Accessibility Overhaul & Amazon A+ Premium API Progress: Icecat Studio – Sprint 102 Release Notes

Sprint 102 delivered on two fronts that will shape the platform for quarters to come. We shipped the first major wave of our accessibility and interface quality program — making Icecat Studio operable end-to-end without a mouse and perceivable to screen-reader users — and reached a key milestone in the Amazon A+ Premium API beta. On top of that, the Story Builder gained an Array editor and a reengineered Advanced Preview, and our new MCP server entered development.

Amazon A+ Premium API – Beta Milestone

Our integration with the Amazon A+ Premium API took a significant step forward: it now performs on par with the A+ Standard API, with all premium module types matched and configured on the Icecat side. We expect to unlock the full Amazon A+ Premium content flow soon.

Story Builder – Array Editor & Advanced Preview on the New Framework

Two Story Builder upgrades released this sprint:

• Array editor: A new editing experience for array-based content structures within stories.
• Advanced Preview – new framework: The Advanced Preview has been migrated to the Product Story Framework v2 and now runs on the same rendering engine as the Builder and Player. Users get a preview that precisely matches the published result, while we retire legacy preview code.

Accessibility & Interface Quality Program – First Major Delivery

Following an in-depth interface analysis against leading market standards — WCAG 2.2, Google Material 3, Microsoft Fluent 2, and Apple HIG — we ran code-level accessibility audits on our key pages, combining automated Lighthouse/axe scans with live keyboard and screen reader testing. The findings were consolidated into a prioritized backlog of ~25 tickets across six epics, with one powerful insight: roughly 80% of all violations trace back to six shared component patterns, meaning every component-level fix propagates across the entire platform.

Sprint 102 delivered four major remediation packages:

• Accessible names & semantic structure: Around 200 previously unlabeled controls — icon buttons, favorite toggles, toolbar actions — received accessible names. Admin pages gained proper heading structure, the submission stepper now communicates step state by name rather than color alone, and 12+ dialogs announce their titles on open. Audited pages went from a baseline of ~200 violations to zero on the targeted axe rules.
• Live announcements: Icecat Studio is a heavily asynchronous product — status polling, ZIP generation, bulk actions, AI jobs. Previously, all of it was communicated through toasts and color only. Now every toast is mirrored into a live region, AI job stages and bulk-action progress are announced, and disabled controls expose the reason they are disabled programmatically — all with zero visual changes for sighted users.
• Full keyboard operability for the Sandbox Editor & Story Builder: The editing flow was previously mouse-only at every step. Now the entire journey works from the keyboard: the Layers panel acts as a keyboard object model; objects can be placed and nudged with keys; crop handles are focusable for arrow-key resizing; and the module library and tree support a complete keyboard pass. You can open the editor, crop an image, and build a story without touching a mouse.
• Visual AA — contrast, sizes & reflow: Work progressed on the visual compliance layer: color contrast, interactive target sizes, and content reflow.

This is a systematic march toward WCAG 2.2 AA as our mandatory baseline — with the release of these improvements to production planned for the next sprint.

MCP Server – From Investigation to Development

We started building the Icecat Studio MCP server, which will expose platform capabilities to external AI agents and assistants via the Model Context Protocol. The architecture follows two principles:

• One shared tool layer: The MCP server reuses the same tool registry that powers our AI Assistant — no MCP-only copies. Every new assistant capability becomes automatically available to external agents.
• Scoped authentication: Access is authorized via API keys with a dedicated MCP scope, validated on every request and resolved to the user’s account context — so existing permissions and limits apply automatically.

Sandbox – SVG Import & Rendering

• Gradient feature – testing: Gradient support — the key blocker for importing Figma-exported designs — reached the test environment, with production deployment expected shortly.
• Rendering fixes: Font rendering inconsistencies between editing mode and final output have been resolved and ship with gradients. The majority of SVG import errors have been fixed, with the remainder under analysis based on partner-provided sample files.
• Canva import: Investigation of Canva import support (SVG & PDF) continued.