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>
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>
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>
Both applications show values a person reads or types character by
character: SealShare's share tokens and recovery codes, ReStride's file
and folder names and its keyboard shortcuts. The package wrote the
platform's monospace stack twice (the reset's <code>, <kbd>, <samp> and
the mono field) and gave everything else nothing, so an app copied the
stack by hand. --md-ref-typeface-mono now holds it, read by the reset,
the field and a new md-mono text class; the guard points font-mono at
<code> or md-mono. Not an M3 token: M3 names a brand and a plain
typeface and nothing for code. Decided with the user (plan step 46).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Plan step 36 review. 52fd65a3 replaced the md-visually-hidden class on
field's spoken counter, search's result count and a bound filter chip's
native checkbox with hooks carrying copies of its clip, on the reading
that a package view writes no class but the interaction classes. The
brief and step 34 let a view write the text classes too, and one shared
source beats three copies: the class is back, the copies and the
data-md-field-live and data-md-chip-native hooks are gone
(data-md-search-status predates them and stays). chip.css keeps its
tooltip.css import.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36 (inputs group, last of the "not started" pair): <x-timepicker>
renders data-md-timepicker-* throughout (data-md-mode, data-md-cycle,
data-md-view, data-md-inner, data-md-dragging, data-md-disabled,
data-md-error and data-md-value replace their unprefixed hooks), and
timepicker.css moves into material.components, importing field.css,
icon.css and button.css. The dialog's Tailwind utility class (position,
the scrim, the spatial-fast pop and its @starting-style) becomes plain
declarations under data-md-timepicker-dialog, matching the datepicker's
modal dialog. The landscape rule stays keyed on orientation and viewport
height, in px (560/364/346), never a breakpoint, as M3 asks.
The hour and minute boxes and the period buttons swap their Tailwind
`state-layer focus-ring` utilities for `md-state-layer md-focus-ring`:
same declarations, so nothing else in the stylesheet changes for them.
The dial swaps in `md-focus-ring` for its own hand-rolled, byte-identical
outline-on-:focus-visible rule, which timepicker.css no longer carries.
The period selector was already the radio-in-a-list M3's labelling table
asks for (role="radio" in a role="radiogroup", moved by the arrow keys),
and Shift+M/Y already reached the date picker's dropdowns — both
predate this rewrite.
field.css's data-md-field-button note is tidied now that every component
that selects it (input, password, datepicker, timepicker) is rewritten.
Browser tests written but not run (the inputs group's Chromium run
follows this stream): the period radios moving with the arrow keys, and
the dial lying on its side below the landscape height threshold.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36, step 0 of the inputs stream: mirrors ActionStylesheetsTest's
five checks over the seventeen inputs components already rewritten, plus
pagination (its four views live under resources/views/pagination/**, not
components/, so it gets its own small versions of the two view-reading
checks). The "no class list" check also recognises the wrapper fields'
`:class="$attributes->get('class')"` (Blade's own prop-forwarding syntax,
the only way a component with no root element of its own hands the caller's
class to its field) and a `'class' => '…'` attribute-bag entry, and exempts
field.css's week-long autofill transition from the motion-token check.
The new test caught three already-rewritten components writing a bare
`md-visually-hidden` class, which a package view no longer does (only the
three interaction classes, or the caller's own class, reach a view): field's
character-count live region, search's result-count live region, and chip's
native checkbox under a bound filter chip. Fixed by giving each its own
data-md-* hook with the same visually-hidden declarations in its stylesheet,
and chip.css was also missing its `tooltip.css` import for the tooltip
chip.blade.php renders.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
Plan step 36 audit. field.css's trailing button (clear, copy, reveal)
scales its state layer with the field's own icon size and doubles it as
the touch target, which the foundation's fixed-size interaction classes
cannot draw, and the hook stays data-md-field-button rather than a class
because datepicker and timepicker, not yet rewritten, still select it —
so it keeps its own declarations. It never hid the layer on a disabled
button, unlike every rewritten component's copy; it now does.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
<x-file> marks its field root data-md-file, and the tonal
::file-selector-button with its hover and disabled states moves from
field.css to components/file.css on the state tokens (plan step 36). A
browser test pins the field's 56px box and 16px padding and the pill.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
<x-select> marks its field root data-md-select, and what a <select>
changes in the field — covering the whole box, the customizable
select's closed button, the focused look and turned arrow while open —
moves from field.css to components/select.css (plan step 36). The open
list stays in menu.css until the menu leaves Tailwind.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
<x-textarea> marks its field root data-md-textarea, and what a
<textarea> changes in the field — the grip in the corner, the label on
the first line, autogrow between rows and max-rows, the filled offsets —
moves from field.css to components/textarea.css (plan step 36).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
<x-field> renders data-md-field with its props and parts as data-md-*
attributes (box, control, label, outline, support, counter, trailing
buttons), and field.css moves into material.components on tokens: px
geometry, spacing and state tokens, the 600px breakpoint (plan step 36).
Its icons take a size prop, 24/20/16 by the field's size.
Every control the field wraps now marks itself data-md-field-control,
so the inputs, the pickers, choices, field.js, menu.css's select and
listbox rules and timepicker.css follow the renamed hooks. Browser tests
cover the error icon, the disabled field's hover, the counter and the
width bound from 600px.
Co-Authored-By: Claude Opus 5 (1M context) <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
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
The hover declaration sat on .field-box while the disabled edge is
inherited from .field, and a direct declaration beats an inherited one
whatever the selectors weigh — so a disabled outlined field lit a
full-opacity on-surface outline. It is now scoped past a disabled control
and, like every other state layer here, kept inside @media (hover: hover)
so a tap does not leave it stuck. M3 gives a disabled control no state.
Plan step 13, finding IN-02.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qwx5USif3wFFmxtHg5U1g9
M3's assist, filter, input and suggestion chips with chip sets; choices
as filter chips or a searchable combobox whose list is an anchored
popover; and the search bar that opens into a docked or full-screen
search view.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01V9NnLxnPp8vaaurb3Z1MFy
Outlined and filled text fields (input, password, auto-growing textarea,
native select with the customizable select menu, file), checkbox with an
indeterminate state, radio buttons, the M3 switch and form, ported from
ReStride and extended. The first half of Phase 6.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01V9NnLxnPp8vaaurb3Z1MFy
