Phase A++ — Play Fidelity, In-Play Ranking, and Variant URL Safety

Plan of record for the next slice of @dididecks/shell, picking up where [[Phase-A-Plus-In-Deck-Ranking-Shared-Nav-and-Play-Runtime]] left off. Closes the three gaps the founder named within minutes of opening pnpm dev against the Phase A+ shell. Still operates in workspace-link mode; orthogonal to Phase A.7’s publish.

What Phase A+ accomplished (smoke-tested, working)

The Phase A+ session landed the entire scope of its plan. As of the 2026-05-12 smoke test:

• ✅ Global nav <DididecksNav> auto-mounted on /toc/* and /play/* shell routes, with aria-current="page" active state, brand slot, and consumer-side opt-in import path (@dididecks/shell/components/DididecksNav.astro). Reads the consumer’s DECKS registry at render time so Scroll / TOC / Play links resolve to the first deck’s first variant.
• ✅ <SlideRankPill> floating overlay mounted in chroma’s /scroll/pitch/enhanced-v3/. Uses IntersectionObserver to detect the active section[data-slot], hydrates the local rank cache from GET /api/slide-rank, posts back optimistically on click. Founder confirmed this works in-browser — “now I see it the floating rank pill.”
• ✅ /play/[deckSlug]/[variantSlug]/[slot]/ route injected. Renders the per-slide component if it exists at src/components/slides/{variant}/{slot}-{slug}.astro via import.meta.glob, falls back to <DecomposeFirstPlaceholder> if not. Build emits all 16 enhanced-v3 slot pages.
• ✅ /play/[deckSlug]/[variantSlug]/ redirect route to slot 01.
• ✅ PlayChrome wrapper with the keyboard contract: ← / → / Space / Home / End / F / C / T / Esc. Bottom strip shows {slot}/{total} + variant label + ←/TOC/→ buttons.
• ✅ <DecomposeFirstPlaceholder> empty-slot fallback with link back to TOC (?focus={slot} query param).
• ✅ Slot 01 — Cover (faithful recreation) at client-sites/chroma-decks/src/components/slides/enhanced-v3/01-cover.astro. Self-contained styles + import "../../../styles/global.css" so v3-* tokens resolve in standalone play context.
• ✅ Slot 15 — Ask (deliberate redesign) at 15-ask.astro. Mega-numeral $12M hero, two-column Round / Status <dl> blocks, milestones footer. No inline <pre> blocks. Plan deviation: original plan said “slot 16 — ask”; the page’s actual section ordering placed ask at section 15 with COLOPHON at 16. Slides.ts reconciled to match the page.
• ✅ TOC [view →] links added on rows whose per-slide file exists. Verified 3 links (slots 01, 05, 15).
• ✅ slides.ts reconciled with the actual 16 page sections (renames: category-size → market-context; vs-pinecone slot 14 → slot 13; ask slot 16 → slot 15; new colophon at slot 16). Existing decompose stub at 05-bottleneck.astro untouched. Audit registry key pitch/enhanced-v3/05 continues to map cleanly.
• ✅ Shell version bumped from 0.0.1 to 0.1.0-rc.0 to signal in-flight pre-release.
• ✅ Typecheck + build green. Build emits 16 slot-page HTMLs + 4 variant-index redirects + the scroll + toc routes, all bytestable against pre-A+ except for the additive pill mount + data-attrs.

The founder verified in browser: the pill works, the nav renders, the play route loads. Phase A+ is functionally complete.

What Phase A+ did NOT accomplish (the founder’s three notes)

Direct quotes from the 2026-05-12 smoke test:

Note 1. “Play 404 at /play/pitch/proto/01/.”

The /play/[deckSlug]/[variantSlug]/index.astro redirect was authored unconditionally — it 302s to {...}/01/ regardless of whether that slot exists. The [slot].astro route only emits paths for slot entries present in SLOTS[variantSlug]. Proto, enhanced-v1, and enhanced-v2 have no entries (only enhanced-v3 does). Result: the redirect target 404s for those variants.

Note 2. “Play works, doesn’t have an elegant UI like calmdecks, needs more fidelity to that UI we worked on it hard.”

The Phase A+ PlayChrome is intentionally minimal — a thin top bar + bottom strip wrapping a <slot />. Calmstorm-decks has a mature /play runtime built across [[../explorations/Chroma-Parity-and-the-Path-to-a-Shared-Deck-UI-Module]] §“Mature primitives in calmstorm” with PageAsDeckWrapper, SlideCanvas, ContentFit, tier-aware MetaTags, DeckHeader, DeckNav, GateScript, and the brand-mark slot pattern. The shell shipped none of those in A+. The visual register gap is real and the founder named it specifically.

Note 3. “Play doesn’t seem to have the enum ranking UI to prioritize improvements.”

The <SlideRankPill> lives only in /scroll/. The Phase A+ plan didn’t include mounting it in /play/ — the assumption was that ranking happens during scroll-reading. But the founder uses Play to review slides at presentation fidelity; that’s exactly when “this one needs work” judgments form. The pill belongs in /play/ too.

Note 4 (implicit). “TOC is there, needs iteration.”

Unspecified scope. Captured as an open question; Phase A++ does not pre-commit to specific TOC changes. The founder iterates on /toc/ next session with concrete pointers.

Scope boundary

In scope:

• Variant URL safety. /play/[deck]/[variant]/ index route gates its redirect on SLOTS[variantSlug] existing and non-empty. Empty-SLOTS variant routes render a small “no slots yet” page that links back to /toc/{deck}/{variant}/ (which itself shows the “Create slides.ts entry” hint).
• Mount <SlideRankPill> inside /play/. The pill renders in PlayChrome’s chrome area (above the bottom strip, or as a separate corner overlay). Same component, same /api/slide-rank round-trip. The pill watches the currently-rendered slot in Play (no IntersectionObserver needed — exactly one slot is on screen at a time).
• Lift SlideCanvas + ContentFit from calmstorm into the shell. Two related primitives that together make a slide render at a fixed aspect ratio (typically 16:9) and auto-fit content to viewport with a scale transform. These are the minimum visual-fidelity lift to make /play/ feel like calmstorm-decks’s play runtime, not a developer preview.
• Refactor PlayChrome to wrap its slot content in SlideCanvas. Adds the dotted-grid backdrop, the aspect-ratio container, the content-fit scale, and the (now standard) “slide-as-presentation” feel.
• Keep the keyboard contract identical. Founder muscle-memory must transfer between Phase A+ and Phase A++ without retraining.
• Verify end-to-end in pnpm dev and pnpm build. Static output should still emit 16 slot pages; chroma’s scroll deck should be byte-stable.

Out of scope (explicit defers):

• Phase A.7 publish. Still orthogonal. Phase A++ remains in workspace-link mode. When A.7 unblocks, A+ and A++ ship together as 0.1.0.
• The rest of Phase B’s calmstorm primitive lift. PageAsDeckWrapper, tier-aware MetaTags, DeckHeader, DeckNav, GateScript, the brand-mark slot pattern, the mode-switcher behavior. Phase B territory. Phase A++ lifts just SlideCanvas + ContentFit because they are the visual-fidelity bottleneck the founder named — not the full mature primitive set.
• Calmstorm-decks’s own migration onto the shell. Phase B.
• TOC iteration. Out until the founder provides specifics. The Phase A++ session may surface a TOC sub-plan if the founder shares pointers during execution.
• Auth + telemetry + admin. Phase D.
• Per-slide content authoring beyond slots 01 + 15. Founder runs the rank → decompose → recreate loop at her own pace.
• The /play/[deck]/ deck-chooser route. The shell’s /play/index.astro redirect logic is fine for now; the deck-chooser surface lives in the consumer’s src/pages/play/index.astro (chroma already has this).

Preconditions

1. Phase A+ is shipped and verified (as of 2026-05-12 smoke-test). All deliverables above marked ✅ are working in pnpm dev.
2. apps/deck-shell/package.json is at 0.1.0-rc.0. Phase A++ will bump to 0.1.0-rc.1 at the end.
3. Founder is available for one stop-and-show check after A++.3 lands (the first SlideCanvas-wrapped Play render is the moment where “does this look like calmstorm?” is answerable).
4. Calmstorm-decks is on its development branch with the existing SlideCanvas, ContentFit, and PageAsDeckWrapper in client-sites/calmstorm-decks/src/components/ reachable from the dididecks-ai pseudomonorepo. (Don’t move; just read.)

Architecture decisions to lock before code

A. Variant URL safety — gate the redirect, render a “no slots” page if absent

Current behavior (A+):

GET /play/pitch/proto/         → 302 /play/pitch/proto/01/
GET /play/pitch/proto/01/      → 404 (slot was never emitted by getStaticPaths)

New behavior (A++):

GET /play/pitch/proto/         → 200 "no slots yet" page if SLOTS[proto] empty/undefined
GET /play/pitch/proto/         → 302 /play/pitch/proto/01/ if SLOTS[proto] has entries
GET /play/pitch/proto/01/      → 404 unchanged (the slot doesn't exist; correct)

The “no slots yet” page mirrors the existing <DecomposeFirstPlaceholder> aesthetically: shell-styled, links to /toc/{deck}/{variant}/ (which surfaces the “create slides.ts entry” affordance), explains the contract.

Rationale: the redirect lying to the user is worse than no redirect. A 404 from a static deploy is jarring; a friendly explanation page is a known-state surface. Render-time check (in index.astro frontmatter) on SLOTS[variantSlug]?.length ?? 0.

B. In-play ranking — mount the existing <SlideRankPill> inside PlayChrome

The same component, re-mounted. Differences from the scroll-deck mount:

• No IntersectionObserver needed. Exactly one slot is on screen at a time. The pill reads slot and variantSlug from PlayChrome’s known render context.
• Position. Bottom-right is taken by the play bottom-strip (slot counter, ←/TOC/→ buttons). Place the play-mode pill at top-right, or as an inline element in the bottom strip (small icon + status).
• Visibility. Hideable via C (chrome toggle) along with the rest of the chrome.
• No fetch on every navigation. Cache the audit registry once per page load; surface the current slot’s rank from cache.

Component reuse strategy: add a mode: "scroll" | "play" prop to <SlideRankPill>. The mode prop:

• scroll (default): existing behavior — IntersectionObserver-driven, fixed bottom-right
• play: takes the active slot from a prop (activeSlot: string), positions in the play-mode location

This keeps the component logic and the API surface stable; the consumer just passes mode="play" and an activeSlot.

C. Lift SlideCanvas + ContentFit — minimum-viable calmstorm primitive lift

SlideCanvas (calmstorm) provides:

• Fixed aspect-ratio container (typically 16:9)
• Optional dotted-grid backdrop
• A <ContentFit> child that auto-scales its contents to fit the canvas via CSS scale transform + ResizeObserver

ContentFit (calmstorm) provides:

• Measures children at natural size
• Computes the scale factor that fits inside the parent’s aspect-ratio box
• Applies transform: scale(N) translate(...) to center

Together they make a slide that’s authored at “natural” size (typically a 1280×720 or 1920×1080 conceptual canvas) render correctly at any viewport. This is what makes calmstorm’s /play feel like a finished presentation and not a webpage with a header.

Lifting strategy:

1. Read client-sites/calmstorm-decks/src/components/SlideCanvas.astro and ContentFit.astro (or wherever they live).
2. Copy into apps/deck-shell/src/components/SlideCanvas.astro + ContentFit.astro. Strip any calmstorm-specific brand assets / mode toggles (those are Phase B).
3. Export from @dididecks/shell/components/SlideCanvas.astro (and ContentFit.astro).
4. Refactor PlayChrome to wrap its <slot /> in <SlideCanvas><ContentFit>...</ContentFit></SlideCanvas>.

The seed slides update accordingly. 01-cover.astro and 15-ask.astro are currently authored as min-height-100vh sections. Inside SlideCanvas, the canvas controls min-height; the slide’s content should drop the min-height-100vh and let ContentFit handle sizing. Two-file edit.

Calmstorm-decks itself is NOT migrated in this plan. Calmstorm continues using its own local copies. The shell version becomes the canonical version; calmstorm migrates to the shell version in Phase B. (This is the standard “fork, stabilize, replace” lift pattern.)

D. Visual-fidelity acceptance criterion

After Phase A++ lands, the founder opens /play/pitch/enhanced-v3/01/ and /play/pitch/enhanced-v3/15/ and says “yes, this looks like a presentation” — or names a specific further gap. The plan’s success criterion is the founder confirming the visual register lifted enough — not a pixel-perfect match to calmstorm.

If after A++.3 + A++.4 the founder says “still doesn’t look like calmstorm,” the followup is Phase B’s larger primitive lift (PageAsDeckWrapper, mode-switcher, brand-mark, DeckHeader, DeckNav). That’s a separate plan, not a continuation of A++.

E. pnpm dev discipline — kill background servers between iterations

Phase A+ surfaced a workflow trap: leaving pnpm dev running across edits caused HMR caching that masked real bugs (the slot-child rendering quirk took an extra hour to diagnose because the dev server cached intermediate states). Phase A++ session discipline:

• One pnpm dev at a time. Kill before edits to shell/components.
• After major edits to the shell, hard-restart dev (kill + restart).
• Use pnpm build for final verification — caching doesn’t apply there.

Phase A++.1 — Variant URL safety

Goal: /play/{deck}/{variant}/ never redirects to a slot that doesn’t exist.

1. Refactor apps/deck-shell/src/routes/play/[deckSlug]/[variantSlug]/index.astro to read SLOTS[variantSlug] and branch:
   • If non-empty: redirect to slot 01 as today.
   • If empty/undefined: render an inline “no slots yet for this variant” panel using the shell’s nav + a small explainer + a link to the TOC.
2. Smoke-test: /play/pitch/proto/, /play/pitch/enhanced-v1/, /play/pitch/enhanced-v2/ all render the empty-state panel; /play/pitch/enhanced-v3/ still redirects to 01/.
3. Build verification: pnpm build. The three “no slots” variant indexes emit as static HTML.

Acceptance: no 404 reachable from a click anywhere on the site.

Phase A++.2 — <SlideRankPill> mode prop and play-mode mount

Goal: rank a slot from within /play/ without leaving the page.

1. Add mode prop to <SlideRankPill> with default "scroll". Implement the "play" branch:
   • Skip the IntersectionObserver setup.
   • Accept activeSlot prop (string, e.g. "01").
   • Position via top: 16px; right: 16px; (top-right) when mode is "play".
   • Otherwise same UI, same status buttons, same /api/slide-rank POST behavior.
2. Update PlayChrome to mount <SlideRankPill mode="play" deckSlug={deckSlug} variantSlug={variantSlug} activeSlot={slot} /> next to the nav (or in the chrome top-right corner).
3. Hide via C chrome toggle — already wrapped in .dididecks-play__chrome, so it toggles automatically.
4. Smoke-test: open /play/pitch/enhanced-v3/05/, the pill shows current rank (urgent-redo — already set in audits). Click a different status; navigate away and back; rank persisted.

Acceptance: pill renders in Play, ranking from Play persists into the same audits file the scroll and TOC read.

Phase A++.3 — Lift SlideCanvas + ContentFit

Goal: the shell carries calmstorm-grade slide-canvas primitives, usable by any future Play wrapper.

1. Read calmstorm’s implementations at client-sites/calmstorm-decks/src/components/SlideCanvas.astro and ContentFit.astro (or wherever they live — discover first).
2. Copy + strip into apps/deck-shell/src/components/SlideCanvas.astro and ContentFit.astro. Remove any calmstorm-specific brand / mode wiring. Keep:
   • The aspect-ratio container math.
   • The dotted-grid backdrop (theme-tokenized — var(--color-border) for dots).
   • The ContentFit ResizeObserver + scale-transform pattern.
3. Add aspectRatio prop (default "16/9") to SlideCanvas for forward-compatibility.
4. Export via @dididecks/shell/components/SlideCanvas.astro and ContentFit.astro.
5. Typecheck.

Acceptance: the two components exist, are exported, are typesafe, and render a 16:9 box around their children with the children scaled to fit.

Phase A++.4 — PlayChrome consumes SlideCanvas

Goal: /play/ renders slides at presentation fidelity, not as webpage sections.

1. Refactor PlayChrome.astro to wrap its <slot /> in <SlideCanvas><ContentFit>...</ContentFit></SlideCanvas>. The stage area becomes a fixed-aspect canvas with the slide content scaled.
2. Update seed slides (01-cover.astro, 15-ask.astro):
   • Drop min-height: 100vh from the section (canvas controls height).
   • Set explicit “natural” dimensions if ContentFit requires them (calmstorm pattern likely needs a sized inner element).
   • Verify visual fidelity in dev.
3. Update DecomposeFirstPlaceholder similarly so it lives inside the canvas too — placeholders should look like a presentation slot saying “this is empty,” not a half-page message.
4. Stop-and-show point: founder opens /play/pitch/enhanced-v3/01/ and /play/pitch/enhanced-v3/15/. Founder confirms “looks like calmstorm” or names a specific further gap.

Acceptance: Play renders with fixed aspect ratio, scaled content, calmstorm-grade visual register. Founder verifies the gap is closed (or identifies the next slice for Phase B).

Phase A++.5 — Cross-cutting verification + version bump

Goal: ship A++ cleanly with no regressions.

1. Scroll-deck byte-stability. pnpm build; compare dist/scroll/pitch/enhanced-v3/index.html size to pre-A++ baseline. Should be unchanged (no edits to scroll-side files in A++).
2. TOC unchanged. Same nav, same [view →] links, same rank pills. No regressions.
3. /play/pitch/{proto,enhanced-v1,enhanced-v2}/ all show the empty-state panel; no 404 reachable.
4. /play/pitch/enhanced-v3/{01..16}/ all render. Slot 01 + slot 15 with real content (inside SlideCanvas), slots 02–14 + 16 with placeholder (inside SlideCanvas).
5. In-play ranking persistence. Rank slot 16 via Play, navigate to TOC, see the rank reflected.
6. Bump version in apps/deck-shell/package.json from 0.1.0-rc.0 to 0.1.0-rc.1.
7. Commit shape:
   • add(deck-shell, play): variant URL safety + no-slots empty-state page (A++.1)
   • add(deck-shell, slide-rank-pill): mode=play prop + mount in PlayChrome (A++.2)
   • add(deck-shell, slide-canvas): lift SlideCanvas + ContentFit from calmstorm (A++.3)
   • update(deck-shell, play-chrome): wrap slot in SlideCanvas for presentation fidelity (A++.4)
   • update(chroma-decks, slides/enhanced-v3): drop min-height-100vh from cover + ask for canvas-wrapped render (submodule side, A++.4)
   • Submodule pointer bump in dididecks-ai.

Acceptance: build green, dev round-trips all five surfaces, founder confirms /play/ fidelity.

Stop-and-show points

1. End of A++.1 — variant URL safety. The founder clicks /play/pitch/proto/ and sees a friendly panel, not a 404. Quick check; trivial.
2. End of A++.2 — in-play ranking works. The founder ranks a slot from Play; the rank shows up in TOC. Visible signal of “this works.”
3. End of A++.4 — the big one. The founder opens /play/pitch/enhanced-v3/01/ and either says “yes, looks like calmstorm now” or names the next visual gap.

Open questions to resolve in flight

1. SlideCanvas aspect ratio default. 16:9 is the safe default for fundraise decks. But chroma’s enhanced-v3 has a long-form vertical feel — does that translate to 4:3 instead? Open: stay 16:9 default; consumer-overridable per-slot if needed (Phase B).
2. ContentFit for overflowing content. Some slides may have content taller than the natural canvas. ContentFit’s scale-transform shrinks fonts; what if shrinking makes them illegible? Open: cap minimum scale at e.g. 0.5; below that, allow scrolling within the canvas? Decide in flight.
3. Position of the in-play rank pill. Top-right vs. inline-in-bottom-strip vs. floating-overlay-mode-aware. Decide in flight; default to top-right and iterate if the founder dislikes.
4. TOC iteration scope. Captured as TBD. The founder will name specifics; A++ does not pre-commit. A separate small plan (or an A++.6 step) can fold in late if the founder shares pointers mid-session.
5. <SlideRankPill> in production builds. Phase A+ left this as “keep visible; revisit if disliked.” Still open. With the play-mode mount, the pill is now on more pages in production; the question of dev-only vs. always-on may bind sooner.

Follow-up plans this work queues

These are separate plans Phase A++ makes possible:

1. Phase B — full calmstorm primitive lift. PageAsDeckWrapper, tier-aware MetaTags, DeckHeader, DeckNav, GateScript, the mode-switcher behavior, the brand-mark slot pattern. Calmstorm-decks itself migrates onto the shell in this phase. Larger scope, requires a quiet calmstorm window. A++.3 + A++.4 are the seed of this lift.
2. Phase A.7 — publish. Still blocked on the org-name decision. A.7 unblocks independently of A++.
3. TOC iteration sub-plan. Authored once the founder shares specifics. May be a one-pager.
4. Per-slide content sprint. Founder fills in slots 02–14 + 16 at her own pace via the rank → decompose → recreate loop. Not a “plan” — it’s the framework’s iteration rhythm. The deck-iteration-workflow skill is the doc for it.
5. Componentize-Slides destination retarget. One-paragraph edit; still queued; doesn’t block.

Cross-references

• [[Phase-A-Plus-In-Deck-Ranking-Shared-Nav-and-Play-Runtime]] — Phase A+. This plan’s preconditions. All A+ deliverables are now ✅.
• [[Stand-Up-Dididecks-Shell-and-Ship-Chroma-TOC-Ranking]] — Phase A. A.1–A.6 complete; A.7 still blocked on org name.
• [[../explorations/Chroma-Parity-and-the-Path-to-a-Shared-Deck-UI-Module]] — the architecture exploration. Phase A++ pulls forward the first slice of Phase B (SlideCanvas + ContentFit) because the founder named it as the visual-fidelity bottleneck.
• [[../explorations/Plans-Inventory-and-Phase-A-Outcome]] — the gap-analysis exploration that framed A+.
• [[../specs/Dididecks-AI-Slide-Decks-as-Code]] — parent spec; NS-2’s “player so good they present live” is what SlideCanvas lift starts unlocking.
• deck-iteration-workflow skill — the rhythm A++ continues operationalizing (rank → decompose → recreate → present).
• astro-knots skill — framework prohibitions A++ inherits.
• theme-system skill — relevant for the SlideCanvas dotted-grid backdrop tokenization.

Status / next step

Status: Draft, ready for the next agent (or this one in a fresh session) to execute.

Immediate next step on approval: Begin Phase A++.1 — refactor /play/[deck]/[variant]/index.astro to gate the redirect on SLOTS[variantSlug] existing. Smallest, most contained change; ships in under 30 minutes.

Total estimated effort: A++.1 + A++.2 are small (under an hour each). A++.3 is the calmstorm-read + copy-and-strip — half a day if SlideCanvas and ContentFit are well-isolated in calmstorm, longer if they have deep dependencies that also need to come over. A++.4 is the integration — half a day with the founder’s eye on visual fidelity. A++.5 is verification — under an hour. Total: roughly one focused day of agent-augmented work.

Required from the user during execution: (a) confirm the no-slots empty-state copy after A++.1; (b) confirm the in-play pill position after A++.2; (c) the big one — confirm the visual fidelity after A++.4. The founder is the only one who can sign off on “this looks like calmstorm now.”

Remaining work (as of 2026-05-16)

This plan is partially shipped. What’s done and what’s left:

Shipped

• A++.3 — SlideCanvas lift (partial). apps/deck-shell/src/components/SlideCanvas.astro exists (pure-CSS 16:9 wrapper using container queries). Composed by /play/[slot] route today and slot files self-wrap inside it.
• A++.4 — /play/[slot] consumes SlideCanvas. Wired and rendering for chroma’s three playable variants (proto, enhanced-v2, enhanced-v3).
• A++.5 — version + verification. Shell bumped to 0.1.0-rc.0; pnpm build green; chroma builds + Vercel-previews cleanly.

Not yet shipped

• A++.1 — variant URL safety. /play/[deck]/[variant]/ index.astro still redirects unconditionally. Variants without per-slide files still 404 after the redirect. Side-effect mitigation: the new /play/ chooser (chroma-side, 2026-05-16) AND-intersects file × SLOTS so it can no longer link to those routes — but typed URLs still 404.
• A++.2 — in-play classifier mount. SlideRankPill is reachable in Play-UI structurally via <DeckOverlay--Play-UI> (new, 2026-05-16), but /play/[slot].astro has not been refactored to consume the overlay. So the pill renders in Scroll-UI only today.
• A++.3 — ContentFit lift (the other half). SlideCanvas shipped without its calmstorm sibling ContentFit. Slides whose natural size exceeds the canvas don’t scale-fit; they clip. Lift ContentFit.astro from calmstorm into the shell and wire SlideCanvas to compose it.

Other notes

• The SlideRankPill position prop (bottom-right | top-right) added 2026-05-15 was the structural prerequisite for A++.2. The remaining work is the route refactor.
• The DeckOverlay components landed 2026-05-15 expose the right shape for A++.2 (composes DeckChrome + SlideRankPill with section-wrap adapter); see [[../sitemap/components/DeckOverlay—Play-UI]].