[design-qa][P1] Spacing and radii drift between peers

[cestercian]

Part of the Sep-2026 designer's-eye QA of the 138 "Artistant iOS Light" screens (theme T17, category spacing; epic #161). Full report with evidence and per-screen coverage: docs/DESIGN_QA_2026-09.md; the per-section appendices under docs/design-qa/ carry the code excerpts.

Why it matters. Peer surface3 cards use four radii (16/18/20/24); insets drift between twin screens; two pinned bars have different tailroom; spacers stack on spacedBy; the tab bar ships 15–39 dp over §2's 88; provider rows are 52 beside a 54 CTA.

Fix direction. One radius per object class, one inset per relationship, one tailroom, and a tab-bar height that matches its own KDoc.

Findings (11: P1 ×1, P2 ×1, P3 ×9)

• F-GS-13 (P3) — Provider rows are 52 dp beside a 54 dp CTA
  • Screens: 12
  • Where: feature/signup/SignupAuthScreen.kt:322 (dimens.size.ctaTall = 52, Dimens.kt:187); designsystem/component/PrimaryButton.kt:86 (dimens.component.cta = 54, Dimens.kt:223)
  • Fix: .height(dimens.component.cta).
• F-GS-21 (P3) — Peer surface3 cards use four radii (16 / 18 / 20 / 24)
  • Screens: 118, 27, 13, 30, 11
  • Where: feature/signup/WelcomeScreen.kt:140; feature/signup/CommunityCommitmentScreen.kt:112; feature/signup/NotifPermissionScreen.kt:139; feature/signup/DoneScreen.kt:134; feature/signup/RoleScreen.kt:159
  • Fix: radii.lg on all five (keep xl on 11 only if the markup measures 24).
• F-DS-17 (P3) — Peer insets drift: header top and list tailroom
  • Screens: 02, 03, 14, 32
  • Where: feature/discover/DiscoverScreen.kt:116-118 (top = space.sm), :195-198 (bottom listTailroom = 56); feature/search/SearchScreen.kt:138-140 (top = space.md), :381-386, :580-583 (bottom contentTailroom + listTailroom = 16 + 56)
  • Fix: One header-top token for tab roots; tab roots share one tailroom expression; pushed lists take listTailroom alone.
• F-DS-25 (P3) — The result card is surface3 on page with no hairline; the only boxed card in the section
  • Screens: 03
  • Where: feature/search/SearchScreen.kt:666-672
  • Fix: surface fill with a 1dp hairline border, or the Artist-list row anatomy (hairline-separated rows, no box).
• F-AP-22 (P3) — Spacers stack with spacedBy, and the top of 04 stacks up to three notices with double padding
  • Screens: 04, 100, 103, 50
  • Where: feature/artist/ArtistProfileScreen.kt:381, :391, :406, :417, :463, :499, :1170, :1186; feature/score/ScoreExplainerScreen.kt:212, :234
  • Fix: contentPadding-style top/bottom on the column, no per-child top padding, and one notice slot at the top (self note or failure, the report banner replacing it while active).
• F-BC-04 (P2) — Two pinned bars with different tailroom; the funnel's dock makes every caller re-pad its actions
  • Screens: 05, 06, 07, 94, 132, 17
  • Where: feature/booking/BookingChrome.kt:171, :186-194 (CtaBar) vs designsystem/component/BottomActionBar.kt:46-53; designsystem/theme/Dimens.kt:197; callers BookingScreen.kt:260, ConfirmedScreen.kt:200, :235, MatchConfirmedScreen.kt
  • Fix: Make CtaBar a thin wrapper over BottomActionBar (one tailroom token, spacedBy(md)) and remove the per-caller padding(top = …).
• F-BN-19 (P3) — Four radii for three classes of object on one page
  • Screens: 18, 96, 52, 20
  • Where: feature/booking/BookingDetailScreen.kt:546 (card, radii.card 20), :783 and :1271 (cards, radii.buttonLg 16), :1410 and :1428 (buttons, radii.md 12), :1206 (CTA, radii.buttonLg 16); feature/bookings/BookingsScreen.kt:390
  • Fix: cards on radii.lg (18), all buttons on radii.buttonLg (16), and stop using buttonLg for cards.
• F-AS-14 (P3) — Manage availability pads by 24 and 16; the rest of the section pads by the 20 gutter
  • Screens: 22, 105, 106
  • Where: ManageAvailabilityScreen.kt:105, :187, :300 (space.xl = 24) and :138, :183 (space.lg = 16); Dimens.kt:12, :11, :221
  • Fix: dimens.component.gutter on the prose blocks and the dock.
• F-AS-20 (P3) — "N waiting" touches the section header it hangs from
  • Screens: 09
  • Where: ArtistHomeScreen.kt:246-257; designsystem/component/SectionHeader.kt:37-40
  • Fix: add a subtitle: String? slot to SectionHeader and pass the count through it.
• F-SH-22 (P3) — Two pushed SH screens, two tailrooms under the last row
  • Screens: 63, 123
  • Where: feature/system/ActivityScreen.kt:196 (chrome.contentTailroom = 16dp) vs feature/system/HelpCentreScreen.kt:133 (size.listTailroom = 56dp)
  • Fix: one token for "last row to bottom edge on a bar-less pushed screen"; use Spacer(Modifier.height(...)).
• F-DSYS-05 (P1) — The tab bar's own arithmetic contradicts its KDoc by 24dp; it ships 15–39dp over §2's 88
  • Screens: —
  • Where: designsystem/component/LightTabBar.kt:98-104 (the claim), :190 (the cell), :287-291 (the measurement); tokens at theme/Dimens.kt:406,408,412
  • Fix: pick one — either cut barTopPad/barBottomPad so hairline + pads + 48 + inset lands on the design's band, or keep the 48 cell and correct the KDoc so nobody re-derives the wrong number. Do not leave the file asserting two different heights.

Acceptance

• Every finding above is closed on the cited lines (or refuted in a comment with the reason).
• ./gradlew :app:assembleDebug, :app:testDebugUnitTest and :app:lintDevDebug green.
• The screens listed above walked on the artistant AVD with the debug harness (RELEASE.md §10) and the fix confirmed visually — this audit was code-grounded, not device-walked.