The notebook layout is now 100% responsive. The shop can hold twice the upgrades and sorts itself by what you can actually afford. Both came out of pre-launch stress-testing for the L5–L9 upgrade pack.
LANDSCAPE PIVOT COMPLETE
All five root-level Controls (Main, DeskBG, NotebookUI, FailureOverlay, EventBanner) now use Anchors + Full Rect, and the Spine between the two pages uses FILL+EXPAND to stretch the full HBox height. The legacy _apply_layout_for_resolution() function is gone — the scene tree now self-lays-out at any resolution.
A custom audit script reports 0 free-standing Position nodes across 15 .tscn files. Everything that needs to be responsive is on Anchors; everything that's a child of a Container stays Position (which is correct, because the Container controls its children). The most common source of false positives in naive layout audits — flagging Container children as broken — is filtered out by the script.
The visual change: nothing. The functional change: the layout no longer breaks if the user resizes the window or plays at a non-1280×720 resolution. This was the last bit of technical debt from the portrait era.
SHOP DENSITY: SCROLL + AFFORDABILITY SORT
The shop used to show 3 random upgrades from your unlocked pool. That works for the early game (L1–L4) where you have ~20 upgrades total, but with the L5–L9 growth pack incoming (40+ more upgrades), the random pick started to feel stingy. You're scrolling through screams, and the upgrade you actually want to buy is locked behind a random pick that doesn't include it.
Two changes:
- 3 → 6 displayed upgrades per refresh. Same random-pick from the unlocked pool (so it's still a fresh shop each refresh), just twice the surface.
- Sort by affordability. Affordable upgrades (cost ≤ your current screams) float to the top, unaffordable/locked sit below. Stable sort preserves the random-pick order within each bucket so the shop still feels intentional.
The container is now a ScrollContainer with an inner VBoxContainer, so the 6 cards scroll naturally when the panel is smaller than their combined height. Horizontal scroll is disabled — only vertical, and only when needed (auto mode).
Why scroll, not pagination or categories: this is the standard pattern from Cookie Clicker, Adventure Capitalist, and Egg Inc — same trade-off they made. For an idle clicker where momentum matters, scroll is the right call. Pagination breaks the flow (Next/Prev on every refresh feels clunky) and category tabs add UI complexity for a problem that "show me what I can afford right now" already solves.
Shop with affordability sort — affordable upgrades (Hockey Mask, Rusty Knife) float to the top, more expensive ones (Leather Gloves, Piano Wire, Bear Trap) sit below.
TOOLING
One thing that came out of this session is a small reusable asset: a godot-ux-ui-master skill that encodes the responsive layout lessons (anchors vs containers vs position, size flags, common mistakes) for future agents and projects. It includes a scripts/audit_layout.js Node tool that scans a project and reports free-standing Position nodes vs in-container ones, with the right filtering for Container children.
Even if I never touch another .tscn file, the audit script is a useful safety net: node scripts/audit_layout.js <project-root> tells you in 2 seconds whether your responsive layout is actually responsive.
WHAT'S NEXT
Growth workstream is next: the L5–L9 upgrade pack (40+ new upgrades themed by level — Suburbia, Hospital, etc.), AchievementData expansion (8 → 50+), and the Prestige skill tree expansion (7 → 30+ nodes across 4 branches). The shop density work was the prerequisite for all of that — without the ScrollContainer, adding 40 more upgrade cards would have broken the right page entirely.
Steam app registration is the long-pole. Once the build is closer to launch-ready I'll do the Steam page and announcement cycle.