stack-598 - Release Notes

Release Date: 2026-07-19

Summary

Stack-598 is a substantive release touching four strategic areas: continued refinement of the Synapse entity page redesign, significant advances in AI-powered data curation infrastructure, expanded data governance capabilities, and meaningful improvements to the Curator and portal search experience.

The entity page redesign takes its next major step with the introduction of a collapsible sidebar that organizes entity metadata properties in a cleaner layout, along with breadcrumb updates and a responsive fix that ensures the action bar remains fully usable on tablet and narrower desktop screens. In parallel, the platform team has completed foundational AI infrastructure work — integrating AWS Bedrock and AgentCore — that will power the next generation of the Curator GridBot. On the governance side, a new electronic Data Use Certificate (e-DUC) signing service advances the effort to make data access requests fully digital. And a meaningful data platform capability lands: RecordSets can now be included as sources in Materialized Views, enabling richer cross-dataset queries for portal teams.

Altogether, 11 tickets were completed in this release across the Synapse Platform (PLFM) and Synapse Web Client (SWC) projects.

Affected Users & Systems

Deprecated or Breaking Changes

New Features

New Entity Pages Now Have a Sidebar for Metadata Properties

The ongoing entity page redesign takes its most significant step yet with the introduction of a dedicated sidebar panel. Entity metadata properties — such as storage location, annotations, and other contextual details — have been moved into this sidebar, which keeps the main content area cleaner and allows users to reference key properties without losing sight of the entity content itself.

This continues the redesign work delivered in stacks 597 and 598, and sets the stage for additional panels (such as access and permissions, mentions, and citations) to be added to the sidebar in future releases.

Who benefits: All Synapse users who work with projects, files, datasets, and tables.

New RecordSets Can Now Be Used in Materialized Views

Data teams can now reference a RecordSet as a source when building a Materialized View in Synapse. This means curated datasets maintained through the Curator workflow can be joined, filtered, and aggregated alongside other Synapse table types — enabling richer, unified data views without duplicating data manually.

This is a foundational capability for portal teams who manage complex multi-source datasets and need a single queryable view across curation outputs and other metadata sources.

Who benefits: Data curators, portal engineering teams, and researchers who build or depend on Materialized Views for cross-dataset analysis and portal Explore pages.

New Electronic Data Use Certificate (e-DUC) Signing — Backend Foundation

A new platform service has been delivered that initiates the electronic signing process for Data Use Certificates attached to Data Access Requests. When a researcher submits a DAR that requires an e-DUC, the platform now creates a signing envelope, populates it with the relevant DAR content, and routes it for signature — putting the DAR into an "e-DUC mode" that will guide it through a fully digital approval workflow.

This is the backend foundation for a broader initiative to eliminate paper-based data use agreements. The researcher-facing UI for this process will follow in a future release. Access & Compliance Team (ACT) members should expect to see this capability surface in the access request workflow in the coming stacks.

Who benefits: Researchers requesting access to controlled datasets, and ACT/DAC reviewers who manage Data Use Certificate approvals.

Platform AI Infrastructure for Next-Generation GridBot (AWS Bedrock + AgentCore)

Two foundational platform components have been completed in support of the upcoming multi-agent GridBot — an AI assistant that helps data contributors automate and validate curation workflows in the Curator tool.

First, the platform now integrates with AWS Bedrock via the Spring AI framework, establishing a configurable AI model layer that can be used by multiple Synapse services going forward. Second, integrations with AWS AgentCore are in place, providing a secure sandbox for executing AI-generated code and a short-term memory layer that allows agents to maintain context across a curation session.

These capabilities are not yet user-facing — they are the infrastructure that future GridBot features will be built on. Implementation of the end-user experience will follow in upcoming releases.

Who benefits: Data contributors and curators will benefit when future GridBot capabilities ship; platform teams now have a reusable AI infrastructure layer available across Synapse services.

Fixes & Improvements

Multi-Word Search Terms Now Resolve Correctly from Shared Links

A bug was fixed where searching for a multi-word term via a direct URL (such as a shared or bookmarked search link) would display the raw encoded form of the query — for example, showing "multi%20word%20term" instead of "multi word term" in the search box. The search results were also affected, as the query was not being interpreted correctly. This has been corrected and shared search links now work as expected.

Who benefits: Any Synapse user who shares or receives search links, or who has bookmarked search queries.

Entity Page Action Bar Now Fully Visible on Tablets and Narrow Screens

Following the action bar redesign delivered in stack-597, a follow-on fix was applied to ensure that all action buttons remain accessible on tablet-sized and moderately reduced desktop viewports. Previously, the wider labeled buttons introduced in the redesign could overflow and become hidden at intermediate screen widths. The layout now adapts gracefully, stacking to a mobile-friendly format at narrower widths while remaining compact on full desktop displays.

Who benefits: Synapse users on tablets or working in reduced-width browser windows.

Entity Page Breadcrumb Icons Updated

The breadcrumb navigation at the top of entity pages received a small visual polish update — icon weights have been adjusted to match the new design language introduced in the broader entity page redesign. No functional changes.

Curation Task API Supports Filtering by Task ID

The platform API for listing curation tasks now accepts an optional task ID filter. This will support future UI improvements to add filtering of tasks in the dashboard.

Who benefits: Data curators and Curator tool users; primarily a platform improvement that enables richer UI capabilities.

OpenSearch Portal Search Now Enforces Row-Level Access Controls

A significant security and correctness improvement has been made to the portal search infrastructure. Previously, search indexes built on Amazon OpenSearch had no per-row access filtering — every indexed document was equally visible to all queries regardless of the viewer's permissions. This release introduces row-level benefactor tracking in the search index, so that search results are now automatically filtered to show only data the requesting user is permitted to see.

This is a critical step in the migration from the legacy search system and ensures that access-controlled data cannot surface in portal Explore searches for unauthorized users.

Who benefits: All portal users — particularly those on portals that index datasets with mixed access levels. Researchers can trust that search results reflect their actual access rights.

Production Deployment Stability Fix

A deployment failure that caused the stack-597 production environment to enter an unhealthy state was diagnosed and resolved as part of this release cycle. The issue was in the server startup sequence and did not result in data loss or user-visible downtime beyond the deployment window. No user action required.