Lossless Changelog
product

reach-edu-hub onboards to @dididecks/shell — and a deck-collections pattern worth lifting

reach-edu-hub becomes the shell's fourth consumer, joining chroma, humain, and lossless. The onboarding also produced two reusable ideas the shell doesn't have yet: deck collections (a menu layer above single-deck convergence) and a token-driven deck-primitives system — both prototyped in reach as lift candidates. The pass also surfaced one concrete shell limitation: the slot registry is keyed by variant slug alone, so sibling decks can't reuse a variant name.

Michael StatonMusic Editoraugmented with Claude Code on Claude Opus 4.8 (1M context)

Why Care?

reach-edu-hub was the last client-site standing outside the shared shell — its own deck chrome, not in the workspace, zero @dididecks/shell references. That meant every shell improvement (review TOC, in-deck ranking, the play runtime) skipped it, and any fix had to be made twice.

It now consumes @dididecks/shell like the other three client-sites. That's the headline: four consumers on one shell. The full onboarding lives in reach's own repo and changelog; this entry records the platform-level milestone and the two patterns the work generated that the shell should eventually absorb.

What's New?

reach is in the workspace. Added client-sites/reach-edu-hub to pnpm-workspace.yaml (so workspace:* resolves the shell) and bumped the submodule pointer to reach's onboarding commit. The reach repo carries the actual integration — registries, /scroll/[deck]/[variant] routes, section annotations.

A plan of record. context-v/plans/Integrate-Reach-Edu-Hub-into-Dididecks-Shell.md documents the onboarding (phases A–D, all shipped + verified) and the one design change forced along the way: client-sites consuming the shell must use Astro output: 'server', because the shell's injected routes run esbuild.transform during prerender, which a static build crashes on. (Other consumers used server for auth reasons, which masked that it's actually a shell requirement.)

Patterns to lift into the shell

Two things prototyped in reach belong in @dididecks/shell once they settle:

  • Deck collections — a menu layer above single-deck convergence. The shell today perfects one deck through variants; collections group several peer decks (different narratives) behind a menu. Reach implements collections.ts + a DeckCollectionMenu component + per-collection routes. Concept doc: client-sites/reach-edu-hub/context-v/explorations/Deck-Collections-A-Menu-Layer-Above-Single-Deck-Convergence.md.

  • deck-primitives.css — a token-driven, mode-aware section vocabulary that reads only Tier-2 semantic + --fx-* tokens, so decks stay on-brand and respond to light/dark/vibrant with no per-element variant classes. The drag-and-drop convergence target.

Limitation surfaced

The shell keys the slot registry by variant slug alone (SLOTS[variantSlug]), not by deck/variant. Sibling decks therefore can't reuse a variant name (every deck wanting v1/baseline collides). reach works around it for now by keeping variant slugs globally unique; the proper fix is to scope SLOTS by deckSlug + variantSlug in apps/deck-shell. Filed as the next shared-tooling task.

What's Next?

  • Scope the shell's SLOTS lookup by deck+variant.

  • Promote DeckCollectionMenu, the collections.ts shape, and deck-primitives.css into the shell after reach exercises them.