An over-engineering audit of the whole tree, applied in five reviewed
batches. Behaviour stays the same except where UPGRADE.md says otherwise.
PHP: the showcase and error-page stylesheets are prebuilt into
resources/dist by bin/stylesheets.mjs, through Vite's own postcss-import
(first occurrence kept, the order an application's build gives), instead
of Stylesheets::bundle() inlining imports on every request; only the
import walk DesignGuard needs stays. SchemeStylesheet::withProfiles()
replaces three copies of the scheme-plus-profiles loop, material:scheme
leaves spec and contrast checks to the node script that already made
them, and the error page's scheme cache, the hashed view namespace, the
translations path with no lang/ folder and DesignGuard's 1.x-name hints
are gone.
JS: the androidx shape port progress.js and both bin scripts each carried
lives once in resources/js/shapes.js (the generated SVGs are unchanged);
util.js holds ringIndex(), ms(), reopenGuard() and remember(), which
were written out several times; listeners are released through
AbortController; tooltip.js's hoverPopover() serves the rich tooltip too.
CSS: every rule for an element inside the navigation rail queries
`--md-navigation-rail-value` instead of repeating the seven collapsed
conditions under five media branches; badge, alert, progress, slider and
button read one non-inheriting colour-role table (components/color.css);
the dialog chrome, the submenu's popover chrome, the chip's state layer
and touch target, and the visually-hidden inputs use the shared rules
they copied; foundation/tokens.css is folded into foundation.css.
Views: Support\Field and Support\Link replace the error-key, bound-value
and link-attribute blocks copied into the fields and link components;
the timepicker period group, the menu filter and the showcase head are
partials; the datepicker's steppers and entry fields are loops; component
docblocks no longer restate SKILL.md.
Tests and tooling: one dataset-driven ComponentStylesheetsTest replaces
four per-group files, DesignGuardTest and the layout-component tests use
datasets, browser tests share one ready() helper, CSS parsing lives in
ComponentStylesheet alone. docs/audits and the finding IDs citing it are
removed, as are pestphp/pest-plugin-laravel, the unused composer scripts
and check:font; the lint job runs in the feature job, which now installs
node packages so the prebuilt-stylesheet staleness test runs in CI.
Feature suite 1177 passed, Chrome browser suite 299 passed.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`data-md-list-row` on an application's own `<li>`, `<div>`, `<tr>` or
`<x-row>` draws its hover, focus and press state layer, its cursor and
its `data-md-selected` fill from list-item.css, but missingStylesheets()
and unusedStylesheets() read only component tags and `->links()`.
ReStride had the stylesheet only because table.css imports it: had its
last `<x-table>` gone, every hand-made row would have lost its states
and its fill without a finding, and importing list-item.css for the
rows directly was reported unused.
A Blade view outside the package's own `resources/views` that writes a
hook in the new `HOOK_STYLESHEETS` now needs that hook's stylesheet in
both checks, named at the earliest line that needs it. The one entry is
`data-md-list-row` → `components/list-item.css`, except on `<x-card>`
(any spelling), whose row card.css draws; the card's opening tag is
blanked out before the hook is read, and a selector such as
`[data-md-list-row]` or a name in backticks writes no row. Nothing else
list-rows.js, the skill or the README has an application write needs a
stylesheet of its own: `data-md-list-open` and a row's
`data-md-selected` are drawn only through the row, a selected row in
`<x-table>` by table.css, `data-md-dragged` by foundation.css's
`md-state-layer` or card.css, `data-md-field-control` by field.css
inside `<x-field>`. The constant's docblock says so, and the guard's
docblocks, the README, the skill and UPGRADE.md describe the check.
A fixture test reads a hand-made `<li>` and `<tr>` row and a card-only
view (plain and namespaced `<x-card data-md-list-row>`, and a selector
in a script) against both checks; it fails without the change. A
second test scans the package's list-item and card views and expects
no row finding; it fails when package views are read.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A modal side sheet, bottom sheet and the modal rail move the focus to
their first control as they open (x-trap), and `<x-modal>` does in
`showModal()`. On a page loaded with the sheet already open (a
`wire:model` set from `?workout=` in the URL) the browsers count that
focus as `:focus-visible`, and so they do for a sheet or dialog opened
from the keyboard, so tooltip.js showed the close button's "Close"
tooltip the moment the layer appeared, over its title, with no keyboard
use. Confirmed in Chrome, Firefox and Safari for the deep-linked sheet
and a full-screen dialog opened with Enter; a deep-linked `<x-modal>`'s
own focus was not counted as keyboard focus. M3 lets a tooltip appear
on hover or focus and says nothing that asks for one on a focus the
component moves itself.
layers.js gains `openingFocus(event)`: a focus is a modal layer's own
when the nearest open dialog, `aria-modal` panel or modal rail around
its target is entered from outside it, or from nowhere, and no Tab is
moving it (a Tab moves the focus while its keydown is handled, by the
browser or by focus-trap wrapping round the panel). tooltip.js and
rich-tooltip.js leave such a focus out; focus moved within the layer,
a Tab onto the control and a focus returned to it from a menu inside
still show the tooltip. The docblocks, the skill and UPGRADE.md say so.
A browser test loads a page with a modal sheet open at 393px, opens a
full-screen dialog and a dialog whose first control is a rich tooltip's
trigger from the keyboard, and checks that no tooltip is up on the
focused control, then that Tab wrapping round to the close button shows
its tooltip; it fails without the change in Chrome, Firefox and Safari,
each part on its own.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The searchable choice put its `anchor-name` on a `<div>` around its
field, and menu.css sizes the popover list with
`anchor-size(width)`. The wrapper fills its container, but from
`medium` the field inside it stops at 40rem, so in a wider pane the
list ran past the field's end across the whole container: 1168px under
a 640px field. Found while giving the choice `full`, which only hid it.
The field itself carries the anchor now and the wrapper is gone, so the
list is as wide as the field, bounded or `full`, and still hangs under
the field's supporting text as before.
A browser test opens a bounded and a `full` searchable choice in a
1200px column and compares each list's width and start with its field;
it fails without the change in Chrome, Firefox and Safari.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`full` takes the 40rem bound off `[data-md-field]` from `medium`, and
`<x-input>`, `<x-password>`, `<x-textarea>`, `<x-select>` and `<x-file>`
pass it to the field they draw. `<x-choices searchable>`,
`<x-datepicker>` and `<x-timepicker>` draw the same field but did not
declare the prop, and their roots only forward `class` and `style`, so
`full` reached no element and the field stopped at 40rem: ReStride's
time zone choice stood narrower than the rows filling its card.
The three now declare `full` and pass it to their field (the date
picker's own field, not the dialog's entry fields). `<x-choices>`' chips
are not a field and have no bound, so they take `full` and render
nothing for it, and a call site can switch `searchable` on and off
without it. The docblocks and the skill list every component that takes
it.
The browser test that bounds a field at 600px now measures a bounded
and a `full` searchable choice, date picker and time picker too; it
fails without the change in Chrome, Firefox and Safari. A feature test
renders `full` on every field-shaped component and none on the chips.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`<x-button fab>` is an extended FAB below `medium` and a filled button
from there, but a disabled one was still drawn as the FAB below
`medium`: the compact FAB's rule set no state of its own, so the button
stood greyed out and fixed over the content. M3 never disables a FAB:
"if its action is unavailable, remove the FAB entirely".
A `fab` given `disabled` (a `<button disabled>`, or a link's
`aria-disabled="true"`) now renders `data-md-unavailable`, and
button.css draws it `display: none` below `medium`, which also takes it
out of the accessibility tree and the Tab order; from `medium` it is
the disabled filled button it was. The mark comes from the prop, not
from `:disabled`, because a `spinner` puts `disabled` on its button
while the action runs, and that FAB is busy rather than unavailable:
hidden, it would vanish instead of showing its loading indicator. The
docblocks and the skill's button table say so.
A browser test renders an enabled fab with a slow spinner action, a
disabled one and a disabled link on a 393px window: the disabled two
are not rendered, the busy one stays on screen with its indicator, and
at 600px both disabled ones are the disabled button again; it fails
without the change in Chrome, Firefox and Safari (and in Chrome with
`:disabled` alone as the condition, on the busy FAB). A feature test
reads the mark and the rule.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`[data-md-toggle] { min-width: 0 }` let the switch's root shrink below
its 52px track as a flex item. In `<x-row justify="between">` beside a
paragraph, flex shrinking weighs each item by its width, so the root of
a bare switch gave up a share of its 52px, and a switch with a long
label of its own gave its label a column one word wide and less; the
track, which does not shrink inside the root, ran out of it and past a
card's edge on a phone. `<x-checkbox>`'s root did the same with its box.
Both roots now keep their automatic minimum, the control and, beside a
label, its longest word, so the text beside the control, or the
control's own label, wraps instead. The toggle's docblock and the skill
say so.
A browser test lays a bare switch, a labelled switch and a bare
checkbox beside long text in rows on a 393px window and checks each
control's size and that it stays inside its row, and that the switch's
own label wraps inside its root; it fails without the change in Chrome,
Firefox and Safari.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Where the browser has the customizable select, select.css made the
select `display: flex` with `text-overflow: ellipsis`, but the closed
select draws its value in a block of its user-agent shadow tree: a flex
item at its automatic minimum width that takes no `text-overflow` from
the select, and no author style reaches it (`display: block` on the
select does not help either). So a long option in a narrow field, or
at `size="sm"` and `xs`, ran on past the end padding under the arrow,
which `overflow: hidden` only clips at the select's edge. The native
select in Firefox cut the value at the padding with no ellipsis.
The view now gives the select a first child `<button type="button"
wire:ignore>` holding `<selectedcontent>`, the customizable select's own
copy of the chosen option, and select.css makes that the flex item that
shrinks and ends in an ellipsis; the native select takes `nowrap`,
`overflow: hidden` and the ellipsis on itself. `wire:ignore`, because a
Livewire render parses the server's empty element into the first
option's label and morphed it back over the chosen one. A browser
without the customizable select never renders the button; `type` keeps
it from being a form's default button.
A browser test measures the value against the arrow in md, sm, xs,
filled and icon selects 240px wide, and again after choosing another
option re-renders the component; it fails without the change in Chrome
and Firefox.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A closed standard sheet is `inline-size: 0` with a negative margin as
wide as its row's gap, but the view writes that gap (`--md-drawer-gap`)
and `data-md-drawer-collapsed` only once Alpine runs. Until then each
closed sheet was a zero-wide flex item that still cost the row one gap,
and a sheet bound to an open Livewire property had no width at all, so
the content beside them changed width when the script started - two
closed sheets beside a column made it 48px narrower, then it jumped.
Until the view has settled, drawer.css now takes a standard sheet that
is not `data-md-open` out of the layout, and the view renders
`data-md-open` on a standard sheet whose `wire:model` property is open
(truthy as Alpine reads it), so it stands at its width from the first
paint. Transitions still wait for `data-md-drawer-settled`. A browser
test measures a column beside two closed sheets (one bound to Livewire,
one to Alpine) and an open one before Alpine starts and once settled;
it fails without the change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The standard sheet was sticky at `top: 0` and `100dvh` tall, so in an
installed app with a status bar its head row and close button slid under
the status bar once the page scrolled, where the modal sheet already
starts at `--material-safe-top`. Inside `<x-scaffold>` a sticky
`<x-app-bar>` in the `top` slot spans the content region, sheet and all,
and covered it the same way, safe area or not. The sheet now sticks at
`--drawer-top`, the top safe area, or under a scaffold's sticky bar that
safe area plus the bar's 64px row (just the row, if taller than the safe
area, for a medium or large bar, which collapses to it), and is as tall
as the window below that. Every side sheet keeps the bottom safe area
inside its bottom padding. A browser test scrolls a standard sheet in a
scaffold with no bar, a small bar and a large bar at a 47px top and 34px
bottom inset and measures where it sticks, its foot, its close button
and its padding; it fails without the change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A modal side sheet, bottom sheet or the modal rail closed on any Escape
the window heard, so a dialog opened from a sheet, a menu, select list
or searchable choice inside one, a sheet opened from a sheet, and a
sheet inside a dialog each closed two layers on one press. And a dialog
or a second sheet rendered elsewhere on the page sat inside the
`aria-hidden` the first sheet's `x-trap.inert` put on its siblings, so a
screen reader could not read it, while the sheet's focus trap took every
Tab inside the dialog back to the inert sheet.
resources/js/layers.js adds `x-layer`, on each of those panels beside
its `x-trap`. An Escape is the panel's only when nothing has handled it
and the nearest open layer around its target is the panel itself - not
an open dialog, popover or customizable select, nor a panel inside it;
the panel claims it with preventDefault(), which also keeps a dialog
around it from cancelling, and dispatches `material-escape`, which the
views close on. A panel that opens lifts `aria-hidden` from its own
ancestors and puts it back on close only where a panel still open hides
them; materialShowModal() does the same for `<x-modal>`, whose new
`x-trap.noautofocus.noreturn` pauses the sheet's focus trap while it is
open and moves no focus of its own. The searchable choice, the search
view and the supporting pane's sheet now preventDefault() the Escape
they act on, so the dialog or sheet around them stays.
Four browser tests stack the layers every way above and fail without
the change in Chrome.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The hover rule's `:not(:has([data-md-field-control]:disabled))` weighs
as much as its argument, so hover weighed (0,5,0) against focus's
(0,2,0), and the error's hover (0,6,0) against the error's focus
(0,4,0): a focused field with the pointer on it drew a 2px on-surface
edge instead of primary, a field in error on-error-container instead of
error, in both variants. A customizable select with its menu open lost
the same way, since focus is then on an option in the top layer and
only select.css's `:open` rule (0,3,1) draws the focus edge. M3 layers
focus over hover. The disabled condition is now weightless inside
`:where()`, still keeping hover off a disabled field, and the focus
rule gains the field ancestor: focus (0,3,0) outweighs hover (0,2,0),
the error's focus (0,4,0) the error's hover (0,3,0), and the open
select both. An open select in error drew primary, not error, and now
draws error. A browser test hovers and clicks an outlined, a filled and
an invalid field and opens a select and a select in error, comparing
each edge with its role; it fails without the change (the error select
without select.css's rule).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A mail component is drawn by the mail theme, not by the application's
CSS entry, and the theme's classes share Tailwind's names: ReStride's
`x-mail::sessions` wraps its table in `<div class="table">`, the class
the package's own mail theme styles, and the guard reported it as
Tailwind's `display: table`. A view under a path `mail.markdown.paths`
names (Laravel's `resources/views/vendor/mail` by default) now skips
family (i) and the breakpoint, scale, palette and colour-value checks;
its icon names, the directives in its component tags, forbidColours()
and forbid() are still read. A mail theme stylesheet under that path,
which has to write literal values, is left out of check (iii) and of
the exempt classes, so its `.table` no longer hides `table` in the
application's other views either. The guard's docblock, the README and
the skill's "Testing the design" say so; a fixture test reads the same
mail view and theme with and without the path configured.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Google publishes 67 symbols per cut (auto_awesome, insights,
tips_and_updates, battery_50 ...) with width and height but no viewBox,
their paths in 24 units, or 20 in the 20 cut, where every other symbol
draws in `0 -960 960 960`. bin/fetch-symbols dropped the size so CSS
decides it, which left these with no coordinate system: `<x-icon>`
drew them at 1:1, whole only at exactly 24px and cut off or misplaced
at any other size, the 20 cut every small button uses included.
normalise() now turns a file's width and height into its viewBox when
it has none, before dropping them. The 268 files are regenerated from
Google's originals at the recorded SOURCE commit with the fixed step
(each shipped file matched the old step's output first; the other
16,272 do not change), and every path lies inside its new viewBox.
IconTest checks that each shipped symbol has a viewBox and that
`<x-icon name="auto_awesome">` renders in 20 units at size 16 and in 24
at size 32; both fail without the change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
1.x drew `data-selected` on any `data-list-row` that was not a card in
secondary-container with on-secondary-container ink, whatever element
the row was. 2.0.0 kept that only for `<x-list-item selected>` and a
table row's `aria-selected="true"`, so an application's own `<li>`,
`<div>` or `<tr>` with `data-md-list-row` and `data-md-selected` - the
contract list-rows.js still describes - lost its selected state without
a word. list-item.css now fills such a row, cards and list items
excluded, and names the fill `--md-list-row-fill`, so hover, focus and
press tint it as they tint a selected list item rather than replacing
it (1.x's selected fill hid them). In `<x-table>`, `data-md-selected`
selects a plain row too, beside `aria-selected="true"`, which ARIA
allows on a row only in a grid. A browser test compares a hand-made
selected `<li>` and a plain selected `<tr>` with the role pair in the
light and dark themes, checks that a selected card keeps its own fill
and that hover tints the row's fill; it fails without the change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The guard knew M3's roles and Tailwind's palette, so `bg-off-plan`,
`text-sport-run`, `from-brand` or `border-l-zone-4` - colours an
application's own Tailwind theme named before 2.0.0 - were the leftovers
its count never showed, though none has compiled since. After every
other family, a class-list token shaped like a colour utility on a name
that starts with a letter is now reported with the application's own
`var(--...)` as its replacement (`color-mix()` with an opacity), unless
the application's CSS declares the class.
So that none of Tailwind's own utilities on those prefixes reads as a
theme colour, each one Tailwind 4.2 lists is now named by its family:
`border-collapse`, `border-spacing-*`, the block borders `border-bs`/
`border-be`, `bg-blend-*` and the other background positions and
angles, gradient stops, `fill-none`/`stroke-*`, `text-shadow-*`,
`shadow-inner`, `accent-auto`, `inherit`/`initial` as colour keywords,
and the `mauve`, `olive`, `mist` and `taupe` palettes; only
`inset-ring-*` is still not read. BoostVocabularyTest no longer reads
a span holding a single property, prop or role name (`border-color`,
`placeholder-value`, `outline-variant`) as a class list. A fixture
test covers the theme colours and the Tailwind names beside them.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
toggle() was set(!collapsed), and while nothing is stored `collapsed`
is only `rail.default`: an adaptive rail between 840 and 1199px is
drawn collapsed whatever that says, so an application's shortcut
collapsed a collapsed rail and the first press changed nothing. Each
interactive rail now registers its root, and toggle() asks the first
one on the page, the way its menu button asks itself: an open rail is
hidden, a rail with no room to expand (modal, or adaptive below
expanded) opens over its scrim without storing anything, and any
other rail stores the opposite of what it draws, so a rail that hid
itself docks back into the layout as before. With no rail on the
page it still flips the choice. collapse() and expand() stay the
choice itself. The browser tests toggle the scaffold's rail at 960
and 1512px (in place) and at 768 and 393px (the modal), and a hiding
rail back into the layout; the first two fail without the change.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Two gaps found rewriting the Boost skill for 2.0's plain CSS, each
closed the way M3 describes it.
`<x-loading size="96">`: M3 gives the loading indicator a responsive
size, 48dp by default and flexible from 24 to 240dp ("never exceed that
range"), with the container and the active shape in a fixed ratio. The
size had no prop, so an application wrote a width and height of its own;
`size` now takes a whole number of px in that range, written as
`--md-loading-size`, and the SVG keeps the 48:38 ratio as it scales. A
value outside the range is ignored, as `<x-icon>` and `<x-shape>` ignore
theirs, and an application's own width and height still win.
`--md-navigation-rail-value`: M3 Expressive's rail has two values,
collapsed and expanded (Compose's WideNavigationRailValue), and content
in a rail follows it. Without a hook an application copied the rail's
seven conditions — mode, `data-rail`, `data-rail-auto`, open, and the
window band — out of navigation-rail.css. The rail now publishes the
answer from the same branches that narrow it: `expanded` by default,
`collapsed` wherever it is drawn collapsed, so a style query in the
application's CSS switches at the first paint and at the same moment
as the rail's own items. A rail open over a scrim reads `expanded`, and
outside a rail the property is unset. A JS attribute would have missed
the first paint; a width container query would have lagged the collapse
animation.
Style queries on a custom property need Firefox 151, so the browser
floor moves from Firefox 147 to 151 (README, the CI note, UPGRADE's new
2.1.0 section); Chrome 125 and Safari 18.4 are unchanged.
Browser tests pin both in Chrome, Firefox and Safari: the indicator's
drawn box at 96 and 32px, and the rail's value — with a style query
acting on it — at the first paint for fixed modes, across the window
classes and the menu button for an adaptive rail, and open and closed
for a modal one. NavigationRailTest pins the value in each of the five
collapsed branches.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2.0.0 took Tailwind out of the stack, but the texts Boost copies into
every application still taught its utilities: the core guideline said
"built on Tailwind CSS", the always-on material-3 guideline wrote every
rule as `bg-primary`, `type-*`, `rounded-corner-*`, `state-layer` and
`medium:`, and both skills' tables and examples did the same. None of
those classes exists in 2.0's stylesheets, so an agent following the
guideline wrote markup that compiled to nothing. Found moving ReStride
onto 2.0: its CLAUDE.md, and SealShare's, carry these texts.
The guidelines and skills now teach what 2.0 has: component and layout
component props (`gap="space200"`, `hide-from="medium"`), the `md-type-*`,
`md-ink-*` and interaction classes, and `--md-sys-*` tokens in the
application's own CSS, with breakpoints as range media queries.
UPGRADE.md §1, §2 and §4 describe the finished move instead of the
in-between state, and header comments that pointed at the removed
tokens/utilities.css and tailwind.css, or called a component "still
Tailwind", say what is true now.
BoostVocabularyTest runs DesignGuard over the code in every shipped
guideline and skill (fenced Blade and CSS, and each inline class list),
so a text that teaches a class the stylesheets do not define fails the
suite; it finds 268 violations in the texts as 2.0.0 shipped them. The
guard's own table of what it reports is exempt. BoostResourcesTest now
asks the design skill for each breakpoint's prop value and media query
instead of the removed `medium:` variants.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
missingStylesheets() names an import the views need, but nothing named
one they had stopped needing: when SealShare's last card left its views,
card.css stayed in every page until a reviewer noticed by hand.
unusedStylesheets($cssEntry) reads the entry's direct package imports
and reports each one no scanned view needs, counting what a needed
stylesheet imports itself and foundation.css always; an entry that
imports all.css is left alone. Opt-in, beside missingStylesheets()
(plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
<x-stack> stretches its children across by default and <x-form> is a
one-track grid, so a button written directly in either drew as a
full-width pill. M3 keeps a button's width "dynamic to fit label" and
says not to "stretch buttons into long flat shapes on large windows".
A button there now sits at the start edge at its own width; a stack
aligned start, centre or end is untouched, and an application's own
width still wins (a cap on the button itself was rejected for exactly
that: it would have limited an app's inline-size, which CascadeTest
guards). Chromium test in both directions, failing without the rules.
Decided with the user (plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The showcase and the error pages set grayscale font smoothing on
themselves, but the foundation did not, so an application that dropped
Tailwind's antialiased class rendered heavier text on macOS than the
package's own pages, silently. base.css now sets it on the page; the
showcase's and error page's copies go, and the guard tells a leftover
antialiased class to go too. Decided with the user (plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
SealShare's only navigation is a floating toolbar at place="bottom". In
1.x it set --material-bottom-bar to lift the snackbar over it; in 2.0.0
the toolbar reads that variable to place itself, so an application can
no longer set it, and the snackbar landed on the toolbar. M3 nudges a
snackbar "upward to avoid overlapping FABs/docked toolbars" and never
puts one in front of navigation.
A bottom-placed toolbar now publishes --material-bottom-toolbar, the
distance from the window's bottom edge to its top: floating 64px (as
tall as an md or lg FAB beside it), docked 64px plus the safe area it
pads, rounded docked floating from 840px. It is declared on :root and
again on the scaffold's root, the only place its formula sees the bar.
The snackbar clears whichever of the bar and the toolbar reaches higher,
and a page pads its end with it. Two Chromium tests, both failing on the
old formula: a docked toolbar on a scaffold's navigation bar, and a
floating toolbar without a scaffold (plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The note named only the scaffold's and the theme toggle's hook changes.
202 of 1.2.0's 237 hooks gained the prefix and nothing else; the table
lists the rest, whose inner parts now carry their component's name. A
negative assertion on a 1.x hook passes vacuously against 2.0.0, which
SealShare's toolbar test did (plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Plan step 41 review: README and the development skill describe the API
that exists after the review, briefly — what scan() reads, the family
table with the M3 spacing step and 1.x utilities, missingStylesheets()
reading imports only, forbidColours() on the 2.0.0 vocabulary — and
UPGRADE.md lists the removed checks and methods and the retargeted
forbidColours() (the user, 2026-09-15).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 39 (part 6). README's Requirements, Installation and
Stylesheet-and-script sections drop Tailwind CSS 4, the second Vite
entry, @source and the Tailwind-replaces-utilities framing; the plain
CSS entry keeps the layer statement (matching every package
stylesheet), and the Layout example's <body> loses classes
foundation/base.css already applies globally (surface, on-surface,
the brand typeface) since Tailwind is no longer there to write them.
UPGRADE.md §6 gets the same Tailwind-free setup in place of the
interim two-entry note step 37 left. The development skill's install
snippet was already Tailwind-free; its one stale paragraph describing
a second Tailwind entry is deleted (fixed with the build in the
"Delete Tailwind's half" commit). DesignGuard's own Tailwind-detection
docs stay: applications can still carry leftover Tailwind classes for
the guard to catch, which is unrelated to the build.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 37 review. Removing material.css left README.md, the
development skill, UPGRADE.md and the tailwind.css and foundation.css
headers telling an application to import foundation.css, tailwindcss and
tailwind.css in its place. material.css was also what brought layout.css
and components.css, so that install rendered every component unstyled,
and without a layer statement Tailwind's preflight would have outranked
the package's layers.
The skill's snippet is now the Tailwind-free one 2.0.0 ends with: the
foundation, then the stylesheet of each component the views render, or
all.css. README.md, whose layout and prose still assume Tailwind, shows
all.css in its own entry and Tailwind in a second, both opening with the
layer statement, the shape the Workbench now builds; the skill and
UPGRADE.md describe that interim in a sentence.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 37: an application on 1.x-style imports now imports
foundation.css before @import 'tailwindcss' and tailwind.css after it,
the way foundation.css's own header has described since Phase F
started — material.css only ever saved writing those two lines.
Updates every place that named it: tailwind.css's and
foundation/hidden.css's and tokens/utilities.css's header comments,
the development skill's install snippet and Tokens section, README.md's
install snippet, and a new line in UPGRADE.md's migration notes.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36, navigation review. navigation-rail.css claimed each FAB
override beats fab.css by specificity alone; the collapsed FAB's gap
and minimum width tie with fab.css's extended-size rules and win by
order, which the file's own import of fab.css guarantees. The rail
view pointed at one file for the collapsed conditions that four files
write out. UPGRADE.md now says applications lose the `rail-collapsed:`
variant the development skill used to teach.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36 (navigation group, third batch): <x-scaffold>'s own
styling moves into resources/css/layout/scaffold.css (the content
region, the bar and rail row, the banner, the actions row and its
rail-collapsed column layout, --material-bottom-bar and
--material-margin publishing, the skip link) alongside step 35's FAB
and content-margin rules already there. Every data-app-shell* hook
becomes data-md-scaffold-* (data-app-shell-bar, -actions, -banner);
the skip link is data-md-skip-link; data-app-shell itself is dropped,
data-md-scaffold already named the root.
The actions row's rail-collapsed:flex-col is written out branch for
branch as the navigation rail's own rewrite did for its internal
parts: the three width-independent conditions in one :where() group,
the four width-gated ones each in their own @media block. With that
gone, resources/css/tailwind.css's rail-collapsed custom-variant
shim (its last use) is removed; tailwind.css now carries only
tokens/theme.css and tokens/utilities.css, which the showcase still
needs until step 38.
navigation-bar.css's hide-on-scroll rule reading --material-bottom-bar
stayed unlayered only because <x-scaffold> published that variable
through a Tailwind utility, which no layered rule could outrank; now
scaffold.css sets it itself in material.layout, a layer
navigation-bar.css's own material.components always beats, so the
rule moves into the layer and the file fits one
@layer material.components block like every other navigation
stylesheet. navigation-bar rejoins NavigationStylesheetsTest.php's
dataset and NavigationBarTest.php's own duplicate shape test is
retired in favour of it.
Browser tests added (docs/plans/material-3-browser-tests.md): the
scaffold's FAB dropping the bar's own height once hide-bar-on-scroll
slides it away, at the trailing edge in a right-to-left document, and
clearing a safe area an application sets on its inline-end and bottom
edges.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36 review. interaction.css read `data-dragged` beside
`data-md-dragged` while <x-card> still set the old name; the card now
sets the new one and nothing in the package renders the old one on an
element with the class, which 2.0.0 introduces. The Tailwind
`state-layer` utility keeps `data-dragged` until step 39 deletes it.
UPGRADE.md's card line named the old `data-card` hook.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36 (containment group): <x-drawer>'s class lists move into
resources/css/components/drawer.css, keyed on data-md-drawer (with
data-md-open, data-md-standard) and data-md-drawer-scrim/-sheet
(data-md-side)/-head/-head-row/-heading/-title/-subtitle/-close/-body/
-actions. The modal sheet slides in on the emphasized-decelerate
easing and out on emphasized-accelerate (an asymmetric CSS transition
expressed by giving the open-state selector its own
transition-duration/-timing-function, which the "after-change style"
rule picks up); the sheet stays x-show-driven, with @starting-style +
transition-behavior: allow-discrete taking the place of x-transition's
enter/leave classes, matching resources/css/components/search.css's
already-rewritten scrim/view. A standard sheet drops the scrim and
trap from 840px and only fades in place (an in-flow box does not
travel); data-md-drawer-collapsed (from the view's `wide` Alpine
state) takes it out of the layout while closed there.
<x-drawer pane> is removed: step 35's <x-list-detail> is M3's form of
that job (the plan's decision, docs/plans/material-3-alignment.md
"Decisions"). Removed the prop, its CSS classes and its tests
(OverlayTest's two pane-only tests deleted, the others' pane cases
dropped); pointed each mention at <x-list-detail> in the material-3
guideline, the material-3-design and livewire-material-development
skills, and added a line to UPGRADE.md's 2.0.0 notes. No other living
file named `<x-drawer pane>` (grepped views, tests, README, boost,
workbench excluding build output); docs/plans and docs/audits keep
their history unchanged.
Imported from the Containment block of components.css, importing
button.css and divider.css for what the view renders.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 35: <x-app-shell> becomes <x-scaffold>, M3's and Compose's name
for the structure of bars, rails and panes, with no alias; every package
view, the showcase, the tests, the skills, README and UPGRADE follow. Its
markup and Tailwind classes stay for the navigation rewrite. The new fab
slot places the FAB as Compose's Scaffold does - bottom-end, 16px from the
edges below medium and 24px from it, clear of the navigation bar and of a
snackbar on screen (resources/css/layout/scaffold.css) - and the content
region tells a pane inside that the margin is already drawn.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Text fields bounded from medium, dialog scroll dividers and the new meaning of
separator, the two-line snackbar, bottom sheet stops, chip set scroll buttons,
and a list of the configurations 2.0.0 adds.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Breakpoints, cleared scales, the regenerated scheme with contrast levels, inks as
roles, and every changed default and hook the component streams reported.
Plan step 31, to finish after Phase E.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
