skill'. The name of the plugin should change accordingly to 'ServiceNow Call Now Assist Skill Step Plugin'.

    Flows (Family Channel)

    PRB1874640

    There's a FlowSkeletonGlideRepository ::isAccessibleTable() error if accessed in a background script

    Flows (Family Channel)

    PRB1875963

    There's a missing image attachment and prompt on Get Flow API This issue occurs when using 'Build with Now Assist' in the Workflow Studio.
    1. Log in to a ServiceNow instance.
    2. Impersonate a delegated Gen AI user with NAC.
    3. Navigate to Workflow Studio.
    4. Create a flow.
    5. Use Build with Now Assist.
    6. Upload an image.
    7. Create a flow.
    8. Refresh the page once the flow is created.
    9. Observe the Get Flow API response.

    Expected behavior: The attachment ID and image prompt from the image generation can be seen.

    Actual behavior: There is no image prompt or attachment ID.

    Flows

    PRB1868224

    In Flow Designer, a dynamic choice isn't using a child alias As a result, the first dynamic choice field value doesn't display as selected. This issue is happening in Xanadu and Yokohama versions.
    1. Create a connection alias with a parent and child alias.
    2. Consider the Create issue action, or anything that uses dynamic choices.

    In the flow override aliases, observe that the override to the child alias doesn't happen in the cases of dynamic choices.

    Form Builder

    PRB1828559

    Changing a field type doesn't surface errors on saving
    1. Navigate to the ServiceNow studio home page.
    2. Open a table in Table Builder.
    3. Select a field.
    4. Change the type of the field to reference from string when some data is in a record in that field.

    Expected behavior: An error message telling the user why the save failed.

    Actual behavior: The save silently fails.

    Form Builder

    PRB1836580

    Selecting Edit in Original Scope from the Form Builder banner enables the override properties of the field for editing When a field is inherited from a parent table, it has the ability to override field properties. Previously, override properties were controlled by isFormReadOnly. However, with this PR update, the implementation has been refined how readability is determined.

    Form Controller

    PRB1876936

    GFORM#ADD_DECORATION isn't working in forms workspace like in they are in UI16 as g_form.addDecoration

    Hermes (Family)

    PRB1821659

    The full topic name for longer instance names and/or topic names are truncated in the cluster_topic_name field
    1. Install an application that requires Hermes (StreamConnect as an example).
    2. Navigate to sys_kafka_topic.
    3. Create a topic with a long name value.
    4. The Name field has a 100 character limit.
    5. View the cluster_topic_name field to confirm the truncation.

    Hermes (Family)

    PRB1833479

    Upgrade a fix by removing previous sys_choice records for the 'Application Id' list In Washington DC, a release application ID was originally configured as a list that relied on sys_choice list. In the later releases, this was changed to be a list that relied a new table that defined the configured applications which are associated to its respective Hermes service.
    1. Configure an instance on a Washington DC release.
    2. Note sys_choice list entries for 'Application ID'.
    3. Upgrade the instance to.

    Note that the 'Application ID' list uses the reference table, but the sys_choice_list entries still exist on the instance, but they aren't used.

    Hermes (Family)

    PRB1846241

    The Hermes metrics dashboard displays data off by 1
    1. Open an instance with data in the hermes_usage_metrics table.
    2. Switch the preference timezone to GMT in preference settings.
    3. Compare the data points on the graph vs the table.

    Expected behavior: It shows the same date as the table.

    Actual behavior: The date is one day behind.

    Hermes (Family)

    PRB1847005

    The 'Hermes Metric Aggregation' job is joining dates together due to a job skew The job should be moved further into the day with the GMT timezone to avoid a skew. Also, add limits so that the aggregation is only picking updates of the previous day's start and end and not have it depend on when the job is run. There should be similar limits for clean-up jobs and monthly aggregation.
    1. Have an instance with Hermes installed.
    2. Create a topic and produce and consume it.
    3. Verify usage metrics are logged every 1 hr.
    4. The next day, verify that all hourly metrics are aggregated to the daily metrics of the previous day and that hourly metrics are clear.

    Hermes (Family)

    PRB1856021

    Data in throughput shown on hermes_usage_metrics when producing data in inconsistent patterns Inaccuracies in 'Total MB Value Per Hour' when data flows in an inconsistent pattern. When producing data at a relatively constant rate, hermes_usage_metrics values closely matched the actual value, but when producing data in large spikes, the values in hermes_usage_metrics are significantly above or below the actual value.
    1. Set up Kafka producer on a local machine using Kafka Java APIs.
    2. Produce a large batch of messages.
    3. Wait until the hourly aggregation job runs to populate hermes_usage_metrics.
    4. Check the corresponding value in 'Total MB Value Per Hour'.

    Hermes (Family)

    PRB1876312

    There's a missing call to refresh clients when clusters are removed When clusters are removed during a 'Retire' request, the deletion should also refresh the clients. A recently added function was missing that refresh call.

    Horizon Component Library

    PRB1862345

    A component needs to be deregistered on disconnect when using now-trigger-library's deregisterComponent Not deregistering a component results in detached DOM nodes/memory leaks, as the registry object still retains a reference to the element even should it be removed from the DOM.
    1. Navigate to a page in a workspace that has the element in question on it (now-alert-list).
    2. Use the console to check the componentRegistry that exists in window.uxfTriggerLibrary.
    3. Navigate away from the tab in question or perform any action that removes the element from the DOM.

    Expected behavior: As the element no longer exists in the DOM, its entry in the componentRegistry should no longer be there.

    Actual behavior: The entry for the element in the componentRegistry persists, as it was not deregistered on COMPONENT_DISCONNECT.

    Horizon Component Library

    PRB1867018

    Update Now Assist Panel (NAP) customization in the 'Tools and Admin' theme NAP choice picker background colors don't allow for good color contrast.

    Horizon Component Library

    PRB1881075

    HDS net new components: Animated carousel and Animated button This is a product update.

    HR Service Delivery

    PRB1831301

    Renaming an attachment isn't working for an agent
    1. Provision an instance with sn_hr_agent_ws 3.3.1/4.0.0-SNAPSHOT installed.
    2. Log in as an agent.
    3. Open an HR case in HR Agent Workspace.
    4. Attach a document in the attachments side panel.
    5. Try to rename it.

    Observe that the attachment isn't renamed.

    HR Service Delivery

    PRB1831691

    KB2043348

    HTML dynamic parameters on HR templates don't work as expected There's an issue in Xanadu and Yokohama where HTML dynamic parameters aren't replaced with their actual values when used in links. The issue arises in the HTML-sanitizer enabled HTML and translated HTML fields during template replacement with HTML data.

    Refer to the listed KB article for details.

    HR Service Delivery

    PRB1858070

    There's an email client template visibility issue in HR Agent Workspace A global ACL isn't working. sys_email_client_template extends sys_metadata (application file) so if the sys_email_client_template record is created in ER scope, it runs the ACL in ER scope. If there's no ER scope ACL, it can inherit the global ACL by ACL inheritance, otherwise, access could be blocked. If there's no ER scope ACL on sys_email_client_template, it might need to be added.

    Incident Communications Management

    PRB1849039

    Update the email client template from the mail script

    Incident Communications Management

    PRB1849469

    Filter Major Incident Management (MIM) email templates in MIM Comm based on a template condition in Classic MIM templates follow a reference qualifier and don't respect conditions set within the templates.
    1. Open major incident in Service Operations Workspace.
    2. Select the 'Communicate' tab.
    3. Select the Compose button on a communication plan.

    Expected behavior: A user can all see all templates

    Actual behavior: Based on a condition within the templates, filtering should be applied. It currently follows the reference qualifier.

    Incident Communications Management

    PRB1860756

    Resolving variables isn't working in UI16
    1. Install com.snc.incident.mim and sn_itsm_gen_ai plugins.
    2. Open any major incident communication plan from UI16.
    3. Select Compose.

    Expected behavior: The email should display as per the variables defined under the email client template.

    Actual behavior: No email is displayed in UI16.

    Integration Hub

    PRB1852454

    com.service_now.mid. connections.jdbc. JDBCConnection splits a property on equal signs The '==' value is removed when the code saves to the property, which caused a save to an incorrect value string.
    1. Set up a JDBC data source.
    2. Add the string 'dGVzdCBzdHJpbmchIQ==' as a property.

    Notice that when it's processed, the '==' are stripped.

    Integration Hub

    PRB1878255

    Suggested actions are erroring out on the Yokohama Now Assist table This issue appears to have been caused by an underlying script, including call adding to the scope stack, which exposed a problem in how the flow engine cleans up the scope stack after a flow execution.
    1. Create a flow that adds to the scope stack but does not clean it up.
    2. Execute the flow in the background.

    Expected behavior: Flow executes without exception.

    Actual behavior: An error is thrown when the flow completes.

    Integration Hub Remote Process Sync

    PRB1874002

    Remote Process Sync (RPS) is triggered by changes from non-captured fields RPS captures changes to a record when any changes are made to the fields selected by the user as part of the CDC Definition (change data capture). Even though changes are being made to fields that are not part of the CDC Definition, RPS still captures these changes.
    1. Set up RPS.
    2. Create a record that syncs.
    3. Modify a field on that record that isn't part of the CDC definition.

    Observe that this creates an update record on cdc_queue_ih, which should not be created.

    Key Management Framework (KMF)

    PRB1854256

    Legacy SEK rekey fails due to the password2 field size not being large enough Instance rekey use sys_dictionary.max_length for field size detection, leading to rekey failure.
    1. Provision an instance.
    2. Populate the oauth_credential. token_received with a large value.
    3. Clone the instance from step 1.

    Expected behavior: Legacy SEK rekey completes successfully.

    Actual behavior: Legacy SEK complains that the field isn't large enough.

    Lifecycle Events

    PRB1848546

    In Yokohama, when the property use_flow is false, creating a LE case using the 'LE Testing' functionality causes the case to be reopened This doesn't happen on Yokohama when the 'sn_hr_le.use_flow' property is set to true, and it doesn't happen on Xanadu.
    1. On a Yokohama instance, as a user with elevated privileges, activate the 'HR Xanadu Deprecated Workflow' [com.sn_hr_xanadu_deprecated_wf] plugin.
    2. Set the 'sn_hr_le.use_flow' property to false.
    3. Open any existing LE in the LE Builder.
    4. Select the Test functionality.
    5. Create a case.

    Actual behavior: In the case worknotes, there's a message 'Case reopened by...'. Review the case history and the user can see that the Active field is updated twice; once from empty to false, and then from empty to true.

    Expected behavior: The Active field should only be updated once, from empty to true, and therefore the worknote 'Case reopened by...' shouldn't be logged.

    List Administration

    PRB1862344

    Component needs to be deregistered on disconnect using now-trigger-library's deregisterComponent This deregistering issue occurs for the component now-record-list-header.
    1. Navigate to a page in a workspace that has the element now-record-list-header.
    2. Use the console to check the componentRegistry that exists in window.uxfTriggerLibrary.
    3. Find an entry for the element.
    4. Navigate away from the page.

    Expected behavior: As the element no longer exists in the DOM, its entry in the componentRegistry should no longer be there.

    Actual behavior: The entry for the element in the componentRegistry persists, as it was not deregistered on COMPONENT_DISCONNECT.

    List Filters

    PRB1751177

    Inconsistency in 'Presentational 'list filters There's an inconsistency between the data displayed and the filter applied for empty values. In one scenario, there's inconsistency when applying filters on the string column and the number column. If the filter text value is empty, the Apply button is turned off for the string column, but still enabled for the number column. Selecting Apply still retains the older filter. In another scenario, there's also an inconsistency between the data shown and the filter applied for empty values. Fields, which are empty in a table, are displayed as 0 in the presentational list based on the data type. When they are filtered, they should be considered empty or should be considered as 0 if it is shown as zero.

    MID Server

    PRB1840424

    KB1772428

    MID Server ECCSender thread gives an 'Invalid byte 2 of 4-byte UTF-8 sequence' error, blocking sending of valid ecc_queue inputs to an instance This is most likely to be seen in LDAP user imports, or Import Set JDBC Data Sources, where large data is involved. Large data is likely to include extended unicode characters such as emojis. For LDAP probes, emojis have been seen in OU/CN group names and user details data. For REST integrations synching incidents, emojis have been seen in comments of cases. They could potentially appear anywhere, for any feature's probe result data.

    Refer to the listed KB article for details.

    Mobile Platform

    PRB1886492

    (MAB) card generation with Now Assist is failing on Yokohama Now Assist based card generation is currently failing on the Y track due to a missing required column. The 'Creation Method' column, needed for handling records created by Now Assist is missing, resulting in insertion failures during the card generation process. Now Assist-based card generation doesn't work for users.
    1. Provision a Yokohama instance with Now Assist for Creator installed.
    2. Enable mobile card generation skill.
    3. Open the web to mobile flow for a record screen.
    4. Select the Generate with Now Assist option.

    Observe an error message: 'unable to insert record into table sys_sg_view_config - setValue called for unknown field creation_method in table 'sys_sg_view_config''.

    Mobile Virtual Agent

    PRB1881114

    Mobile NASS GenAI visual uplift This is a product update.

    Multi-factor Authentication (MFA)

    PRB1862612

    The 'More information' link on MFA registration has the old documentation link structure There's an incorrect documentation link structure present in the page 'multi_factor_auth_setup_page'.
    1. Log in to an instance that has MFA setup but the user isn't enrolled into MFA yet.
    2. After logging in during MFA enrollment, users would see the 'More information' hyperlink.
    3. Verify that on selecting the link, it redirects to documentation home page.

    Multi-Instance Framework - Core

    PRB1849745

    Turn off collect_metadata and recurring_diagnostics in sn_mif record on hermes_app_services The collect_metadata and recurring_diagnostics flags aren't set in the hermes_app_services record. By default in the table dictionary, those flags are true. That means that the instance automatically connects to the Hermes configured in that record, creates metadata and diagnostics topics, and starts sending traffic. For some Hermes running, this data isn't needed or used and observations show that this activity puts significant load on the cluster.
    1. Check the sn_mif record in hermes_app_services.
    2. Note that both collect cluster metadata and recurring diagnostics are checked.

    Next Experience Unified Navigation

    PRB1862338

    A component must be deregistered on disconnect using now-trigger-library's deregisterComponent

    Now Assist Panel

    PRB1883320

    KB search sometimes isn't working after upgrading from Xanadu to Yokohama
    1. Set up a Xanadu instance and ensure that the KB search is working.
    2. Upgrade to Yokohama.
    3. Select NAP.
    4. Enter: 'what is spam'.

    Expected behavior: The KB should display a result.

    Actual behavior: A 'No answer found' message is displayed.

    Now Assist Panel

    PRB1885517

    A user isn't able to add an attachment in playground and Now Assist Panel (NAP) The icon to upload is missing.
    1. Create any agent with the catalog tool.
    2. Select the option to add an attachment after providing required details.

    Observe that the user is asked to add an attachment, but there's no icon to upload.

    Performance Analytics

    PRB1857024

    On some nodes, fields become read-only on after an upgrade This issue is intermittent and depends on the load order. It can deny access with an error message.

    Predictive Intelligence

    PRB1812468

    Classification training capability doesn't return consistent results when doing batch predictions Whenever users train a classification solution on Java, whenever they do single record predict, it by default returns 10 outputs per record and is filtered on the Glide side to return the one with highest confidence, which is more than the threshold. But when doing a batch predict, it returns only one output per record. If the confidence for that output is less than the threshold, then it doesn't return any output.
    1. Train a classification solution on Java.
    2. Select a record such that users get an inconsistent result for a single and a batch test.

    Predictive Intelligence

    PRB1858849

    An empty row_count field in the ml_solution table leads to an error message being logged while parsing an integer from empty string from the ml_solution row_count attribute When the row_count field is empty while training a solution, the following error message is logged: 'Exception caught while building data Description list: For input string: "": no thrown error'. This occurs only for Java-based solutions.

    Procurement

    PRB1859220

    Update 'Procurement Process Flow - Auto allocation enabled' workflow to skip auto allocation when Now Assist for Software Asset Management (SAM) is active Agentic solution for true-up in Yokohama.

    Project Management

    PRB1850637

    Duplicate resource_aggregate_weekly and resource_aggregate_monthly records are created intermittently Code in the UpdateActualsFromTimeCard function of the ResourceActuals SI raises an event to update aggregates for each day when a time card is approved or recalled, causing duplicate resource_aggregate_weekly and resource_aggregate_monthly records to be created intermittently.

    Reporting

    PRB1853349

    Visualizations in non-global domains aren't displayed in the list of visualizations When trying to add some dashboard to a library, it's not saved in any sub domains. It's getting saved only in the global domain. In the sub domains, when they try to save the dashboard, it says it's saved but it's not.

    Schedule Optimization

    PRB1860079

    Performance impact due to the 'Optimization' business rule A 'Optimization Metadata processor' business rule is made in sys_attachment async.

    Schedule Optimization

    PRB1875270

    Prevent the fallback of retrieving the schedule from the sys_user record when there's no schedule in the 'Agent schedule' table Remove the 'Agent schedule' but add a schedule to the sys_user record. The agent shouldn't be returned in the getQualifierData response. Add the 'Agent schedule' only, it should be returned in the getQualifierData response.
    1. Enable a territory model.
    2. For the qualifying agent, verify that no schedule is present in cmn_schedule.
    3. Add a schedule at the agent level in the sys_user record.
    4. Create SO batch with the 'Territory' qualifier.
    5. Perform Graph QL for agents.

    Expected behavior: No schedule should be returned.

    Actual behavior: A schedule from the sys_user record is returned.

    Security Incident Response integration with Splunk Enterprise Event Ingestion

    PRB1865568

    CMDB_CI mapping fails on the Configuration Item field

    Server-side scripts

    PRB1848232

    TD instance gets hung up and unresponsive due to recursive calls while capturing Generic Collection Framework (GCF) metics While publishing the GCF metrics for scripts, MInMaxPriorityQueue is used for collecting 20 longest executions. It attempts to access indices more than 20, causing ArrayIndexOutofBound exceptions.

    Server-side scripts

    PRB1864630

    A fetch process hangs in an infinite loop if the header is returned as a list This is a product update.

    Service Catalog

    PRB1835397

    Users are unable to clear a value on a date type question from a client script/UI Policy

    Service Level Management

    PRB1843470

    KB1777517

    SLA engine reset condition may fail in certain circumstances SLA Engine reset condition fails if the advanced conditions changes, changesFrom or changesTo are used in conjunction with the table in com.snc.sla.get_ task_from_db and the walker is CHECKPOINT.
    1. Create an SLA Definition on the table sc_req_item:
      1. start condition active=true
      2. stop condition active=false
      3. reset condition due_date changes (advanced condition)
    2. Check that the property com.snc.sla.get_task_from_db contains sc_req_item.
    3. Check that property com.snc.walker.default=CHECKPOINT.
    4. Update any value on an existing sc_req_item or create one and verify task_sla is created for SLA Definition created in #1.
    5. Update the due_date field to anything if it is empty or change it to any value if it has a value.

    Expected behavior: task_sla is canceled and new task_sla is created.

    Actual behavior: task_sla isn't canceled. The error log contains 'CheckpointHistoryWalker does not support walk to update: #'.

    Service Level Management

    PRB1849627

    Improve SLA Calculator's handling of very large sets of Task SLA records during trigger processing Under certain circumstances, bulk SLA calculations can cause node memory issues if the set of Task SLAs being calculated is very large.
    1. Open the script include SLACalculatorNG.
    2. Look at the function SLACalculatorNG. calculateSLArange.

    Note that it first processes the list of Task SLA and adds them to an array. This is consumes memory unnecessarily.

    ServiceNow IDE (Family Release)

    PRB1878166

    Integrated development environment (IDE) source files should not be deleted using a business rule on sys_app table This business rule is triggered on every delete action for a sys_app. All associated IDE source files could potentially be deleted unintentionally from the instance, which is irreversible.
    1. Create a ServiceNow instance.
    2. Install ServiceNow IDE 2.0.4.
    3. Bump IDE to 2.1.0.
    4. Create an app in Service Now IDE.
    5. Build and deploy the newly created app.
    6. Observe all the source files being shown in IDE.
    7. Navigate to the sys_app record.
    8. Delete the sys_app record.

    Expected: ServiceNow IDE should not show any source files when app is deleted.

    Actual: A purge occurs due to a business rule.

    Service Portal

    PRB1839297

    KB1925168

    SPEntry page error 'Execute operation on script include 'SPEntryPage'' from scope occurs Switching the scope and refreshing the page gives the error, 'Execute operation on script include 'SPEntryPage' from scope 'Knowledge Management - Service Portal' was denied'.
    1. Log in to a Xanadu instance.
    2. Create the glide.entry.first.page.script system property.
    3. Give it the value: new SPEntryPage().getFirstPageURL().
    4. Make 'Knowledge Management - Service Portal' as the current scope.
    5. Refresh the browser page.
    6. Notice the error, 'Execute operation on script include 'SPEntryPage' from scope 'Knowledge Management - Service Portal' was denied. The application 'Knowledge Management - Service Portal' must declare a cross scope access privilege. Please contact the application author to update their privilege requests.'
    7. Repeat step 2 and step 3 with the scopes 'CI Landing Experience' and 'Conversational Interfaces - Diagnostics'.

    Notice that the same error message occurs.

    Sidebar (Family Release)

    PRB1849937

    When sn_hr_core.impersonateCheck is set to 'true', users are unable to add users to the sidebar When sn_hr_core.impersonateCheck is set to 'true', users can't add users to the sidebar discussion for sn_hr_core_case_payroll records. There's instead an error message: 'Participants can't join the discussion because they don't have access to this record.'
    1. Install the sn_hr_core plugin with demo data.
    2. Set the system property sn_hr_core.impersonateCheck to true.
    3. Navigate to any sn_hr_core_case_total_rewards record.
    4. Select the Discuss UI action.
    5. Select Add Participants.
    6. Select any user (such as Abel Tuter) with access to the parent record.

    Notice the error message appears.

    Software Asset Management

    PRB1878273

    Databroker changes to call prompt This is a product update.

    Software Asset Normalization

    PRB1840383

    For a software product type other than the licensable normalization engine, it keeps the version empty and marks it as normalized using only the product map For a software product type other than the licensable normalization engine, it doesn't pick a pattern rule that has a potential to stamp the version, and rather keeps the version empty and marks it as normalized using only the product map. When a discovery model undergoes normalization, in cases for software product types other than licensable, the normalization engine doesn't select a pattern rule that could assign a version. Instead, it leaves the version field empty and designates it as normalized using only the product map.
    1. Have a DM for a non licensable product. It should be product normalized.
    2. Have a pattern rule for the particular product.
    3. Run rormalization.

    Expected behavior: The product becomes pattern normalized, and the version is stamped in.

    Actual behavior: The product remains product normalized with no version stamped.

    Syntax Editor

    PRB1533082

    The fonts defined on the system property 'glide.ui.html.editor .v4.font.collection' aren't displayed on the font list of the HTML editor on Agent Workspace The list should show only the fonts defined in the value of system property 'glide.ui.html.editor .v4.font.collection, but it shows all the fonts mentioned on the system property 'Description'. On the native UI HTML editor, it's working as expected.

    Table Administration and Data Management

    PRB1852726

    KB2068808

    Users get an error: 'Syntax Error or Access Rule Violation detected by database ((conn=316832) Unknown column 'ba.a_ref_10' in 'on clause')'

    Refer to the listed KB article for details.

    Third-party Software

    PRB1793365

    Commons-beanutils hits BEANUTILS-509, causing severe default thread contention, leading to 429s In a few cases, this caused a node to completely hang indefinitely because all the threads were attempting to enter a synchronized block of a WeakHashMap that was being hogged by a thread stuck in an infinite loop.

    Transaction Management

    PRB1866307

    Sanitizing URLs in GlideElementURL For performance, the sanitized URL can be cached rather than computing the same value repeatedly.

    UI Builder (Family Channel)

    PRB1874490

    PROXY_DATA_UPDATED is undefined This is a product update.

    UI Form Administration

    PRB1843602

    There's an issue with the date picker position The date picker position doesn't change when scrolling through the page.
    1. Open a form with a Date field.
    2. Select the date picker.
    3. Scroll the page.

    Notice that the date picker remains in the same position on the screen even when scrolling.

    UI Form Administration

    PRB1847059

    Formatting on the special handling notes header is aligned to the left in Xanadu Special handling notes aren't aligned properly and are formatted completely to the left.

    UI Form Administration

    PRB1859837

    UX view rule configuration is not applied in the Workspace when a record is opened from a reference field UX view rule configurations are not applied when a record is opened in a Workspace from a reference field on a form, but when opening the same record from a related list in a Workspace, the UX view rule configuration is applied.
    1. Open a cmdb_ci_computer table.
    2. Create a custom view.
    3. Create a workspace view rule on the sysrule_view_workspace table with the values:
      • Table: cmdb_ci_computer
      • View: Use the Custom View created
      • Experience Restricted: Uncheck
      • Workspace: Empty
      • Execution Order: 1
    4. Create a UX view rule configuration on the sys_ux_view_rules_configuration table with the value: Active: True
    5. Navigate to Workspace View Rules > Related list.
    6. Add the Workspace view rule created.
    7. Open the Service Operations Workspace.
    8. Create new incident where the Configuration Item field references a cmdb_ci_computer record.
    9. Save the new incident.
    10. Navigate to Incident > Details > Open Record in the Configuration Item field.
    11. Select Open Record on the modal that appears.

    Observe that the Custom View created in is not applied.

    UI Form Administration

    PRB1868782

    While composing an email in Vendor Operations, the cursor occasionally jumps, making it difficult to continue typing When the user types rapidly, the position event is triggered multiple times, resulting in the cursor unexpectedly jumping to the end.
    1. On a base instance, enable email reply recommendations on any table.
    2. Add a line to the end of the email's last line.
    3. Start typing in the first line quickly.

    Observe that the cursor jumps to the end of the file.

    UXF Components

    PRB1831642

    Contents of UI Builder (UIB) components can't inherit the height of the container component it is placed into Inside of the macroponent element used to render UIB components, two divs wrap internal contents - a plain div, and a div with display: contents; set on it. As a result child elements of the UIB component are unable to inherit the height of the container it is placed into.
    1. Create a component in UI Builder.
    2. Set the height and min-height to 100% on the component element.
    3. Add the resizable panes component inside.
    4. Populate it with any content in left and right panes.
    5. Set the height and min-height of a resizable component to 100%.
    6. Save the UIB component.
    7. In an experience built from Workspace App Shell, create a page.
    8. Add the previously created component to the page.
    9. Set the height and min-height of the body to 100%.
    10. Set the height and min-height of component to 100%.

    Expected behavior: It is possible to set up a component built in UIB to take the full height of the container or page it is placed into.

    Actual behavior: The component doesn't stretch to the full height of the page.

    UX Framework

    PRB1853498

    Backslashes in UX Library's asset's source map URL causes an empty screen It is possible to create custom components that have a source map URL path with backslashes. In the bundling logic, it looks for each asset source map URL line and remove it. That logic isn't full matching the strings that include backslashes and it leaves a partial string in the JavaScript content. This partial string is what's causing the empty screen, since the JavaScript content is malformed.

    Modify the source map URL path with blackslashes instead of forward slashes in any UX library asset that is included in a core bundle.

    Expected behavior: The instance should display pages.

    Actual behavior: Empty screens are displayed.

    UX Framework

    PRB1860878

    sn-canvas-badge-connected fetches the count and then subscribes to the channel provided as part of a UX page property Updates in between aren't captured by the badge.

    UX Framework

    PRB1869757

    There's a performance degradation when loading the home page on Xanadu When the system property glide.ux.user_ criteria_enabled=true, there's a performance degradation between Washington DC and Xanadu when loading the home page.
    1. On a Washington DC instance, set glide.ux.user_ criteria_enabled to true.
    2. Navigate to the instance home page.
    3. Measure the page load times.

      Notice that the average page load time is less than five seconds.

    4. On an Xanadu instance, set glide.ux.user_ criteria_enabled to true.
    5. Navigate to the instance home page.
    6. Measure the page load times.

      Notice that the average page load time is between eight and ten seconds.

    Observe the performance degradation between the two versions when the system property glide.ux.user_ criteria_enabled is set to true.

    Virtual Agent Actionable Notifications

    PRB1870525

    Actions are not shown to the user if notification actions are passed in event parameters Actions should be shown to the user, with the actions mapped to the content statically followed by the actions from the event parameters.
    1. Set up batch notification with the value of system property com.glide.cs. notification_batch _user_size less than the number of recipients on notification.
    2. Setup an actionable notification with an event trigger.
    3. Add some actions, which may or may not be linked to the content.
    4. Select the checkbox to pass in actions as event parameters in Link actions to content record.
    5. Trigger the notification by passing in some actions as event parameters (gs.eventQueue(, , )) assuming the event parameters contain action sys IDs.

    Expected behavior: Actions should be shown, with the order being first the ones mapped to the content statically, followed by actions specified in the event parameters.

    Actual behavior: Actions are not shown along with the notification.

    Virtual Agent Actionable Notifications

    PRB1881238

    Non-actionable notifications aren't triggered It isn't working with new or existing users. There's errors in the notification logs.
    1. Have 1 non-actionable notification setup or 1 non-actionable and 1 actionable.
    2. Create an incident for a user.
    3. Open Virtual Agent.

    Observe there's no notification.

    Virtual Agent Designer Legacy

    PRB1880139

    L10n warning in Now Assist in Yokohama Localization warning in Now Assist in Yokohama.

    Virtual Agent

    PRB1847804

    The script tool in skill-kit results is null and doesn't evaluate the script A skill that calls a script tool to filter RAG results based on custom filtering does not process the script, causing the script to return null.

    Virtual Agent

    PRB1853608

    Now Assist Panel (NAP) isn't loading after 10+ cases notification have been triggered

    Virtual Agent

    PRB1854580

    In the AutoEval Proxy2 skill, the grounded prompt is inserted incorrectly The context is put into the hypothesis although the metric is mapped correctly.

    Virtual Agent

    PRB1854684

    Metric results override batch results when the batch run's run type is 'Auto Prompt Tuning' Metric results aren't created in sys_one_extend_eval_metric_result and override the batch results (sys_one_extend_batch_result).
    1. Create a skill.
    2. Trigger a prompt optimization run.

    Notice that the optimized evaluation run triggered by prompt assist (of runType: Auto Prompt Tuning) results are incorrect. Metric results aren't created in sys_one_extend_ eval_metric_result and override the batch results.

    Virtual Agent

    PRB1859774

    Create sys_auto_flush records to auto delete one_api_variable_store records one_api_variable_store is a run time record and that does not need to continue indefinitely. one_api_variable_store records with sys_created_on with a specific timestamp need to be auto deleted by the instance.

    Virtual Agent

    PRB1862370

    Skill executions are stuck when Now Assist guardian settings (prompt injection, offensiveness) are enabled Skill test runs aren't working from the skill kit, whereas executions outside of the skill kit (for example: OneExtend.execute() in async mode) are working fine for the same settings. This is due an unsupported operation exception coming from EvaluationServiceImpl.java.
    1. Create a skill from the Now Assist Skill Kit.
    2. Enable prompt injection from the Now Assist Guardian settings (use option as 'Block').
    3. Select Run Test on that skill.

    Expected behavior: Run Test should work irrespective of the guardian setting because users can execute the same One Extend capability outside of the skill kit.

    Actual behavior: Run Test is stuck due to an exception (Error Execution of evaluation request failed for batch result: fbb3afc4c348 e2149e83d62 f0501312b due to error: null).

    Virtual Agent

    PRB1863302

    Eval metric result records are not created when metric execution is skipped because of an applicability script condition Eval metric result record should be created with status of 'Skipped'.
    1. Add a condition to a skill in an applicability script for any metric in an extended resource mapping record.
    2. Trigger an evaluation run by selecting the above changed metric.

    In the 'Eval metric result' table, notice that the records for the above metric is not created.

    Virtual Agent

    PRB1864394

    When a user selects the None of these option on the 'Submit Request (Template)' topic in Virtual Agent, it generates empty catalog item page When the user selects the option None of these and selects the link, they are directed to an empty catalog item page. The user should not be asked to submit a request after selecting the option None of these.
    1. Hop into a base Xanadu instance.
    2. Ensure the plugin 'Customer Service Virtual Agent Conversations' is installed on the instance.
    3. Ensure the topic 'Submit Request (Template)' is published and visible in Virtual Agent.
    4. Navigate to the Customer Service Management (CSM) portal.
    5. Select Show me everything > Submit Request (Template).
    6. Enter Create case.
    7. Select None of these.
    8. Select Yes for the query 'Do you want to proceed with?'.
    9. Select the link.

    Expected behavior: The user selects the link and they are either directed to the correct catalog item page, a case is created, or for 'None of these' options they are not asked to submit a request.

    Actual behavior: The user is able to submit the 'None of these' option, and it gives an empty link for the catalog item.

    Virtual Agent

    PRB1865096

    There's a NotFoundExecption when a sys_cs_session record isn't available for a user In the error log: 'Unable to find session in db. Script execution error: Script Identifier: null.null.script, Error Description: com.glide.rest.util. NotFoundException, Script ES Level: 0. Couldn't decipher the stack trace resulting from the following JavaScriptException: com.glide.rest.util. NotFoundException: org.mozilla.javascript. JavaScriptException: com.glide.rest.util .NotFoundException: org.mozilla.javascript. Context.makeJavaScript Exception(Context.java:2626)...'

    Virtual Agent

    PRB1866728

    The user is not able to asynchronously process a dataset and trigger an eval run A sys trigger can't process a dataset and an eval run at the same time.

    Virtual Agent

    PRB1867769

    Users can't activate a trigger for a Virtual Agent channel and profile Instead of calling getChannelByName and hardcoding the channel, it should be grabbed by the channel ID. There's also a missing context profile.

    Virtual Agent

    PRB1868677

    Added a fix script for auto_prompt_tuning internal in sys_one_extend_batch_run This is a product update.

    Virtual Agent

    PRB1869860

    System property 'proxy=false' with a carousel output does not return a Genius Response Genius Response is stuck on 'Looking into your request' after updating the 'web client' output type to '= carousel' and updating the value of sn_vad_genai. com.glide.cs. one_extend. auto_proxy_enabled to 'false'.
    1. Navigate to now_assist_va_ search_results_output_type table.
    2. Update the 'web client' output type to '= carousel'.
    3. Navigate to sys_properties table.
    4. Update the value of sn_vad_genai. com.glide.cs.one_ extend.auto_ proxy_enabled to 'false'.
    5. Start a new conversation with Now Assist Virtual Agent on an /ESC portal.
    6. Enter 'What is a cookie?'.

    Expected behavior: The response appears as a carousel.

    Actual behavior: The response is stuck on 'Looking into your request'.

    Virtual Agent

    PRB1871270

    Additional read of conversation in worker.lockAndProcess should be moved A new call to retrieve the conversation and read values form the context before every lock and process operation in Worker for the AI Agents. This adds unnecessary overhead to any transaction that is not LLM or Agent AI related.

    Virtual Agent

    PRB1872922

    Domain separation issue when running AI Agents When retrieving a record from a separate domain using an AI agent, no is data returned.

    Virtual Agent

    PRB1872958

    Virtual Agent search result showing issues after upgrade Virtual Agent results returned were misaligned after upgrading to Yokohama.
    1. Log in to an Employee Service Center (ESC) portal.
    2. Impersonate a user.
    3. Initiate a Virtual Agent chat.
    4. Enter bereavement policy.

    Notice that the search results are shown in misaligned order.

    Virtual Agent

    PRB1874300

    A grammatically incorrect message is thrown by the API Part of an error message from Now Assist skill capability is grammatically incorrect: 'Failed to execute the feature due to an error while executing it's depends on feature'.

    Virtual Agent

    PRB1876378

    A KB article is not returned in NAVA during a Yokohama Now Assist upgrade When a plugin stage user types 'what is spam' during an upgrade, the conversation gets stuck.

    Virtual Agent

    PRB1877176

    Enabling KG isn't making unified slot fill calls This causes slot filling to not work.
    1. Ensure KG is enabled at the instance level as well as for a conversation catalog.
    2. Impersonate a user.
    3. Request the 'Employee Details' catalog item.

    Expected behavior: Slot fill should happen for most of the questions, and the unified slot fill call happens.

    Actual behavior: Slot fill doesn't happen for questions and the unified slot fill call didn't happen.

    Virtual Agent

    PRB1877432

    Dynamic loading message control (on Virtual Agent chat client) should contain non-agentic 'processing messages' Non-agentic messages should appear in the chat client response.
    1. Set up Agentic.
    2. Perform a search, such as 'What is spam?'.

    Expected behavior: In the VA chat client, the 'View AI Agent Processing Steps' collapsible control should contain all update messages.

    Actual behavior: Only the first two messages ('Figuring out next steps', 'Executing agent Search QnA Agent') that were initiated from the Agentic frameworks are included inside the 'View AI Agent Processing Steps' collapsible control on the VA chat client. The next two status update messages ('Finding the needed info', 'Reviewing the details') appear as standalone messages.

    Virtual Agent

    PRB1878226

    When running AI Agent Studio on a non-English language instance, the stacked message renders until the conversation moves into the error topic During the updateDynamicProgressMessage scriptable path, one of the progressMessages in the DynamicLoaderRichControl has a message of 'null'. This causes an exception.
    1. Provision and instance with a language plugin installed and enable DT for that language.
    2. Set up NowAssist and AI Agents.
    3. Ensure that the DT optimizer is on and the sys_cs_dynamic_translation_text table is cleared or that the optimizer is off.
    4. Change the user's language to a language other than English.
    5. Start a conversation in Agent Studio.

    Expected behavior: The stacked messages should all be translated/localized.

    Actual behavior: Conversation ends abruptly.

    Virtual Agent

    PRB1880022

    The skill applicability URL condition isn't working The skill applicability is broken and doesn't apply correctly. The root cause is confirmed to be a URL query failing.
    1. Navigate to CMDB Workspace.
    2. Open the Now Assist Panel (NAP).

    Expected behavior: 'Diagnose a service graph connector' should be a recommended topic.

    Actual behavior: The Diagnose a service graph connector topic button doesn't appear.

    Virtual Agent

    PRB1880277

    Dynamic Translation (DT) in the full page experience does not translate all elements Some elements are not translated in the user's language in the full page experience.
    1. Enable DT.
    2. Index sources for KB, Catalog, and skills.
    3. Impersonate a user.
    4. Change the language to Spanish.
    5. Search for KB, Catalog, or topics.
    6. Open the left panel.
    7. Check for the translation of elements.

    Expected behavior: All elements in full page should be translated to the user's language.

    Actual behavior: Some objects are not translated to the user's language.

    Virtual Agent

    PRB1880594

    The 'sys_gen_ai_log_metadata_list' caller column is empty The 'sys_gen_ai_log_metadata_list' caller column is empty even if the caller column in 'sys_generative_ai_log_list' contains data.
    1. Run a dynamic capability.
    2. See if the caller is present on sys_generative_ai_log but not on sys_gen_ai_log_metadata.

    Virtual Agent

    PRB1881057

    Java Proxy should use the LLM language as the source instead of the system language Synthesized results should appear in the non-English language set by glide.sys.language when executing a search in Now Assist.
    1. Setup Now Assist Setup DT for VA Install language plugins.
    2. Set the LLM prop (com.glide.cs.llm.default.Language) to English.
    3. Set the NLU prop (glide.sys.language) to a non-English language as a user with their language preference set to same value as 'glide.sys.language'.
    4. Execute a search in Now Assist.

    Expected behavior: Synthesized results should appear in the user's language.

    Actual behavior: Synthesized results appear in English.

    Virtual Agent

    PRB1881106

    Global changes for auto chat in Glide This is a product update.

    Virtual Agent

    PRB1881107

    Enhance the async mode error response format in one API This is a product update.

    Virtual Agent

    PRB1881108

    A capability should have a new type called 'Predictive Intelligence' This is a product update.

    Virtual Agent

    PRB1881109

    Deep clone of some skills and multiple default prompt bug fixes This is a product update.

    Virtual Agent

    PRB1881110

    Truncation support for tool outputs This is a product update.

    Virtual Agent

    PRB1881111

    Support 'Cancel' for AutoEval Run This is a product update.

    Virtual Agent

    PRB1881112

    Enhance APIs to support for creating a dataset and processing a dataset to avoid creating dataset skill mapping This is a product update.

    Virtual Agent

    PRB1881113

    Auto-prompt tuning check due to a backend name change This is a product update.

    Virtual Agent

    PRB1881115

    Extend Metric Result's schema and API to persist results per group This is a product update.

    Virtual Agent

    PRB1881116

    Clone Agentic Eval Run This is a product update.

    Virtual Agent

    PRB1881118

    Support Agentic AI Eval execution without a proxy skill This is a product update.

    Virtual Agent

    PRB1881121

    API support for new schema changes This is a product update.

    Virtual Agent

    PRB1881123

    Adding to an update set should generate child update sets with a unique name This is a product update.

    Virtual Agent

    PRB1881125

    As an AI Practitioner, users should be able to select a solution, and be able to navigate to additional configurations sections This is a product update.

    Virtual Agent

    PRB1881126

    Create a new API for getTestDatasets for agentic-type datasets This is a product update.

    Virtual Agent

    PRB1882397

    ReAct Search Agent stuck in a loop This issue occurs with a search query and a meeting agent. Synthesized answers are returned in the logs, but are not displayed.
    1. Navigate to /ESC.
    2. Open the NASS query, 'Let's get info on password reset then set up meeting'.

    Notice that the search agent is stuck in a loop with the message, 'Sorry no agents are available', and a synthesized answer is returned in logs but is not displayed.

    Virtual Agent

    PRB1882692

    Assistant-based agents functionality is missing

    Virtual Agent

    PRB1883231

    Sync calls are happening every 5 seconds when creating a new chat
    1. Navigate to dynamic window/FPE.
    2. Open the browser console.
    3. Create a conversation.

    Expected behavior: '/sync' calls should be happening as often as the 'syncIntervalTime' is set. Ensure that 'syncIntervalTime' isn't displaying in the /sync calls response when initiating a new conversation.

    Actual behavior: that '/sync' calls are happening every 5 seconds.

    Virtual Agent

    PRB1883338

    Skill discovery for a non-agent topic (caller's assets) redirects to a AI Agent workflow in Now Assist Panel (NAP) In NAP, skill discovery sometimes redirects to the default Virtual Agent workflow (AI Agent) instead of starting the caller assets skill.

    Virtual Agent

    PRB1883376

    Responses aren't coming when a user summarizes a record after providing an incident ID An error happens in MessageBatchingSession.escapeList(), which escapes a list of strings before calling DT. The problem is the list is immutable, so it fails to update the list with the escaped values.
    1. Log in to the instance as Beth Anglin with the Spanish language
    2. Select the 'NAP' icon.
    3. Search for 'Resumir un registro'.
    4. Once the response comes back, provide the incident ID as INC0009009.

    Expected behavior: The incident summarization should come in Spanish.

    Actual behavior: There's no response in the chat, and an error in the log.

    Virtual Agent

    PRB1883466

    Parity with FD in NowLLM invocation

    Virtual Agent

    PRB1883829

    Verify 'Summarize a Record from Record View'
    1. Navigate to the 'Incident' table.
    2. Select any incident number.
    3. Navigate to the Now Assist Panel (NAP) panel by selecting the NAP icon in the top right corner of the unified navigation.
    4. Select the Summarize a record skill pill.

    Expected behavior: The summary should be displayed.

    Actual behavior: A 'What is the record number? 'question is displayed.

    Virtual Agent

    PRB1884082

    Input text messages from agents are displaying with a lower sequence number than they should This affects the conversation history.
    1. Navigate to an instance with the 'Ticket status' agent.
    2. Type 'What is the status of my tickets?'.

    Expected behavior: The messages when sorted by sequence should reflect the actual flow of the conversation.

    Actual behavior: In the sys_cs_message table, if users sort the messages by the sequence number, they appear to be out of order.

    Virtual Agent

    PRB1884615

    Citations in NAVA and Now Assist Panel (NAP) for non-English conversation languages are displaying incorrectly
    1. Navigate to the SP portal.
    2. In the NAVA chatbot, enter 'Quels sont quelques monuments célèbres à Paris ?'.

    Observe that in the returned Genius Results, the citation at the end is displayed as just [1]. In the English session, users can type 'what is spam' to see expected response format.

    Virtual Agent

    PRB1884694

    Proactive triggers with topics don't launch in dynamic window (DW)
    1. Set up proactive triggers for the portals with topics that can be launched.
    2. Navigate to the portal on DW and wait for the proactive trigger.
    3. Launch the topic in proactive trigger.

    Expected behavior: The topic should launch in DW.

    Actual behavior: The topic doesn't launch and the user lands on a new chat in DW.

    Virtual Agent

    PRB1887021

    A conversation call is taking too long after an update There's no 'IN' clause on a query on the sys_cs_message table.
    1. Load Now Assist Panel (NAP).
    2. Check the conversation call form network.

    Observe that the times are more than 1 second.

    Virtual Agent Web Client

    PRB1820930

    The Show More button isn't working consistently in iOS Safari The Show More button in a NAVA synthesized response is shown despite not having more responses to display.
    1. Have NAVA enabled on an instance.
    2. Create a user to log in using a mobile browser.
    3. Enter 'What should I know about direct deposit?'.
    4. Select the Show More button on the response.

    Expected behavior: More of the SR should show without any issues.

    Actual behavior: SR is rendered incorrectly with the Show Less button showing behind the rest of the message.

    Virtual Agent Web Client

    PRB1869818

    Toggling between the search and Now Assist isn't retaining the search term
    1. Navigate to Service Portal.
    2. Enter 'what is spam' in the portal search test box.
    3. Select the New chat button in the 'NASS' page.
    4. Select the search toggle.

    Expected behavior: An empty search should be displayed.

    Actual behavior: The search term from step 2 is retained.

    Virtual Agent Web Client

    PRB1871159

    The 'DW' icon is visible and popups for citations aren't clickable
    1. Navigate to dynamic window (DW)/Full Page (FP).
    2. Open in floating mode.

    Expected behavior: The 'DW' icon shouldn't be available.

    Actual behavior: The 'DW' icon is available and clickable.

    Virtual Agent Web Client

    PRB1871232

    Chats aren't in sync when navigating multiple portal pages
    1. Enable full page (FP) for Service Portal.
    2. Navigate to dynamic window (DW).
    3. Run a few conversations such that there are a few chats in the 'Active' section.
    4. Perform another search in the search bar on FP.
    5. Switch to another chat on FP.
    6. Navigate to Knowledge, Catalogs, Requests, System Status, Tours and other pages on the portal.

    Expected behavior: The chat should remain in sync when the user opens DW in these pages.

    Actual behavior: The chat isn't in sync when the user goes through multiple pages and there's content from another chat on DW.

    Virtual Agent Web Client

    PRB1877908

    A document upload fails the first time

    Try to upload an attachment to Virtual Agent.

    Notice that it gets stuck in loading. If the user refreshes and uploads again, it works.

    Virtual Agent Web Client

    PRB1878155

    Icons for different sources aren't visible in the 'Sources' section
    1. Navigate to NAVA.
    2. Search for a person and KB/catalog 'who is abel tuter and what is spam' / 'who is abel tuter and how can i order a laptop'.
    3. Expand the 'Sources' section to view citations.

    Expected behavior: Each source should have a corresponding icon beside the citation number.

    Actual behavior: No icons are visible except for the KB and the 'People' card. The current icon for the 'People' card is incorrect. It should be a generic icon instead of a user profile picture.

    Virtual Agent Web Client

    PRB1879738

    Chats are not in sync when navigating multiple portal pages After enabling Full Page (FP), switching between multiple chats and navigating away to different Portal pages in the Dispatcher Workspace, the chats are not in sync when the user returns to them.
    1. Enable Full Page (FP) for Service Portal.
    2. Navigate to the Dispatcher Workspace.
    3. Run a few conversations so that there are a few chats in the Active section.
    4. Perform another search in the search bar on FP.
    5. Switch to another chat on FP.
    6. Navigate to other pages on Portal such as Knowledge, Catalogs, or Requests.

    Expected behavior: The chat should remain in sync when user opens dynamic window (DW) in these pages.

    Actual behavior: The chat is not in sync when user goes through multiple pages, and content from another chat on DW can be seen.

    Virtual Agent Web Client

    PRB1880258

    Focus isn't on the input box in dynamic window (DW)
    1. Navigate to SP portal in DW mode.
    2. Open DW.

    Expected behavior: The focus should be on an input text area where the user can start typing.

    Actual behavior: The focus isn't anywhere.

    Virtual Agent Web Client

    PRB1880490

    Loading a document with .doc ends document QnA with an error Now Assist gives a message: 'Something went wrong. I'm not able to process the uploaded document right now. I will end this Document Q&A. You can start a new session by uploading the document again, or you can try uploading something else.'

    Virtual Agent Web Client

    PRB1880493

    The Send button isn't enabled after a document Q&A topic exits with an error

    Virtual Agent Web Client

    PRB1880507

    Input and output topics are not rendered properly in dynamic window (DW) and Full Page (FP) Topics will have to be imported to VA Designer with a selected assistant.
    1. Import attached topics or create topics with all input/output controls
    2. Run both in DW and FP.

    Expected behavior: All topics should run correctly.

    Actual behavior: Input and output topics are not rendered correctly in both DW and FP, and an assistant must be selected to import them to VA Designer.

    Virtual Agent Web Client

    PRB1880552

    Agentic reasoning messages continue loading after topic completion
    1. Enable AI agents in NAVA/DW.
    2. Run the laptop ordering assistant.
    3. Complete all steps for a topic.

    Expected behavior: All agentic reasoning messages should have a checkmark showing completion.

    Actual behavior: Some messages are still loading after an AI agent topic is complete.

    Virtual Agent Web Client

    PRB1880646

    People card citations are grayed out and occasionally return duplicates Searching for People cards in the Dispatcher Workspace returns a synthesized response that is grayed out, and with occasional duplicates.
    1. Setup Knowledge Graph and People card on the instance.
    2. Open Dispatcher Workspace.
    3. Enter 'who is Abel Tuter'.

    Expected behavior: Synthesized response with single citation people card for Abel Tuter is displayed and it is not grayed out.

    Actual behavior: Synthesized response with duplicate citation links are displayed sometimes and the citation is grayed out even before selecting it.

    Virtual Agent Web Client

    PRB1881744

    An attachment icon is enabled in the middle of a conversation on a page refresh
    1. Start a topic execution on a chat client.

      Notice that the attachment icon no longer appears as expected.

    2. Refresh the browser page.

      Notice that the attachment appears again.

    Expected behavior: The attachment icon should no longer appear on page refresh in the middle of the conversation.

    Actual behavior: The attachment icon shows up on a refresh.

    Virtual Agent Web Client

    PRB1881830

    Float mode of dynamic window (DW) is not resizable after minimizing from a pinned state A resizing issue of DW occurs in float mode.
    1. Open DW.
    2. Notice it's resizable.
    3. Pin it and then minimize.
    4. Select the tab to display the DW again.

    Expected behavior: Float mode is always resizable irrespective of other state changes.

    Actual behavior: Observe it is not resizable anymore.

    Virtual Agent Web Client

    PRB1881836

    The UIs for chats are mixed up when switching between chats Components are overlapping in the UI and dynamic window (DW) can't be minimized past 420px.

    Components are overlapping in the UI:

    1. Navigate to DW and run a few conversations.
    2. Do a search on Full Page.
    3. Return to DW's home.
    4. Try switching between the active chats.

    DW cannot be minimized past 420px:

    1. Navigate to Service Portal with DW configured.
    2. Select the 'FAB' icon.
    3. Try to drag the side of DW to make it a smaller chat window.

    Virtual Agent Web Client

    PRB1881936

    Elements in dynamic window (DW)/Full Page (FP) aren't translated to the user selected language Untranslated items: titles in the left panel, show/hide sources, the Skip button, tooltips, the Live agent button in support, toggles and titles in the settings of the left panel, topic descriptions, and the Submit button.
    1. Install a language.
    2. Switch the profile language to that language.
    3. Configure a full window/enhanced chat NAVA in a portal.
    4. Navigate to the chat location with the nowassistselfservice ID passed to see the full page experience.
    5. Select around the UI.
    6. Run a catalog/KB/topic by searching 'prestamista de ordenadores portatiles' 'que es el spam' 'pedir cafe'.

    Expected behavior: All elements on the page are translated to the user selected language.

    Actual behavior: All the elements aren't displayed in the user language.

    Virtual Agent Web Client

    PRB1883412

    'Show more' is removed after streaming is done, and for non-streaming, it's not displayed
    1. Navigate to Service Portal.
    2. Enter 'what is cookie and as follow how to set it in browser'.

    Expected behavior: 'Show more' shouldn't be displayed.

    Actual behavior: 'Show more' is displayed during streaming and then it's removed.

    Virtual Agent Web Client

    PRB1883668

    The Attachment button comes after a page refresh when in doc QnA mode
    1. Open the Virtual Agent chat.
    2. Attach a document to it.
    3. Ensure QnA mode is entered.
    4. Observe the Attachment button is turned off.
    5. Refresh the page.
    6. Observe that now the Attachment button is enabled.

    Virtual Agent Web Client

    PRB1883927

    Turned off text is unreadable
    1. Set up Now Assist for May.
    2. Set up a full page.
    3. Navigate to /esc.
    4. Type 'order laptop'.
    5. Start an execution.

      The link displays as turned off text.

    Notice that it shows as unreadable.

    Virtual Agent Web Client

    PRB1883963

    Context data doesn't send start_conversation _immediately: true when going from the Portal to FPE
    1. Open a portal page with 'Enhanced Chat' setup.
    2. Enter a search term into a search combo box.

    Expected behavior: It should send 'start_conversation_immediately: true' so that the server knows to immediately start the conversation in the session call and not wait for a sync call to kick off the state machine.

    Actual behavior: The '/session' call doesn't contain 'start_conversation_immediately: true' in the contextData.

    Virtual Agent Web Client

    PRB1884003

    The 'People' icon is flashing when going from a streaming response to a final response
    1. Open a portal page with 'Enhanced Chat' setup.
    2. Enter a search term into the search combo box that results in a people search result, such as: 'who reports to Abel Tuter'.

    See that the 'People' icon gets shown, goes away, and comes back.

    Virtual Agent Web Client

    PRB1884112

    The dynamic window (DW) should occupy the entire screen for a browser window width less than 420px
    1. Open DW on a portal.
    2. Reduce the browser window width size to smaller than 421px.

    Expected behavior: DW should fill the entire screen for a width to smaller than or equal to 420px.

    Actual behavior: DW is cut off and not able to occupy the entire screen.

    Virtual Agent Web Client

    PRB1884123

    The 'Attachment' icon isn't appearing on refresh on a new chat in the middle of a topic
    1. Open a Virtual Agent chat.
    2. Start any topic.
    3. Select on a new chat.
    4. Refresh the page.

    Observe that the 'Attachment' icon isn't coming.

    Walk-Up Experience

    PRB1873610

    KB2046788

    After an Yokohama upgrade, the walk-up queue isn't displaying any users who are checked in for any location The issue occurs because the Rest API contains the snc_internal role and as the user has snc_external role, so it disqualifies the walkup user and doesn't render any users in the queue.

    Refer to the listed KB article for details.

    Work Order Management

    PRB1853346

    Upgrading to the 'New Places' API

    Work Order Management

    PRB1855599

    A work order task (WOT) isn't assigned to a vendor/contractor when assigning a WOT to a vendor/contractor group A work order task (WOT) isn't assigned to a vendor/contractor when assigning a WOT to a vendor/contractor group, so the Assigned to field is empty in CSM/FSM Configurable Workspace.
    1. Navigate to Dispatcher workspaceWorkspace setting.
    2. Check 'Enable task assignment modal' to ensure an assignment model opens when drag and dropping.
    3. Install the 'Field service territory planning' plugin.
    4. Navigate to a territory model module and set the territory model as true.
    5. Navigate to the WOT Form page > Form layout.
    6. Add a Territory field in the 'DispatcherWorkspace' view if the it isn't already added.
    7. Create a WOT and assign it to some territory that manages at least one contractor group.
    8. Move it in a pending dispatch. Don't give values to assignment group or assigned_to fields.
    9. Impersonate as a dispatcher that manages the territory where users assign a work order task.
    10. Select the territory from the territory filter.
    11. Drag and drop the WOT created in step 7 to the contractor group under 'Contractor Groups'.

    Expected behavior: The Assigned_to field should be populated with the contractor manager.

    Actual behavior: The Assigned_to field comes up as empty.

    Work Order Management

    PRB1859719

    Upgrade to the new Google Maps Places API for Platform and Portal Using the new Places API to query and display suggestions on the NDS typeahead component.

    Work Order Management

    PRB1880140

    Getting a pop-up error while opening the Work Order Task (WOT) on Portal The error 'There is a JavaScript error in your browser console' pops up.
    1. Hop to any instance.
    2. Attempt to open any WOT on the portal.

    Notice that the error message pops up.

    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.