/*  Checker 24. Red and green mean a verdict and nothing else; everything
  structural is navy. One neutral ramp. Cursor and selection only on what
  is meant to be copied. Icons and images carry width and height in the
  markup. */

/*  Sahel, SIL Open Font License 1.1, self-hosted like everything else here.
  Vazirmatn is the family a machine reaches for and the owner recognised it as
  one. The measured reason is in docs/TYPE_AND_MOTION.md §2.4: Vazirmatn's
  tabular figures have no Latin coverage, so Persian and Latin digits land at
  different widths and never line up in a table. This site prints prices, IMEIs
  and serials in tables.

  Two weights. The SemiBold file answers for 600 and 700 both, so the many
  rules in this file that ask for 700 resolve to it without being rewritten and
  it is still downloaded once.

  These URLs are literal and are deliberately not run through the ?v=
  fingerprint helper, so the filename is what protects the swap: no browser can
  serve a stale Vazirmatn for a name that no longer exists. */
@font-face {
  font-family: Sahel;
  src: url(/static/fonts/sahel-regular.woff2) format("woff2");
  font-weight: 400;
  font-style: normal;
  font-display: swap;
}

@font-face {
  font-family: Sahel;
  src: url(/static/fonts/sahel-bold.woff2) format("woff2");
  font-weight: 600 700;   /* Sahel ships one bold, so both resolve to it */
  font-style: normal;
  font-display: swap;
}

/*  ===========================================================================
  The Apple layer. docs/APPLE_DESIGN.md is the source; docs/APPLE.md holds the
  measurements taken off apple.com's own stylesheets that back it up.

  Three things were taken and they are the whole change:

  1. Sections are FULL-BLEED and alternate their background. The colour change
     is the divider - no rule, no border, no gap. This is the answer to «کل
     صفحه خالیه»: at 2000px the panel is 2000px wide and only the words inside
     it are 980px. A narrow column on a white page is what looked empty.
  2. The type scale is dramatic. 56px against 17px is 3.3x; this site had 26
     against 16, which is 1.6x, and that ratio is most of why it read as a form
     rather than as a product.
  3. There is exactly ONE shadow in Apple's entire system and it is under a
     photograph of a product. Never a card, never a button. Every card shadow
     here is gone, and cards are separated from the page by their background
     instead - white on wash, or wash on white.
  ========================================================================= */

:root {
  --navy-900: #0a1f44;
  --navy-700: #123a6b;
  --navy-100: #e8edf5;
  --muted: #4a5a73;
  --line: #d7dee9;
  --paper: #ffffff;
  --wash: #f5f8fc;
  --ok: #0f7b3d;
  --bad: #b3261e;

  /*  The verdict states. docs/SCAMMINDER.md §5 item 2, minus the fifth rung:
    `--v-risk` is gone because a Find My check cannot produce a "risky" - there
    is the answer, the absence of one, and the age of one. A colour kept for a
    state the product cannot reach is the same mistake as a gauge with no
    quantity behind it, one layer down.
    ``--ok`` and ``--bad`` are unchanged and still mean a verdict and nothing else.
    The three middle states deliberately avoid both, which is how the scale gets
    four steps without spending red or green on any of them - the same split
    Scamminder arrived at independently, where amber and orange carry the middle
    and slate carries "never seen".

    Measured against both surfaces this site draws a verdict on, because a
    contrast figure that is only true on white is not a figure:

      --v-clean    #0f7b3d   white 5.36:1   parchment 4.92:1
      --v-caution  #8a5200   white 6.39:1   parchment 5.87:1
      --v-unknown  #4a5a73   white 7.00:1   parchment 6.43:1
      --v-blocked  #b3261e   white 6.54:1   parchment 6.00:1

    All five clear 4.5:1 on both. --v-clean is the weakest at 4.92:1 on
    parchment - it passes, with little room, so it is not lightened and it is
    never drawn on a surface darker than parchment. */
  --v-clean: var(--ok);
  --v-caution: #8a5200;
  --v-unknown: #4a5a73;
  --v-blocked: var(--bad);

  /*  Apple's neutral ramp, verbatim from docs/APPLE_DESIGN.md. Five values and
    no more. `--wash` above stays for the rules that already name it; new work
    uses --parchment, which is Apple's own #f5f5f7. */
  --parchment: #f5f5f7;   /* the alternating light band          */
  --pearl: #fafafc;       /* a ghost button, the top rail        */
  --hairline: #e0e0e0;    /* the 1px border on a card            */
  --ink-48: #7a7a7a;      /* fine print                          */

  /*  The dark band. #272729 rather than black: Apple never uses pure black for
    a tile, only for the nav and for video. Verdict colours are NEVER drawn on
    this surface - see the note above .tile-dark. */
  --tile-dark: #272729;
  --on-dark: #ffffff;
  --on-dark-muted: #cccccc;

  /*  The type scale. Apple's sizes exactly; the leading is ours, because
    Persian descenders and the dots below the baseline collide at Latin
    leading. docs/APPLE.md §3. These are the phone values; the two media
    queries at the foot of this file raise them.

    Letter-spacing is 0 and stays 0. Apple tracks headlines to -0.374px and
    that is the one part of their system that cannot be copied: Persian is a
    joined script and tracking breaks the joins between letters. */
  --t-display: 32px;      /* the one line at the top of the home page */
  --t-head: 28px;         /* a section heading                        */
  --t-sub: 21px;          /* a card heading                           */
  --t-lead: 19px;         /* the line under a display heading         */
  --t-body: 17px;         /* Apple runs body at 17, not 16            */
  --t-small: 14px;
  --t-fine: 12px;

  --lh-display: 1.25;     /* Apple 1.07, raised for Persian           */
  --lh-head: 1.35;
  --lh-body: 1.70;        /* Apple 1.47, raised for Persian           */

  /*  Section rhythm. Apple pads a tile 80px top and bottom at desktop and
    tightens to 48 on a phone. The bottom is always a little larger than the
    top - optical centring, text sits slightly high in its panel. */
  --sec-y: 48px;
  --sec-y-end: 54px;
  /* The shape system. Five radii, two border weights, three shadow depths, and
     nothing outside them. A component that wants a value it has not got gets
     one added here with a name; a number written into a single rule is how the
     site came to have thirteen different corners. */
  --r-xs: 6px;        /* a focus ring, a mark inside a control */
  --r-sm: 10px;       /* an input, a chip, a small button */
  --r-md: 14px;       /* a button, a card inside a card */
  --r-lg: 18px;       /* a card, a panel, anything that holds others */
  --r-pill: 999px;    /* a badge or a track, where the ends are round */
  --r-round: 50%;     /* only where the thing really is a circle */

  --line-w: 1px;      /* every border */
  --line-w-strong: 2px;   /* the one card that is meant to stand forward */
  --cap-w: 4px;       /* the coloured bar that caps a verdict */

  /*  All three are `none`, deliberately, and the names are kept so the rules
    that already ask for them resolve without being rewritten.

    docs/APPLE_DESIGN.md, "Shadow philosophy": Apple uses exactly ONE
    drop-shadow in the entire system - rgba(0,0,0,.22) 3px 5px 30px - and it
    is applied to a photograph of a product resting on a surface. Never a
    card. Never a button. Never text. Elevation comes from the surface colour
    changing, and from scale(0.95) on a press.

    The owner said the boxes were "the worst and most rubbish kind in the
    world". Three stacked shadow depths on every card is a large part of what
    he was looking at: it is the visual language of an admin dashboard, which
    is exactly what this is not meant to be. */
  --shadow-1: none;
  --shadow-2: none;
  --shadow-3: none;

  /*  The one real shadow, for the one thing entitled to it: the generated
    result image, which is a picture resting on a surface. */
  --shadow-product: rgba(0, 0, 0, 0.22) 3px 5px 30px 0;
  --ease: cubic-bezier(0.2, 0, 0, 1);

  /*  Three durations, measured off live Iranian sites rather than chosen:
    docs/TYPE_AND_MOTION.md §3.1. Nothing in this file animates for a length
    that is not one of these. */
  --dur-press: 90ms;    /* the press itself */
  --dur-quick: 160ms;   /* a button colour, a border, a link */
  --dur-lift: 220ms;    /* a card lifting, and its shadow */
}

*, *::before, *::after { box-sizing: border-box; }

/*  [hidden] loses to any author rule setting display, and copy/print are
  hidden because they do nothing until a script wires them up. */
[hidden] { display: none !important; }

html { -webkit-text-size-adjust: 100%; }

body {
  margin: 0;
  /*  "Segoe UI" is gone: it sat ahead of Tahoma and has no Persian coverage at
    all, so for this audience it contributed nothing. */
  font-family: Sahel, Tahoma, sans-serif;
  /*  17px, not 16. docs/APPLE_DESIGN.md: "Body copy at 17px, not 16px. Apple
    breaks the SaaS convention and runs paragraph text at 17px. The extra pixel
    gives the page an unmistakable reading, not scanning pace." */
  font-size: var(--t-body);
  line-height: var(--lh-body);
  color: var(--navy-900);
  background: var(--paper);
  cursor: default;
  -webkit-user-select: none;
  user-select: none;
  overflow-wrap: break-word;
}

/*  Weight 600, not 700. docs/APPLE_DESIGN.md: "Weight 600, not 700, for
  headlines"; the whole ladder is 300/400/600/700 with 500 deliberately absent
  and 700 kept for one 21px tagline. The SemiBold file answers for both, so
  this costs no extra download either way. */
h1, h2, h3 { margin: 0 0 8px; font-weight: 600; }
h1 { font-size: var(--t-display); line-height: var(--lh-display); }
h2 { font-size: var(--t-head); line-height: var(--lh-head); }
h3 { font-size: var(--t-sub); line-height: var(--lh-head); }
/*  Space belongs to the type size, so it scales with the ladder on its own.
  Apple states it the same way: --sk-paragraph-plus-headline-margin: 1.6em. */
p { margin: 0 0 0.8em; }
h1 + p, h2 + p, h3 + p { margin-top: 0; }
p + h2, p + h3 { margin-top: 1.6em; }
ul, ol, dl { margin: 0; padding: 0; list-style: none; }
a { color: var(--navy-700); text-decoration: none; cursor: pointer; }
a:hover { text-decoration: underline; }
button, summary, label, a { cursor: pointer; }
img { max-width: 100%; }
/*  One set, one stroke weight. 1.75 rather than 2: Apple's own glyphs are
  drawn lighter than the default every icon library ships, and a 2px stroke on
  a 20px icon is what makes a line drawing read as clip-art. The owner said
  «استیکرها و لوگوها افتضاح» and the weight is half of that; the other half was
  the mark, which was a shield glyph inside a filled navy square - two shapes
  doing one job. It is one shape now. */
svg:not(.sprite) {
  flex: 0 0 auto;
  fill: none;
  stroke: currentColor;
  stroke-width: 1.75;
  stroke-linecap: round;
  stroke-linejoin: round;
}
.sprite { position: absolute; width: 0; height: 0; overflow: hidden; }

/*  An icon never shrinks, and never overlaps the words beside it.
  ----------------------------------------------------------------------------
  Every icon on this site is an <svg> sitting inside a flex row next to its
  label. A flex item's default is to shrink when the row runs out of space, and
  an <svg> given less width than its viewBox needs does not clip - it scales
  its contents down and keeps drawing, so the glyph slides over the text beside
  it.

  The owner saw exactly this on the home page in a real phone browser and our
  screenshots never reproduced it. That is the important part: a headless
  browser measured the Persian text slightly narrower than his phone did, so
  the row fitted here and did not fit there. The defect was never in the
  markup, and no amount of looking at our own pictures would have found it.

  The width and height attributes in the markup say how big an icon is; this
  says nothing is allowed to take that away. */
svg { flex: none; }

/*  And the label beside it is the thing that gives way instead: it may wrap or
  be cut, but it may not push the icon out of its own space. */
.risk, .chip, .btn, .nav a, .rows li, .f-long, .f-short {
  min-width: 0;
}

:focus-visible { outline: 3px solid var(--navy-700); outline-offset: 2px; border-radius: var(--r-xs); }

/*  Sahel's Latin digits are proportional by default, so a figure that is
  really read against something else asks for tabular ones: an IMEI checked
  digit by digit against the back of a box, a tracking code read down a
  telephone, a card number, the exact amount to transfer, a column of amounts.

  Not everything Latin. A model name is a name, and giving its digits a full
  column each puts gaps inside "iPhone 13 (A2634)". Not a single large figure
  either: tabular spacing gives its thousands separator a whole digit's width
  and it reads as a gap. `lining-nums` is stated so old style figures can never
  be picked up from the font's own defaults. */
.delta, .exact {
  /*  Only where figures are genuinely read against each other down a column:
    the amounts in a statement, and the exact sum to transfer. Sahel's
    tabular digits are about a third wider than its proportional ones, so
    asking for them anywhere else spaces a fifteen-digit IMEI out into
    "3 5 3 5 9 ..." and turns a price into "۴۹۹ , ۰۰۰". That is what the
    paragraph above warns about, and it was done anyway. */
  font-variant-numeric: tabular-nums lining-nums;
}

/* Latin values sit isolated, or their characters swap places. */
.ltr {
  direction: ltr;
  unicode-bidi: isolate;
  text-align: left;
}

.num { unicode-bidi: isolate; font-variant-numeric: tabular-nums; }

/* A text cursor on a heading is something he named by name. */
.copyable, input, textarea, .code, .code-url, .exact, .card-no,
.found-imei, .cert-model {
  -webkit-user-select: text;
  user-select: text;
  cursor: text;
}
a.copyable, .btn { cursor: pointer; }

/*  ---------------------------------------------------------------------------
  The two containers, and the reason there are two.

  Apple runs a NARROW one for reading and a WIDE one for structure:

      .viewport-content { width: 980px }               @>=1069
      .viewport-content { width: 692px }               @<=1068
      .viewport-content { width: 87.5% }               @<=734
      .responsive-content { max-width: 1680px; width: 87.5% }

  Copying only the narrow one is exactly how this site ended up with a 690px
  column centred in his 2000px window with nothing either side of it - 692px is
  Apple's TABLET width and it was being used at desktop. The column is not the
  bug. The bug is that the column sat on a plain white page instead of inside a
  full-bleed panel. See .tile-* below.
  --------------------------------------------------------------------------- */
.wrap { width: 100%; max-width: 980px; margin: 0 auto; padding: 0 16px; }
.wrap-wide { width: 87.5%; max-width: 1440px; margin: 0 auto; }
.narrow .wrap { max-width: 620px; }

/*  ---------------------------------------------------------------------------
  Full-bleed tiles. The single most important rule in this file.

  docs/APPLE_DESIGN.md: "Alternating full-bleed tile sections: white/parchment
  <-> near-black, with the colour change itself acting as the section divider",
  and "tiles stack edge-to-edge with 0 gap". No rule, no border, no margin
  between them - the change of colour IS the break.

  A section is therefore the full width of the window at every width, and only
  the .wrap inside it is 980px. At 2000px the panel is 2000px wide. There is no
  width at which the page can look empty, because there is no width at which
  the background stops.
  --------------------------------------------------------------------------- */
section { padding: var(--sec-y) 0 var(--sec-y-end); }
section + section { margin-top: 0; }

.tile-white { background: var(--paper); color: var(--navy-900); }
.tile-wash { background: var(--parchment); color: var(--navy-900); }

/*  The dark band. Verdict colours are never drawn on this surface: --ok is
  #0f7b3d, which against #272729 measures about 1.6:1 and is unreadable, and a
  site whose whole subject is whether a phone is clean cannot afford a verdict
  that cannot be read. So a dark tile carries a statement and a link and
  nothing that reports a finding. */
.tile-dark { background: var(--tile-dark); color: var(--on-dark); }
.tile-dark h1, .tile-dark h2, .tile-dark h3 { color: var(--on-dark); }
.tile-dark p { color: var(--on-dark-muted); }
.tile-dark a { color: var(--on-dark); }

.band { background: var(--parchment); }

/*  The shell every inner page sits in. A form on a bare white page is the
  "empty" he complained about at its worst, because there is nothing on the
  screen but a column. On parchment, the white card reads as an object resting
  on a surface - which is the whole of Apple's elevation model, without a
  shadow. */
.narrow { background: var(--parchment); padding: var(--sec-y) 0 var(--sec-y-end); }

/*  A card is white on parchment, or parchment on white, with a hairline and no
  shadow. docs/APPLE_DESIGN.md store-utility-card: 18px radius, 1px hairline
  #e0e0e0, 24px padding, "No shadow by default". */
.card {
  background: var(--paper);
  border: var(--line-w) solid var(--hairline);
  border-radius: var(--r-lg);
  padding: 24px;
  margin: 0 0 14px;
}
/*  On white, a white card with a hairline is nearly invisible, so it takes the
  parchment fill instead and the surface still does the separating. */
.tile-white .card { background: var(--parchment); border-color: transparent; }
.card:last-child { margin-bottom: 0; }

/*  ---------------------------------------------------------------------------
  The top bar. «نوار بالا خالیه و بی کاربرد» - a mark on one end, two links on
  the other, and a vast gap between them.

  Apple's answer is not to space two links further apart. Their rail is 44px
  tall, its type is 12px - far SMALLER than body text - and it carries twelve
  items spread with justify-content: space-between: the logo, nine product
  links, search and bag. The rail is full because the whole site is in it.

  So this one carries the whole site too: the check, the prices, the sample
  certificate, the tutorial, the questions, and the account. 13px rather than
  Apple's 12 because Persian sets smaller than Latin at a given pixel size.

  Dark, like Apple's own nav, which is the one place pure black appears on
  their page. Ours is the brand navy rather than black. It also gives the top
  of every page a full-bleed band of colour, which is the same medicine as the
  tiles below it.
  --------------------------------------------------------------------------- */
.top {
  background: var(--navy-900);
  color: #fff;
  position: sticky;
  top: 0;
  z-index: 40;
}
/*  docs/SCAMMINDER.md §5 item 5, and his own words: «لوگو و نام چسبیده به لبه
  راست نوار، موجودی چسبیده به لبه چپ». Not centred.

  It already had `space-between`, so the mark and the balance were already at
  opposite ends - the container was the only thing holding them in. `.wrap` is
  980px, so at 2000px the mark sat 510px from the edge of the window. A
  percentage rail leaves a 4% gutter at every width instead, which is 80px at
  2000px. Deliberately NOT `.wrap-wide`: a bar wants a different maximum from a
  block of content. */
.top-in {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 10px;
  min-height: 48px;   /* Apple: 48 on a phone, 44 on a desktop */
  width: min(1680px, 92%);
  max-width: none;
  margin-inline: auto;
}
.mark { display: inline-flex; align-items: center; gap: 8px; color: #fff; }
.mark:hover { text-decoration: none; }
.mark-badge {
  display: inline-flex;
  align-items: center;
  justify-content: center;
  width: 28px;
  height: 28px;
  color: #fff;
}
.mark-name { font-weight: 600; font-size: 17px; }

/*  The link row. Hidden on a phone, where the disclosure below replaces it. */
.nav { display: none; }
.nav a {
  display: inline-flex;
  align-items: center;
  gap: 5px;
  min-height: 44px;
  padding: 0 9px;
  font-size: 13px;
  color: rgba(255, 255, 255, 0.82);
  white-space: nowrap;
  transition: color var(--dur-quick) var(--ease);
}
.nav a:hover { color: #fff; text-decoration: none; }
.nav-me { font-weight: 600; color: #fff; }
/* A Telegram name can be anything, so it gets a ceiling. */
.nav-name { display: none; max-width: 10em; overflow: hidden; text-overflow: ellipsis; white-space: nowrap; }
.nav-balance { font-size: 13px; }

/*  The phone menu. A <details>, so it opens and closes with no script at all -
  rule 4. The summary is a 48px target, which is rule 6. */
.menu { position: relative; }
.menu > summary {
  display: inline-flex;
  align-items: center;
  justify-content: center;
  width: 48px;
  height: 48px;
  color: #fff;
  list-style: none;
}
.menu > summary::-webkit-details-marker { display: none; }
.menu > summary::marker { content: ""; }
/*  Hidden by `display`, driven by [open], and NOT left to the browser's own
  closed-<details> handling. This was a real defect and it was on every page.

  A closed <details> hides its contents by skipping the content slot for
  painting, not by taking it out of layout. A child in normal flow is skipped
  with the slot; a child with `position: absolute` escapes it, because an
  out-of-flow box is laid out against its containing block - here `.menu`,
  which is outside the slot. Measured on an iPhone 12 profile with the menu
  shut: `details.menu` reported `open: false` while `.menu-panel` computed to
  `display: block; visibility: visible; opacity: 1`, a live 240x305 box at
  y=48, with six anchors inside it.

  The consequences were all real. It put six focusable links over the first
  screen; it is what `scripts/phone.py` reported as 113 things overlapping on
  the home page, which is almost certainly the «شروع استعلام» button "inside the
  risk chips" the owner reported; and it made every one of the 42 phone/page
  combinations fail, which buried every other finding underneath it.

  So the open state is an author rule on [open] and nothing is inferred. */
.menu-panel {
  display: none;
  position: absolute;
  top: 100%;
  inset-inline-end: 0;
  min-width: 15rem;
  padding: 8px 0;
  background: var(--navy-900);
  border-top: var(--line-w) solid rgba(255, 255, 255, 0.14);
  border-radius: 0 0 var(--r-lg) var(--r-lg);
}
.menu[open] > .menu-panel { display: block; }
.menu-panel a {
  display: flex;
  align-items: center;
  gap: 8px;
  min-height: 48px;
  padding: 0 20px;
  font-size: var(--t-body);
  color: rgba(255, 255, 255, 0.88);
}
.menu-panel a:hover { color: #fff; background: rgba(255, 255, 255, 0.07); text-decoration: none; }

/*  ---------------------------------------------------------------------------
  The way back. «گزینه بازگشت در هیچ جا نیست» - there was none anywhere, and he
  was using the browser's own button.

  Apple's sub-nav is the model: a thin strip that sticks below the global nav,
  52px tall, parchment, carrying the name of where you are on one side. Ours
  carries the way out of where you are instead.

  It is drawn by base.html on every page, in the same place, with the same
  words, so it is never something to look for. The home page is the one page
  that overrides it away, because there is nowhere above home to go.
  --------------------------------------------------------------------------- */
.subnav {
  background: var(--parchment);
  border-bottom: var(--line-w) solid var(--hairline);
}
.subnav-in {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  min-height: 52px;
}
.back-link {
  display: inline-flex;
  align-items: center;
  gap: 6px;
  min-height: 48px;
  font-size: var(--t-small);
  color: var(--navy-700);
}
/*  The chevron points the way it travels. In a right-to-left page "back" is to
  the right, and the icon is drawn pointing left, so it is mirrored. Getting
  this wrong is the same class of defect as the plan card markers. */
.back-link svg { transform: scaleX(-1); }
.back-link:hover { text-decoration: none; color: var(--navy-900); }
.subnav-title {
  font-size: var(--t-small);
  font-weight: 600;
  color: var(--muted);
  overflow: hidden;
  text-overflow: ellipsis;
  white-space: nowrap;
}

.foot { border-top: var(--line-w) solid var(--line); background: var(--wash); padding: 22px 0; }
.foot-in { display: flex; flex-wrap: wrap; gap: 16px; align-items: flex-start; justify-content: space-between; }
.foot-tag { font-size: 14px; color: var(--muted); margin: 4px 0 0; max-width: 420px; }
.foot-links { display: flex; flex-wrap: wrap; gap: 6px 16px; }
.foot-links a { font-size: 15px; min-height: 44px; display: inline-flex; align-items: center; }
/*  Reserved at its real size so the seal drops in without a relayout.
  Nothing is drawn: an empty framed box reads as a failed image. */
.seal-slot { width: 90px; height: 90px; }

.btn {
  display: flex;
  align-items: center;
  justify-content: center;
  gap: 8px;
  width: 100%;
  min-height: 52px;
  padding: 12px 18px;
  border: var(--line-w) solid transparent;
  border-radius: var(--r-md);
  font: inherit;
  font-size: 17px;
  font-weight: 700;
  text-align: center;
  margin: 10px 0 0;
}
.btn:hover { text-decoration: none; }
.btn-go { background: var(--navy-900); color: #fff; }
.btn-go:hover { background: var(--navy-700); }
/*  Press feedback belongs on the buttons that cost money, not on
  navigation. */
.btn-go:active { transform: scale(0.95); }
/*  The colour eases; the press does not. A press that eases feels like lag,
  which is the one thing a button must never feel like. */
.btn-go { transition: background var(--dur-quick) var(--ease), transform 0s; }
.btn-plain { background: var(--paper); color: var(--navy-900); border-color: var(--line); }
.btn-plain:hover { border-color: var(--navy-700); background: var(--wash); }
.btn-plain {
  transition: background var(--dur-quick) var(--ease),
              border-color var(--dur-quick) var(--ease),
              transform 0s;
}
.btn-plain:active { transform: scale(0.95); }
.btn-quiet {
  display: inline-flex;
  align-items: center;
  gap: 6px;
  min-height: 44px;
  padding: 6px 12px;
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-sm);
  background: var(--paper);
  color: var(--navy-700);
  font: inherit;
  font-size: 14px;
}
.btn-quiet-wide {
  display: block;
  width: 100%;
  min-height: 48px;
  margin-top: 6px;
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-md);
  background: transparent;
  color: var(--muted);
  font: inherit;
  font-size: 15px;
}

/*  The promise and the box, one object on a tinted panel (§1.2). One column
  on a phone; the two-column row is a wide-screen rule below. */
/*  No background and no padding of its own: the tile class on the section
  sets the first and --sec-y sets the second, so this band cannot drift out of
  step with the other five. It used to carry #f5f8fc, a faintly blue white,
  where every other wash on the page is Apple's #f5f5f7. */
/*  No rule of its own at all. The tile class on the section sets the
  background and `section { padding: var(--sec-y) 0 var(--sec-y-end) }` sets
  the rhythm, so this band cannot drift out of step with the other five. It
  used to carry #f5f8fc, a faintly blue white, where every other wash on the
  page is Apple's #f5f5f7 - two different off-whites nobody chose. */
/*  The size comes from the ladder. This rule used to set 27px here, 36px at
  735 and 40px at 1069, all written before the scale changed - so the one line
  on the site that most needed to be enormous was still on the old scale and
  measured 40px against a 17px body, a ratio of 2.35. Apple's is 3.3. Only the
  measure and the space below it belong to the hero. */
.hero h1 { margin-bottom: 16px; max-width: 16em; }
.hero-sub { font-size: var(--t-lead); color: var(--muted); max-width: 30em; margin-bottom: 24px; }
.hero-btn { max-width: 320px; }
.hero-box { margin-top: 22px; }
.risk { border: var(--line-w) solid #dbe3ef; }

.risks { padding: 0 0 24px; }
.risks-in { display: grid; grid-template-columns: 1fr 1fr; gap: 8px; }
.risk {
  display: inline-flex;
  align-items: center;
  gap: 6px;
  padding: 8px 13px;
  border-radius: var(--r-pill);
  background: var(--navy-100);
  color: var(--navy-900);
  font-size: 14px;
  font-weight: 700;
}

/*  Never under 16px: iOS Safari zooms the page when a smaller one is
  tapped. */
input, select, textarea { font-family: inherit; font-size: 16px; }

.check-card { max-width: 620px; margin-inline: auto; margin-bottom: 0; }

.lab { display: block; font-size: 15px; font-weight: 700; margin-bottom: 8px; }
.box {
  display: block;
  width: 100%;
  min-height: 52px;
  padding: 12px 14px;
  font: inherit;
  font-size: 18px;
  color: var(--navy-900);
  background: var(--paper);
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-md);
  letter-spacing: 0.04em;
}
.box:focus { border-color: var(--navy-700); }
.hint { font-size: 14px; color: var(--muted); margin: 12px 0 6px; }
.under { font-size: 13px; color: var(--muted); margin: 10px 0 0; text-align: center; }

.dial { display: flex; align-items: center; gap: 10px; margin-bottom: 4px; }
.dial .ltr {
  font-size: 17px;
  font-weight: 700;
  padding: 8px 14px;
  border-radius: var(--r-sm);
  background: var(--navy-100);
  min-height: 44px;
  display: inline-flex;
  align-items: center;
}

.free { display: flex; align-items: center; justify-content: center; gap: 8px; margin: 14px 0 0; }
.free-n {
  font-size: 18px;
  font-weight: 700;
  min-width: 32px;
  height: 32px;
  display: inline-flex;
  align-items: center;
  justify-content: center;
  border-radius: var(--r-pill);
  background: var(--navy-100);
}
.free span { font-size: 13px; color: var(--muted); }

/*  A mistake about what was typed is navy and bold with a glyph. It is
  never red: red spent here is red missing from "قفل است" later. */
.notice {
  display: flex;
  align-items: flex-start;
  gap: 8px;
  margin: 0 0 14px;
  padding: 12px 14px;
  border-radius: var(--r-md);
  background: var(--navy-100);
  color: var(--navy-900);
  font-weight: 700;
  font-size: 15px;
  line-height: 1.6;
}

.steps li {
  display: flex;
  align-items: center;
  gap: 10px;
  min-height: 42px;
  color: var(--muted);
  font-size: 16px;
}
.steps li::before {
  content: "";
  width: 20px;
  height: 20px;
  flex: 0 0 auto;
  border: var(--line-w-strong) solid var(--line);
  border-radius: var(--r-round);
}
.steps li.done { color: var(--navy-900); font-weight: 700; }
.steps li.done::before {
  content: "\2713";
  border-color: var(--navy-900);
  background: var(--navy-900);
  color: #fff;
  font-size: 12px;
  line-height: 16px;
  text-align: center;
  animation: pop 180ms var(--ease);
}
@keyframes pop { from { transform: scale(0.8); } to { transform: scale(1); } }

.plans-h { font-size: 18px; margin-bottom: 14px; }
.anchor-a {
  text-align: center;
  font-size: 18px;
  font-weight: 700;
  line-height: 1.6;
  margin: 0 auto 6px;
  max-width: 22em;
}
.anchor-b {
  text-align: center;
  font-size: 15px;
  color: var(--muted);
  margin: 0 auto 22px;
  max-width: 24em;
}

/*  Three across at every width. DESIGN_RESEARCH.md §3.6 and §4: 340px less
  two 7px gaps over 3.2 gives 102/122/102. Fractions, so no extra
  breakpoint. */
.plans .wrap { padding: 0 10px; }

.podium {
  display: grid;
  grid-template-columns: 1fr 1.2fr 1fr;
  gap: 7px;
  align-items: start;
}

/*  One class used twice: nothing may change one side and not the other.
  border-box absorbs the middle's 2px border against the sides' 1px. */
.plan {
  box-sizing: border-box;
  margin: 8px 0 0;
  padding: 10px 8px;
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-md);
  background: var(--paper);
  box-shadow: var(--shadow-1);
  text-align: center;

  /*  The owner: "when you move the mouse over the three check options it has
    to be obvious the pointer is on it. There should be an animation - it grows,
    or it comes up under the pointer."

    The trap: the middle card already carries a static translateY as its podium
    lift. `transform` is one property, so a hover rule that set it would replace
    the podium offset and the middle card would jump DOWNWARDS under the
    pointer. The two offsets are composed inside one calc() instead, and the
    transition is declared on `transform` - a custom property does not
    interpolate on its own unless it is registered, but substituting it changes
    the computed transform and the browser eases between the two.

    Only transform, box-shadow and border-color move. No layout property is in
    this transition and there is no `transition: all`. BUILD_RULES §11. */
  --rest: 0px;   /* the podium offset, set per card, per width */
  --lift: 0px;   /* the pointer offset, set by :hover and :active */
  transform: translateY(calc(var(--rest) + var(--lift)));
  transition: transform var(--dur-lift) var(--ease),
              box-shadow var(--dur-lift) var(--ease),
              border-color var(--dur-lift) var(--ease);
  -webkit-tap-highlight-color: transparent;
}

/*  Guarded, so the lift cannot stick after a tap on a phone - where hover does
  not really exist and a browser emulates it until the next tap elsewhere.
  Most of his customers are on a phone, and they get the press below. */
@media (hover: hover) and (pointer: fine) {
  .plan:hover {
    --lift: -6px;
    box-shadow: var(--shadow-2);
    border-color: var(--navy-700);
  }
  .plan--best:hover { box-shadow: var(--shadow-3); }
}

.plan:active { --lift: -2px; transition-duration: var(--dur-press); }
.plan:focus-within { box-shadow: var(--shadow-2); border-color: var(--navy-700); }
/* One box for all three names, so the prices sit on one line. */
.name {
  display: flex;
  align-items: center;
  justify-content: center;
  min-height: 32px;
  margin: 0;
  font-size: 12px;
  font-weight: 700;
  line-height: 1.35;
}
.price {
  margin: 4px 0 0;
  font-size: 15px;
  font-weight: 700;
  line-height: 1.2;
}
/*  The figure and the word are two things, not one word. Without this they
  are printed hard against each other on every page that names a price with its
  unit beside it. The margin is logical rather than left or right, so it stays
  on the correct side of the number in a right-to-left page. */
.price > span {
  margin-inline-start: 6px;
  font-size: 14px;
  font-weight: 400;
  color: var(--muted);
}

.unit { margin: 0; font-size: 10px; color: var(--muted); line-height: 1.6; }

.rows { margin: 8px 0 10px; text-align: right; }
.rows li { display: flex; align-items: flex-start; gap: 4px; padding: 1px 0; font-size: 11px; line-height: 15px; }
.rows li > span { flex: 1 1 auto; }
.rows .f-on { color: var(--navy-900); }
.rows .f-off, .rows .f-na { color: var(--muted); }
.f-long { display: none; }

/*  48px, not §4's 44: rule 6 asks for 48. All three slots match - emphasis
  is fill and weight, never height. */
.cta {
  display: flex;
  align-items: center;
  justify-content: center;
  height: 48px;
  padding: 2px 4px;
  border: var(--line-w) solid var(--navy-900);
  border-radius: var(--r-sm);
  background: var(--paper);
  color: var(--navy-900);
  font-size: 11px;
  font-weight: 700;
  line-height: 1.25;
}
.cta {
  transition: background var(--dur-quick) var(--ease),
              border-color var(--dur-quick) var(--ease),
              transform 0s;
}
.cta:hover { text-decoration: none; background: var(--wash); }
.cta:active { transform: scale(0.95); }

/*  The same slot holding a fact, not a control: this row cannot be bought
  from (OFFER §10.3), and a control that could only apologise is never
  drawn. */
.cta-note {
  display: flex;
  align-items: center;
  justify-content: center;
  height: 48px;
  margin: 0;
  padding: 2px 4px;
  font-size: 11px;
  line-height: 1.3;
  color: var(--muted);
}

/* The middle. Every difference is listed here and nothing else differs. */
.plan--best {
  position: relative;
  padding: 12px 8px;
  border-width: var(--line-w-strong);
  border-color: var(--navy-900);
  background: var(--wash);
  box-shadow: var(--shadow-2);
  /*  The podium offset at this width. It sets --rest rather than transform:
    `transform` is one property, and setting it here would have replaced the
    composed value on .plan and left the hover doing nothing at all - which is
    what it did, and which no test in this project could have seen.

    docs/TYPE_AND_MOTION.md §4.5 names two of these offsets; this is the third,
    and it is the one that was overriding the other two. */
  --rest: -6px;
}
.plan--best .name { font-size: 13px; font-weight: 700; }
.plan--best .price { font-size: 18px; }
.plan--best .cta { background: var(--navy-900); color: #fff; font-size: 13px; }
.plan--best .cta:hover { background: var(--navy-700); }

/* Outside the card box, so it costs none of the middle column's width. */
.badge {
  position: absolute;
  top: -9px;
  left: 50%;
  transform: translateX(-50%);
  padding: 2px 6px;
  border-radius: var(--r-pill);
  background: var(--navy-900);
  color: #fff;
  font-size: 9px;
  font-weight: 700;
  line-height: 1.6;
  white-space: nowrap;
}

.step-line { font-size: 12px; color: var(--navy-900); font-weight: 700; margin: 14px 0 0; text-align: center; }
.free-note { font-size: 11px; color: var(--muted); margin: 4px 0 0; text-align: center; }


.moat-card { background: var(--navy-900); border-color: var(--navy-900); color: #fff; }
.moat-a { font-size: 18px; font-weight: 700; margin-bottom: 6px; }
.moat-b { font-size: 15px; color: #c7d3e6; margin-bottom: 14px; }
.sample {
  display: inline-flex;
  align-items: center;
  gap: 8px;
  min-height: 44px;
  padding: 8px 14px;
  border-radius: var(--r-sm);
  background: rgba(255, 255, 255, 0.12);
  color: #fff;
  font-size: 15px;
}
.sample:hover { background: rgba(255, 255, 255, 0.2); text-decoration: none; }

.live { background: var(--paper); }
.small-h { font-size: 16px; color: var(--muted); font-weight: 700; margin-bottom: 12px; }
.live-list { display: flex; flex-wrap: wrap; gap: 8px; }
.live-list li {
  padding: 7px 12px;
  border-radius: var(--r-pill);
  border: var(--line-w) solid var(--line);
  font-size: 14px;
}
/*  The strip of model names above this is gone - the owner could not tell what
  it was for. What is left is the one figure that needs no explaining, and it
  is set at the top of the ladder because it is the whole content of its band
  now rather than a footnote under a list. */
.counter {
  display: flex;
  flex-direction: column;
  align-items: center;
  gap: 4px;
  margin: 0;
  text-align: center;
}
.counter b { font-size: var(--t-display); line-height: var(--lh-display); font-weight: 600; }
.counter span { font-size: var(--t-lead); color: var(--muted); }

.trust { border-top: var(--line-w) solid var(--line); background: var(--paper); }
.trust-in { display: flex; flex-direction: column; gap: 4px; }
.trust-row { display: flex; align-items: center; gap: 10px; margin: 0; min-height: 44px; font-size: 16px; }
.trust-row a { display: inline-flex; align-items: center; min-height: 44px; }

.faq { background: var(--paper); }
.faq details { border-bottom: var(--line-w) solid var(--line); }
.faq summary {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  min-height: 52px;
  padding: 8px 0;
  font-weight: 700;
  font-size: 16px;
  list-style: none;
}
.faq summary::-webkit-details-marker { display: none; }
.faq summary::after {
  content: "";
  flex: 0 0 auto;
  width: 9px;
  height: 9px;
  border-right: var(--line-w-strong) solid var(--muted);
  border-bottom: var(--line-w-strong) solid var(--muted);
  transform: rotate(45deg);
  transition: transform 200ms var(--ease);
}
.faq details[open] summary::after { transform: rotate(-135deg); }
.faq details p { font-size: 15px; color: var(--muted); margin: 0 0 12px; max-width: 34em; }
.faq details[open] p { animation: fade 120ms 60ms backwards; }
@keyframes fade { from { opacity: 0; } to { opacity: 1; } }

.verdict { border-top: var(--cap-w) solid var(--line); text-align: center; }
.verdict h1 { font-size: 22px; }
.verdict-mark { margin: 0 0 6px; }
.verdict-body { font-size: 15px; color: var(--muted); margin: 0; max-width: 30em; margin-inline: auto; }
.verdict-word { margin-right: 8px; }

/*  One rule per state, each setting the three co-ordinated values Scamminder
  uses: the arc colour, the cap, and the colour of the mark and the word.
  `.ok` and `.bad` are kept as aliases of the two ends so nothing that still
  names them breaks. docs/SCAMMINDER.md §5 item 2. */
.verdict.clean, .verdict.ok    { border-top-color: var(--v-clean); }
.verdict.caution               { border-top-color: var(--v-caution); }
.verdict.unknown               { border-top-color: var(--v-unknown); }
.verdict.blocked, .verdict.bad { border-top-color: var(--v-blocked); }

/*  The mark takes the verdict colour. The word is in the heading beside it, so
  a reader who cannot tell the colours apart still gets the answer. */
.verdict.clean .verdict-mark, .verdict.ok .verdict-mark { color: var(--v-clean); }
.verdict.caution .verdict-mark { color: var(--v-caution); }
.verdict.unknown .verdict-mark { color: var(--v-unknown); }
.verdict.blocked .verdict-mark, .verdict.bad .verdict-mark { color: var(--v-blocked); }

.kicker { font-size: 14px; color: var(--muted); margin-bottom: 10px; }

/*  The picture, and the owner's own instruction about it:
  «وقتی سایت عکس رو نشون میده، طرفین سایت خالی نباشه و خالی با یه عکس کوچیک وسط
  صفحه». It was a small picture in the middle of an empty page: 420px wide,
  centred, in a 620px column, on a 1920px window.

  So the ceiling is gone. It fills the column it is given, and on a wide screen
  that column is the leading half of `.result-in` with the facts beside it -
  §5 item 4 - rather than a ribbon with nothing on either side.

  The box is declared before the file arrives (§5 item 13): 1/1 because the file
  really is 1080x1080, so there is no layout shift on the one page a paying
  customer looks at. `object-position: top` means a tall image is cropped from
  the bottom, so the part a person recognises is always the part they see. */
.shot-link { display: block; width: 100%; margin: 0 auto 14px; }
.shot-link:hover { text-decoration: none; }
/*  The card carries the border and the radius; the picture inside it is square
  and uncropped, so the bar that bleeds to the PNG's edge is never cut. No
  `overflow: hidden` - there is nothing to clip, and clipping is what caused the
  wedges. */
.shot-box {
  padding: 10px;
  background: var(--parchment);
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-lg);
}
/*  `aspect-ratio` with the width and height attributes in the markup reserves
  the box before the file arrives, so there is no layout shift on the one page a
  paying customer looks at - §5 item 13. The file really is 1080x1080, so 1/1 is
  a fact rather than a crop, and `object-fit: contain` guarantees nothing is cut
  even if that ever changes. */
.shot {
  display: block;
  width: 100%;
  height: auto;
  aspect-ratio: 1 / 1;
  object-fit: contain;
}

.facts dl > div, .facts .f {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  padding: 9px 0;
  border-bottom: var(--line-w) solid var(--line);
}
.facts dl > div:last-child, .facts .f:last-child { border-bottom: 0; }
.facts dt { font-size: 14px; color: var(--muted); flex: 0 0 auto; }
.facts dd { margin: 0; font-size: 15px; font-weight: 700; text-align: left; }
.facts dd.ltr { direction: ltr; }
.facts .f dd { display: inline-flex; align-items: center; gap: 6px; }

/* What a finding costs the buyer, printed under the finding itself and
   only when the finding is adverse. It reads as a footnote to the row
   above rather than as a second row, so the table stays a table. */
.facts .why {
  margin: -2px 0 0;
  padding: 0 0 12px;
  font-size: 13px;
  line-height: 1.7;
  color: var(--danger);
  border-bottom: var(--line-w) solid var(--line);
}

.code-label { font-size: 14px; color: var(--muted); margin: 0 0 2px; }
.code { font-size: 22px; font-weight: 700; letter-spacing: 0.02em; margin: 0 0 6px; }
.code-url { font-size: 14px; color: var(--muted); margin: 0 0 4px; }
.share-btns { display: flex; flex-direction: column; gap: 0; }

/* Controls, not cards: a row with a chevron that turns. */
.branch details {
  margin: 0 0 10px;
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-md);
  background: var(--paper);
}
.branch summary {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  min-height: 52px;
  padding: 0 14px;
  font-weight: 700;
  font-size: 16px;
  list-style: none;
}
.branch summary::-webkit-details-marker { display: none; }
.branch summary::after {
  content: "";
  flex: 0 0 auto;
  width: 9px;
  height: 9px;
  border-right: var(--line-w-strong) solid var(--muted);
  border-bottom: var(--line-w-strong) solid var(--muted);
  transform: rotate(45deg);
  transition: transform 200ms var(--ease);
}
.branch details[open] summary::after { transform: rotate(-135deg); }
.branch-in { padding: 12px 14px 14px; border-top: var(--line-w) solid var(--line); }

.reach { text-align: center; }
.reach p { font-size: 15px; margin: 0; }
.loud { font-size: 18px; font-weight: 700; margin: 10px 0; }

.tickbox { display: flex; align-items: flex-start; gap: 10px; min-height: 48px; padding: 6px 0; font-size: 15px; }
.tickbox input { width: 22px; height: 22px; flex: 0 0 auto; margin-top: 6px; accent-color: var(--navy-900); }

/*  The two confirmations used to be enforced from here: the quote was hidden
  until both boxes were ticked. A stylesheet may not be what stands between a
  visitor and a price - with this file blocked the price was simply visible -
  and a checkbox leaves no record, which is what the owner asked for. It is a
  form post now, and the server decides. */
.quote { border-top: var(--line-w) solid var(--line); margin-top: 10px; padding-top: 12px; }
.quote-h { font-size: 17px; font-weight: 700; margin-bottom: 2px; }
.quote-list li { display: flex; align-items: baseline; justify-content: space-between; gap: 12px; padding: 7px 0; font-size: 15px; }
.quote-list b { font-weight: 700; white-space: nowrap; }

.offer { border-color: var(--navy-900); }
.offer-a { font-size: 17px; font-weight: 700; margin-bottom: 4px; }
.offer-b { font-size: 15px; color: var(--muted); }
.chips { display: flex; flex-wrap: wrap; gap: 8px; margin: 12px 0 14px; }
.chip {
  display: inline-flex;
  align-items: center;
  gap: 6px;
  padding: 7px 12px;
  border-radius: var(--r-pill);
  border: var(--line-w) dashed var(--line);
  color: var(--muted);
  font-size: 14px;
}
.offer-price { display: flex; align-items: baseline; gap: 6px; margin: 0; }
.offer-price b { font-size: 26px; font-weight: 700; }
.offer-price span { font-size: 14px; color: var(--muted); }

.back { margin: 18px 0 0; text-align: center; }
.back a { display: inline-flex; align-items: center; gap: 6px; min-height: 44px; font-size: 15px; color: var(--muted); }

.paper .card { padding: 18px; }
.cert-head { text-align: center; }
.cert-model { font-size: 17px; font-weight: 700; margin: 0 0 12px; }
.warnings { border-color: var(--bad); }
.warn { display: flex; align-items: center; gap: 8px; margin: 0 0 8px; font-weight: 700; color: var(--bad); font-size: 15px; }
.warn:last-child { margin-bottom: 0; }

/*  Three states, and they must not be able to be mistaken for one another.
  A flag that came back unknown is the one that must never read as good. */
.f-ok dd { color: var(--ok); }
.f-bad dd { color: var(--bad); }
.f-unknown dd { color: var(--muted); font-weight: 400; }
.f-unknown dd span { border-bottom: var(--line-w) dashed var(--muted); }
.footnote { font-size: 13px; color: var(--muted); margin: 12px 0 0; }

.spec { max-width: 520px; margin: 22px auto 0; }
.spec-h { font-size: 15px; font-weight: 700; margin: 0 0 6px; text-align: center; }

.money { text-align: center; }
.who { font-size: 15px; color: var(--muted); margin-bottom: 14px; }
.balance-label { font-size: 14px; color: var(--muted); margin: 0; }
.balance { display: flex; align-items: baseline; justify-content: center; gap: 6px; margin: 0 0 6px; }
.balance b { font-size: 30px; font-weight: 700; }
.balance span { font-size: 14px; color: var(--muted); }
.two { display: flex; flex-direction: column; gap: 0; }

.rowlist li { border-bottom: var(--line-w) solid var(--line); }
.rowlist li:last-child { border-bottom: 0; }
.rowlist li > *, .rowlist li a {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  min-height: 48px;
  font-size: 15px;
  color: var(--navy-900);
}
.rowlist li a:hover { text-decoration: none; color: var(--navy-700); }
.ledger li { display: flex; align-items: center; justify-content: space-between; gap: 12px; min-height: 46px; }
.ledger .reason { font-size: 14px; color: var(--muted); }
.rowlist .what, .ledger .reason { display: flex; flex-direction: column; gap: 0; }
.when { font-size: 12px; color: var(--muted); line-height: 1.5; }
.ledger .delta { font-size: 15px; font-weight: 700; unicode-bidi: isolate; }
.empty { font-size: 15px; color: var(--muted); margin: 0; }

.tag {
  /*  §5 item 8. `.tag.ok` and `.tag.bad` coloured text and stopped there, and a
    colour is not a verdict: it fails for a colour-blind reader and it fails in a
    screenshot sent over WhatsApp. The border is `currentColor` rather than a
    tint, because a tint light enough to carry 700-weight text would be a fifth
    colour to justify. The text must always contain the word. */
  display: inline-flex;
  align-items: center;
  gap: 6px;
  padding: 4px 10px;
  border-radius: var(--r-sm);
  font-size: 13px;
  font-weight: 700;
  border: var(--line-w) solid currentColor;
  background: var(--paper);
}
.tag.ok { color: var(--ok); }
.tag.bad { color: var(--bad); }
.tag.done, .tag.no, .tag.wait { color: var(--muted); font-weight: 400; }

.pay { text-align: center; }
.pay > p:first-of-type { font-size: 15px; color: var(--muted); text-align: right; }
.exact-label, .card-label, .short-label { font-size: 14px; color: var(--muted); margin: 14px 0 0; }
.exact { font-size: 32px; font-weight: 700; margin: 0; letter-spacing: 0.01em; }
.card-no { font-size: 20px; font-weight: 700; margin: 0; letter-spacing: 0.04em; text-align: center; }
.holder { font-size: 15px; margin: 2px 0 0; }
.promise {
  display: flex;
  align-items: flex-start;
  gap: 8px;
  margin: 18px 0 0;
  padding: 12px 14px;
  border-radius: var(--r-md);
  background: var(--navy-100);
  font-size: 14px;
  text-align: right;
  line-height: 1.7;
}
.form-rows .box { margin-bottom: 14px; }
.pay-line { display: flex; align-items: baseline; justify-content: space-between; gap: 12px; margin: 0 0 6px; font-size: 15px; }
.pay-line b { font-weight: 700; }
.wallet-line { display: flex; align-items: baseline; justify-content: space-between; gap: 12px; font-size: 15px; color: var(--muted); margin: 14px 0 0; }
.wallet-line b { color: var(--navy-900); }

.confirm { text-align: center; }
.ask { font-size: 22px; }
.found { font-size: 20px; font-weight: 700; margin: 12px 0 2px; }
.found-imei { font-size: 15px; color: var(--muted); margin: 0 0 4px; }
.missing h1, .problem h1, .spent h1 { font-size: 21px; }
.missing, .problem, .spent, .enter { text-align: center; }
.missing p, .problem p, .spent p, .enter p { font-size: 15px; color: var(--muted); max-width: 30em; margin-inline: auto; }
.maybe { margin: 12px 0; }
.maybe li { font-size: 15px; color: var(--muted); padding: 5px 0; }
.enter .code { font-size: 30px; letter-spacing: 0.14em; margin: 14px 0 6px; }
.loop { text-align: center; }

/* --- the tutorial, and the player it plays in ---------------------------- */

/*  The seek bar reads left to right inside a page that reads right to left,
  and that is a decision rather than an oversight. A timeline is not a
  sentence: it is the same number line that Telegram, Instagram, YouTube and
  Aparat all fill to the right, and an Iranian viewer has years of practice
  reading it that way. The buttons under it do follow the page - play on the
  right, where the thumb starts - because a button row *is* a sentence. */

.tut-what { max-width: 34em; margin: 0 0 14px; }
.tut-rows { margin: 0 0 16px; }
.tut-rows li {
  display: flex;
  align-items: center;
  gap: 8px;
  padding: 6px 0;
  font-size: 15px;
  border-bottom: var(--line-w) solid var(--line);
}
.tut-rows li:last-child { border-bottom: 0; }
.tut-rows li > span { flex: 1 1 auto; }
.tut-rows li > b { font-weight: 700; }
/*  Rule 9: green means a verdict. A selling point is not one. */
.tut-rows svg { color: var(--navy-700); }

.kicker { display: flex; align-items: center; gap: 8px; }

.vp-card { padding: 12px; }

.vp {
  display: flex;
  flex-direction: column;
  background: #000;
  border-radius: var(--r-sm);
  overflow: hidden;
}

.vp-stage { position: relative; background: #000; line-height: 0; }

.vp-video {
  display: block;
  width: 100%;
  height: auto;
  max-height: 68vh;
  background: #000;
}

/*  The tap everybody tries first. Hidden in the markup and revealed only once
  a script exists to answer it, so with no JavaScript there is no dead surface
  sitting on top of the browser's own controls. */
.vp-tap {
  position: absolute;
  inset: 0;
  width: 100%;
  height: 100%;
  padding: 0;
  border: 0;
  background: transparent;
  cursor: pointer;
}
.vp-tap:focus-visible { outline: 3px solid #fff; outline-offset: -6px; border-radius: 0; }

.vp-nope { color: #fff; padding: 20px; font-size: 15px; line-height: 1.7; }

.vp-bar {
  background: #0d1830;
  color: #fff;
  padding: 4px 10px 8px;
  -webkit-user-select: none;
  user-select: none;
}

/*  Left to right: elapsed, the bar, total. See the note at the top. */
.vp-line { display: flex; align-items: center; gap: 10px; direction: ltr; }

.vp-time {
  flex: 0 0 auto;
  min-width: 46px;
  font-size: 13px;
  line-height: 1;
  color: #c7d3e6;
  text-align: center;
  font-variant-numeric: tabular-nums;
}

.vp-seek {
  position: relative;
  flex: 1 1 auto;
  height: 44px;
  direction: ltr;
  cursor: pointer;
  touch-action: none;
}
/*  The track. The 44px above is the target a thumb has to hit; this is the
  6px a person actually sees. */
.vp-seek::before {
  content: "";
  position: absolute;
  left: 0;
  right: 0;
  top: 50%;
  height: 6px;
  margin-top: -3px;
  border-radius: var(--r-pill);
  background: rgba(255, 255, 255, 0.22);
}
.vp-buffered, .vp-played {
  position: absolute;
  left: 0;
  top: 50%;
  height: 6px;
  margin-top: -3px;
  border-radius: var(--r-pill);
  width: var(--w, 0%);
  pointer-events: none;
}
.vp-buffered { background: rgba(255, 255, 255, 0.38); }
.vp-played { background: #fff; }
.vp-knob {
  position: absolute;
  top: 50%;
  left: var(--at, 0%);
  width: 14px;
  height: 14px;
  margin: -7px 0 0 -7px;
  border-radius: var(--r-round);
  background: #fff;
  pointer-events: none;
}
.vp-seek:focus-visible { outline: 3px solid #fff; outline-offset: -4px; border-radius: var(--r-xs); }

.vp-btns { display: flex; align-items: center; gap: 2px; }
.vp-gap { flex: 1 1 auto; }

.vp-btn {
  min-width: 48px;
  height: 48px;
  display: inline-flex;
  align-items: center;
  justify-content: center;
  padding: 0;
  border: 0;
  border-radius: var(--r-sm);
  background: transparent;
  color: #fff;
  font-family: inherit;
}
.vp-btn > span { display: inline-flex; }
.vp-btn:hover { background: rgba(255, 255, 255, 0.14); }
.vp-btn:focus-visible { outline: 3px solid #fff; outline-offset: -3px; }
/*  Left to right, so it reads as the figure followed by its sign. In a
  right-to-left run the multiplication sign is a neutral character and jumps
  to the other side of the number, which turns "1x" into "x1". */
/* The primary control of the player. It used to be primary only because
   its icon was two pixels bigger than its neighbours', which is not a
   system - it is a coincidence. Every icon in this bar is now one size
   and the play button is the one that is filled. */
.vp-btn-main { background: rgba(255, 255, 255, 0.16); }
.vp-btn-main:hover { background: rgba(255, 255, 255, 0.26); }

.vp-btn-speed { min-width: 60px; font-size: 14px; font-weight: 700; }
.vp-btn-speed .num { direction: ltr; unicode-bidi: isolate; }

/*  Left to right for the same reason the seek bar is: inside a right-to-left
  page a range input reverses, which puts full volume at the left-hand end and
  reads to anyone who has used any other player as though the sound is off.
  Louder is to the right, here as everywhere else.

  The 16px is not about reading it - a range has no text. It is there because
  iOS Safari zooms the whole page the instant any control under 16px is
  touched, and because the measurement script checks every input. */
.vp-vol {
  flex: 0 0 auto;
  width: 90px;
  height: 48px;
  margin: 0 4px;
  font-size: 16px;
  direction: ltr;
  background: transparent;
  accent-color: #fff;
}
.vp-vol:focus-visible { outline: 3px solid #fff; outline-offset: -3px; border-radius: var(--r-xs); }

/*  The keyboard shortcuts, as rows rather than as a sentence: a Persian name
  on one side, a Latin key on the other, each in a box of its own so the two
  scripts never share a text run. Gone on a phone, which has no keyboard. */
.vp-keys { display: none; }
@media (min-width: 735px) {
  .vp-keys {
    display: grid;
    grid-template-columns: 1fr 1fr;
    gap: 0 24px;
    margin: 12px 0 0;
  }
  .vp-keys li {
    display: flex;
    align-items: baseline;
    justify-content: space-between;
    gap: 12px;
    padding: 4px 0;
    font-size: 13px;
    color: var(--muted);
  }
  .vp-keys b {
    flex: 0 0 auto;
    font-weight: 700;
    color: var(--navy-900);
    font-size: 12px;
    padding: 1px 7px;
    border: var(--line-w) solid var(--line);
    border-radius: var(--r-xs);
    background: var(--wash);
  }
}

/*  Fullscreen. The wrapper goes full screen rather than the video, so the
  controls built here stay on screen instead of being replaced by the
  browser's. iOS Safari is the exception - it only ever fullscreens the video
  element itself - and there the native controls are what appears, which is
  correct for that platform and is what site.js falls back to. */
.vp:fullscreen { height: 100%; justify-content: center; border-radius: 0; }
.vp:fullscreen .vp-stage {
  flex: 1 1 auto;
  min-height: 0;
  display: flex;
  align-items: center;
  justify-content: center;
}
.vp:fullscreen .vp-video { height: 100%; width: 100%; max-height: none; object-fit: contain; }
.vp:-webkit-full-screen { height: 100%; border-radius: 0; }
.vp:-webkit-full-screen .vp-video { height: 100%; width: 100%; max-height: none; object-fit: contain; }

/*  On a phone the hardware keys set the volume better than anything drawn
  here, so the slider goes and the mute button - the one genuinely wanted on a
  phone - stays. */
@media (max-width: 559px) {
  .vp-vol { display: none; }
}

@media (prefers-reduced-motion: reduce) {
  /*  Blanket, not a list. Capping only the animations left every transition
    running at its full length for somebody who asked for less movement, and a
    transition added later would have been missed entirely.

    The colour, border and shadow changes still happen - they simply arrive at
    once. Somebody who asked for less movement still sees which card the
    pointer is on, which is the whole point of the hover. */
  *, *::before, *::after {
    animation-duration: 1ms !important;
    animation-iteration-count: 1 !important;
    transition-duration: 1ms !important;
    scroll-behavior: auto !important;
  }
  /*  Feedback is replaced, never removed: no answer to a press reads as
  broken, and broken reads as untrustworthy. */
  .btn-go:active { transform: none; background: var(--navy-700); }
  .faq summary::after { transition: none; }

  /*  NOT optional, and this is the reason it is spelled out rather than left to
    the wildcard above: these five elements START invisible. `opacity: 0` with
    the animation cancelled is a card that never appears at all, so the reduced
    motion setting would empty the page. §5 item 7. */
  .rise, .rise-1, .rise-2, .rise-3, .rise-4 {
    animation: none !important;
    opacity: 1 !important;
    transform: none !important;
  }
}

@media (min-width: 735px) {
  /*  The tokens step, not the rules. Everything that names --t-* or --sec-*
    moves with them, which is why the ladder can be changed in one place.
    Apple: hero 56 -> 40 at 1068 -> 34 at 640 -> 28 at 419. */
  :root {
    --t-display: 40px;
    --t-head: 32px;
    --t-sub: 24px;
    --t-lead: 21px;
    --sec-y: 64px;
    --sec-y-end: 72px;
  }
  .wrap { width: 87.5%; max-width: 692px; padding: 0; }

  /*  The rail fills. Apple's is 44px here rather than 48, and the links spread
    across the whole width with space-between rather than bunching at one end -
    which is the thing he was looking at when he called the bar empty. */
  .top-in { min-height: 44px; }
  .menu { display: none; }
  .nav { display: flex; align-items: center; flex: 1 1 auto; justify-content: space-between; gap: 2px; }
  .nav-name { display: inline; }

  .risks-in { display: flex; flex-wrap: wrap; }
  /* Real room here, so the settled wording returns and the type steps up. */
  .plans .wrap { padding: 0 16px; }
  .podium { gap: 14px; }
  .plan { padding: 16px 12px; border-radius: var(--r-lg); margin-top: 14px; }
  .plan--best { padding: 20px 14px; --rest: -10px; }
  .name { font-size: 14px; min-height: 38px; }
  .plan--best .name { font-size: 16px; }
  .price { font-size: 20px; }
  .plan--best .price { font-size: 24px; }
  .unit { font-size: 12px; }
  .rows { margin: 12px 0 14px; }
  .rows li { font-size: 13px; line-height: 19px; gap: 6px; }
  .f-long { display: inline; }
  .f-short { display: none; }
  .cta { height: 48px; font-size: 13px; border-radius: var(--r-sm); }
  .plan--best .cta { font-size: 14px; }
  .cta-note { height: 48px; font-size: 12px; }
  .badge { top: -12px; padding: 4px 12px; font-size: 12px; }
  .step-line { font-size: 14px; }
  .free-note { font-size: 13px; }

  .nav { gap: 8px; }
  .nav > a { padding: 0 12px; font-size: 16px; }
  .nav-name { display: inline; }
  /*  Removed: it made every button shrink to 220px between 735 and 1068, so
    the submit button sat at 220px inside a 620px card at 768 and at the full
    card width at 1280. A control that is a different shape at three widths is
    the kind of thing the owner spots immediately. */
  .btn, .two .btn { display: inline-flex; }
  .two { flex-direction: row; gap: 10px; }
  .two .btn { flex: 1 1 0; }
  .share-btns { flex-direction: row; flex-wrap: wrap; gap: 10px; }
  .share-btns .btn { flex: 1 1 200px; margin-top: 10px; }
  .price b { font-size: 30px; }
  .anchor-a { font-size: 20px; }
  .exact { font-size: 38px; }
  .trust-in { flex-direction: row; flex-wrap: wrap; gap: 8px 28px; }
}

@media (min-width: 1069px) {
  /*  The top of the ladder. 56px against a 17px body is Apple's 3.3x ratio,
    and it is the single number that separates a page that reads as a product
    from a page that reads as a form. This site was at 1.6x.

    Apple pads a tile 80px here; 72/80 is that, with the bottom larger than the
    top for optical centring. */
  :root {
    --t-display: 56px;
    --t-head: 40px;
    --t-sub: 28px;
    --t-lead: 24px;
    --sec-y: 72px;
    --sec-y-end: 80px;
  }
  /*  Apple's .viewport-content is exactly 980px at 1069 and above and does not
    grow again, not even past 1441. The panel around it does. */
  .wrap { width: 100%; max-width: 980px; padding: 0; }

  section { padding: 44px 0; }

  /*  Words on the right where the eye enters; the box in the half that was
  empty. The four risks run under both. */
  .hero-in {
    display: grid;
    grid-template-columns: 1fr 1fr;
    gap: 30px 44px;
    align-items: start;
  }
  .hero-words { grid-column: 1; grid-row: 1; align-self: center; }
  /*  The box spans both rows, and the four risk chips move up under the words
    instead of running across the bottom. Measured at 1920 before this: the
    headline column was about 200px tall beside a 740px card, so the whole area
    under the headline was empty and the chips sat alone at the foot of it -
    which is the empty-space complaint inside the hero rather than beside it. */
  .hero-box { grid-column: 2; grid-row: 1 / span 2; margin-top: 0; }
  .risks-in { grid-column: 1; grid-row: 2; align-self: start; }
  
  .hero-btn { display: none; }
  
  .hero-bare .hero-in { display: block; }
  .hero-bare .hero-box { margin: 0; }

  /* The eye enters at the right: ساده sets the reference, کامل is badged. */
  .podium {
    flex-direction: row;
    align-items: flex-end;
    justify-content: center;
    gap: 20px;
    padding-top: 44px;
  }
  .plan { margin: 0; }
  /*  Sides 335px, middle 382px (+14%) at a 1100px row. The middle also lifts
  16px and carries a 2px border, so it needs no more. §4.2. */
  .podium { grid-template-columns: 1fr 1.14fr 1fr; gap: 24px; max-width: 1100px; margin: 0 auto; }
  .plan { padding: 24px 20px; border-radius: var(--r-lg); margin-top: 18px; }
  .plan--best { padding: 30px 20px; --rest: -16px; box-shadow: var(--shadow-3); }
  .name { font-size: 16px; min-height: 46px; }
  .plan--best .name { font-size: 18px; }
  .price { font-size: 24px; }
  .plan--best .price { font-size: 34px; }
  .unit { font-size: 13px; }
  .rows li { font-size: 14px; line-height: 22px; }
  .cta { height: 52px; font-size: 14px; }
  .plan--best .cta { font-size: 16px; }
  .cta-note { height: 52px; font-size: 13px; }
  .badge { top: -14px; padding: 5px 14px; font-size: 12px; }
  .plans-h { font-size: 20px; }
  .anchor-a { max-width: 34em; }
  .anchor-b { max-width: 34em; }
  .btn { width: 100%; }
}

@media print {
  .top, .foot, .back, .share-btns, .btn { display: none !important; }
  body { background: #fff; color: #000; font-size: 12pt; }
  .narrow, .band { background: #fff; padding: 0; }
  .card { border: var(--line-w) solid #999; break-inside: avoid; margin-bottom: 8pt; }
}

/* ===========================================================================
   ACCOUNTS — sign in, register, verify, reset, and the account section
   ---------------------------------------------------------------------------
   Added with the accounts work. Kept in one marked block so a later pass over
   the typeface and the motion can restyle it without unpicking the markup.
   Every value here comes from the tokens above; nothing invents a corner, a
   border weight or a shadow of its own.
   ========================================================================= */

/*  A form inside a centred card is still a form. The card centres its heading
  and its sentences, which is right; centring a field's own label over a
  full-width box is not, and it reads as something a machine laid out. Labels,
  hints and the forgot link all start where the text starts.

  The heading, the notice and the line that offers the other form stay centred,
  because each of those is about the card rather than about one field. */
.enter .form-rows,
.enter .form-rows .lab,
.enter .form-rows .hint,
.enter .form-rows .aside { text-align: right; }
.enter .form-rows p { margin-inline: 0; max-width: none; }

/*  The rule under the password box belongs to the box above it, not to the
  label below it, and the spacing has to say so. */
.enter .form-rows .hint { margin: -6px 0 14px; }

/*  A link that has to read as a link. The owner asked for ثبت نام in blue
   beside the sentence that offers it, and it is the only blue on the site. */
/*  48px is on the base rule, not on one of its three homes. It used to be on
  `.swap .link-blue` alone, so the identical link inside `.under` on the home
  page measured 20px - which shots.py caught at 360. A touch target that is
  only large in the place somebody remembered to make it large is the same
  defect waiting to come back the next time the link is reused. */
.link-blue {
  display: inline-flex;
  align-items: center;
  min-height: 48px;
  padding: 0 4px;
  color: var(--navy-700);
  text-decoration: underline;
  text-underline-offset: 3px;
  font-weight: 700;
}

/* The line under a form that offers the other one. */
.swap {
  margin: 14px 0 0;
  font-size: 15px;
  color: var(--muted);
  text-align: center;
  line-height: 28px;
}
/*  48px, like every other thing on this site a thumb has to hit. It measured
  24 and it is the one control the owner named by name. */
.swap .link-blue {
  display: inline-flex;
  align-items: center;
  min-height: 48px;
  padding: 0 8px;
  margin-inline-start: 2px;
}

/* «رمز را فراموش کرده‌ام», beside the password box rather than under the
   button, where it would compete with the thing the visitor came to press. */
.aside {
  margin: -4px 0 6px;
  font-size: 14px;
}
.aside a {
  display: inline-flex;
  align-items: center;
  min-height: 48px;
  padding: 0 4px;
}

/* «یا» between the two doors. A rule with a word sitting in it. */
.rule {
  display: flex;
  align-items: center;
  gap: 12px;
  margin: 18px 0 14px;
  color: var(--muted);
  font-size: 14px;
}
.rule::before,
.rule::after {
  content: "";
  flex: 1 1 auto;
  height: var(--line-w);
  background: var(--line);
}

/* A notice that is not a problem. The same shape as .notice, in the verdict
   green, because a confirmation that looks like a warning reads as one. */
.ok-note { border-color: var(--ok); color: var(--ok); }
.ok-note svg { color: var(--ok); }

/* What the account holds: the address, the Telegram link, the phone. */
.idents { margin: 14px 0 4px; }
.idents > div {
  display: flex;
  align-items: baseline;
  justify-content: space-between;
  gap: 12px;
  padding: 10px 0;
  border-top: var(--line-w) solid var(--line);
}
.idents dt { color: var(--muted); font-size: 14px; }
.idents dd {
  margin: 0;
  font-size: 15px;
  display: flex;
  align-items: center;
  gap: 8px;
  flex-wrap: wrap;
  justify-content: flex-end;
}

/*  The one field on the site that takes a tracking code. It sits in a card of
  its own under the check, quieter than the thing the screen is for. */
/*  The same ceiling as the card above it. They are stacked, and at 768 one
  measured 620px and the other 672px - two boxes in a column that do not line
  up with each other, which is the class of defect the owner has rejected a
  whole interface over before. */
.code-card { margin-top: 12px; max-width: 620px; margin-inline: auto; }
.code-h {
  margin: 0 0 8px;
  font-size: 15px;
  font-weight: 600;
}
.code-check { margin-top: 0; }
.code-check .box { text-transform: uppercase; }

/* What an adverse verdict costs the person reading it, on the one page a
   stranger opens. Never beside good news, where it would be padding. */
.means {
  margin: 10px 0 0;
  font-size: 15px;
  line-height: 26px;
  color: var(--muted);
  max-width: 34em;
  margin-inline: auto;
}

/* Signing out, and signing out of everything. */
.leave { margin-top: 14px; }

/* A disabled control is still drawn, with the seconds on it. A control that
   vanishes and comes back is a control nobody trusts. */
.btn[disabled] {
  opacity: 0.55;
  cursor: default;
  pointer-events: none;
}

/* ===========================================================================
   THE APPLE PASS — components added or rebuilt for it
   ---------------------------------------------------------------------------
   Everything below is new work and is kept in one block so it can be read as
   a unit. It obeys the same rules as the rest of the file: only transform and
   opacity are animated, every touch target is 48px, nothing is loaded from
   another host, and no rule here needs a script to be true.
   ========================================================================= */

/*  --- the pill ------------------------------------------------------------
  docs/APPLE_DESIGN.md, button-primary: "rounded {rounded.pill} (full pill —
  capsule-shaped), padding 11px × 22px. The full-pill radius IS the brand
  action signal." And the press state, system-wide, is scale(0.95).

  So every button on this site is a capsule now, and the press is the same
  gesture everywhere rather than being reserved for the buttons that cost
  money. A press that only some buttons answer reads as some buttons being
  broken. */
.btn, .cta, .btn-plain, .btn-quiet, .btn-quiet-wide {
  border-radius: var(--r-pill);
}
.btn { font-weight: 600; font-size: var(--t-body); }
.btn:active, .cta:active, .btn-plain:active, .btn-quiet:active {
  transform: scale(0.95);
}
/*  The colour eases; the press does not. Declared on transform explicitly -
  never `all`, which would catch a layout property somebody adds later. */
.btn, .cta, .btn-plain, .btn-quiet {
  transition: background var(--dur-quick) var(--ease),
              border-color var(--dur-quick) var(--ease),
              color var(--dur-quick) var(--ease),
              transform 0s;
}

/*  --- the plan cards, and the reveal he asked for twice -------------------

  He asked for this in his own words: touching a plan on a phone, or bringing
  the pointer over it on a desktop, should show a sample of that plan
  underneath, with an animation.

  How it works with JavaScript off, which is the requirement that shaped it:

  * The trigger is a <summary> inside a <details>. A disclosure is a control
    the browser itself owns - it opens on a tap with no script on the page at
    all. That is the phone case, and most of his customers are on a phone.
  * The pointer case is a plain :hover rule, guarded by (hover: hover) so it
    can never stick after a tap on a touch screen.
  * With the stylesheet gone as well, the panel is content in normal flow:
    readable, in order, below its card. Rule 1.

  The panel is positioned against .podium rather than against its own card, so
  it opens at the width of the whole row. At 360px a card is about 95px wide
  and a panel that size could hold nothing worth reading.

  It is absolutely positioned, so opening it moves nothing on the page - which
  is what lets the whole reveal be transform and opacity, per rule 11. Nothing
  here animates a width, a height or a margin. */
.podium { position: relative; }

/*  The checkbox carries the open state. It stays in the page rather than being
  display:none, so it can still be reached and toggled from a keyboard; it is
  clipped to a pixel instead. */
.peek-toggle {
  position: absolute;
  width: 1px;
  height: 1px;
  margin: 0;
  padding: 0;
  opacity: 0;
  pointer-events: none;
}

/*  The trigger, inside its card. A <label> pointing at that checkbox, at the
  48px this site gives everything a thumb has to hit. */
.peek-btn {
  display: flex;
  align-items: center;
  justify-content: center;
  gap: 4px;
  min-height: 48px;
  margin-top: 6px;
  font-size: var(--t-fine);
  color: var(--muted);
  border-radius: var(--r-pill);
}
.peek-btn:hover { color: var(--navy-900); }
.peek-btn svg { transition: transform var(--dur-quick) var(--ease); }

/*  The panel. A sibling of the cards, not a child of one, so no card's
  transform can trap it - and positioned against .podium, so it opens at the
  width of the whole row at every width. */
.peek-panel {
  position: absolute;
  top: 100%;
  inset-inline: 0;
  z-index: 20;
  margin-top: 12px;
  padding: 20px;
  background: var(--paper);
  border: var(--line-w) solid var(--hairline);
  border-radius: var(--r-lg);
  text-align: right;

  /*  Closed means NOT LAID OUT, not `opacity: 0`.

    With opacity alone the three panels stayed in the layout, stacked on each
    other and on the section below - which is docs/BUGS.md B1, «the sample panel
    opens and falls on top of the section below it». It also made
    scripts/phone.py report 105 overlapping elements on the home page of every
    phone profile, because that probe reads an element's OWN opacity and these
    inherit theirs from the panel: real findings were buried under a hundred
    phantoms.

    Exactly the same mistake as the menu panel at the top of this file, and it
    gets the same answer. The entrance keeps its animation - an animation runs
    when an element becomes displayed, where a transition cannot. */
  display: none;
}

/*  Open by tap - a checked checkbox, which needs no script. */
#peek-basic:checked ~ .peek-basic,
#peek-full:checked ~ .peek-full,
#peek-fmi:checked ~ .peek-fmi {
  display: block;
  animation: rise var(--dur-lift) var(--ease);
}
#peek-basic:checked ~ .plan-basic .peek-btn svg,
#peek-full:checked ~ .plan-full .peek-btn svg,
#peek-fmi:checked ~ .plan-fmi .peek-btn svg { transform: rotate(180deg); }
#peek-basic:checked ~ .plan-basic .peek-btn,
#peek-full:checked ~ .plan-full .peek-btn,
#peek-fmi:checked ~ .plan-fmi .peek-btn { color: var(--navy-900); }

/*  Open by pointer. Guarded, so it can never stick after a tap on a phone -
  which is the trap docs/TYPE_AND_MOTION.md names for every hover rule here.
  The keyboard gets the same thing through :focus-within on the card. */
@media (hover: hover) and (pointer: fine) {
  .plan-basic:hover ~ .peek-basic,
  .plan-full:hover ~ .peek-full,
  .plan-fmi:hover ~ .peek-fmi,
  .plan-basic:focus-within ~ .peek-basic,
  .plan-full:focus-within ~ .peek-full,
  .plan-fmi:focus-within ~ .peek-fmi {
    display: block;
    animation: rise var(--dur-lift) var(--ease);
  }
}

.peek-h {
  margin: 0 0 12px;
  font-size: var(--t-small);
  font-weight: 600;
  color: var(--muted);
}
.peek-rows { margin: 0; }
.peek-rows .f {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 12px;
  padding: 7px 0;
  border-bottom: var(--line-w) solid var(--parchment);
  font-size: var(--t-small);
}
.peek-rows .f:last-child { border-bottom: 0; }
.peek-rows dt { color: var(--muted); }
.peek-rows dd { margin: 0; display: inline-flex; align-items: center; gap: 5px; }
.peek-rows .f-ok dd { color: var(--ok); }
.peek-rows .f-bad dd { color: var(--bad); }

/*  With no stylesheet the panel must not be a floating box over the page. It
  is not: every property that lifts it is in this file, so it falls back to
  ordinary flow. This rule is only about the open state being obvious when the
  CSS IS there. */
.peek[open] > summary { color: var(--navy-900); }

/*  --- the dead .f-mark ----------------------------------------------------
  The em-dash span is gone from the markup; all three feature states draw an
  <svg> now. The old rules are removed rather than left to rot, and this note
  is here so nobody reintroduces a <span> marker: `.rows li > span` has a
  specificity of (0,1,2) and will beat any `.f-mark` rule that tries to stop it
  growing, which is exactly how the markers ended up on the wrong side. */
.rows .f-na { color: var(--ink-48); }

/*  --- the result image ----------------------------------------------------

  This rule used to read `.shot img, .result-shot img`, and it matched nothing
  at all: `.shot` IS the <img>, so `.shot img` asks for an image inside an image.
  The radius and the product shadow it intended have therefore never applied to
  anything since the day they were written. Found by measuring rather than by
  reading, which is the only way this class of defect is ever found.

  And the radius is deliberately NOT restored on the picture. At 1920 the
  full-bleed bar across the top of the generated PNG was cut into wedges by a
  rounded corner - the bar bleeds to the file's own edge, so any rounding of the
  image clips it. The template is frozen and the owner likes it, so the answer is
  on this side: the rounded shape is the CARD, the picture inside it is square
  and whole, and the pad between them is what makes the corner read as a corner.

  The shadow does apply now, and this is the one thing in the system entitled to
  it - a picture resting on a surface. docs/APPLE_DESIGN.md allows exactly one
  drop shadow and this is it. */
.shot-box .shot {
  box-shadow: var(--shadow-product);
}

/* ===========================================================================
   WHAT LOOKING AT THE 1920 PICTURES CHANGED
   ---------------------------------------------------------------------------
   Everything above measured clean at 1920 and still had four things wrong with
   it that only a picture shows. This block is those four, and it is last in
   the file so it wins.
   ========================================================================= */

/*  1. A band is not allowed to set its own background any more. `.trust` said
  `background: var(--paper)` and `.live` said the same, which quietly beat the
  tile class on the same element and broke the alternation - two white bands in
  a row with a rule drawn between them. Apple puts no rule between tiles at
  all: "the colour change itself acting as the section divider". */
.trust { background: none; border-top: 0; }
.live { background: none; }

/*  2. A card inside the dark band was drawing a navy box on a near-black band:
  two dark surfaces fighting, which is the "boxes are the worst" complaint in
  its purest form. On a dark tile there is no card - the band IS the card. */
.tile-dark .card,
.tile-dark .moat-card {
  background: none;
  border: 0;
  padding: 0;
  color: var(--on-dark);
}
.tile-dark .moat-a { color: var(--on-dark); font-size: var(--t-head); line-height: var(--lh-head); }
.tile-dark .moat-b { color: var(--on-dark-muted); font-size: var(--t-lead); }
.tile-dark .sample {
  display: inline-flex;
  align-items: center;
  gap: 8px;
  min-height: 48px;
  padding: 0 22px;
  border: var(--line-w) solid rgba(255, 255, 255, 0.4);
  border-radius: var(--r-pill);
  color: var(--on-dark);
}
.tile-dark .sample:hover { background: rgba(255, 255, 255, 0.1); text-decoration: none; }

/*  3. The section headings were still on the old ladder - 18px and 20px, set
  before the scale changed, so the one part of the page that most needed to be
  big stayed small. They are on the ladder now like everything else. */
.plans-h { font-size: var(--t-head); line-height: var(--lh-head); margin-bottom: 16px; }
.anchor-a { font-size: var(--t-lead); font-weight: 600; }
.anchor-b { font-size: var(--t-body); }

@media (min-width: 1069px) {
  /*  4. The plans band was a 1100px podium and a 520px specimen adrift in a
    1920px window - the same "empty" the hero had, one band lower. Apple runs a
    WIDE container for a grid and a NARROW one for text, and this band has
    both, so it gets both: the row of cards takes the wide one, and every piece
    of prose in the band is pulled back to the reading width underneath it. */
  .plans .wrap { max-width: 1280px; }
  .podium { max-width: none; }
  .plans-h, .anchor-a, .anchor-b, .step-line, .free-note {
    max-width: 980px;
    margin-inline: auto;
  }
  .spec { max-width: 760px; }

  /*  The hero's two columns were top-aligned, so the words ended halfway up a
    card that is much taller than they are and left a hole under them. */
  .hero-in { align-items: center; }
}

/*  The contact band held one short line, right-aligned, in a band as wide as
  the window - which at 1920 is a sentence adrift in a field. It is centred and
  on the ladder now, and the rows sit as a row of three rather than a stack, so
  the band has something in it at every width. */
.trust-in { align-items: center; justify-content: center; text-align: center; }
.trust-row { font-size: var(--t-lead); }
@media (min-width: 735px) {
  .trust-in { gap: 12px 48px; }
}


/*  A short page left a band of bare white under the footer at 1920x1080 - the
  body's own background showing below everything. It is the "empty page" in its
  smallest form and the fix is the ordinary one: the page is a column as tall as
  the window, and the middle of it takes up the slack. */
body { display: flex; flex-direction: column; min-height: 100vh; }
main { flex: 1 0 auto; }
.foot { flex: 0 0 auto; background: var(--parchment); border-top: 0; }


/* ===========================================================================
   THE SCAMMINDER PASS
   ---------------------------------------------------------------------------
   docs/SCAMMINDER.md §5. Every rule here works with JavaScript disabled and
   loads nothing from another host.

   What is taken is PRINCIPLES - the type scale, a sentence under every value,
   columns that fill the width, an entrance that arrives in order, a page that
   is not flat white. Those transfer to any product.

   What is deliberately NOT taken is that site's gauge, and the reasoning is
   written out where it used to live: a component whose shape means "how much"
   cannot carry an answer that is yes or no. If anything else here ever starts
   implying a number, a score or a proportion this product does not measure, it
   goes the same way.

   The one sentence behind all of it is his:
   «باید یکم روح داشته باشه سایت» - the site needs some life in it.
   ========================================================================= */

/*  --- the verdict is a status, not a measurement ------------------------

  What stood here was `@property --gauge-angle`, a conic-gradient arc, a punched
  hole, a 1200ms fill and 21 step classes - Scamminder's gauge, adapted.

  It is deleted, and the reason is worth keeping so nobody rebuilds it. That
  ring depicts a quantity: their product is a trust score out of a hundred and a
  ring that fills is exactly right for it. Ours is yes or no. A ring sitting half
  filled above «این دستگاه قفل iCloud دارد» tells the reader the phone is partly
  locked, and there is no such state - so it was not a component that merely
  failed to fit, it was one that made a false statement on the page whose whole
  value is being believed. docs/SCAMMINDER.md notes in §2.8 that a Find My check
  produces no score; that was the moment to stop rather than to adapt.

  A status gets the shape a status has: one mark, one colour, generous size,
  with the heading carrying the word so the colour is never alone.

  Everything else taken from that research stays - the type scale, the evidence
  tile with a sentence under every value, the two-column layout, the staggered
  entrance, the gradient. Those are ideas. The gauge was a component with a
  meaning of its own, and the meaning was not ours. */
.verdict-mark-big {
  margin: 0 0 14px;
  line-height: 1;
}
.verdict-mark-big svg {
  width: 56px;
  height: 56px;
}
@media (min-width: 1069px) {
  .verdict-mark-big svg { width: 72px; height: 72px; }
}
@media (max-width: 559px) {
  .verdict-mark-big svg { width: 48px; height: 48px; }
}

/*  --- 10. the provenance under the verdict -------------------------------
  A verdict with a date attached is a record; a verdict without one is an
  opinion. Two lines of CSS, and a large part of why their page is believable. */
.verdict-meta { font-size: var(--t-fine); color: var(--ink-48); margin: 10px 0 0; }

/*  --- 11. a route out of an adverse verdict ------------------------------
  Drawn ONLY for an adverse or unknown verdict. A dispute link under a clean
  verdict is a control that exists to be ignored, which is the defect he has
  named more than once. Body text with a link, not a button: it is a route, not
  the action of the page. */
.verdict-out { font-size: 14px; color: var(--muted); margin: 12px 0 0; }

/*  --- 3. the evidence tile, and a sentence under every value ------------

  Their three-line tile, with `text-transform: uppercase` and
  `letter-spacing: 0.08em` both dropped - §4.3. Persian is a joined script:
  tracking breaks the joins between letters, and uppercase does nothing at all.
  The label is distinguished by size and colour instead.

  `minmax(200px, 1fr)` rather than their 180px, because a Persian label needs
  more room than "Final score". `auto-fit` is what fills whatever width the tile
  is given with no media query at all. */
.evidence {
  display: grid;
  grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
  gap: 12px;
}
.ev {
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-md);
  padding: 12px 14px;
  background: var(--parchment);
}
.ev-label { font-size: var(--t-fine); color: var(--ink-48); margin: 0 0 4px; }
.ev-value { font-size: 17px; font-weight: 700; margin: 0; }
/*  The whole point of the component: where "what this means for the buyer"
  lives, which was written down as a rule and never built. */
.ev-note { font-size: 13px; line-height: 1.6; color: var(--muted); margin: 6px 0 0; }
.ev.ev-clean .ev-value { color: var(--v-clean); }
.ev.ev-bad .ev-value { color: var(--v-blocked); }
.ev.ev-unknown .ev-value { color: var(--v-unknown); }

/*  --- 7. the cards arrive in sequence ------------------------------------

  Five classes, no JavaScript, and it is most of what "the site should have some
  soul" means in practice. The 8px translateY is added to what they do: fading
  alone reads as a slow page, moving reads as arriving.

  Four steps and no more - by 250ms a visitor is waiting. */
@keyframes rise {
  from { opacity: 0; transform: translateY(8px); }
  to   { opacity: 1; transform: none; }
}
.rise   { animation: rise 300ms var(--ease) forwards; }
.rise-1 { animation: rise 300ms var(--ease) 50ms  forwards; opacity: 0; }
.rise-2 { animation: rise 300ms var(--ease) 100ms forwards; opacity: 0; }
.rise-3 { animation: rise 300ms var(--ease) 150ms forwards; opacity: 0; }
.rise-4 { animation: rise 300ms var(--ease) 200ms forwards; opacity: 0; }

/*  --- 6. the page is not flat white --------------------------------------

  The cheapest line in the document. On the hero tile only: the result page stays
  flat, because a report should not be decorated.

  The stops are derived from our own palette rather than copied - #eef2fa is
  --navy-100 lightened and #f4f6fb is --wash cooled. Not their indigo and purple:
  we have a navy brand and a purple stop would introduce a colour the site does
  not own. 160deg rather than `to bottom right`, so the wash reads the same in
  RTL, where a corner-to-corner gradient flips. */
.hero.tile-wash,
.tile-white.hero {
  background: linear-gradient(160deg, #eef2fa 0%, var(--paper) 48%, #f4f6fb 100%);
}

/*  --- 14. a surface answers a pointer, without a shadow ------------------
  A 2px transparent border that gains a colour, plus a 2px lift, so nothing
  shifts. No `hover:shadow-lg` - §4.1. The coarse-pointer guard matters: a lift
  on hover on a touch screen fires on tap and sticks. */
.tile-link {
  border: var(--line-w-strong) solid transparent;
  border-radius: var(--r-lg);
  transition: border-color var(--dur-quick) var(--ease),
              transform var(--dur-lift) var(--ease);
}
.tile-link:hover { border-color: var(--navy-700); transform: translateY(-2px); }
@media (pointer: coarse) { .tile-link:hover { transform: none; } }

/*  --- 12. touch targets, enforced once ----------------------------------
  Their block, once, rather than component by component. `summary` is in the
  list because the phone menu is a <details> and its summary is the single
  most-tapped target on the site. If :has() is unsupported the second rule is
  skipped and nothing breaks. */
@media (pointer: coarse) {
  button, [role="button"], input[type="submit"], input[type="button"],
  select, summary, .btn, .nav a, .menu-panel a { min-height: 44px; }
  button:has(> svg:only-child), a:has(> svg:only-child) {
    min-width: 44px;
    min-height: 44px;
    display: inline-flex;
    align-items: center;
    justify-content: center;
  }
}

/*  --- 16. the focus ring -------------------------------------------------
  Our --navy-700 in place of their indigo. The `:focus:not(:focus-visible)` line
  is what stops a ring appearing on a mouse click while keeping it for the
  keyboard. */
:focus-visible {
  outline: 2px solid var(--navy-700);
  outline-offset: 2px;
  border-radius: var(--r-xs);
}
:focus:not(:focus-visible) { outline: none; }
@media (prefers-contrast: high) {
  :focus-visible { outline: 3px solid currentColor; outline-offset: 3px; }
}

/*  --- 15. the performance rules -----------------------------------------
  Free, and they matter on one small server. `contain-intrinsic-size` is set
  close to a real section height: get it badly wrong and the scrollbar jumps as
  sections render. */
section { content-visibility: auto; contain-intrinsic-size: auto 600px; }
img, video { max-width: 100%; height: auto; display: block; }

/*  --- when money moves: the notice (docs/LOGIC2.md R91-R95) -------------

  R91: pinned to the top, on the same side as the balance in the bar, on every
  page - so the eye travels from the sentence to the figure it is about. A notice
  that appears on the right on one page and the left on another is the class of
  defect he has rejected by name.

  R92: `position: fixed`, and only transform and opacity animate. No layout
  property in any transition, no blur, no shadow animation.

  R93: it never blocks anything. `pointer-events` is off on the container and
  back on for each card, so the page underneath stays clickable everywhere the
  notice is not. On a phone it is a full-width strip at the top rather than a
  floating box, so it cannot land on the IMEI field or steal the caret. */
/*  On a phone it is IN THE FLOW, directly under the bar, and this is R93 read
  literally: «on a 360px screen it sits above the fold's content, not over the
  form». Two pictures decided it. Pinned at 8px it covered the whole top bar -
  the mark, the menu and the balance. Moved below the bar it still covered the
  page's own heading, because a 374px card 156px tall on a 390px screen is not a
  toast, it is a wall.

  In the flow it covers nothing, it cannot reach an input, and the server has
  already drawn it before the page paints - so on load there is no shift at all.
  A notice that arrives later does move the page down, and that is the right
  trade on a phone: R92 forbids ANIMATING a layout property, which this does not
  do, and being unmissable is the whole point of the feature.

  Above 735px there is room beside the content, so it floats in the corner where
  the balance is - R91 - and covers nothing there either. */
.notices {
  position: static;
  margin: 8px auto 0;
  width: min(100%, 980px);
  padding: 0 12px;
  display: flex;
  flex-direction: column;
  gap: 8px;
}
.notice-card {
  pointer-events: auto;
  background: var(--paper);
  border: var(--line-w) solid var(--line);
  border-radius: var(--r-lg);
  padding: 12px 14px;
  max-width: 420px;
  margin-inline-start: auto;
  margin-inline-end: 0;
}
.notice-t { margin: 0 0 4px; font-size: 15px; font-weight: 700; }
.notice-b { margin: 0; font-size: 14px; line-height: 1.6; color: var(--muted); }
.notice-row {
  display: flex;
  align-items: center;
  justify-content: space-between;
  gap: 10px;
  margin-top: 8px;
}
/*  48px, keyboard reachable, and it is a real control rather than an icon
  somebody has to aim at. */
.notice-x {
  display: inline-flex;
  align-items: center;
  justify-content: center;
  min-width: 48px;
  min-height: 48px;
  margin: -12px -14px -12px 0;
  border: 0;
  background: none;
  color: var(--muted);
  font-size: 20px;
  line-height: 1;
}
.notice-x:hover { color: var(--navy-900); text-decoration: none; }

@media (min-width: 735px) {
  .notices {
    position: fixed;
    top: calc(44px + 12px);
    inset-inline-start: auto;
    inset-inline-end: 4%;
    width: 380px;
    margin: 0;
    padding: 0;
    z-index: 60;
    /*  Off for the container and back on for each card, so the page underneath
      stays clickable everywhere the notice is not. */
    pointer-events: none;
  }
  .notice-card { max-width: none; }
}
@media (min-width: 1069px) {
  .notices { top: calc(56px + 12px); }
}


/* ===========================================================================
   NO EMPTY SIDES, ANYWHERE
   ---------------------------------------------------------------------------
   He has said this five times and called himself sensitive about it:

     «من رو این موضوع حساس هستم»
     «اصلا نمیخوام هیچ جای سایت طرفینش خالی باشه»

   Measured before this block, at 1920: the content column was 620px on the
   result page and 980px everywhere else, so between 650px and 470px of each
   side was doing nothing at all, on ten pages.

   The fix is two things and NOT simply a bigger number, because a 1900px line of
   Persian text is worse than a short one, not better:

   1. The RAIL grows with the window, as a percentage with a ceiling. A tile's
      background already ran edge to edge (§4.4), so what was empty was the
      content inside it, and that is what widens.
   2. The CONTENT inside the rail becomes columns, so the extra width holds
      something. `repeat(auto-fit, minmax(...))` does that with no media query,
      and `minmax(320px, ...)` is the floor that stops any column becoming a
      ribbon again.

   Prose keeps its own ceiling in `em` wherever it is set, which is why widening
   the rail does not produce unreadable lines: a paragraph stops at 30-34em on
   its own.
   ========================================================================= */

@media (min-width: 1069px) {
  /*  980px was Apple's `.viewport-content`, and copying it was right for a
    page of prose and wrong for this site at 1920. 90% with a 1440 ceiling keeps
    a 5% gutter on each side at every width: 96px at 1920 rather than 470. */
  .wrap { width: 90%; max-width: 1440px; padding: 0; }

  /*  The bar is `<div class="wrap top-in">` - ONE element with both classes -
    so the rule above was overriding `.top-in`'s own width simply by coming
    later in the file, and the mark sat 240px from the edge at 1920 instead of
    the 120 it asks for. Measured, not assumed. `.top .top-in` outranks `.wrap`
    on specificity, so the bar keeps its own rail wherever it is written. */
  .top .top-in { width: min(1680px, 96%); max-width: none; }

  /*  A page that is one short form - sign in, register, a code, the recovery
    screen - keeps a narrow column, because a password field 1440px wide is
    absurd. But 620px on a 1920px window is the ribbon he objects to, so the
    column grows with the window up to a sane ceiling. It is still the narrowest
    thing on the site and it is no longer a third of it. */
  .narrow .wrap { max-width: 860px; }

  /*  The result page. §5 item 4: the verdict, the dial and the picture in the
    leading column; the facts, the evidence and the code in the trailing one.

    `minmax(0, 1.15fr)` and not `1.15fr` - the `0` floor is what stops a long
    unbroken IMEI or a serial from forcing the grid wider than the window. And
    `minmax(320px, ...)` on the trailing column is the floor that stops it
    collapsing into a ribbon. Below 1069 it is one column in source order, which
    is already the right order on a phone. */
  .result-in {
    display: grid;
    grid-template-columns: minmax(0, 1.15fr) minmax(320px, 0.85fr);
    gap: 32px;
    align-items: start;
  }
  .result-lead { grid-column: 1; }
  .result-side { grid-column: 2; }
  /*  What to do next spans both, because it is about the whole page rather than
    about either column. */
  .result-next { grid-column: 1 / -1; }

  /*  The evidence fills the trailing column: two tiles across at this width
    without a single media query, because `auto-fit` works it out from the
    200px floor. */
}

/*  A wide screen deserves the picture larger, but not enormous: past a point a
  1080px PNG is being scaled up and goes soft. The leading column is where it
  lives and the column is what bounds it. */
@media (min-width: 1440px) {
  .shot-link { max-width: 640px; }
}

@media (max-width: 559px) {
  /*  §5 item 18. `.wrap` padded 16px at every width, and 16px each side of a
    360px phone is 9% of the screen spent on nothing - on the page with the most
    to show. 12px and not their 8px: 8px makes a card look as though it is
    falling off the screen, and our cards have visible borders where theirs rely
    on a shadow. */
  .wrap { padding: 0 12px; }
  .result-in .card { padding: 14px; }
}

/*  --- the other pages, so this is not one page's fix ---------------------

  Every block on the site that is a LIST of things becomes a grid that fills its
  rail. Nothing here needs a media query: `auto-fit` collapses to one column on a
  phone by itself, and each floor is chosen so a column never becomes a ribbon.
  docs/SCAMMINDER.md §3.4 - `auto-fit` is the cheap answer and it is the one that
  scales to every width rather than to the two somebody photographed. */

/*  The account page: three lists that were three stacked full-width blocks with
  nothing beside them. */
@media (min-width: 1069px) {
  .me-grid {
    display: grid;
    grid-template-columns: repeat(auto-fit, minmax(330px, 1fr));
    gap: 20px;
    align-items: start;
  }
  /*  The wallet, the same way: the balance and the form beside the history
    rather than above it with a column of white to the left. */
  .wallet-grid {
    display: grid;
    grid-template-columns: minmax(0, 1fr) minmax(320px, 0.8fr);
    gap: 28px;
    align-items: start;
  }
}

/*  The rows inside a fact list. docs/REVIEW.md U2 and docs/BUGS.md A7: «قفل
  iCloud» on one side and «روشن» a hundred steps away on the other, which is a
  two-column row stretched to the full width of a wide container. Widening the
  rail would have made that worse, so the row gets a ceiling of its own and the
  value sits near its label. */
.facts dl, .facts .cert { max-width: 46em; }
@media (min-width: 1069px) {
  .facts dl > div, .facts .f { gap: 20px; }
  /*  A list of label-and-value rows is exactly the thing that should be in two
    columns on a wide screen rather than one tall thin strip. */
  .facts .cert {
    max-width: none;
    display: grid;
    grid-template-columns: repeat(auto-fit, minmax(340px, 1fr));
    column-gap: 32px;
  }
}
