Summer 2025 Early Access

Fixes

• Web entry links redirected users to the home page instead of the configured screen when using SAML with Default SSO enabled. The system now correctly redirects authenticated users to the appropriate web entry screen.

• The Task column appeared empty in CSV reports sent via email from saved searches, despite showing correctly in the UI. The export now includes all task data correctly, ensuring consistency between the saved search view and the downloaded report.

• The system logged a “Case not found” error during script task execution, even though the case and task existed and progressed correctly. The logging behavior has been corrected to prevent false CaseException entries when script tasks execute as expected.

• When editing a dashboard, the assigned screen did not appear in the interface, making it unclear which screen was linked. The system now correctly displays the assigned screen when editing dashboards.

• Reloading the page after applying the "Completed" status filter in the Inbox reset the filter to “All clear”. The system now preserves the selected status filter when the page is refreshed.

• In task saved searches, clicking "Reset to default" did not properly restore the default column configuration and incorrectly kept REQUEST ID, PROCESS, and ASSIGNEE COMPLETED as active columns. The reset action now correctly moves these variables back to the available columns list.

• Saved searches in the Task and Request views were displaying duplicate variables in the Available Columns section after resetting to default and re-saving column configurations. This duplication caused confusion when customizing views. The issue has been fixed to ensure each variable appears only once in the list.

• Users without permissions were unable to view process documentation from the Launchpad and received a "Not Authorized" error. The system now correctly allows access to viewable documentation or hides the option entirely based on the user's permissions.

• Settings options in devLink bundles were not synchronized correctly between servers, causing inconsistencies in displayed and marked options. The system now accurately retains and displays selected options across both source and target servers during bundle import and synchronization.

• The AI Assistant displayed a false "AI is currently offline" message with a 500 error when using the "Image to Process" feature, even though the process generated successfully. The interface now handles this flow correctly, and the error no longer appears during image-based process generation.

• The Task section displayed an empty column that served no functional purpose. The interface now removes this unused column, ensuring a cleaner and more accurate layout in task-related views.

• Selecting the "By email" option in the Two Step Authentication settings triggered an error, even when the default email server was properly configured. The system now correctly accepts and applies the "By email" option without errors when a valid email server is set.

• FlowGenie system prompts disappeared after saving and reopening a genie configuration. The system now properly retains and displays the defined prompt when the genie is revisited.

• FlowGenie returned a timeout error when processing prompts that required more than 60 seconds, such as parsing complex PDF content. The system now supports longer execution times through a configurable AI_GENIE_CLIENT_TIMEOUT environment variable, allowing extended processing without triggering Axios timeout errors.

• FlowGenie no longer returns a 400 Bad Request error when processing valid PDFs with complex prompts. The system now correctly handles structured extraction scenarios like shareholding compositions without requiring manual timeout changes.

• Added an index on cases_participated(case_number) to improve update performance for slow queries targeting the case_number field. Updates on this column now execute significantly faster.

• Fixed an issue where available columns in Saved Search configuration could be duplicated each time the settings were reopened. Now, columns appear only once in the configuration panel.

• Resolved an issue where web entry redirections failed due to screen cache desynchronization. The cache is now properly invalidated on import, ensuring the latest screen versions are used.

• Resolved an inconsistency where the RAG collection type was not available when creating a new Collection, yet remained selectable in FlowGenie. The interface now correctly displays the RAG collection option only when enabled.

• Scripts can now rebuild correctly by updating the PHP executor version. The fix uses a newer image that solves problems some users were having when running scripts.

• Parallel subprocess threads no longer reload each other when one is submitted. Each thread now runs independently, preventing unwanted redirection or reloads in parent cases.

• The "Pending translations" modal now disappears as expected when all translations finish, even with multiple tabs open. The modal no longer persists unnecessarily, resolving the desynchronization issue.

• Translations now complete and display correctly without requiring a page refresh. The interface updates automatically once the translation reaches 100%, resolving the issue caused by delayed callback synchronization.

• When trying to add a new instance in the DevLink tab, the system returned a server error due to a missing client_secret field in the database. The application failed to provide a default value during the insert operation, breaking the expected behavior. This issue blocked users from linking new environments and was identified as a regression. The fix ensures that DevLink correctly saves all necessary fields when creating a new connection.

• The “Save and Finish” button was appearing visually distorted when creating a process from a guide template due to conflicting custom CSS in the wizard-template component, specifically when using certain fonts like "Comic Sans." This issue has been corrected to ensure consistent rendering across all guide templates.

• SCIM group provisioning failed to assign users to groups due to the system rejecting PATCH requests with a “MethodNotAllowed” error. This occurred when identity providers attempted to update group membership during provisioning. The issue has been addressed to ensure SCIM-compatible group and user assignments function as expected.

• The IFrame Whitelist accepted URLs with unsupported protocols such as ftp://, bypassing security expectations. This issue has been resolved by restricting entries to http:// and https:// protocols only, ensuring stricter validation and improved iframe security.

• The setting that blocked embedded content (like Web Entries) was turned on by default, which caused confusion and made setup harder. It is now turned off by default so everything works as expected unless stricter security is needed.

• The IFrame Whitelist allowed saving invalid URLs (like plain text or numbers) when editing entries. This validation issue has now been corrected to ensure only valid URLs can be saved.

• URLs added to the IFrame whitelist were not being properly validated, allowing invalid entries that caused console errors when embedding pages. Validation has been added to ensure only well-formed URLs (e.g., domains, subdomains) are accepted, preventing misconfiguration issues.

• The category edit modal was not appearing in the Projects section, preventing users from editing categories. This issue has been fixed, and categories can now be edited as expected.

• Case creation failed on customer-specific environment because a slow database query prevented updates to the cases_participated table during process initiation. The delay stemmed from a missing index, which caused significant lag when updating case status. The issue has been resolved by adding the appropriate index to ensure cases can start without performance issues.

• Tasks that used "Action by Email" failed with a "Script execution failed with exit code" error because required settings were missing. Specifically, the “Actions by Email” configuration had not been enabled in the environment, preventing the process from completing after an email response. This has been clarified, and proper configuration now ensures email-based actions run as expected.

Saved Searches

• Saved searches based on self-service task filters caused a server error when creating charts. The system now properly loads these searches and allows chart creation without triggering errors.

• Saved Search reports with large datasets failed to send via email, especially when containing more than 8,000 records. The system now successfully generates and sends reports regardless of the number of records included.

• The advanced filter in the PROCESS column of task saved searches returned no results due to an incorrect reference to the process name field. The filter now correctly targets the appropriate table, allowing accurate filtering by process name.

Collections

• CSV imports into collections excluded checkbox fields, preventing complete data insertion. The import process now includes checkbox values, ensuring all collection fields are populated correctly.

• Collections showed unexpected behavior when a column label or user name included a single quote ('). The system now properly handles quotes, allowing labels and names like “Shareholder’s Name” or “O’Brien”.