May 31, 2026

In this edition

Agent Roster & Discovery

Enriched, discoverable specialist profiles for Nova, Polaris, Vega, and Nexus, with clear roles, descriptions, and core questions to help people pick the right assistant for the job; Nexus remains a background automation agent rather than a directly selectable conversation partner.
Crew page polish for a smoother discovery experience: improved avatar cropping, keyboard accessibility, and reduced empty-state noise.

Updated radial app launcher with Polaris as a selectable specialist, a compact dynamic radial layout (defaulting lower-right), improved keyboard navigation, and quieter drag/persistence behavior for more predictable placement.
New per-agent customization and layout controls — including color assignment, spread tuning, and allowed-overlap settings — plus light-theme styling for consistent visibility across themes.

Introduced the searchable Methods Library catalog with category rails and target-layer badges across 31 template methods, making it faster to find and compare experiment approaches.
Streamlined experiment creation with a unified two-step wizard: Step 1 selects a method from the catalog; Step 2 uses a premium split-pane editor that shows method visuals and metadata alongside the experiment form, with a "Start without template" fallback.
Upgraded visuals and theme support: template cards and previews now include custom blueprint-style illustrations and native light/dark theme transitions for clearer, more engaging method previews.
Tighter integration across the workflow: visual target-layer badges now appear on Validation Tree nodes and Experiments Kanban cards to make intent and scope visible at-a-glance.
Safer handling of legacy or corrupted templates: experiments created without a template or with missing snapshot data now show a friendly "Custom Template" fallback and a resilient detail view rather than breaking the layout.

Theme Learning Synthesis reliability improvements: resolved an error that could prevent saving auto-synthesized theme learnings to ensure learnings persist reliably.
Agent Theme Controls: agents (for example, Nova) can now update a theme's lifecycle status and disposition when instructed, simplifying automated theme administration.