New ✨

Scorecard Override Toggle — scorecards now support a per-entry override switch. Admins can toggle individual scorecard values on or off without modifying the underlying metric, making exception handling and QA review significantly faster

MCP File Import API — the Milemarker MCP server now exposes a file import tool, letting agents upload and attach files directly to platform records. Enables document-driven workflows and removes the need for manual file upload steps in automation sequences

Improvements 🛠️

12-Column Layout — the platform's grid system migrates to a standardized 12-column layout, giving page authors more precise placement control and delivering consistent spacing across dashboard sizes and breakpoints

Platform Metric Aggregation by Grouping Columns — metrics that define grouping columns can now aggregate those groups natively at the platform layer, reducing the amount of aggregation SQL you need to write upstream in Snowflake

Org Admin Impersonation — org admins can now impersonate end users directly from the frontend to reproduce reported issues without requiring the user to be on the call

Admin Console: User Entitlement Query View — a new entitlement query panel in Admin Console lets admins inspect which users can access specific data objects without leaving the console

Bug fixes 🐛

Navigator Access Group Labels — corrected naming inconsistencies in Navigator access group display labels that caused confusion when multiple groups shared similar names

MCP forms_update Description Field — forms_update now correctly saves the description field. Previously the tool echoed the stale description back as if the update had succeeded while silently dropping the change

View Icons — all view-type icons are now solid-fill, matching the platform's established icon style

Donut Chart Total — the center total on donut charts was missing in certain configurations; it now renders correctly in all cases

Grouped Chart Series Rendering — fixed an issue where grouped chart series could render out of order and legend sort sequence could appear shuffled relative to the data