Synapse Docs
Breadcrumbs

stack-599 - Release Notes

Release Date:

Summary

Stack-599 delivers 28 tickets across the Synapse Platform (PLFM) and Synapse Web Client (SWC). It marks a major milestone on several fronts simultaneously: the AI-powered GridBot multi-agent system reaches functional completion, the Curator dashboard receives its most comprehensive task management upgrade yet, the entity page redesign continues with meaningful UX improvements, and the electronic Data Use Certificate (e-DUC) backend is now fully built out.

For researchers and data contributors, the most visible changes are in the Curator tool — a completely overhauled Tasks & Actions experience with rich task cards, due date tracking, direct task linking, and the first release of Compute Task creation and execution in the UI. For platform stakeholders, the GridBot multi-agent architecture is now operational end-to-end, and the e-DUC signing workflow has the full set of backend services needed to support a complete digital data access agreement process.

No breaking changes or required user actions in this release. All changes are additive. The GridBot AI capabilities are available within the Curator tool for supported workflows. The e-DUC signing UI for researchers is not yet available — the backend is complete and the UI will follow in a future release.

Affected Users & Systems

Who / What

Impact

All Synapse users

Entity pages: Synapse ID now visible and copyable in the header bar; redesigned Project Information dialog; improved empty provenance state

Data contributors & curators (Curator)

Fully redesigned Tasks & Actions tab with rich task cards, status tracking, due dates, and direct task linking via URL

Data contributors using Compute Tasks

Can now create and trigger AI-powered Compute Tasks (e.g. sample sheet generation) directly from the Curator UI

DCC teams & portal data managers

RecordSet schema synchronization now supported in both platform and UI; agent-assisted RecordSet generation PoC complete

Researchers requesting controlled data access

e-DUC backend fully operational — services for listing DARs, checking signature status, retrieving signed documents, and cancelling signatures are all in place

NIH-affiliated researchers

OIDC identity linking to Synapse account now fully resolved and operational

Platform & AI infrastructure

GridBot multi-agent system fully assembled — metadata specialist, compute dispatcher, sample sheet supervisor, and async job worker all operational

Deprecated or Breaking Changes

There are no deprecated or breaking changes in this release. No user action is required.

New Features

New Curator Tasks & Actions — Complete Redesign

The Tasks & Actions tab in the Curator dashboard introduces an interactive task card view designed to give contributors and curators a clear picture of their work at a glance.

Each task card now displays:

  • Project name and task ID as inline identifiers

  • Expandable instructions — clicking the task name reveals full guidance without leaving the dashboard

  • Task Status pill — color-coded as Not Started (gray), In Progress (yellow), or Complete (green); editable by users with project edit access

  • Due Date chip — color-coded by urgency: gray (30+ days), yellow (within 30 days), red (overdue)

Project managers and support staff can also now link directly to a specific task using a URL parameter, eliminating the need for instructions like "find the third item in the list." This makes it significantly easier to direct contributors to exactly the task they need to complete.

Who benefits: Data contributors, data curators, project coordinators, and DCC staff who manage curation workflows in Curator.

New Compute Tasks — Create and Execute AI-Powered Tasks from the Curator UI

This feature is currently in experimental mode. The Curator UI now supports the full lifecycle of Compute Tasks — a new task type that allows contributors to trigger automated, AI-powered workflows directly from the dashboard. In this release, the primary Compute Task available is sample sheet generation: contributors can initiate the process with a single action, and the system will automatically extract and transform their data into a properly formatted sample sheet.

Both task creation and execution are now available in the UI. A routing bug that caused a 404 error after creating a sample sheet task has also been resolved.

Who benefits: Data contributors and curators working with datasets that require sample sheet preparation, particularly those using automated nf-core or similar bioinformatics pipelines.

New RecordSet Schema Synchronization

Synapse now supports synchronizing a RecordSet's schema — keeping it consistent as the underlying data source evolves. Both the platform API and the Curator UI reflect this capability, allowing data teams to trigger a sync that updates the RecordSet structure without requiring manual recreation of the dataset.

If a RecordSet schema is updated, users will see the following button in the Curator grid.

image-20260728-161956.png

Who benefits: Data curators, DCC teams, and portal engineering teams managing evolving datasets.

New e-DUC Signing Backend — Now Fully Complete

The backend services for the electronic Data Use Certificate (e-DUC) signing process are now fully operational. Building on the envelope creation service delivered in stack-598, this release adds four additional services that complete the signing lifecycle:

  • List DARs — retrieve all Data Access Requests associated with an e-DUC signing process

  • Signature status — check the current state of a signing envelope, including delivery failures surfaced via DocuSign webhooks

  • Retrieve signed document — obtain the file handle for the completed, signed DUC PDF

  • Cancel signature — revoke an in-flight e-DUC request and return the DAR to non-e-DUC mode

The researcher-facing UI for initiating and tracking e-DUC signatures will be delivered in a future release. ACT and DAC staff can expect this workflow to surface in the access request management interface once UI work is complete.

Who benefits: Researchers requesting access to controlled data; ACT/DAC reviewers managing Data Use Certificate approvals.

New Entity Pages — Synapse ID in Header Bar & Project Information Dialog

Three further improvements land in the ongoing entity page redesign:

  • The Synapse ID (e.g. syn12345678) for any entity is now displayed directly in the blue header bar, with a button to copy it to the clipboard — making it significantly easier to reference or share a specific entity.

  • The Project Information dialog has been redesigned with a cleaner layout consistent with the new entity page design system.

  • The empty state of the provenance graph has been updated with a more informative and visually consistent placeholder for entities that do not yet have provenance recorded.

Who benefits: All Synapse users, particularly those who frequently share or reference entities by their Synapse ID.

Platform GridBot Multi-Agent System — Fully Assembled

The GridBot multi-agent AI system — first introduced in stack-598 with foundational AWS Bedrock and AgentCore integrations — is now fully assembled. This release delivers the four remaining components that complete the end-to-end architecture:

  • SYNAPSE_CORE_METADATA Specialist — an AI agent that retrieves and interprets entity metadata, annotations, and schema information to inform downstream processing decisions

  • ComputeTaskDispatcher — an async job worker that receives executable curation tasks and routes them to the appropriate type-specific sub-worker

  • Sample Sheet Supervisor — an autonomous multi-agent orchestrator that coordinates specialist agents to produce a complete, validated sample sheet from project metadata

  • SampleSheetGenerationSubWorker — manages the async job lifecycle and handles all Synapse state changes as a sample sheet generation job progresses

A proof-of-concept for agent-assisted RecordSet generation has also been completed and validated — demonstrating that the system can automatically extract, transform, and load data from a contributor-uploaded folder into a Curator RecordSet matching a target schema.

The GridBot capabilities are now available to Compute Task workflows within Curator, surfacing through the Compute Task UI delivered in this same release.

Who benefits: Data contributors and DCC teams using Curator for complex curation workflows; platform teams now have a production-ready multi-agent AI infrastructure.

Fixes & Improvements

The entity page sidebar introduced in stack-598 was inconsistently remembering whether the user had collapsed or expanded it. The sidebar state is now maintained globally and consistently across navigation, so users who prefer a collapsed sidebar no longer have to repeatedly close it.

Curator Upsert Key Column Style Corrected

The upsert key column in the Curator grid was displaying with a blue background across the entire column header instead of the correct treatment — a key icon with a blue background, with a standard gray column header. This has been corrected to match the intended design.

AWS AgentCore Permission Fixed for Compute Tasks

A missing AWS IAM permission was preventing the Compute Task system from launching AI code interpreter sessions in production. The required permission has been added, enabling Compute Task execution to function as designed.

CI Build Test Context Loading Failure Resolved

A build system failure was causing test context files to fail to load on Windows build environments, with truncated error logs making the root cause difficult to diagnose. The issue has been resolved, restoring reliable CI builds across all environments. No user-facing impact.