M3 § Chips, Accessibility asks that a row which overflows horizontally carry a
visible affordance; `<x-chip-set scroll>` faded its edges and left it there
(plan step 24, audit docs/audits/m3-alignment/inputs.md § Missing). Where the
pointer is fine, and so there is no swipe to reach for, a button now sits over
each fading edge and scrolls the row by most of its width. They are pointer
affordances only — no tab stops, since the arrow keys already walk every chip —
and the row's scroll padding grows on a fine pointer so a chip the keyboard
reaches clears the buttons as well as the fade. The scroller became a ref,
because the buttons stand outside it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 § Search names three entry points and only the bar existed; suggestions
before the first keystroke were missing too (plan step 24, audit
docs/audits/m3-alignment/inputs.md § Missing). `trigger="icon"` is the search
icon button — search as a secondary action, one 48px button that expands into
the full-screen view at any width, since an icon button has nothing to dock
under, and takes its focus back on close. The `suggestions` slot stands where
the results do until something is typed; the live region counts whichever of
the two is on screen and names suggestions as suggestions.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 § Date Pickers, Anatomy describes a 14-element full-screen range picker for
compact breakpoints; `<x-datepicker range>` always used the 360px modal dialog,
which on a phone is cramped for a two-month scroll (plan step 24, audit
docs/audits/m3-alignment/inputs.md § Missing). A third presentation, "full",
joins docked and modal: the same dialog grown to the screen, an app bar with a
close button and Save, the supporting text and the range as the headline over a
divider, the weekday labels held at the top, and the months in one vertically
scrolling list, each under its own label. The grid now draws from one `rows`
getter, which is this month's weeks everywhere else, so the docked and
single-date pickers render exactly as before. The list is a window of months
grown by the scroll and by the keyboard, since nothing here is lazy.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 Expressive's configuration table lists a vertical orientation the library
had no prop for (plan step 24, audit docs/audits/m3-alignment/inputs.md
§ Missing). The drawing is the horizontal one, laid out as long as the slider
is tall inside a size container and turned a quarter anticlockwise, so the
geometry, the sizes and the tokens are untouched; the value label and the inset
icon are turned back so they read across. The keyboard stays the native range's
— Up and Right raise, Down and Left lower — and the inputs say
aria-orientation. A vertical slider takes its length from its wrapper, so the
header, the SKILL.md entry and the showcase all say to give it a height. M3
keeps range sliders horizontal, so `range` wins over `orientation`.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 § Checkbox, Behaviour asks that from expanded (840px) related checkboxes be
gathered into a contained region rather than left as one long column. That is a
rule about the page around the control, not about the control, so it is written
down rather than built: the component's header and its SKILL.md entry now say
it, and the showcase lays a group out in an `expanded:grid-cols-2` card (plan
step 24, audit docs/audits/m3-alignment/inputs.md § Missing).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
The paragraphs restate the material-3 guideline's rules beside the tokens they
govern (plan step 28); the README's AI agents section names the second guideline
and skill (step 30).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 § Text Fields, Behaviour: compact may let a field span full width, but
medium and expanded must bound it — "never let it span the full width of a
large screen". Nothing in the package capped a field (plan step 24, audit
docs/audits/m3-alignment/inputs.md § Missing). From `medium` a field now stops
at 40rem — the site names no number, so the header says where this one comes
from — which a `max-w-*` class beats and `full` takes off. The search bar
already carries M3's own 720px.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 lists "supporting text + character counter" as a configuration of both
text-field variants; nothing in the package drew one (plan step 24, audit
docs/audits/m3-alignment/inputs.md § Missing). `counter` on `<x-input>` and
`<x-textarea>` now puts `n/max` at the end of the supporting-text row, counted
from the control on every input, in the error colour past the maximum, and said
as M3's own "Character count, 5/20" from a polite region once typing settles.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
The scrim is --md-sys-color-scrim, black in every scheme, theme and contrast level,
so the ink over it is the absolute the theme keeps rather than inverse-on-surface,
which is dark in a dark scheme (containment C-25, revisited).
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
The actions stream has since given the small extended FAB an 80px minimum width,
which the morph would have kept once the label closed. Dropping that minimum and
squaring the box against the height its size already sets gives back the 56, 80
or 96px FAB without this file naming any of those numbers.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-23).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-11. The full-screen layout was another
horizontal row with a 28px corner and a mask, where M3's "shows one
edge-to-edge large item at a time and scrolls vertically". It is now a
vertical scroll-snap column: items fill the row with no corner and no
mask, 16dp apart, no end padding, the previous/next buttons point up and
down and the arrow keys are Up and Down. The row is capped at the 840px
medium window, which with the portrait rule in the header comment is as
far as CSS can hold M3's "compact and medium, portrait only". The ported
FullScreenCarouselStrategy keylines go with it: nothing measures them
any more.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 asks the rail to animate its FAB into an extended FAB as it expands; the
documented pattern was two FABs swapped by `display`, so the collapsed one
appeared at once while the rail's width was still springing. The header now takes
one <x-fab label icon> whose label springs open and shut on the rail's own
spring — and names it at both widths, where the collapsed FAB had needed a
tooltip to have a name at all.
M3 also lists two container treatments as optional that the library had neither
of: `divider` draws the vertical divider on the page's edge, which is its own
answer to a page scrolling under a fixed rail, and `fill="false"` drops the
container colour. Neither reaches a rail open over a scrim. And a `collapsible`
rail, whose width is the visitor's choice rather than the window's, is held to
the collapsed 96px below `medium`, where M3 asks for a navigation bar instead.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-22, N-23, N-24).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-18. The row was the focusable `region`
and the items were not focusable at all, which is the thing M3's
accessibility page draws a Don't for: "use Tab to place initial focus on
the first carousel item", "avoid focusing on the carousel container".
Each item is now `tabindex="0"` with the focus ring drawn inside it, the
row is out of the tab order, and from a focused item the arrows move one
item (moving focus with them), Home and End go to the ends, and Space or
Enter opens one that is not fully in view.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
A plain tooltip vanished the instant the pointer left, where M3 keeps one for
1.5s and shows only one at a time — these are popover="manual", so nothing was
closing the last one. A rich tooltip's bubble was never associated with its
trigger at all, so the explanation it exists to give was never announced; the
script now writes aria-describedby (and aria-haspopup/aria-expanded when it is
persistent) onto the focusable control in the trigger slot, and again after a
morph. Plan step 18, actions.md ACT-22, ACT-25.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Shift+M and Shift+Y reach the month and year dropdowns, which is M3's
date picker keyboard table; Home and End go to the ends of an open
combobox list, and stay the field's own caret keys when it is closed. The
password reveal drops aria-pressed and keeps only its flipping label, so
a screen reader hears the state once rather than twice and inverted. The
radio's `inline` now says that M3 cautions against a row and wants an
option chosen on load.
Plan step 20, findings IN-22, IN-23, IN-28 and IN-29.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
The trigger had the focus ring but no hover and no pressed feedback at all,
where every comparable trigger in the library carries `state-layer` and M3 asks
for two visual indicators per state. An avatar image sits behind the layer, so
the wash shows over a photograph too.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-15).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-12 and C-25. `padding` defaulted to 0
where M3's specs table gives every layout 16dp of leading and trailing
padding — leading only for `uncontained`, none for the edge-to-edge
`full-screen` — so the default is 16 and each layout now says which
ends it applies to (`data-padding`, `data-padding-end`). The row gains
the 8dp above and below that goes with it, without shortening the
items. The overlay label is `inverse-on-surface` instead of a literal
`white`, which no scheme or contrast profile follows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
A floating dot sat 2px inside the icon's top-trailing corner where M3 puts it
flush, and a floating count 4px above it where M3 says 2px. An `outline` status
label drew its edge in outline-variant, the decorative role dividers use, which
is about 1.5:1 on surface — M3 asks a badge for 3:1, so it takes `outline`.
Plan step 18, actions.md ACT-23, ACT-24.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3's scaffold is bars, then rails, then panes, and the shell had no slot that
rendered above the rail: an application-wide bar could only be a pane bar beside
it. `banner` is that slot — the shell is a column now, with the rail and the page
as one row inside it — and the rail sticks under a pinned banner through
--material-banner rather than behind it. `top` still means the page's own bar.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-14).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
An error or warning alert was role="alert", an assertive live region — but the
usual alert is part of the page as it renders, which some screen readers then
announce over the page title, and a Livewire morph re-announces. It is now
role="status" whatever its colour, with an `assertive` prop for a notice put on
screen in answer to something the person just did. Its 40px dismiss button takes
the shared touch-target. Plan step 18, actions.md ACT-18, ACT-36.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-27 and C-28. `.inner-body` was 24px, a
value not on M3's 0/4/8/12/16/20/28/32/48 scale; it is 28px, the
extra-large corner of the dialog and bottom-sheet surface it stands in
for. `.header a` is title-large at weight 400, which its own section
comment and the `h2` under it already said.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 wants two visual indicators per state and names the pair for a text
field: the error colours and a trailing error icon. The field now draws
one, labelled "Error", unless the caller already trails it or the field
is xs. A read-only field loses its dashed outline: M3 keeps "the same
visual style as an editable field", and the native readonly attribute is
what tells a screen reader. The data-readonly hook stays.
Plan step 20, findings IN-18 and IN-19.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 says to avoid an icon in a snackbar and gives the supporting text no fourth
colour, so the state glyph and the 80% description are gone: a type now only
picks the announcement role. The 40px action and close buttons take the shared
touch-target, Escape dismisses a snackbar that holds the focus, and the host
publishes --material-snackbar-height on <html> so a `fab` button sits above the
snackbar instead of under it. Plan step 18, actions.md ACT-17, ACT-18, ACT-20,
ACT-34, ACT-35.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-24. `interpolate-size: allow-keywords`
was set and nothing transitioned, so the content snapped open. A new
resources/css/components/collapse.css transitions `block-size` on
`::details-content` over the fast spatial spring, with
`content-visibility` `allow-discrete` beside it so the section stays
rendered while it closes. The `<details>` carries `data-collapse` for
the rule to find.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
NavigationBarTokens names one label font for both icon positions, label-medium;
the horizontal layout had taken label-large from the rail's item, a different
component with a token of its own. And the item's state layer washed in
on-surface over a secondary-container pill, where the only state-layer tokens M3
states for a navigation item — the rail's, which the rail already follows — are
on-secondary-container for all six states.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-08, N-19).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Density was applied to every fine-pointer device with no way out, which
M3 forbids outright: "don't apply density by default; offer an explicit
density opt-in instead, keeping opt-out targets at >= 48x48 CSS px". A
row is now 52px and `dense` is a prop that tightens it to 36. The row
divider also takes outline-variant itself rather than 60% of it, since M3
keeps opacity for state layers and disabled.
Plan step 20, findings IN-16 and IN-30.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-10. The action row was right-aligned with
24px above it; the side-sheet specs table puts bottom actions on the
left in a 72dp row with 16dp above and 24dp below. A dialog's actions
stay trailing-aligned — M3 specifies the two differently.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 asks a vertical toolbar for a 24dp margin from the window edge where a
horizontal one keeps 16, and its standard-toolbar colour list ends "Standard
button (Primary)" — the vibrant row of the same list was already honoured, so
the two were being read inconsistently. Both rules are unlayered, because an
icon button's ink is a utility.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-12, N-13).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-19 and C-09. A modal sheet opened at 90%
of the viewport where M3 caps the initial position at 50% and lets the
person pull it up; `height` now defaults to `50dvh` and whatever is
passed is held under a ceiling of the screen less M3's 72dp top margin.
The drag handle is `on-surface-variant` undiluted, as
SheetBottomTokens' DockedDragHandleColor and the component's own header
comment already said.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
The centred headline was absolutely positioned 56px in from each edge — room for
exactly one of M3's two trailing icon buttons, so a long title ellipsised behind
them. It is now the middle column of a three-column row and shrinks against
whatever is really beside it. The search container follows M3's curve, filling
the row to 312dp and then only half of what is left, so the showcase drops the
hand-written max-width it had been working around it with.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-09, N-11).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Its trigger holds two aria-hidden icons and wrote aria-label only when a caller
passed `label`, so an unnamed button was the default; `label` now falls back to
"Toggle menu" and is always written. The list had no ceiling, so six items ran
off a landscape phone — it now scrolls under the FAB, which stays put — and its
items rose into place but vanished on close, because the popover took its
display away before they could move. Plan step 18, actions.md ACT-21, ACT-32,
ACT-33.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-13, C-23 and the directly actionable card
of C-13/C-18. A card had one elevation and never changed it, and its
hover replaced the elevated card's shadow with the state layer; it now
names its variant in `data-card` and moves a step on hover (elevated
1 to 2, filled and outlined 0 to 1) with the layer over it. The 12 to 16
corner morph on hover is gone: M3 gives a card one shape and lists shape
morph for buttons, FABs and list items only. New `actionable` prop for
M3's directly actionable card — the card is the one tab stop, with
`role="button"` (or `link`), its `title` as the name, and Enter or Space
reaching the opener from resources/js/list-rows.js.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Both were 40px or less, which is M3's state-layer size, not its 48px
target: the paginator's steps and page numbers keep their 40px circle and
now catch presses over 48, and the sort button — a title-small line and a
16px arrow, about 20px tall — does the same, since the cell's padding
belongs to the cell and not to the button inside it.
Plan step 20, findings IN-15 and IN-17.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Five or more sections wrapped onto a 3- or 4-column grid, which left the tab
bar's divider under the last row only and stranded every upper row's active
indicator against nothing. From five they are now what M3 prescribes and its
accessibility page defends: a scrollable tab bar, each tab as wide as its label,
offset 52dp. Four or fewer still share the row. The 20px icons take the 20 cut.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-16).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3's chip keyboard table makes a set one tab stop that the arrows move
through; ten filter chips cost ten Tab presses today. The set now keeps a
roving tabindex — re-applied whenever a Livewire morph rewrites the chips
— and walks left, right, Home and End, following the writing direction.
The input chip's remove button also grows to a full 48x48 target, where
it was 34px wide.
Plan step 20, findings IN-13 and IN-14.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3 is explicit twice over — never disable a FAB, remove it instead — and the
prop painted the full disabled treatment with nothing warning against it. It is
gone; hiding the FAB is the documented answer. The extended FAB's icon-to-label
gap goes to M3's 16dp at md and 20dp at lg, the small one takes the baseline's
80dp minimum width, and the root gains a `data-fab` hook for a place that draws
a nested FAB its own way. Plan step 18, actions.md ACT-08, ACT-16, ACT-31.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
A scrollable set now starts 52dp in, so it reads as scrollable; a primary
indicator is inset 2dp at each side while a secondary one still spans the tab;
a link marked aria-current takes the active label colour the indicator already
followed; and the focus ring is 2px outside the tab, as everywhere else in the
package, paid for with 5px of room around the bar that the state layer and the
indicator reach back into.
Plan: docs/plans/material-3-alignment.md, step 21 (navigation N-10, N-18,
N-20, N-21).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-14, C-15 and C-26. The leading and
trailing gap was 12px where ListItem.kt's LeadingContentEndPadding and
TrailingContentStartPadding are 16dp; a three-line item was
middle-aligned and padded 10px where M3 top-aligns an 88dp item and
Compose pads it 12; and a segmented list's icons stayed at the baseline
24dp instead of ListTokens' expressive 20dp, which the item now reads
from its list with `@aware` and draws from the 20 optical cut.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Items were 44px tall with the baseline menu's 12px sides, the separator had half
its 8dp padding, a disabled item was skipped by the keyboard entirely, a chosen
item was told by colour and shape alone, and the list simply faded in place. Now:
48px rows with 16px either side, an 8px separator, disabled items focusable but
never activatable, a trailing check on a selected item, and an enter-exit that
scales the list out of the corner nearest its trigger — the scale on the spatial
spring, the fade on the effects one. Plan step 18, actions.md ACT-11, ACT-12,
ACT-13, ACT-26, ACT-27, ACT-28.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 19, containment.md C-08, C-21, C-22 and C-16. The segmented
fill was declared unlayered, so it beat every state layer in
@layer components and a segmented row lit up for nothing: it now sits in
the layer, `:where()`d to weigh nothing, ahead of the state rules. Its
tone is `surface`, which is ListTokens' ItemSegmentedContainerColor, and
a keyboard-focused row morphs to the large corner like a pressed one.
`<x-list dividers>` draws ListTokens' 16dp-inset rule instead of a
full-bleed `divide-y`.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
