/* ===================================================================
   mission.css — /our-mission
   ===================================================================

   PORTED FROM AN EXTERNAL HANDOFF, 11 OCT 2026. Guillaume designed
   this page with ChatGPT; it arrived as a self-contained static site
   (CoRoute-Claude-Handoff, source commit ba160b2d). The approved
   visual design is the handoff's, and the rules below are its
   mission.css with ONE mechanical change applied throughout.

   ===================================================================
   EVERY SELECTOR IS SCOPED UNDER .mission, AND THAT IS NOT TIDYING.
   ===================================================================

   The handoff is a whole page and styles like one. Its stylesheet
   carries these GLOBAL selectors:

       *  html  body  a  button  h1  h2  h3  p  p+p
       .wrap  .hero  .brand

   Every one of those already means something else in app.css —
   .wrap is 75rem there and 1180px here; body is the system font
   stack there and 'DM Sans' here; .hero is the homepage's purple
   panel. Loading both stylesheets unscoped makes the result depend on
   which <link> came last, and the SHARED CHROME would be restyled by
   a stylesheet that knows nothing about it: .topbanner (the C1/D5
   bar), .flownav and .sitefooter all contain <p> and <a> elements
   this file has opinions about.

   So the page's own markup sits inside <div class="mission"> and
   every rule here is prefixed with it. body{} became .mission{},
   h2{} became .mission h2{}, and so on. The chrome sits OUTSIDE that
   div and is therefore untouchable from here — which is the property
   that lets the mandatory banner and the footer keep their approved
   appearance on a page with a completely different design language.

   DO NOT ADD AN UNSCOPED SELECTOR TO THIS FILE. A single `h2 { }` at
   the top level would reach into the chrome and into every other
   page that happens to be open in the same document, and the fault
   would show up as "the footer looks wrong on one page".

   WHAT WAS DELIBERATELY NOT SCOPED, and why:

     :root{...}        became .mission{...}. The custom properties
                       (--purple, --cream, --orange …) are the
                       handoff's palette and they MUST NOT leak onto
                       app.css's tokens, which use the same kind of
                       names for different values. Declared on the
                       scope element, var() still resolves for
                       everything inside it.

     *{box-sizing}     became `.mission, .mission *`. The scope
                       element needs it too.

     html{scroll-behavior:smooth; scroll-padding-top:35px}
                       re-declared below rather than dropped: a
                       document-level concern that cannot be scoped
                       to a div. Guarded by prefers-reduced-motion, as
                       in the handoff.

                       THE 35px IS NOW 0 — see the rule below. This
                       line records the handoff's value, not the live
                       one. It was meant to clear "the sticky chrome",
                       and there is no sticky chrome on this page:
                       nothing here is position:fixed or sticky at
                       1280 or 393, so it only held every anchor 35px
                       clear of nothing and showed that much of the
                       previous colour band.

   ===================================================================
   THE FONTS ARE SELF-HOSTED, AND THE CSP IS WHY.
   ===================================================================

   The handoff opened with

       @import url('https://fonts.googleapis.com/css2?family=DM+Sans…')

   That import is REMOVED. This site is served under

       Content-Security-Policy: default-src 'self'; style-src 'self'

   (platform/security.go) with no external style or font source, so
   the import is blocked by the browser and the two typefaces simply
   never arrive. Adding a font-src exception was considered and
   REJECTED: a request to fonts.gstatic.com hands the visitor's IP
   address to Google on every page view, and the privacy notice
   promises no third-party tracking. Nothing on this page is worth
   that.

   So the two faces are served from our own origin, under
   /static/fonts/. Both are SIL Open Font License 1.1 — verified
   11 Oct 2026 against google/fonts METADATA.pb (`license: "OFL"`)
   and the OFL.txt shipped beside each — which permits redistribution
   of the unmodified font software. The licence texts are committed
   next to the files as OFL-Manrope.txt and OFL-DMSans.txt, which is
   the OFL's own condition for redistribution.

   LATIN AND LATIN-EXT ONLY. Google serves these as a dozen subsets
   (cyrillic, greek, vietnamese…). This page is en-GB, so the four
   files here are the two latin subsets of each family: 95KB in
   total, against roughly 500KB for the full set. The unicode-range
   on each @font-face is Google's own, so a character outside those
   ranges falls back to the system stack rather than rendering from a
   font that lacks the glyph.

   VARIABLE FONTS, one file per family per subset. `font-weight:
   400 800` on the @font-face is not a typo — it declares the range
   the variable axis covers, and the 400/500/600/700/800 weights the
   page asks for are all instances of that one file.

   font-display:swap, as the handoff had it: the text paints
   immediately in the fallback and reflows when the face arrives.
   Measured with the fonts absent, this page's layout is nearly
   identical in the system stack — the line breaks in the hero are
   the same — so the swap is a change of letterform, not of layout.
   =================================================================== */

/* The four faces. Paths are absolute from the site root because this
   stylesheet is served from /static/ and the fonts from
   /static/fonts/ — a relative url() would resolve correctly today and
   break the moment the stylesheet moves. */
@font-face {
  font-family: 'Manrope';
  font-style: normal;
  font-weight: 400 800;
  font-display: swap;
  src: url(/static/fonts/manrope-latin.woff2) format('woff2');
  unicode-range: U+0000-00FF, U+0131, U+0152-0153, U+02BB-02BC, U+02C6,
    U+02DA, U+02DC, U+0304, U+0308, U+0329, U+2000-206F, U+20AC, U+2122,
    U+2191, U+2193, U+2212, U+2215, U+FEFF, U+FFFD;
}
@font-face {
  font-family: 'Manrope';
  font-style: normal;
  font-weight: 400 800;
  font-display: swap;
  src: url(/static/fonts/manrope-latin-ext.woff2) format('woff2');
  unicode-range: U+0100-02BA, U+02BD-02C5, U+02C7-02CC, U+02CE-02D7,
    U+02DD-02FF, U+0304, U+0308, U+0329, U+1D00-1DBF, U+1E00-1E9F,
    U+1EF2-1EFF, U+2020, U+20A0-20AB, U+20AD-20C4, U+2113, U+2C60-2C7F,
    U+A720-A7FF;
}
@font-face {
  font-family: 'DM Sans';
  font-style: normal;
  font-weight: 400 700;
  font-display: swap;
  src: url(/static/fonts/dmsans-latin.woff2) format('woff2');
  unicode-range: U+0000-00FF, U+0131, U+0152-0153, U+02BB-02BC, U+02C6,
    U+02DA, U+02DC, U+0304, U+0308, U+0329, U+2000-206F, U+20AC, U+2122,
    U+2191, U+2193, U+2212, U+2215, U+FEFF, U+FFFD;
}
@font-face {
  font-family: 'DM Sans';
  font-style: normal;
  font-weight: 400 700;
  font-display: swap;
  src: url(/static/fonts/dmsans-latin-ext.woff2) format('woff2');
  unicode-range: U+0100-02BA, U+02BD-02C5, U+02C7-02CC, U+02CE-02D7,
    U+02DD-02FF, U+0304, U+0308, U+0329, U+1D00-1DBF, U+1E00-1E9F,
    U+1EF2-1EFF, U+2020, U+20A0-20AB, U+20AD-20C4, U+2113, U+2C60-2C7F,
    U+A720-A7FF;
}

/* THE ONE DOCUMENT-LEVEL RULE, lifted out of the scope because a div
   cannot carry it. scroll-padding-top keeps an in-page anchor from
   landing underneath the chrome; smooth scrolling is the handoff's
   and is switched off for anyone who asked for less motion.

   35px CLEARED THE CHROME AND NOTHING ELSE, which was the bug.
   15 Oct 2026, from an external review: "when I click on the link,
   I'd expect to scroll fully into the white section, not see a bit of
   the yellow one above."

   He was right, and it was every anchor, not one. Measured at 1280:
   #why, #cover and #approach all landed with their section top at
   ~35px, because this value equals the chrome it is clearing and no
   target carried a scroll-margin-top of its own. So exactly 35px of
   the PREVIOUS section stayed on screen — at #cover that is the cream
   .coverage panel (#F0E7DB) sitting above white content, which is the
   strip he described. On a phone it was worse: his screenshot of
   coroute.co shows the previous section's "Help drivers improve"
   heading still legible across the top after the jump.

   THE FIX IS TO REMOVE THE OFFSET, AND THE FIRST ATTEMPT WENT THE
   WRONG WAY. A scroll-margin-top on the targets was tried first, on
   the reasoning that the anchor should "land clear of the section
   above". That is backwards: scroll-margin-top pushes the target
   DOWN the viewport, so it shows MORE of the previous section, not
   less. Measured: it took the strip from 35px to 124px, and on a
   phone the heading "03 Make cover clear" from the section being left
   became legible at the top. Reverted.

   THE CHROME THIS WAS CLEARING DOES NOT EXIST. Checked at 1280 and
   393: nothing on this page is `position: fixed` or `sticky` — the
   banner and nav scroll away with the document. So the 35px was
   holding the anchor 35px clear of nothing, and 35px of the previous
   colour band is precisely what that bought.

   0 is therefore correct rather than merely smaller: the section's
   own 65px of padding-top gives the heading its air, and the band
   above ends exactly at the top of the viewport.

   IF A STICKY HEADER IS EVER ADDED TO THIS PAGE, this must become its
   height — and at that point scroll-padding-top is still the right
   property, because the header would belong to the document. */
html:has(.mission) { scroll-behavior: smooth; scroll-padding-top: 0; }

@media (prefers-reduced-motion: reduce) {
  html:has(.mission) { scroll-behavior: auto; }
}

/* ===================================================================
   Below this line: the handoff's mission.css, scoped. Its own
   section comments are preserved.
   =================================================================== */

/* ===================================================================
   THE SCOPED PALETTE, NOW PER THEME — v2, 22 Oct 2026
   ===================================================================

   THE NAMES ARE THE HANDOFF'S AND THEY STAY. --purple no longer
   names a purple and --cream no longer always names a cream, which
   is a real cost and is paid deliberately: every one of the ~54
   var() references below reads one of these seven, and renaming them
   would touch every rule in a 2,900-line file for no behavioural
   gain. What each one MEANS is written beside it instead, which is
   the thing a reader actually needs.

     --purple   a PANEL GROUND only, since 22 Oct 2026. It backs the
                closing panel, the chat bubble and the spark tile:
                #142B49 light, #182B43 dark.

                ⚠️ IT USED TO DO TWO JOBS AND THAT WAS THE BUG. The
                same token was also the colour of seven pieces of
                TEXT (.mission-strip b, .chat-status, the question
                chips and four rules at 2063+), which works while one
                value serves both -- a navy that is a dark ground in
                one place and dark ink in another, both on light
                surroundings. It stops working the moment the token
                goes per-theme: in dark mode #182B43 is correct as a
                card and 1.06:1 as text on the page. Measured on the
                rendered page: the five strip numbers, "£1,860", the
                retention aim and the AI answer heading were all
                invisible.

                A token cannot be per-theme and mean two things, so
                the ink uses moved to --strong-ink below. --purple is
                now only ever a background.
     --strong-ink  the emphatic text colour -- navy on light, near-
                white on dark. What --purple used to be when it was
                being used as ink. 13.64:1 / 14.21:1 on --cream.
     --navy     the contextual section. Dark in both themes, but NOT
                the same dark: sampled off the rendered boards,
                MissionV2 draws it #142B49 and MissionV2Dark draws it
                #0B1626. The page had the DARK board's value pinned
                in both, which is why it read correctly at night and
                a shade too deep by day. It is per-theme now and its
                values are app.css's --band exactly, so this band and
                the homepage's contextual band cannot drift apart.
     --cream    the PAGE GROUND. Light in light mode, the dark page
                in dark.

                ⚠️ IT IS NEVER AN INK, since 22 Oct 2026. It was used
                as `color` in five rules -- the contextual band, its
                meaning rule, the trust icon, the closing panel and
                the footer wordmark -- which works while the token is
                one cream serving as both the page and as text on a
                dark band. It stops working the moment it is
                per-theme: in dark mode it is #101D30, so the
                contextual h2 drew at 1.07:1 and the drive title at
                1.19:1. All five surfaces are dark in BOTH themes, so
                their ink is the literal #FFF9F1 in both. Same class
                of bug as --purple above, found the same way.
     --orange   the brand orange: the ring, the accents.
     --ink      body text.
     --muted    secondary text.
     --line     hairlines.

   MEASURED, light / dark:
     --ink on --cream            13.64:1 / 14.21:1
     --muted on --cream           5.42:1 /  9.65:1
     --hero-text on --hero-bg    12.08:1 / 11.98:1
     cream text on --navy        17.35:1 / 15.23:1

   WHY THE HERO'S GROUND IS NOT --purple ANY MORE. It cannot be: the
   same token is used as a TEXT colour in nine rules, and a value
   that works as navy ink on cream cannot also be a pale-sky panel.
   The hero takes app.css's --hero-bg / --hero-text directly, which
   are already per-theme and are the same tokens the homepage hero
   uses -- so the two heroes cannot drift apart, which they had
   already done once under the old arrangement. */
.mission{
  --purple:#142B49;      /* a PANEL GROUND only -- see above */
  --strong-ink:#142B49;  /* the emphatic TEXT colour */
  /* THE ACCENT AS *TEXT*, WHICH IS NOT --orange AND NOT --logo.
     --orange and --logo are LOGOTYPE colours: WCAG 1.4.11 exempts a
     brand mark from the contrast requirement, and #E8762A measures
     2.98:1 on white, which is fine for a ring and not for a word.
     This token is the same family pushed dark enough to be read as
     TEXT on a light ground -- #9A4A00, 6.26:1 on --ui-card -- and
     light enough in dark mode -- #FFB264, 8.06:1. It is what the
     "CoRoute AI" wordmark and any other accent SENTENCE uses. */
  --accent-text:#9A4A00;
  --navy:#142B49;     /* the contextual band -- see above */
  --cream:#FFF9F1;    /* the page ground */
  --orange:#E8762A;   /* the brand orange -- logotype and accents */
  --ink:#142B49;
  --muted:#55697d;
  --line:#cbd8e1;
}
/* THE SELECTORS START WITH .mission, WHICH IS NOT A STYLE CHOICE.
   TestMissionStylesheetIsScoped requires every prelude in this file
   to begin with .mission, and it is right to: an unscoped prelude
   here reaches the shared banner, nav and footer, and the symptom
   turns up on a different page entirely.

   The theme condition is an ancestor (<html data-theme>), and CSS
   has no way to put an ancestor AFTER the subject -- so it goes in
   :where(), which matches the ancestor without contributing
   specificity. `.mission:where(:root[data-theme="dark"] *)` reads
   "this .mission, when it is inside a dark-themed root" and is
   (0,1,0) exactly like a bare `.mission`, so the cascade order
   between the light block above and the dark blocks here is decided
   by SOURCE ORDER, which is what a theme override wants. */
@media (prefers-color-scheme: dark){
  .mission:where(:root:not([data-theme="light"]) *){
    --purple:#182B43;
    --navy:#0B1626;
    --strong-ink:#E5ECF3;
    --accent-text:#FFB264;
    --cream:#101D30;
    --orange:#FFB264;
    --ink:#E5ECF3;
    --muted:#B7C5D6;
    --line:#2f4a68;
  }
}
.mission:where(:root[data-theme="dark"] *){
  --purple:#182B43;
  --navy:#0B1626;
  --strong-ink:#E5ECF3;
  --accent-text:#FFB264;
  --cream:#101D30;
  --orange:#FFB264;
  --ink:#E5ECF3;
  --muted:#B7C5D6;
  --line:#2f4a68;
}
/* ---- THE PROSE TYPE SCALE — 11 Oct 2026 --------------------------

   Guillaume, on the approach strip under the hero: "this feels too
   small .. can we have a default text size for the written stuff on
   the mission page", clarifying "for the blabla text at least".

   THE FAULT WAS NOT ONE SIZE, IT WAS FOURTEEN. An audit of this page
   at 1280 — every element with its own run of text 18 characters or
   longer, headings and SVG excluded — found FOURTEEN distinct prose
   sizes between 8px and 37px, with no system. The same kind of body
   copy was 17px in the AI section and 20px in section 01, and
   genuine sentences, including two honesty disclosures, sat at 11px
   and 14px.

   FOUR STEPS, AND PROSE USES ONLY THESE:

     --t-body  19px  the default for prose. A person reading a
                     sentence on this page gets this unless there is
                     a reason.
     --t-small 17px  secondary prose: supporting paragraphs in a
                     narrow or dark column, card copy, step copy.
     --t-cap   15px  captions, notes and disclosures. The floor for
                     anything made of sentences.
     --t-micro 12px  eyebrows and strip labels. NOT for sentences.

   WHY 19px FOR THE DEFAULT. 17px was the most common prose size
   here and Guillaume's note is that the page reads small, so the
   default had to move up; 19px is one step up from that most-common
   value and sits comfortably under the 22px lede without competing
   with it. It is a marketing page read at arm's length, not a dense
   application.

   WHY A 15px FLOOR FOR DISCLOSURES. .hero-note and .demo-foot were
   11px. A pre-launch insurance site whose disclaimer is the hardest
   thing on the page to read is a bad look on its own terms, and the
   claim beside .hero-note was 37px — a 3.4x ratio between a promise
   and its qualification. 15px is not full body size, which would
   over-weight them, but it is legible.

   .hero-note ITSELF IS GONE as of 12 Oct 2026 — canvas v191 dropped
   "CoRoute is pre-launch. This is the future we're building towards."
   from the hero, along with the "Why CoRoute exists" eyebrow, and the
   page follows the board. Its four rules went with it. THE FLOOR
   STILL APPLIES to .demo-foot and to every disclosure added later;
   the reasoning is why the number is 15px, not a note about one
   element. The authorisation disclosure was never this element: it
   is the banner above the hero and the footer's "not yet authorised
   to sell insurance", both untouched.

   WHAT IS DELIBERATELY NOT ON THIS SCALE: text baked into the
   ILLUSTRATIONS — the dashcam HUD, the camera readout, the portal
   chrome, the map credit. Those are 7-10px because they are UI
   chrome inside a picture, and they are meant to read as small
   interface text rather than as the page's own voice. Raising them
   would make the illustrations look wrong, not more readable.
   Headings are not on it either; they keep their display sizes. */
.mission{--t-body:21px;--t-small:17px;--t-cap:15px;--t-micro:12px}
.mission, .mission *{box-sizing:border-box}
.mission{margin:0;background:var(--cream);color:var(--ink);font-family:'DM Sans',sans-serif;font-size:var(--t-body);line-height:1.65}
.mission a{color:inherit}
/* THE BUTTON RESET — WIDENED DURING THE PORT, AND THIS IS WHY.
   =================================================================

   The handoff's rule was exactly:

       button{font:inherit;cursor:pointer}

   which is all a standalone page needs. It is NOT enough here, and
   the difference was found by measuring the built page against the
   handoff's own rendering rather than by reading the CSS.

   app.css styles bare `button` for the quote flow's form controls:

       button { width: 100%; margin-top: 0.5rem; min-height: 44px;
                padding: …; border: 2px solid …; border-radius: …; }
       @media (min-width: 34rem) { button { width: auto;
                                            min-width: 14rem; } }

   Those are full-width form submits, which is right for a
   questionnaire and wrong for everything on this page. Scoping
   mission.css did NOT protect against it: `.mission .tabs button` is
   more specific than `button` and wins on the properties it SETS —
   background, border, padding, radius — but mission.css never set
   width, min-width, min-height or margin-top, so app.css's values
   applied unopposed.

   MEASURED AT 1280px BEFORE THIS RULE EXISTED, every one of the
   eleven buttons on the page came out 224px wide (min-width: 14rem)
   with an 8px top margin:

     "Avoiding a hazard" / "Following too closely"   224px each,
         so the two scenario pills stacked into a column instead of
         sitting side by side on one row — visible in a screenshot,
         and the reason this was caught
     the replay play button                          224px wide at
         44px tall with border-radius:50%, i.e. a stretched ellipse
         where the design draws a circle
     the three coverage question chips               224px each

   Nothing in the markup was wrong and no text comparison could have
   seen any of it. A bar chart still contains the word "Premium".

   SO THE RESET TAKES BACK THE FOUR PROPERTIES app.css SETS AND
   mission.css DOES NOT. It is deliberately a RESET to the browser
   default rather than a set of values: the handoff's own rules then
   apply on top exactly as they did in the delivered page, so this
   restores the approved design rather than reinterpreting it.

   width:auto and min-width:0 — not a width of our choosing. Every
   button here is sized by its own padding and content, which is what
   a pill does.

   min-height:0 — the handoff sets explicit heights where it wants
   them. NOTE the accessibility trade: app.css's 44px is a touch-
   target minimum and this removes it.

   MEASURED 11 OCT at 1280 AND 390 (both identical), because an
   earlier version of this comment guessed and got it wrong — it
   claimed the scenario pills "measure 44px" when they measure 40:

       Avoiding a hazard / Following too closely     40px
       A friend driving / A cracked windscreen /
           A work journey  (the coverage chips)      35px
       the replay play button                        33x42 at 1280,
                                                     32x50 at 390

   SIX controls under the 44px target, not two, and the phone is no
   better than the desktop — which is the case that matters, since
   44px exists for fingers. The moment cards are 62px and fine.

   Raising them would change Guillaume's approved design, so this is
   recorded as a known departure rather than silently fixed or
   silently ignored. It is an OPEN question, not a settled trade.

   WHAT THE TWO NUMBERS ARE. 44px is WCAG 2.2 SC 2.5.5 Target Size
   (Enhanced), Level AAA. The Level AA criterion is SC 2.5.8 Target
   Size (Minimum) and it is 24x24 CSS PIXELS, not 44 — the governing
   dimension being the shorter one, so a wide short pill is judged on
   its height. Each of the six measured above clears 24 and falls
   short of 44: on these measurements they meet AA and miss AAA.

   SCOPED DELIBERATELY TO WHAT WAS MEASURED, and this comment must
   not be widened into "the page meets 2.5.8". Six controls at two
   viewports do not establish that for every target, every state and
   every zoom level; 400% zoom is the case most likely to turn an AA
   pass into a failure and it was NOT tested. Cite the criteria to
   explain what the gap is, not as a verdict on the page.

   Also do not read 44 as an Apple figure borrowed here: Apple's HIG
   uses 44 POINTS and Android 48dp, and the coincidence with 2.5.5's
   44 CSS pixels is why 44 is so often mis-cited as the AA minimum.

   The cheapest fix that keeps the look is padding, not min-height.

   margin-top:0 — app.css's 0.5rem is form-field rhythm. Here it
   pushed every pill 8px down out of its row.

   DO NOT NARROW THIS TO `.mission .tabs button`. The measurement
   above found the same fault on four different button groups; a
   per-component patch would fix the two anyone had looked at and
   leave the rest, which is how the play button stayed an ellipse
   through the first pass. */
.mission button{
  font:inherit;
  cursor:pointer;
  width:auto;
  /* auto, NOT 0. The browser default for min-width on a flex item is
     auto, and 0 lets a flex item shrink below its content — a
     difference that does not show at 1280 and would show as a
     squashed pill in a tight row. Diffed against the handoff's
     computed values rather than assumed. */
  min-width:auto;
  min-height:auto;
  margin-top:0;
}
.mission button:focus-visible,.mission a:focus-visible,.mission summary:focus-visible{outline:3px solid #FFB264;outline-offset:5px}
.mission h1,.mission h2,.mission h3,.mission p{margin:0}
.mission h1,.mission h2,.mission h3{font-family:Manrope,sans-serif;letter-spacing:-.055em;line-height:1.1}
.mission h2{font-size:clamp(34px,4.2vw,58px);font-weight:800}
.mission h3{font-size:25px}
.mission p+p{margin-top:18px}
.mission .wrap{max-width:1180px;margin:auto;padding:0 36px}
.mission .eyebrow{font-size:var(--t-micro);letter-spacing:.17em;text-transform:uppercase;font-weight:700;display:block;margin-bottom:22px}
.mission .dot{display:inline-block;width:8px;height:8px;background:var(--orange);border-radius:50%;margin-right:9px}
/* THE HERO TAKES app.css's --hero-* DIRECTLY, v2, 22 Oct 2026.

   It was `background:var(--purple);color:var(--cream)`, which made
   this hero a dark panel in both themes. MissionV2.dc.html draws the
   LIGHT one as pale sky with navy text, exactly as HomeV2 does, so
   the two heroes now read the SAME tokens and cannot drift apart --
   which they had already managed once, when this file's scoped
   --purple and app.css's --hero-bg were two independent definitions
   of "the hero colour".

   THE RAMP INTO --sky IS THE BOARD'S, and the stops are the mission
   board's own rather than the homepage's: 0->10% hold, 60% ramp,
   against the homepage's 0->22%/56%. The mission hero is taller and
   its art sits lower, so the sky starts sooner. Taken from
   MissionV2.dc.html verbatim.

   CONTRAST, measured against BOTH ends of the ramp because the
   ground is a gradient:
     --hero-text on --hero-bg   12.08:1 light / 11.98:1 dark
     --hero-text on --sky        8.11:1 light / 13.69:1 dark */
.mission .hero{
  background:var(--hero-bg);
  background:linear-gradient(180deg,var(--hero-bg) 0,var(--hero-bg) 10%,var(--sky) 60%,var(--sky) 100%);
  color:var(--hero-text);
  position:relative;overflow:hidden}
.mission .nav{display:flex;justify-content:space-between;align-items:center;gap:24px;padding-top:28px;padding-bottom:28px;position:relative;z-index:2}
/* 19px AND gap:0 — THE BOARD'S NUMBERS, and the same as the homepage.
   Was 28px with a 2px gap further down this file, which rendered as
   "C oRoute" with a visible space and made the two pages' wordmarks
   obviously different. letter-spacing drops from -1.5px to -0.02em:
   -1.5px was tuned for 28px and at 19px it crushed the letters. */
.mission .brand{font-family:Manrope,sans-serif;font-size:19px;font-weight:800;letter-spacing:-0.02em;text-decoration:none;display:inline-flex;align-items:center;gap:0}
/* THE WORDMARK RING — CORRECTED TO THE BOARD, 11 Oct 2026.
   =================================================================

   WAS width:25px; height:28px; border:5px; transform:rotate(-25deg),
   which draws a ROTATED ELLIPSE. The board (Mission.dc.html, canvas
   v180, version 1791751196-14a4) draws an UPRIGHT CIRCLE:

       17px x 17px, border: 3px solid, border-right-color:transparent,
       border-radius: 50%, margin-right: -1px, no transform

   and HomePhone.dc.html draws the same mark at 16px. Guillaume's
   spec for this round confirms the board, so the page was the one
   adrift, not the design.

   WHY 25x28 WAS WRONG AND NOT MERELY DIFFERENT. Unequal width and
   height on a border-radius:50% box is an ellipse, and rotating it
   tilts the gap in the ring away from the 3 o'clock position the
   design puts it at. The mark is the letter C of the wordmark —
   DESIGN-BRIEF.md is explicit that the arc IS the letter, which is
   why no "C" is written beside it — so a tilted ellipse is a
   misdrawn glyph, not a stylistic variant.

   box-sizing IS ALREADY border-box from `.mission, .mission *` at the
   top of this file, so the 3px border is INSIDE the 17px, as it is on
   the board (which states box-sizing:border-box explicitly because it
   is a standalone artboard).

   margin-right:-1px AND gap:0 ON .brand close the optical gap between
   the ring and the "oRoute" that follows it. The gap was 10px while
   the mark was oversized; at 17px the letterforms need to sit against
   it.

   ---- ONE DEFINITION, FIVE PLACES. 11 Oct 2026 -------------------

   Guillaume: "make sure the orange C style looks the same
   everywhere." It did not. This file had FIVE specifications of the
   same mark — header 17/3, header-phone 16/3, portal 18/4, avatar
   20/4, footer 21/4 — three of which had no board behind them, and
   the three 4px ones drew a visibly fatter arc than the homepage's.

   Everything is now THIS rule. The only thing an instance may change
   is --mark-size, the DIAMETER; the 3px stroke is set once, here, and is
   not overridden anywhere. That is what keeps the arc the same
   weight at every size, and it is why the override below reads
   `--mark-size:20px` rather than a fresh width/height/border triple —
   there is no longer a way to change the stroke by accident.

   THE DIAMETERS, AND WHY EACH ONE: 17px everywhere (the board, and
   app.css:2470 on the homepage), 16px on the phone (HomePhone.dc.html),
   and 20px for the AI-card avatar alone, which is argued at its own
   rule below because it is the single exception.

   IT IS --mark-size AND NOT --mark, AND THAT RENAME IS A BUG FIX —
   v2, 23 Oct 2026. The v2 palette brought app.css a token named
   --mark: #FF9E52, the stroke under an accented headline. Both
   stylesheets load on /our-mission, so this file's --mark (a LENGTH)
   and that one (a COLOUR) became the same inherited custom property,
   and the colour won by cascade.

   `width: var(--mark, 17px)` then resolved to `width: #FF9E52`. The
   fallback DID NOT SAVE IT: a fallback applies only when the variable
   is not defined at all, never when it is defined and nonsense for
   the property. Both boxes collapsed to the 3px border alone — a 6px
   orange speck where the ring should be, measured in the browser on
   all three marks (header logo, AI avatar, portal brand).

   Custom properties have no types and no namespaces: two files that
   never knew about each other can collide on a name, and the failure
   is silent, cosmetic and confined to the one page where both load.
   Nothing in the suite can see it — the homepage is unaffected
   because mission.css is not loaded there, and a brace-balance check
   or a hex-literal grep cannot tell a length from a colour. It was
   found by looking at the rendered page. */
/* ⚠️ em, SIZED TO THE CAPITAL HEIGHT. Corrected 23 Oct 2026, with
   the same change on the homepage. Guillaume: "the C seems a bit
   off" — it was 17px inside 19px text (0.895em) against the v2
   spec's 0.74em x 0.75em, a fifth too large in both directions, so
   the ring stood taller than the R beside it and read as an O.

   --mark-size STILL EXISTS, in em now, because the avatar tile
   overrides it. (The name is also load-bearing: it was --mark until
   23 Oct, when it collided with app.css's --mark COLOUR token and
   collapsed every ring on this page to a 6px speck. See
   css_token_collision_test.go.) */
.mission .brandmark{box-sizing:border-box;width:var(--mark-size,0.74em);height:var(--mark-size-h,0.75em);display:inline-block;position:relative;top:.012em;border:0.165em solid var(--orange);border-right-color:transparent;border-radius:50%;background:none;margin-right:-.03em}
/* FORCED COLORS, AND THIS RULE IS NOT OPTIONAL — copied here because
   the fault it fixes is in THIS file's mark too.

   app.css:2487 records it: Chrome repaints EVERY border side to the
   system foreground in a forced palette, the transparent one
   included, so the ring closes into a solid circle and the wordmark
   reads "OoRoute" — the ring stops being a C and becomes an O. It was
   found by LOOKING at a forced-colors screenshot, not by any text
   check.

   /our-mission had no such rule at all until 11 Oct 2026, so the mark
   closed here while the homepage's stayed open. "The orange C must
   look the same everywhere" is not only about diameter and stroke:
   a mark that survives a forced palette on one page and not the other
   is the same inconsistency by a different route.

   The name is still announced either way — the header <a> and the
   portal span both carry aria-label — so this is for people who SEE
   the page in a forced palette. */
@media (forced-colors: active){.mission .brandmark{forced-color-adjust:none;border-color:CanvasText;border-right-color:transparent;background:none}}
/* ⚠️ EVERY LINK CARRIES THE RULE, NOT JUST "Home". 23 Oct 2026,
   Guillaume, with the two navs side by side: this one "should match
   this style".

   WHAT IT WAS, AND WHY IT WAS A 1.4.1 FAULT RATHER THAN ONLY A
   MISMATCH. Only .links-home was underlined; the three in-page
   anchors beside it had no affordance at all. Every item in this row
   is the same colour as its siblings, so a link without the rule was
   identified BY COLOUR ALONE -- which is the exact argument
   app.css's .sitenav-links already writes down, in the comment on
   the current-page rule:

       "the orange rule under every link is a WCAG 1.4.1 affordance,
        not decoration [...] It therefore cannot double as the
        current-page mark."

   The homepage nav had worked this out and this nav had not. So the
   fix is not a new convention, it is adopting the one already
   written: the rule goes under EVERY link, and anything marking one
   item out is carried by weight.

   NOTHING IS CURRENT HERE, so nothing is marked. "Home" leaves to
   "/" and the other three are anchors within /our-mission -- this
   nav does not link to the page you are on, so no item takes
   aria-current and all four are the ordinary state. That mirrors
   the homepage, where both links are ordinary for the same reason.
   .links-home keeps its class as a hook; it no longer styles.

   SIZE AND HOVER MATCH TOO. 13px -> 1rem and the hover moves from
   text-decoration to brightening the rule, because the link is
   already underlined at rest and the available answer is to
   strengthen what is there. Both taken from .sitenav-links. */
.mission .links{display:flex;gap:27px;align-items:center;font-size:1rem}
/* ⚠️ THE NAV OPTS OUT OF THIS PAGE'S TYPEFACE. 23 Oct 2026,
   Guillaume on the first attempt at matching the two navs: "font and
   style look different than the homepage".

   HE WAS RIGHT AND THE FIRST PASS MISSED IT. That pass matched the
   size, the weight, the colour and the rule, measured all four, and
   called it a match -- without ever comparing font-family. The two
   navs were 16px/600 in DIFFERENT TYPEFACES, which is exactly the
   thing a table of matching numbers cannot show.

   WHY THIS PAGE IS DM SANS AT ALL: it is a port, and 'DM Sans' is
   its body font, set on .mission at the top of this file and
   scoped so it does not leak into the shared chrome (see the long
   note at the head of the file). That is deliberate and it STAYS --
   Guillaume's instruction is the nav, not the page.

   SO THE NAV IS THE EXCEPTION, NOT THE PAGE. This row is shared
   furniture: it does the same job as .sitenav-links on the homepage
   and a visitor moving between the two pages sees them as one
   component. The inherited page font is what made them read as two.
   The stack is app.css's, copied rather than referenced because the
   two files share no variable for it.

   ⚠️ UPDATE 7 Nov 2026 — THIS RULE IS NOW A NO-OP, AND THE NOTE ABOVE
   IS KEPT AS HISTORY RATHER THAN AS INSTRUCTION. Guillaume asked for
   one set of fonts across the whole site, so app.css's body is now
   'DM Sans' and its headings Manrope — the pair this page already
   used. The paragraph above that says the page staying DM Sans "is
   deliberate and it STAYS" was right for the ask of 23 Oct, which was
   about the nav; this is the wider ask he did not make then.

   The rule below pins the nav to "the stack app.css uses", and app.css
   no longer uses that stack. It therefore now does the OPPOSITE of
   what it was written to do: it makes this nav the one component on
   the page in a different typeface from everything around it.

   SO IT IS CHANGED TO DM SANS, which is what .mission sets and what
   app.css's nav now resolves to. The two navs still match, which was
   the whole point — they match on the new font instead of the old
   one. The declaration is kept rather than deleted so that this note
   has something to attach to: a future reader comparing the two navs
   should find the reasoning here, not an absence. */
.mission .links a{
  font-family: 'DM Sans', system-ui, -apple-system, "Segoe UI", Roboto, sans-serif;
  text-decoration:none;
  font-weight:600;
  border-bottom:2px solid rgba(255,178,100,.9);
  padding-bottom:2px;
}
.mission .links a:hover{border-bottom-color:rgb(255,178,100)}
/* Waiting for a page that IS one of these destinations, same as the
   homepage's copy of this rule. aria-current carries the meaning,
   the weight carries the look. */
.mission .links a[aria-current="page"]{font-weight:700}
.mission .hero-body{min-height:625px;padding-top:55px;padding-bottom:80px;position:relative;z-index:1}
.mission .hero-copy{max-width:600px;position:relative;z-index:2}
.mission .hero h1{font-size:clamp(46px,5.5vw,76px);font-weight:800;line-height:1.05;max-width:630px}
/* THE TAGLINE ACCENT TAKES --headline-accent, NOT --orange, and that
   is forced by the light hero. --orange is #E8762A, a LOGOTYPE
   colour, exempt from 1.4.11 as a mark but 2.52:1 on pale sky -- and
   this is a SENTENCE, not a mark, so the exemption does not reach
   it. --headline-accent is the per-theme accent-text colour the
   homepage's h1 already uses, #7A3606 light / #FFCB9A dark.
   7.54:1 / 9.69:1 on --hero-bg, 5.06:1 / 11.08:1 at the bottom of
   the sky ramp. */
.mission .hero h1 em{font-style:normal;color:var(--headline-accent)}
/* --hero-soft, not a literal: this paragraph sits on the hero, whose
   ground is now pale sky in light mode. The literal it replaces was a
   pale blue that measured 1.14:1 there. 8.20:1 / 9.99:1, and 5.51:1
   at the bottom of the sky ramp, which is the worst case. */
.mission .intro{max-width:34em;margin-top:27px;font-size:var(--t-body);line-height:1.65;color:var(--hero-soft)}
/* currentColor at 40%, not a white at 40%: on the light hero a white
   rule is invisible. currentColor is --hero-text in both themes, so
   the underline is always a dilution of the text above it. */
.mission .hero-link{display:inline-flex;gap:20px;align-items:center;margin-top:30px;font-size:var(--t-cap);font-weight:700;text-decoration:none;border-bottom:1px solid color-mix(in srgb, currentColor 40%, transparent);padding-bottom:7px}
/* A TOP FADE, v2, 22 Oct 2026, AND THE ART FORCED IT. The v1
   mission art had a transparent sky; day/mission-people-wide.webp is
   0.0% transparent with a painted one, measured on the shipped
   bytes. Without the fade the picture is a hard-edged rectangle
   pasted on the hero -- seen in the first render after the swap, and
   the board draws no such edge.

   0->12%, which is MissionV2.dc.html's own value for this art
   (the homepage's is 36%, because its art is a wide panorama whose
   horizon sits far lower). The geometry -- 840px, right -165px,
   bottom -20px -- is untouched; v2 is colours, and a mask is what
   makes a colour meet another colour. */
.mission .hero-art{position:absolute;width:840px;right:-165px;bottom:-20px;pointer-events:none;
  -webkit-mask-image:linear-gradient(180deg,transparent 0,#000 12%);
          mask-image:linear-gradient(180deg,transparent 0,#000 12%);
  -webkit-mask-repeat:no-repeat;mask-repeat:no-repeat;
  -webkit-mask-size:100% 100%;mask-size:100% 100%}
/* --warm, not a literal: this strip was #f5eadd in both themes, so in
   dark mode it drew --ink (light) on a cream band at 1:1 -- five
   invisible links. --warm is the palette's own warm band, #F4ECDF
   light and #17263B dark. The border follows --line with it. */
.mission .mission-strip{background:var(--warm);border-bottom:1px solid var(--line)}
.mission .mission-strip .wrap{padding-top:24px;padding-bottom:24px;display:flex;justify-content:space-between;gap:20px;font-size:var(--t-small);font-weight:600}
.mission .mission-strip span{display:flex;align-items:center;gap:14px}
/* --accent-text, which is the BOARD'S own token for these numbers:
   MissionV2.dc.html styles `.strip b { color: var(--accent-text) }`
   and its dark twin resolves that to the orange. They became
   --strong-ink when --purple was split into a ground and an ink,
   which was right for the six other uses and wrong for this one --
   the board draws them as an ACCENT, not as emphatic body text.
   Caught in the phone board comparison: the board's numbers were
   orange and the page's were light grey.
   5.34:1 on --warm light, 8.57:1 dark. */
.mission .mission-strip b{color:var(--accent-text);font-size:var(--t-cap);font-weight:500}
.mission .section{padding:105px 0}
.mission .split{display:grid;grid-template-columns:1fr 1fr;gap:90px}
/* ---- #why's COLUMNS ARE NOT EQUAL, 11 Oct 2026 ------------------

   Guillaume, on the body copy at 19px: "font size still feels too
   small to me on my desktop". Raising the step alone would have made
   it WORSE to read: the column is 509px, so 21px type wraps at about
   41 characters and the paragraph becomes a narrow ladder.

   The cap on .body-copy is max-width:34em, which at 21px is 714px —
   so the CAP was never binding; the 1fr column was. Widening the
   text column to 1.15fr lets the em cap do its job and brings the
   measure back to roughly 60 characters, inside the readable band.

   The heading column loses nothing it needs: "You're new to driving."
   is the longest line and still sets on one line at this width. */
@media (min-width: 761px) {
  .mission #why .split { grid-template-columns: .85fr 1.15fr; gap: 72px; }
}
/* ---- WHY #why's RIGHT COLUMN IS CENTRED, 11 Oct 2026 -------------

   Guillaume, looking at section 01: "this text is too high up no?"
   He was right, and the measurement said so plainly. At 1280 the two
   columns are 509px each in one grid row, and the row is as tall as
   the TALLER of them:

       left  (eyebrow, heading, £1,860, source)   499px of content
       right (two paragraphs)                      221px of content,
                                                   277px of EMPTY SPACE

   align-items defaults to `normal`, which stretches both boxes to the
   full row height; the right box then renders its text from the top,
   so more than half of it was blank and the copy read as stranded.

   CENTRED ONLY WHERE THERE ARE TWO COLUMNS. Below 760px .split
   collapses to one column (see the rule further down), and vertical
   centring in a single-column stack means nothing — so this sits in
   the two-column range rather than on the base rule.

   It is #why only. The other .split sections pair text with a figure
   that fills its own box, and centring those would move artwork that
   is already placed deliberately. */
@media (min-width: 761px) {
  /* THE HEADING STARTS THE SECTION; the right column is the longer
     one now that the stat sits under it, so centring BOTH would push
     the h2 down the page away from its own eyebrow. Left starts at
     the top, right centres against it. */
  .mission #why .split > :first-child { align-self: start; }
  .mission #why .split > .body-copy   { align-self: center; }
}
.mission .body-copy{color:var(--muted);font-size:var(--t-body)}
/* ---- SECTION 01'S RIGHT-HAND BODY COPY — 11 Oct 2026 ------------

   Guillaume, looking at this section at desktop width: "the text is
   hard to read on the right". Only the type changes; the wording,
   the heading and the stat block are all untouched.

   WAS 18px / 29.7px leading / 62 characters, in a 509px column. The
   character count was already inside the readable band, so the fault
   was not the measure alone — it was 18px set against a 54px heading
   and a 64px figure in the same eyeline, with leading tight enough
   that the lines greyed together. Next to that much display type,
   18px reads as a caption rather than as the argument of the
   section, which is what it actually is.

   NOW 20px / 1.7 leading / ~62 characters.
     - 20px is the step that makes it read as body copy of the same
       page as its heading, without touching the heading.
     - 1.7 is the leading those longer lines needed; the old 1.65
       at a smaller size was the greying.
     - max-width:36em caps the measure in EMS so it scales WITH the
       font instead of fighting it. At 20px that is ~62 characters,
       inside the 60-70 band; it binds at 1440+ where the column
       grows, which is what stops the larger type simply running
       further across a wider screen.

   THE WHITE SPACE BELOW IS DELIBERATE AND IS NOT A FAULT. The right
   column is shorter than the left because the left carries the stat
   block; Guillaume's note was explicit that the space is fine and
   the line length and size were the problem.

   WHY A SECTION-01 SCOPE AND NOT .body-copy. The class is shared
   with section 03, which sets its own 17px for a narrower column
   beside the AI card. Changing the shared rule would resize a
   section nobody asked about.

   THE COLUMN SPLIT IS LEFT AT 1fr 1fr. The stat block sits in the
   left column under the heading, so changing the proportions would
   reflow the figure and its caption — explicitly not to be touched. */
.mission #why .body-copy{font-size:var(--t-body);line-height:1.75;max-width:34em}
.mission #why .body-copy p+p{margin-top:1.15em}
.mission .body-copy strong{color:var(--ink);font-weight:600}
.mission .statement{font-size:clamp(27px,3vw,39px);line-height:1.35;letter-spacing:-.035em;font-family:Manrope,sans-serif;font-weight:700}
.mission .underlined{position:relative;white-space:nowrap}
.mission .underlined:after{content:'';position:absolute;height:5px;background:var(--orange);bottom:-4px;left:0;right:0;transform:rotate(-2deg);border-radius:50%}
.mission .small-note{font-size:var(--t-cap);line-height:1.65;color:var(--muted)}
.mission .principle{margin-top:65px;padding:28px 0;border-top:1px solid var(--line);border-bottom:1px solid var(--line);display:flex;gap:30px;align-items:center}
.mission .principle .num{font-size:54px;line-height:1;color:#b0cadc}
.mission .principle p{font-size:var(--t-body);max-width:34em}
/* A LITERAL INK, not var(--cream): this band is dark in BOTH themes
   (--navy is #0B1626 either way), so its text must be light in both.
   --cream is the PAGE ground and goes dark; using it here drew the
   section h2 at 1.07:1 in dark mode. 17.35:1 now. */
.mission .context{background:var(--navy);color:#FFF9F1}
.mission .context .eyebrow{color:var(--orange)}
.mission .context-top{display:flex;justify-content:space-between;gap:60px;align-items:end}
.mission .context-top h2{max-width:630px}
.mission .context-top p{max-width:330px;color:#c5d5e3;font-size:var(--t-small)}
.mission .demo{display:grid;grid-template-columns:1.08fr 1fr;margin-top:48px;border:1px solid #ffffff25;border-radius:22px;overflow:hidden;background:#0b1626}
.mission .road-panel{padding:32px;min-height:375px;position:relative;background:#1a3557;display:flex;flex-direction:column}
.mission .demo-label{font-size:10px;font-weight:700;letter-spacing:.13em;text-transform:uppercase;color:#dae9f6}
.mission .sensor{display:flex;justify-content:space-between;align-items:center;margin-top:23px;gap:12px}
.mission .sensor b{font-family:Manrope,sans-serif;font-size:26px;letter-spacing:-1px}
.mission .sensor span{color:#ccdae7;font-size:12px}
.mission .context-detail{padding:32px;display:flex;flex-direction:column;justify-content:center}
.mission .tabs{display:flex;gap:7px;margin-bottom:28px;flex-wrap:wrap}
.mission .tabs button{border:1px solid #ffffff38;background:transparent;color:#e3eef8;padding:9px 14px;border-radius:30px;font-size:12px;font-weight:600}
.mission .tabs button[aria-pressed=true]{background:var(--orange);color:var(--navy);border-color:var(--orange)}
.mission .context-detail h3{font-size:29px;margin-bottom:15px}
.mission .context-detail p{font-size:var(--t-small);color:#d6e1eb}
.mission .context-detail .meaning{border-left:2px solid var(--orange);padding-left:17px;margin-top:24px;color:#FFF9F1;font-size:var(--t-small);line-height:1.6}
.mission .demo-foot{font-size:var(--t-cap);line-height:1.5;color:#c4d4e3;margin-top:16px}
.mission .how-head{max-width:680px}
.mission .impact{margin-top:32px;padding:24px 30px;background:#ecf3f9;border-radius:12px;font-size:var(--t-small);color:#3E5668}
/* --warm, for the same reason as .mission-strip: the whole AI-guide
   band was a pinned beige, so its eyebrow, its h2 and two paragraphs
   all drew light ink on it at 1.03:1 in dark mode. */
.mission .coverage{background:var(--warm)}
.mission .coverage .split{gap:65px;align-items:center}
/* 28px, PER THE BOARD (Mission.dc.html, canvas v203: margin:16px 0 28px
   on the h2). Was 26px. This is the ONLY gap under section 03's
   heading now the .body-copy wrapper has gone, so the number has to
   be right here rather than shared with a paragraph margin. */
.mission .coverage h2{margin-bottom:28px}
/* SECTION 03'S ONE PARAGRAPH, 12-13 Oct 2026. The board sets 20px/1.55
   in --ink, which is NOT --t-body (21px desktop, 17px phone), so it is
   written out rather than taken from the scale. It replaced a
   .body-copy wrapper holding two paragraphs; that rule is gone with
   them, and .body-copy survives on section 01, which still uses it. */
.mission .coverage-lede{margin:0 0 24px;font-size:20px;line-height:1.55;color:var(--ink);max-width:34em}
/* ---- THE THREE PRODUCT MOCK-UPS TAKE --ui-*, v2 ------------------
   .chat, .journey-preview and .driver-portal are pictures of
   software, not sections of this page, and the v2 palette gives
   them their own per-theme surface set for exactly that -- the same
   set the homepage's portal uses, so the two pages' mock-ups cannot
   drift apart.

   THEY WERE `background: white` IN BOTH THEMES, which made them the
   largest single source of unreadable text on this page in dark
   mode: every heading and paragraph inside them inherited --ink
   (light) and landed on white. Measured before this change, 14 pairs
   on white, worst 1.15:1.

   EACH ONE SETS ITS OWN INK WITH ITS FILL. That is the rule this
   project has learned twice: a surface that names its background
   must name every ink token its descendants inherit, or the half it
   forgets is found later by a person. --ui-ink is 14.27:1 on
   --ui-card in light and 12.03:1 in dark. */
.mission .chat{background:var(--ui-card);color:var(--ui-ink);--muted:var(--ui-soft);border-radius:22px;box-shadow:0 18px 50px #142b4910;border:1px solid var(--ui-line);overflow:hidden}
.mission .chat-header{padding:22px 26px;border-bottom:1px solid #eaf4fa;display:flex;align-items:center;gap:12px;font-weight:700;font-size:14px}
.mission .spark{width:34px;height:34px;border-radius:10px;background:var(--purple);color:var(--orange);display:grid;place-items:center}
.mission .chat-header small{display:block;font-size:var(--t-cap);color:var(--muted);font-weight:400}
.mission .chat-body{padding:24px}
.mission .question-options{display:flex;gap:8px;flex-wrap:wrap;margin-bottom:24px}
/* The example-question chips, same card, same reason: pinned
   #f3f7fa with themed ink was 1.11:1 in dark mode. */
.mission .question-options button{background:var(--ui-bg);border:1px solid var(--ui-line);border-radius:20px;padding:8px 14px;font-size:var(--t-cap);color:var(--ui-ink)}
.mission .question-options button[aria-pressed=true]{background:var(--ui-accent-soft);border-color:var(--ui-accent)}
.mission .bubble{background:var(--purple);color:white;border-radius:14px 14px 3px 14px;padding:17px 20px;font-size:var(--t-small);margin:0 0 20px 38px}
.mission .answer{font-size:var(--t-small);line-height:1.6;color:var(--ui-soft);min-height:225px}
.mission .answer strong{display:block;color:var(--ui-ink);margin-bottom:11px;font-size:var(--t-body)}
.mission .answer ul{padding-left:18px;margin:12px 0}
.mission .chat-status{font-size:var(--t-cap);line-height:1.5;color:var(--strong-ink);border-top:1px solid #eaf4fa;padding-top:15px}
/* THE COMMITMENT LINE, re-measured against the board 12-13 Oct 2026
   (canvas v203): padding-left:16px, not 22px, and 17px/600 rather
   than --t-body. It is the last thing in section 03's column now the
   stepper has gone, and the board draws the orange rule tight to the
   text.

   MARGIN-TOP IS 0, NOT 28px, AND THE 24px GAP IS REAL. The board puts
   the space on the paragraph ABOVE (margin:0 0 24px on the lede), not
   on this line. Keeping the old 28px here would have stacked against
   that 24px — these are adjacent siblings in a grid item, so the
   margins do NOT collapse — and the measured gap would have been 52px
   against the board's 24. Measured live after the change: 24.0px at
   every width from 360 to 2560. */
.mission .commitment{font-size:17px;line-height:1.6;font-weight:600;color:var(--ink);border-left:3px solid #ee9c50;padding-left:16px;margin-top:0}
.mission .coverage .small-note{margin-top:22px}
/* ---- THE TRUST SECTION, A FULL-WIDTH LILAC BAND -------------------

   Rebuilt 12 Oct 2026 from Mission.dc.html, canvas v1791836956-6698.
   The board's own comment states the brief: "TRUST, restyled 12 Oct
   (Guillaume: 'make it look better'). A full-width lilac band so the
   page ends on a finished block, not loose cards on cream. Heading
   left (sticky column), 2x2 cards right, filled purple icon discs,
   soft lift instead of borders."

   WHAT IT REPLACED: four hairline-ruled .trust-item blocks on the
   page's cream background, in a 2-column grid with a 70px gutter and
   no cards at all. The copy was also four long sentences; the board
   cut each to a short label and one line, and the cut is the
   board's, not mine — its comment records "The substance of each
   principle is kept, only shorter".

   THE BOARD'S VALUES:

     band        background #F3ECFD, 1px #E0D4F2 top border,
                 full-bleed, running straight into the footer
     inner       max-width 1200, padding 80px 40px 88px
     columns     360px | 1fr, 56px gap, align-items start
     cards       white, 18px radius, NO border,
                 shadow 0 1px 0 rgba(41,20,73,.04),
                        0 12px 28px rgba(75,31,174,.10)
                 padding 26px 24px, 20px grid gap
     icon disc   48px, 50% radius, background #6624C9, cream stroke
     h3          19px / 1.25, weight 800, -0.01em
     p           15px / 1.5, ink-soft

   THE ICON DISC IS #6624C9 ON THE BOARD, which is var(--brand)
   there. This stylesheet's purple token is --purple #4B1FAE, a
   different value: #6624C9 is lighter and reads better on the lilac
   band than the page's own purple, which is why the board uses it.
   It is written as a literal here because the page has no token for
   it, and inventing one for a single use would be worse.

   "RUNNING STRAIGHT INTO THE FOOTER" is why this section drops the
   page's standard 105px .section padding and sets its own: the band
   must meet the footer with no cream gap, or it reads as a card
   rather than as the page's closing block. */
.mission .trust-band {
  padding: 0;
  /* --panel / --line, not literals: this is the page's closing block
     and it carried a pinned pale fill, so in dark mode its eyebrow,
     its 54px h2 and the full stop all drew light ink on pale at
     1.01:1 and 1.51:1. */
  background: var(--panel);
  border-top: 1px solid var(--line);
}
.mission .trust-band .trust-grid {
  max-width: 1200px;
  padding: 80px 40px 88px;
  display: grid;
  grid-template-columns: 360px 1fr;
  gap: 56px;
  align-items: start;
}
.mission .trust-head { max-width: none; }
/* The orange full stop is its own element on the board, so the
   heading can wrap naturally and the stop always follows the last
   word rather than being baked into a hard-coded line break. The old
   markup forced "Trust needs more<br>than a clever model." */
/* The full stop after the trust heading. It is a 54px glyph so the
   bar is 3:1, and --orange measured 2.52:1 on the panel it sits on.
   --accent-text clears it in both themes. The mark is decorative --
   the sentence reads identically without it -- but a decorative mark
   that cannot be seen is not decoration, it is absence. */
.mission .trust-head .trust-stop { color: var(--accent-text); }
.mission .trust-list {
  display: grid;
  grid-template-columns: repeat(2, minmax(0, 1fr));
  gap: 20px;
  margin-top: 0;
}
.mission .trust-item {
  border-top: 0;              /* was a hairline rule; the board uses cards */
  background: var(--cream);
  border-radius: 18px;
  padding: 26px 24px;
  box-shadow: 0 1px 0 rgba(20, 43, 73, .04), 0 12px 28px rgba(20, 43, 73, .10);
}
.mission .trust-icon {
  display: grid;
  place-items: center;
  width: 48px;
  height: 48px;
  border-radius: 50%;
  background: #142B49;
  /* A literal for the same reason as .context: this icon tile is
     navy in both themes, so its glyph is light in both. */
  color: #FFF9F1;
}
.mission .trust-item h3 {
  margin: 18px 0 6px;
  font-size: 19px;
  line-height: 1.25;
  font-weight: 800;
  letter-spacing: -0.01em;
}
.mission .trust-item p { font-size: 15px; line-height: 1.5; color: var(--muted); }

/* THE BAND COLLAPSES IN TWO STAGES, because it has two grids.

   1000px — the OUTER grid goes to one column. 360px of heading plus
   two cards plus two 56px gaps needs about 1000px; below that the
   heading column squeezes the cards before it gives up any of its
   own fixed width. 1000px is an existing breakpoint in this
   stylesheet (see the .hero-body block), not a new one.

   760px — the INNER card grid goes to one column, which the existing
   rule in that block already does. Between the two the heading sits
   above a 2x2 of cards, which is the arrangement the board's own
   phone treatment implies and reads correctly at tablet widths.

   The band keeps its full-bleed background at every width; only the
   inner padding comes in, so it still runs into the footer. */
@media (max-width: 1000px) {
  .mission .trust-band .trust-grid {
    grid-template-columns: 1fr;
    gap: 36px;
    padding: 64px 36px 72px;
  }
}
@media (max-width: 760px) {
  .mission .trust-band .trust-grid { padding: 52px 24px 60px; gap: 28px; }
  .mission .trust-item { padding: 22px 20px; }
}
.mission .closing{padding:77px 0;background:var(--purple);color:#FFF9F1}
.mission .closing .wrap{display:grid;grid-template-columns:1.2fr 1fr;gap:90px;align-items:center}
.mission .closing h2{font-size:46px}
.mission .closing h2 span{color:var(--orange)}
.mission .closing p{color:#e1edf7}
.mission .closing a{display:inline-block;margin-top:22px;font-size:var(--t-cap);text-underline-offset:6px}
.mission .footer{background:#142B49;color:#cfddea;padding:32px 0}
.mission .footer .wrap{display:flex;justify-content:space-between;gap:30px;align-items:center}
.mission .footer .brand{font-size:23px;color:#FFF9F1}
.mission .footer p{font-size:var(--t-cap);max-width:540px;text-align:right}
.mission .preview-badge{position:fixed;bottom:16px;left:16px;padding:7px 12px;border-radius:30px;background:#fff9f1ef;color:var(--ink);border:1px solid #d1dde6;box-shadow:0 2px 12px #0b162615;z-index:10;font-size:10px;backdrop-filter:blur(8px)}
@media(min-width:1450px){.mission .hero-art{right:-110px;width:900px}}
@media(max-width:1000px){.mission .hero-art{width:670px;right:-230px}
.mission .hero-copy{max-width:550px}
.mission .split{gap:45px}
.mission .context-top{gap:30px}
.mission .closing .wrap{gap:40px}}
@media(max-width:760px){/* THE SCALE STEPS DOWN ON THE PHONE, IN ONE PLACE. Every prose rule
   reads these properties, so the phone keeps smaller type without a
   second set of per-rule overrides to drift out of step. The brief
   for this change allows the phone its own values; this is how it
   gets them. --t-cap does NOT drop below 14px: the two honesty
   disclosures use it, and the whole point of raising them was that a
   disclosure should not be the hardest thing on the page to read. */
.mission{--t-body:17px;--t-small:16px;--t-cap:14px;--t-micro:11px}
.mission .wrap{padding-left:24px;padding-right:24px}
/* THE THREE IN-PAGE ANCHORS GO ON PHONE, BUT HOME STAYS.
   They were all hidden together, which left the wordmark as the
   only route off this page on a phone — and a logo is an implicit
   affordance, which is exactly what Guillaume was asking about.
   The section anchors are the ones worth dropping: the page is a
   single scroll and they do not fit a 390px row.
   Home is the one that has somewhere else to go.

   HOME IS NOW THE ONLY CHILD OF THIS ROW, 12 Oct 2026. The
   PRE-LAUNCH pill used to sit beside it, and on a phone the two were
   jammed together at the right edge with the pill clipped by the
   viewport. With the pill gone the row is wordmark-left,
   Home-right — a two-item space-between, which is what the row is
   for. The gap below no longer separates anything at this width; it
   is kept because the hidden anchors are hidden with display:none,
   not removed, and a gap on a one-item flex row costs nothing and is
   correct again the moment a link comes back. */
.mission .links a:not(.links-home){display:none}
.mission .links{gap:14px}
.mission .brand{font-size:19px}
/* 16px on the phone, per HomePhone.dc.html. The ring tracks the
   wordmark's own step down from 28px to 25px. */
/* The phone step is no longer re-stated: an em ring already tracks
   the wordmark down from 28px to 25px by itself. */
.mission .nav{padding-top:22px;padding-bottom:22px}
.mission .hero-body{padding-top:32px;padding-bottom:0;min-height:0}
.mission .hero-copy{max-width:none}
.mission .hero h1{font-size:48px;max-width:530px}
.mission .intro{max-width:460px}
.mission .hero-art{position:relative;width:calc(100% + 48px);right:auto;bottom:-2px;display:block;margin:24px -24px 0}
.mission .mission-strip .wrap{display:grid;grid-template-columns:1fr 1fr;gap:14px;font-size:var(--t-cap)}
.mission .mission-strip span{gap:8px}
.mission .section{padding:65px 0}
.mission .split,.mission .coverage .split{grid-template-columns:1fr;gap:28px}
/* SECTION 03'S LEDE STEPS DOWN WITH EVERYTHING ELSE, 12-13 Oct 2026.

   THE BOARD CANNOT DECIDE THIS: there is no MissionPhone board — the
   canvas has one Mission board, desktop 1280 — so its 20px is a
   DESKTOP figure and nothing on the canvas speaks for 390px.

   WHY IT HAD TO MOVE. 20px is written out literally rather than taken
   from --t-body, so it does NOT follow the phone step-down the way
   every other prose rule on this page does. Measured at 390 before
   this rule existed: section 01's lede 19px, the closing 17px, the
   then-extant stepper's body 15px — and this paragraph 20px, which made it the single
   largest run of body text on the phone. It is not the most important
   sentence on the page; section 01's lede is, and Guillaume raised
   that one to 19px himself.

   19px, not 17px, because 19px is the size he settled on for phone
   prose in section 01 ("font size still feels too small to me") and
   this paragraph is doing the same job. The line-height stays 1.55,
   which is the board's. */
.mission .coverage-lede{font-size:19px}
.mission .principle{margin-top:35px;gap:20px}
.mission .principle .num{font-size:38px}
.mission .context-top{display:block}
.mission .context-top p{max-width:none;margin-top:24px}
.mission .demo{grid-template-columns:1fr;margin-top:30px}
.mission .road-panel{min-height:290px;padding:24px}
.mission .context-detail{padding:24px}
.mission .context-detail h3{font-size:27px}
.mission .impact{padding:20px}
.mission .trust-list{grid-template-columns:1fr}
.mission .closing .wrap{grid-template-columns:1fr;gap:25px}
.mission .closing h2{font-size:38px}
.mission .footer .wrap{display:block}
.mission .footer p{text-align:left;margin-top:18px}
.mission .preview-badge{font-size:9px;bottom:10px;left:10px}
.mission .answer{min-height:0}
.mission .eyebrow{margin-bottom:17px}
.mission .chat-body{padding:20px}
.mission .bubble{margin-left:15px}}
@media(prefers-reduced-motion:reduce){}
/* Mission redesign: independent artwork column and intrinsic image proportions. */
.mission .hero-body{display:grid;grid-template-columns:1fr 1fr;gap:28px;align-items:center;min-height:565px;padding-top:50px;padding-bottom:60px}
.mission .hero-copy{min-width:0;max-width:none}
.mission .hero h1{font-size:clamp(54px,6vw,80px);line-height:1.06;white-space:normal}
.mission .mission-headline{font-family:Manrope,sans-serif;font-size:clamp(26px,3vw,37px);font-weight:700;line-height:1.22;letter-spacing:-.04em;margin-top:24px}
/* --accent-text, NOT --orange, and the same argument as .brandai:
   --orange is the LOGOTYPE orange, exempt from 1.4.11 as a mark and
   2.84:1 on the page as a SENTENCE. This is a tagline, so it takes
   the accent-as-text token: 5.99:1 light, 9.52:1 dark. */
.mission .mission-headline em{font-style:normal;color:var(--accent-text)}
.mission .intro{font-size:17px;margin-top:22px;max-width:440px}
.mission .hero-figure{margin:0;min-width:0;align-self:center}
.mission .hero-art{position:static;display:block;width:100%;height:auto;right:auto;bottom:auto;margin:0;aspect-ratio:4/3;object-fit:contain}
.mission .hero-figure figcaption{text-align:center;font-size:var(--t-cap);color:var(--hero-soft);margin-top:10px}
.mission .design-review{background:#142B49;color:#fff9f1;border-top:1px solid #ffffff24}
.mission .design-review .wrap{display:flex;align-items:center;justify-content:space-between;padding-top:15px;padding-bottom:15px;gap:16px}
.mission .design-review span{font-size:12px;font-weight:700}
.mission .design-review small{font-weight:400;color:#cbdae8;margin-left:12px}
.mission .direction-buttons{display:flex;gap:8px}
.mission .direction-buttons button{border:1px solid #ffffff55;border-radius:30px;background:transparent;color:#fff9f1;padding:9px 16px;font-size:12px;min-height:40px}
.mission .direction-buttons button[aria-pressed=true]{background:#FFB264;color:#142B49;border-color:#FFB264}
.mission .context-diagram{margin-top:26px}
.mission .context-diagram svg{width:100%;height:auto;display:block;margin:12px 0}
.mission .scene-heading,.mission .scene-foot{display:flex;justify-content:space-around;gap:20px;font-size:10px;letter-spacing:.08em;color:#d8e5f0}
.mission .scene-foot{letter-spacing:0;font-size:12px}
.mission .safe-only{display:none}
.mission .scenario-safe .safe-only{display:block}
.mission .scenario-safe .close-only{display:none}
.mission .scene-foot .safe-only{display:none}
.mission .scenario-safe .scene-foot .safe-only{display:inline}
.mission .road-panel{justify-content:center}
.mission .journey-preview{margin-top:44px;background:var(--ui-card);color:var(--ui-ink);--muted:var(--ui-soft);border:1px solid var(--ui-line);border-radius:22px;overflow:hidden;box-shadow:0 14px 35px #142b4906}
.mission .journey-heading{padding:28px 32px;display:flex;justify-content:space-between;gap:18px;align-items:center;border-bottom:1px solid #eaf4fa}
.mission .journey-heading .eyebrow{font-size:9px;letter-spacing:.12em;margin-bottom:10px;color:var(--ui-soft)}
.mission .journey-heading h3{font-size:26px}
.mission .concept-tag{font-size:10px;color:var(--ui-soft);border:1px solid #e0e8ee;border-radius:20px;padding:6px 12px}
.mission .journey-content{display:grid;grid-template-columns:1.1fr 1fr;gap:40px;padding:22px 32px 30px;align-items:center}
.mission .journey-route svg{width:100%;height:auto}
.mission .journey-route>span{display:block;text-align:center;font-size:12px;color:var(--ui-soft)}
.mission .journey-moments>div{display:flex;align-items:center;gap:16px;padding:14px 0;border-bottom:1px solid #eaf4fa}
.mission .journey-moments>div:last-child{border:0}
.mission .journey-moments b{width:29px;height:29px;border-radius:50%;background:#edf4fa;color:#1a4c86;display:grid;place-items:center;font-size:12px;flex:none}
.mission .journey-moments strong{font-size:14px;display:block}
.mission .journey-moments span{display:block;color:var(--ui-soft);font-size:12px}
/* THE ✦ BEFORE EACH TRUST HEADING IS GONE, 12 Oct 2026. The board
   gives every card a 48px purple icon disc above its heading, and a
   star glyph in front of the text as well would be two markers for
   one card. Removed rather than left overridden, so nothing has to
   reason about which wins. */
.mission .preview-badge{display:none}
@media(max-width:1000px){.mission .hero-body{gap:16px}
.mission .hero-art{width:100%;right:auto}
.mission .hero h1{font-size:60px}
.mission .mission-headline{font-size:30px}
.mission .journey-content{gap:20px}}
@media(max-width:760px){.mission .hero-body{grid-template-columns:1fr;padding-top:30px;padding-bottom:28px;gap:22px;min-height:0}
.mission .hero h1{font-size:clamp(46px,11vw,64px)}
.mission .mission-headline{font-size:29px;margin-top:20px}
.mission .hero-figure{width:100%;max-width:520px;justify-self:center}
.mission .hero-art{width:100%;margin:0;position:static;height:auto}
.mission .hero-figure figcaption{font-size:11px;margin-top:5px}
.mission .design-review .wrap{flex-direction:column;align-items:flex-start;gap:10px}
.mission .design-review small{margin-left:8px}
.mission .direction-buttons{width:100%;gap:6px}
.mission .direction-buttons button{flex:1;padding:9px 6px;font-size:11px}
.mission .journey-heading{padding:22px;display:block}
.mission .concept-tag{display:inline-block;margin-top:15px}
.mission .journey-content{grid-template-columns:1fr;padding:18px 22px;gap:14px}
.mission .journey-route{max-width:440px;width:100%;margin:auto}
.mission .context-diagram{margin-top:20px}
.mission .road-panel{padding:24px}
.mission .mission-strip .wrap{gap:14px 9px}
.mission .mission-strip span{align-items:flex-start}
.mission .links{gap:0}
.mission .wrap{min-width:0}
.mission .underlined{white-space:normal}}
/* Journey portal illustration */
.mission .driver-portal{margin-top:40px;border:1px solid var(--ui-line);border-radius:22px;overflow:hidden;background:var(--ui-card);color:var(--ui-ink);--muted:var(--ui-soft);box-shadow:0 22px 65px #142b4910}
.mission .portal-top{padding:22px 28px;background:#142B49;color:#fff9f1;display:flex;align-items:center;justify-content:space-between;gap:16px}
.mission .portal-brand{display:flex;align-items:center;gap:9px;font-family:Manrope,sans-serif;font-size:18px;font-weight:800;letter-spacing:-.7px}
/* ONE MARK, ONE DEFINITION — 11 Oct 2026, Guillaume: "make sure the
   orange C style looks the same everywhere."

   This instance was 18px with a 4px border. The board does not draw
   the portal header's mark at all (it writes a plain "CoRoute"
   wordmark there), so the 18/4 was an invention, and a 4px stroke on
   an 18px ring is a VISIBLY heavier arc than the 3px on 17px the
   header and the homepage use. Back to the one definition. */
/* Likewise the portal: em tracks the portal wordmark on its own. */
.mission .portal-brand>span:last-child{font-family:'DM Sans',sans-serif;letter-spacing:0;font-size:11px;font-weight:400;color:#d3e2ef;border-left:1px solid #ffffff40;margin-left:13px;padding-left:21px}
.mission .portal-demo{font-size:10px;color:#dbe6ef;border:1px solid #ffffff30;padding:6px 10px;border-radius:20px}
.mission .portal-layout{display:grid;grid-template-columns:310px minmax(0,1fr)}
.mission .trip-sidebar{background:#fafcfd;border-right:1px solid #eaf4fa;padding:27px 18px;display:flex;flex-direction:column}
.mission .trip-list-heading{display:flex;align-items:center;justify-content:space-between;padding:0 8px;gap:12px}
.mission .trip-list-heading h3{font-size:20px;letter-spacing:-.04em}
.mission .trip-list-heading>span{font-size:10px;color:var(--ui-soft)}
.mission .trip-hint{font-size:11px;color:var(--ui-soft);padding:8px 8px 21px}
.mission .trip-list{display:grid;gap:10px}
.mission .trip-row{display:grid;grid-template-columns:28px minmax(0,1fr) 12px;gap:10px;text-align:left;background:transparent;border:1px solid #e0e8ee;color:#142b49;border-radius:12px;padding:17px 12px;width:100%;align-items:start}
.mission .trip-row[aria-pressed=true]{background:#ebf3fa;border-color:#6f90b4;box-shadow:inset 3px 0 #1f5fa8}
.mission .trip-row:hover{background:#edf4fa}
.mission .trip-symbol{font-size:22px;color:#1f5fa8;line-height:1.2}
.mission .trip-row strong{font-size:13px;display:block;line-height:1.3}
.mission .trip-row small{font-size:10px;display:block;color:var(--ui-soft);margin:5px 0 9px}
.mission .trip-tag{font-size:10px;display:block;color:var(--ui-soft)}
.mission .trip-chevron{font-size:21px;color:var(--ui-soft);line-height:1}
.mission .portal-intent{margin-top:auto;padding:28px 8px 0;display:flex;gap:12px;align-items:start}
.mission .portal-intent>span{font-size:27px;color:var(--ui-soft);line-height:1.2}
.mission .portal-intent p{font-size:11px;line-height:1.7;color:var(--ui-soft)}
.mission .portal-intent strong{display:block;font-weight:600;color:var(--ui-soft)}
.mission .trip-detail{min-width:0}
.mission .trip-detail-header{padding:25px 28px 20px;display:flex;justify-content:space-between;align-items:center;gap:12px}
.mission .portal-overline{font-size:9px;font-weight:700;letter-spacing:.14em;color:var(--ui-soft);display:block;margin-bottom:8px}
.mission .trip-detail-header h3{font-size:24px}
.mission .trip-detail-header>span{font-size:11px;color:var(--ui-soft)}
.mission .trip-map{position:relative;background:#f1f6f9;border-top:1px solid #eaf4fa;border-bottom:1px solid #eaf4fa}
.mission .trip-map svg{display:block;width:100%;height:auto}
.mission .map-key{position:absolute;bottom:12px;left:16px;display:flex;gap:7px;align-items:center;font-size:9px;line-height:1;background:#fffdf9ef;border:1px solid #e2ebf2;padding:9px 11px;border-radius:20px;color:#3C5469}
.mission .map-key i{width:17px;height:3px;border-radius:3px;background:#1f5fa8}
.mission .map-key b{width:8px;height:8px;border-radius:50%;background:#FFB264;margin-left:7px}
.mission .map-label{position:absolute;bottom:17px;right:17px;font-size:9px;color:var(--ui-soft)}
/* THE DETAIL PANEL — REDRAWN TO THE BOARD, 11 Oct 2026.
   =================================================================
   Guillaume's review, points 1 and 3. Two faults, both structural:

   1. A LARGE NUMBERED CIRCLE repeated the number already on the tab,
      and because it sat in a flex row with the text, the eyebrow and
      the heading started ~45px to the RIGHT of the body copy and the
      "For your next drive" box. Three different left edges in one
      block. The circle is gone (the tab badge is the one number per
      moment) and every child now starts at the SAME left edge,
      because the panel is a plain block, not a flex row.

   2. THE EYEBROW REPEATED THE SKILL NAME. The tab already says
      "Anticipation"; the eyebrow said "GOOD ANTICIPATION". It now
      states the VERDICT — "Went well" or "Room to improve" — which
      is the one thing the tab does not spell out in words, and it is
      coloured by that verdict rather than by a single grey.

   THE 64ch CAP is the board's (max-width:64ch on the detail block).
   It is a READING MEASURE: this is the only running prose inside the
   portal card, and at 1106px the line would otherwise run ~140
   characters. Capping the BLOCK rather than the <p> is deliberate —
   the eyebrow, heading, copy and tip box all stop at the same right
   edge too, not just the paragraph. */
.mission .moment-feedback{padding:4px 28px 26px;max-width:64ch}
.mission .moment-verdict{display:flex;align-items:center;gap:7px;font-size:9px;letter-spacing:.1em;text-transform:uppercase;font-weight:700;margin:0 0 7px}
.mission .moment-verdict i{font-style:normal;font-size:11px;line-height:1}
/* --ok-ink / --warn-ink, not literals: these two sit inside
   .journey-preview, which follows the theme now, and the dark greens
   measured 2.18:1 and 2.93:1 on the dark card. The tokens are the
   per-theme pairs and give 9.17:1 and 7.24:1 there. The MEANING is
   unchanged -- green for "went well", amber for "room to improve" --
   and the icon and the words carry it too, so neither is doing the
   job on colour alone. */
.mission .moment-feedback[data-verdict=good] .moment-verdict{color:var(--ok-ink)}
.mission .moment-feedback[data-verdict=improve] .moment-verdict{color:var(--warn-ink)}
.mission .moment-feedback h4{font-family:Manrope,sans-serif;font-size:22px;font-weight:800;line-height:1.25;letter-spacing:-.02em;margin:0 0 9px}
.mission .moment-copy{font-size:var(--t-small);color:var(--ui-soft);line-height:1.7;margin:0 0 16px;min-height:41px}
/* THE ARROW SITS BESIDE THE TEXT, not above it (review point 3).
   align-items:flex-start keeps it on the first line rather than
   centring it against a two-line paragraph. */
.mission .next-drive{display:flex;gap:12px;align-items:flex-start;background:var(--ui-bg);border-radius:10px;padding:13px 15px;margin-top:0}
.mission .next-drive>span{color:var(--ui-accent);font-size:16px;line-height:1.5;flex:none}
.mission .next-drive p{font-size:var(--t-cap);line-height:1.6;color:var(--ui-soft);margin:0}
.mission .next-drive strong{display:block;color:var(--ui-ink);font-size:var(--t-cap)}
.mission .next-drive p>span{display:block}
/* The portal mock-up's footer strip. --ui-card / --ui-line /
   --ui-soft so it inverts with the rest of the mock-up; it was a
   near-white strip pinned in both themes, sitting under a card
   that does invert. */
.mission .portal-bottom{display:flex;justify-content:space-between;gap:15px;border-top:1px solid var(--ui-line);padding:13px 24px;font-size:9px;color:var(--ui-soft);background:var(--ui-card)}
.mission .portal-bottom span:first-child{display:flex;align-items:center;gap:8px}
.mission .portal-bottom i{width:6px;height:6px;border-radius:50%;background:var(--ui-soft)}
@media(max-width:1000px){.mission .portal-layout{grid-template-columns:270px minmax(0,1fr)}
.mission .portal-top{padding:20px}
.mission .trip-detail-header,.mission .moment-feedback{padding:22px}
.mission .portal-brand>span:last-child{display:none}}
@media(max-width:760px){.mission .driver-portal{margin-top:28px;border-radius:16px}
.mission .portal-top{padding:18px;flex-wrap:wrap;gap:10px}
.mission .portal-demo{font-size:8px;padding:5px 8px}
.mission .portal-layout{grid-template-columns:1fr}
.mission .trip-sidebar{border-right:0;border-bottom:1px solid #eaf4fa;padding:20px 16px}
.mission .trip-list-heading{padding:0}
.mission .trip-hint{padding:6px 0 14px}
.mission .trip-list{grid-template-columns:repeat(3,minmax(0,1fr));gap:6px}
.mission .trip-row{display:block;padding:11px 8px}
.mission .trip-symbol,.mission .trip-chevron,.mission .trip-tag{display:none}
.mission .trip-row strong{font-size:10px;line-height:1.5}
.mission .trip-row small{font-size:9px;line-height:1.5;margin:5px 0 0}
.mission .portal-intent{display:none}
.mission .trip-detail-header{padding:19px}
.mission .trip-detail-header>span{font-size:10px}
.mission .trip-detail-header h3{font-size:22px}
.mission .moment-feedback{padding:20px}
.mission .moment-feedback h4{font-size:17px}
.mission .moment-copy{font-size:var(--t-cap);min-height:0}
.mission .trip-map svg{min-height:175px;object-fit:contain}
.mission .map-label{display:none}
.mission .map-key{left:10px;bottom:8px;font-size:8px;padding:7px 9px}
.mission .portal-bottom{display:block;padding:13px 19px;line-height:1.7}
.mission .portal-bottom span:last-child{display:block;margin-top:4px}
.mission .next-drive p{font-size:11px}}
/* AI coverage agent: clear identity and conversational hierarchy. */
.mission .coverage h2{font-size:clamp(36px,4.1vw,55px);line-height:1.12}
/* 64px, PER THE BOARD (canvas v203 draws gap:64px on section 03's
   grid). Was 58px. THIS is the rule that wins — the 65px above it
   is earlier in the file at the same specificity — so the column gap
   is changed here. align-items:center is the board's too, and with
   the left column now one short paragraph it is what keeps it level
   with the middle of the 520px chat card. */
.mission .coverage .split{gap:64px;align-items:center}
.mission .agent-explainer{display:grid;gap:12px;margin-top:27px}
.mission .agent-explainer span{font-size:var(--t-small);color:var(--ui-soft);display:flex;gap:13px;align-items:center}
.mission .agent-explainer b{font-size:10px;color:#1f5fa8;border:1px solid #ccd9e2;background:#eff6fb;border-radius:50%;width:25px;height:25px;display:grid;place-items:center}
.mission .chat{box-shadow:0 24px 60px #142b4912;border-color:#d2dee8}
.mission .chat-header{background:#142B49;color:#fff9f1;padding:23px 24px;border-bottom:0;gap:12px}
.mission .chat-header strong{font-family:Manrope,sans-serif;font-size:19px;letter-spacing:-.5px}
.mission .chat-header small{font-size:var(--t-cap);color:#d6e4f0;margin-top:2px}
.mission .agent-avatar{width:43px;height:43px;position:relative;background:#142B49;border:1px solid #ffffff25;border-radius:13px;display:grid;place-items:center;flex:none}
/* THE AI-CARD AVATAR RING — THE ONE INSTANCE THAT IS GENUINELY
   LARGER, AND THE ONLY ONE, 11 Oct 2026.

   20px, NOT 17px, and that is a deliberate exception with a reason:
   this mark is not sitting beside a wordmark. It is alone in a 43px
   rounded tile, as the avatar for "CoRoute AI", with a sparkle badge
   overlapping its corner. At 17px it reads as a speck in the middle
   of the tile; the board draws it at 20px for that reason
   (Mission.dc.html:579, inside the 46px gradient tile).

   THE STROKE STAYS 3px. The board writes 4px here, and 4px is what
   this file had — but a 4px stroke on a 20px ring is 20% of the
   diameter against the 17.6% every other instance uses, and side by
   side the avatar's arc reads noticeably fatter than the header's.
   Guillaume's instruction was that the mark look the SAME everywhere;
   holding the stroke at 3px and letting only the diameter change is
   what makes a 20px mark read as the same letter as a 17px one. This
   is the one place this file departs from the board, deliberately,
   and it is recorded rather than silent.

   margin-right:0 because there is no "oRoute" after it to close up
   against — the -1px optical pull exists only next to the wordmark,
   and inside a centred tile it would push the ring off-centre. */
/* THE AVATAR KEEPS PIXELS AND KEEPS ITS OWN HEIGHT. It is a
   centred tile, not a letter beside other letters, so there is no
   capital height to match and no baseline to sit on. */
.mission .agent-avatar .brandmark{--mark-size:20px;--mark-size-h:20px;top:0;margin-right:0}
.mission .agent-label{margin-left:auto;font-size:var(--t-cap);letter-spacing:.12em;border:1px solid #afcde4;border-radius:6px;padding:5px 7px;color:#e7f2fb;white-space:nowrap}
.mission .agent-demo-note{padding:10px 24px;background:#edf4fa;color:#5C7084;font-size:var(--t-cap);display:flex;gap:8px;align-items:center;border-bottom:1px solid #e2ebf2}
.mission .agent-demo-note>span{width:5px;height:5px;border-radius:50%;background:#7190b0}
.mission .example-label{font-size:var(--t-cap);letter-spacing:.11em;font-weight:700;color:var(--ui-soft);margin-bottom:10px}
.mission .question-options{margin-bottom:21px;gap:6px}
.mission .question-options button{font-size:var(--t-cap);padding:10px 15px;min-height:40px}
.mission .conversation-label{font-size:var(--t-cap);letter-spacing:.09em;font-weight:700;color:var(--ui-soft);margin-bottom:8px}
.mission .user-label{text-align:right;margin-right:4px}
/* NOT --brand, though MissionV2Dark.dc.html's `.you` uses it: the
   board's chat card is light in both themes and this one is navy, so
   --brand (#A9DCFA in dark) turned the user's message into the
   brightest slab in the card. MEASURED, seen, reverted. A fixed navy
   one step off the card, with a border to hold its edge. */
.mission .bubble{font-size:var(--t-small);margin-left:26px;margin-bottom:23px;line-height:1.6;background:#1E3552;color:#E5ECF3;border:1px solid #2F4A68}
.mission .ai-label{display:flex;gap:8px;align-items:center;color:var(--ui-soft);margin-top:0}
/* MissionV2Dark.dc.html attributes the reply with a plain uppercase
   label and nothing else -- `<p class="lab">COROUTE AI</p>`. The page
   had grown a sparkle tile, a second ring and a second wordmark in
   this row, so a single card presented the brand three ways: a ringed
   tile with an orange badge, a pale-blue sparkle chip, and a bare
   ring. The chip was also a literal #dceffa, which is a light tile
   that stayed light in dark mode and became the brightest thing in
   the lower half of a navy card. All three are gone; the row is the
   board's label. */
.mission .ai-label{text-transform:uppercase;letter-spacing:.08em;font-size:11px;font-weight:800}
/* --ui-bg / --ui-line, not literals: this panel sits INSIDE .chat,
   which now follows the theme, so a pinned near-white bed put the
   card's light ink on it at 1.15:1 and 1.95:1 in dark mode -- the
   answer heading and every line of the answer. The surface and the
   ink that lands on it have to move together. */
.mission .answer{padding:17px 19px;border:1px solid var(--ui-line);border-radius:3px 14px 14px 14px;background:var(--ui-bg);min-height:0;font-size:var(--t-small);line-height:1.7}
.mission .answer strong{font-size:var(--t-body);margin-bottom:10px}
.mission .answer ul{padding-left:17px;margin:12px 0}
.mission .answer li{padding:2px 0}
.mission .agent-outcomes{display:flex;justify-content:space-between;gap:6px;margin-top:18px;font-size:var(--t-cap);color:var(--ui-soft)}
.mission .agent-outcomes b{font-size:14px;color:var(--ui-soft);font-weight:400;margin-right:4px}
.mission .chat-status{font-size:var(--t-micro);margin-top:16px;padding-top:12px;color:var(--ui-soft);line-height:1.6}
/* WAS a re-set to --t-body, which at this point in the file would
   override the 17px above it and put the line back to 21px. The
   board's figure is 17px, so this rule now agrees with it instead of
   fighting it. Left in place, not deleted: it documents that section
   04's commitment is deliberately NOT on the --t-body scale. */
.mission .coverage .commitment{font-size:17px}
.mission .coverage .small-note{font-size:var(--t-cap)}
@media(max-width:760px){.mission .coverage .split{gap:35px}
.mission .coverage h2{font-size:38px}
.mission .chat-header{padding:19px}
.mission .agent-demo-note{padding:10px 19px;font-size:9px}
.mission .chat-body{padding:19px}
.mission .agent-label{font-size:var(--t-micro)}
.mission .chat-header strong{font-size:18px}
.mission .bubble{margin-left:18px;font-size:13px}
.mission .answer{padding:16px;font-size:12px}
.mission .agent-outcomes{flex-wrap:wrap;gap:9px}
.mission .ai-label small{font-size:8px}
.mission .agent-explainer{margin-top:22px}}
/* Orange C is the first letter of the wordmark. */
/* BASELINE, not centre: centring puts the ring on the text box
   including descender space, which floats a mark that has none.
   The v2 spec says "align-items: baseline". */
.mission .brand{gap:0;align-items:baseline}
.mission .brandmark{flex:none}
.mission .portal-brand{gap:2px}
.mission .portal-brand>span:last-child{margin-left:18px}
/* THE FOOTER MARK IS GONE, AND SO IS THE RULE — 11 Oct 2026.

   It was `width:21px;height:21px;border-width:4px`, the heaviest of
   the five variants, and auditing it turned up something better than
   a size to correct: /our-mission HAS NO .mission .footer. The page
   ends with the shared .sitefooter from _chrome.html, whose mark is
   app.css's .brand-mark at 17px/3px and is already right. Every
   `.mission .footer` rule in this file is dead — verified in the
   browser, 0 matches for .mission .footer .brandmark on the rendered
   page.

   So the honest fix is deletion, not a third number. A rule that
   specifies a mark nobody can see is exactly the kind of claim this
   project keeps getting bitten by: the next person counting the
   variants would have counted this one.

   (The remaining dead .mission .footer rules above are left alone —
   they are outside this round's scope, and this note is the record
   that they are dead.) */
/* Aerial imagery with native, interactive HUD. */
/* ---- THE MANCHESTER BASEMAP IN DARK MODE -------------------------
   v2, 22 Oct 2026. Same file, same route geometry, same markers and
   the same replay logic -- v2rules is explicit that none of those
   change, and none of them do. The raster is a daylight street map
   and in dark mode it is the brightest thing on a dark page, so it
   is DIMMED, not replaced and NOT INVERTED.

   filter: brightness(.78) saturate(.85) -- the brief's values and
   the canvas's. An invert() was explicitly rejected: inverting a map
   turns green parks magenta and blue water orange, which is a
   different map rather than the same map at night, and the route
   drawn on top would then be inverted relative to its own legend.

   THE FILTER IS ON THE <image>, NOT ON THE SVG OR THE CONTAINER.
   Putting it on either would dim the route, the three markers and
   the vehicle along with the basemap -- the exact elements that have
   to stay legible on top of it, and whose colours were chosen
   against the undimmed raster. Scoping it to the raster keeps the
   overlay at full strength, which is what makes the dimming work. */
@media (prefers-color-scheme: dark){
  .mission:where(:root:not([data-theme="light"]) *) #route-map-tiles{
    filter:brightness(.78) saturate(.85)}
}
.mission:where(:root[data-theme="dark"] *) #route-map-tiles{
  filter:brightness(.78) saturate(.85)}
.mission .aerial-map{background:#384236;isolation:isolate}
.mission .aerial-map svg{display:block;width:100%;height:auto;min-height:0}
.mission .hud-top{position:absolute;left:18px;right:18px;top:17px;display:flex;justify-content:space-between;align-items:start;pointer-events:none}
.mission .hud-mode{background:#0e1b2cdf;backdrop-filter:blur(8px);color:#fff;font-size:9px;letter-spacing:.1em;line-height:1.5;padding:11px 13px;border:1px solid #ffffff26;border-radius:9px}
.mission .hud-mode>span{display:inline-block;width:5px;height:5px;border-radius:50%;background:#ffb264;margin-right:6px}
.mission .hud-mode small{font-size:8px;color:#dce5ec;letter-spacing:0;display:block;margin-top:3px}
.mission .hud-compass{background:#0e1b2cdf;border:1px solid #ffffff26;border-radius:20px;color:#fff;font-size:10px;width:30px;text-align:center;line-height:1.3;padding:7px 0}
.mission .hud-metrics{position:absolute;left:18px;bottom:36px;display:flex;background:#0e1b2ce8;backdrop-filter:blur(8px);border:1px solid #ffffff26;border-radius:10px;color:#fff;padding:13px 16px;gap:21px}
.mission .hud-metrics>div+div{border-left:1px solid #ffffff26;padding-left:21px}
.mission .hud-metrics span{font-size:7px;letter-spacing:.12em;color:#c1d1dd;display:block;margin-bottom:4px}
.mission .hud-metrics strong{font-size:13px;line-height:1.3;display:block;font-weight:500}
.mission .hud-metrics>div:first-child strong{font-size:24px;font-weight:700}
.mission .hud-metrics small{font-size:10px;color:#d2dde5;font-weight:400}
.mission .hud-event{position:absolute;right:18px;top:96px;display:flex;align-items:center;gap:9px;border:1px solid #fbd8af;background:#fff9f1f5;color:#142b49;border-radius:10px;box-shadow:0 4px 20px #0003;padding:10px 12px;text-align:left}
.mission .hud-event>span{width:23px;height:23px;background:#ffb264;display:grid;place-items:center;border-radius:50%;font-size:11px;font-weight:700}
/* A LITERAL, LIKE THE PILL AROUND IT. .hud-event pins its own cream
   fill because it sits ON the daylight basemap in both themes, so
   its ink must be pinned with it -- the grey-ink pass converted this
   one and put a themed grey on a pinned cream at 1.82:1. #4A5E72 on
   #fff9f1 is 6.40:1, both themes, because neither moves. */
.mission .hud-event div{font-size:8px;color:#4A5E72}
.mission .hud-event strong{display:block;font-size:11px;color:#142b49;margin-top:2px}
.mission .hud-event>b{font-size:16px;font-weight:400}
.mission .hud-event:hover{background:#FFFFFF}
/* THE OPENSTREETMAP ATTRIBUTION — z-index:2 ADDED 11 Oct 2026, AND IT
   WAS ALREADY BROKEN BEFORE THIS ROUND TOUCHED ANYTHING.

   It is a LICENCE CONDITION, not a credit we chose to give, and the
   board says so twice: it must stay visible and linked. It was
   neither. `.mission .trip-map svg` carries an explicit z-index:1
   (so it sits above the no-JS fallback basemap); this span had
   z-index:auto, so the basemap painted straight over it and the
   attribution was INVISIBLE on the opened-drive card.

   Found while checking that the relocated legend had not covered it —
   elementsFromPoint at its own centre returned image, svg, A,
   SPAN.aerial-credit, in that order, with the attribution last. It
   was never covered BY the legend; it was already underneath the map.

   No text check could see this. The markup was correct, the link was
   right, the contrast was fine, nothing overflowed. Only the stacking
   order was wrong, and only a picture and a hit test showed it. */
.mission .aerial-credit{position:absolute;z-index:2;right:12px;bottom:8px;font-size:8px;color:#fff;background:#0e1b2cbb;padding:3px 6px;border-radius:3px}
/* THE PLAYER BAR — Guillaume's review points 4, 5 and 6, 11 Oct 2026.
   =================================================================

   THE PLAY BUTTON IS THE PAGE'S ONLY INVITATION TO PRESS ANYTHING
   inside this card, and it was a dim grey outline on a dark bar —
   quieter than the static legend sitting beside it. It is now the
   board's control: 36px, solid #FFAA66, ink glyph, with a 3px solid
   bottom edge in #C8682A that reads as a physical key. It remains a
   real <button> with an aria-label that changes with its state; the
   colour is the whole of the change.

   THE SCRUBBER CARRIES THE THREE EVENTS. A bare track said nothing
   about where anything happened, so the dots on the map and the
   position of the playhead were two unrelated facts. The dots sit at
   the SAME 25/53/79% as the map markers, and they are not written
   down twice: mission.js builds them from moments[], the same array
   getPointAtLength() reads to place the markers.

   They are BUTTONS, not decorations. Clicking one selects that
   moment, which is the behaviour the map markers already had; making
   them focusable costs nothing and gives the keyboard the same
   shortcut as the mouse. :focus-visible is explicit here because the
   bar is dark and the UA default ring is nearly invisible on it. */
.mission .replay-controls{display:flex;align-items:center;gap:16px;padding:14px 24px;background:#16283e;color:#e9f2f9}
.mission .replay-controls>button{width:36px;height:36px;border-radius:50%;border:0;background:#FFAA66;color:#142B49;font-size:13px;flex:none;box-shadow:0 3px 0 #C8682A;cursor:pointer;display:grid;place-items:center;padding:0}
.mission .replay-controls>button:focus-visible{outline:2px solid #fff9f1;outline-offset:3px}
.mission .replay-track{flex:1;min-width:0}
.mission .replay-track>div{display:flex;justify-content:space-between;font-size:8px;letter-spacing:.03em;color:#cedbe7;margin-bottom:5px}
/* The dots overlay the range input, so the two share one coordinate
   space and one width. pointer-events on the host is none, restored
   on each dot: the bare track between the dots must still drag.

   THE HOST IS NOT aria-hidden, 12 Oct 2026. It carried
   aria-hidden="true" while mission.js fills it with real <button>s
   that have aria-labels and click handlers. A focusable control
   inside an aria-hidden subtree is hidden from assistive technology
   but still in the tab order, so a screen-reader user tabs onto
   three controls that announce nothing — WCAG 4.1.2 Name, Role,
   Value. The comment above this block already called them "BUTTONS,
   not decorations ... making them focusable costs nothing and gives
   the keyboard the same shortcut as the mouse", which is exactly
   what the attribute undid.

   It is now role="group" with a label, so the three dots announce as
   one named set and each button keeps its own "Jump to moment N"
   label. */
.mission .replay-scrub{position:relative}
.mission .replay-track input{width:100%;display:block;margin:0;accent-color:#ffb264;height:16px;cursor:pointer}
.mission .replay-track input:focus-visible{outline:2px solid #fff9f1;outline-offset:2px}
.mission .replay-dots{position:absolute;inset:0;pointer-events:none}
/* THE DOT IS 10px TO LOOK AT AND 24px TO HIT.
   ------------------------------------------------------------------
   Measured 12 Oct 2026 on the built page: these buttons were 10x10,
   and 12x12 when selected, at BOTH 1280 and 390. WCAG 2.5.8 Target
   Size (Minimum) is Level AA and asks for 24x24 CSS px, so three
   controls on this page failed it. (2.5.5's 44px is Level AAA and is
   NOT the bar being applied here.)

   THE BOARD DRAWS 10px AND 12px AND THAT IS KEPT. The fix is not to
   grow the dot — a 24px disc on a 4px track would dominate the
   scrubber and stop matching Mission.dc.html. Instead the <button>
   itself becomes a 24px transparent square centred on the dot's
   position, and the visible dot is painted by ::before inside it. So
   the picture is the board's and the hit area is the standard's.

   WHY NOT A ::before ON THE DOT INSTEAD, i.e. keep the button 10px
   and hang a larger pseudo-element off it: a pseudo-element does
   enlarge the clickable area, but 2.5.8 is measured on the target,
   and leaving the element 10px makes the success criterion depend on
   reading the pseudo-element's box. This way the button's own
   getBoundingClientRect() is 24x24, which is what an audit measures
   and what the probe above now reports.

   SPACING IS NOT AT RISK. The dots sit at 25/53/79% of a track that
   is 1006px at 1280 and 265px at 390 — 70px apart at the narrowest,
   so three 24px targets never overlap. Checked at 390, the width
   where it would bite first. */
.mission .replay-dot{position:absolute;left:var(--at);top:50%;width:24px;height:24px;margin:-12px 0 0 -12px;padding:0;border:0;background:none;display:grid;place-items:center;pointer-events:auto;cursor:pointer}
.mission .replay-dot::before{content:'';display:block;width:10px;height:10px;border-radius:50%;background:#FFAA66}
.mission .replay-dot.is-passed::before{background:#a9dcfa}
.mission .replay-dot.is-selected::before{width:12px;height:12px;box-shadow:0 0 0 3px #ffaa6673}
.mission .replay-dot.is-passed.is-selected::before{box-shadow:0 0 0 3px #a9dcfa73}
/* The ring follows the visible dot, not the invisible 24px box, or
   the focus indicator would float unexplained around empty track. */
.mission .replay-dot:focus-visible{outline:none}
.mission .replay-dot:focus-visible::before{outline:2px solid #fff9f1;outline-offset:3px}
/* THE LEGEND MOVED OFF THIS BAR AND ONTO THE MAP (review point 6).
   It explains two marks that are BOTH on the map — a purple line and
   an orange dot — and on the player bar it was explaining them from
   a different box, next to a clock it had nothing to do with. The
   board puts it bottom-right on the map as a white chip; it sits
   above the OSM credit, which keeps its own corner.

   z-index:2 IS LOAD-BEARING AND WAS FOUND BY LOOKING. Without it the
   chip is in the DOM, correctly positioned, fully opaque — and
   INVISIBLE, because `.mission .trip-map svg` further down this file
   carries an explicit z-index:1 (it is there to sit above the no-JS
   fallback basemap). An explicit z-index beats an auto one whatever
   the DOM order, so the map painted straight over the chip.

   Every measurement said it was fine: getBoundingClientRect put it
   inside the map, computed visibility was `visible`, opacity 1. Only
   elementsFromPoint at its own centre — and the screenshot — showed
   image and svg stacked ON TOP of it. This is the fault class
   CLAUDE.md describes: a check that cannot see shape reports a pass.

   2, not 1: level with the other overlays and still under nothing.

   bottom:44px, NOT 28px, AND THAT IS THE SECOND THING THE PICTURE
   CAUGHT. At 28px the chip cleared the map fine and sat straight on
   top of the OpenStreetMap attribution, which is pinned bottom-right
   at bottom:8px and is ~23px tall. Raising the chip's own z-index is
   what made it win that overlap — so fixing the first fault created
   the second, and only the screenshot showed it.

   THE ATTRIBUTION IS A LICENCE CONDITION, not a credit we chose to
   give (the board says so twice). It keeps the corner it had; the
   legend stacks above it. 44px = the credit's 8px bottom + its ~23px
   height + 13px of air. */
.mission .replay-key{position:absolute;z-index:2;right:12px;bottom:44px;display:flex;gap:14px;align-items:center;background:#fffffff0;border-radius:8px;padding:7px 12px;font-size:10px;line-height:1;color:#142B49}
.mission .replay-key>span{display:inline-flex;align-items:center;gap:6px}
.mission .replay-key i{width:16px;height:3px;background:#142B49;border-radius:2px}
.mission .replay-key b{width:8px;height:8px;border-radius:50%;background:#FFAA66}
@media(max-width:760px){.mission .hud-top{left:10px;right:10px;top:10px}
.mission .hud-mode{font-size:7px;padding:8px 9px}
.mission .hud-mode small{font-size:7px}
.mission .hud-compass{width:24px;font-size:8px;padding:5px 0}
.mission .hud-event{right:10px;top:69px;padding:7px 8px;gap:6px}
.mission .hud-event strong{font-size:9px}
.mission .hud-event div{font-size:7px}
.mission .hud-event>span{width:19px;height:19px;font-size:9px}
.mission .hud-metrics{left:10px;bottom:27px;gap:12px;padding:9px 11px}
.mission .hud-metrics>div+div{padding-left:12px}
.mission .hud-metrics>div:first-child strong{font-size:19px}
.mission .hud-metrics strong{font-size:10px}
.mission .hud-metrics span{font-size:6px}
.mission .hud-metrics small{font-size:8px}
.mission .aerial-credit{font-size:6px;bottom:6px;right:8px}
.mission .replay-controls{padding:12px 16px;gap:11px}
/* The legend now sits on the map, not on this bar, so the phone rule
   that USED TO HIDE IT is a size rule instead: at 390 the chip still
   fits bottom-right, and hiding it would take the only explanation of
   the purple line and the orange dots with it. */
.mission .replay-key{font-size:8px;gap:9px;padding:5px 8px;right:8px;bottom:30px}
.mission .replay-controls>button{width:32px;height:32px;box-shadow:0 2px 0 #C8682A}
.mission .aerial-map svg{min-height:0}}
/* Opened Recent drives card, matching the user's portal. */
.mission .portal-layout.opened-drive{display:block}
.mission .opened-drive .trip-detail-header{padding:8px 30px 22px}
.mission .drive-breadcrumb{font-size:10px;color:var(--ui-soft);padding:22px 30px 11px}
.mission .drive-breadcrumb span{padding:0 9px;color:var(--ui-soft)}
.mission .opened-drive .trip-detail-header h3{font-size:31px;letter-spacing:-.04em}
.mission .trip-detail-header h3>span{font-size:23px;color:var(--ui-soft);margin:0 4px}
.mission .drive-summary{font-size:12px;color:var(--ui-soft);margin-top:10px}
.mission .drive-summary>span{padding:0 9px;color:var(--ui-soft)}
/* A SELF-CONTAINED GREEN CHIP, so its fill and ink are the --ok-*
   PAIR and move together. It was a pinned light green with themed
   ink, which put --ui-soft on it at 2.97:1 in dark mode. */
.mission .drive-score{padding:10px 18px;background:var(--ok-bg);border:1px solid var(--ok-ink);border-radius:12px;text-align:center}
.mission .drive-score>span{font-size:9px;color:var(--ok-ink);display:block}
.mission .drive-score strong{font-size:29px;color:var(--ok-ink);line-height:1.4}
/* --ok-ink, like the figure above it: these two sit on --ok-bg, so
   they belong to the same pair. The sage greens they replace were
   picked against the old pinned #f1f8f3 and measured 2.90:1 and
   3.07:1 there -- below AA even in the theme they were chosen for.
   6.52:1 / 9.10:1 now. */
.mission .drive-score small{font-size:11px;font-weight:400;color:var(--ok-ink)}
.mission .drive-score b{display:block;font-size:8px;font-weight:400;color:var(--ok-ink)}
.mission .ai-drive-summary{display:flex;gap:12px;align-items:center;padding:14px 30px;background:var(--ui-bg);border-top:1px solid var(--ui-line);border-bottom:1px solid var(--ui-line)}
/* --ui-accent: the sparkle sits on --ui-bg, which follows the theme,
   and the literal was 2.63:1 on the dark one. */
.mission .ai-drive-summary>span{font-size:20px;color:var(--ui-accent)}
.mission .ai-drive-summary p{font-size:var(--t-cap);color:var(--ui-soft);line-height:1.6}
.mission .ai-drive-summary strong{display:block;font-size:10px;color:var(--ui-soft);margin-bottom:2px}
.mission .opened-drive .aerial-map{max-height:none}
.mission .opened-drive .aerial-map svg{max-height:490px;width:100%;object-fit:cover}
/* THE MOMENT TABS — Guillaume's review points 1 and 2, 11 Oct 2026.
   =================================================================

   ONE NUMBER BADGE PER MOMENT, AND IT LIVES HERE. The badge has
   exactly two states and no third: ORANGE when this tab is selected,
   LILAC (the panel background, brand text) when it is not. The
   board's tab row draws precisely that.

   THE VERDICT IS VISIBLE BEFORE ANY CLICK. This is the substantive
   half of the change, not styling: previously you had to select each
   moment in turn to discover which one went badly. A small round
   icon on the right of every tab answers that for all three at once
   — a tick on pale green for "went well", an arrow on pale orange
   for "room to improve" — and the subtitle naming the skill is
   coloured to match, so the verdict reads even if the icon does not.

   WHY data-verdict RATHER THAN :nth-child. The verdict belongs to
   the MOMENT, and moments[] in mission.js is its source. An
   nth-child rule would hard-code "the third one is the bad one" into
   the stylesheet, and would silently colour the wrong tab the first
   time the order changed. */
.mission .drive-moments{display:grid;grid-template-columns:repeat(3,minmax(0,1fr));gap:12px;padding:20px 28px 14px}
/* The drive-moment cards sit INSIDE the portal mock-up, so they take
   the --ui-* set like the rest of it. They were white tiles with
   #33495E ink pinned in both themes, which left three bright cards on
   a navy card in dark mode. MEASURED 23 Oct. */
.mission .drive-moments button{border:1px solid var(--ui-line);border-radius:12px;background:var(--ui-card);color:var(--ui-soft);padding:12px 14px;display:flex;gap:12px;align-items:center;text-align:left;font-size:11px}
.mission .drive-moments button b{display:grid;place-items:center;width:26px;height:26px;border-radius:50%;background:var(--ui-accent-soft);color:var(--ui-ink);font-size:13px;font-weight:800;flex:none}
.mission .drive-moments button>span{flex:1;min-width:0}
.mission .drive-moments button>span{font-size:13px;font-weight:700}
.mission .drive-moments button small{display:block;font-size:11px;font-weight:600;margin-top:3px}
.mission .drive-moments button[data-verdict=good] small{color:var(--ok-ink)}
/* No --ui-warn exists; these two keep a measured fixed pair. */
.mission .drive-moments button[data-verdict=improve] small{color:var(--up)}
/* The verdict icon. 22px, round, glyph in the verdict's own ink on
   its own pale ground — the board's #E3F4EA/#0F6B3F and
   #FFEBD9/#B4561A pairs.

   THE AMBER IS DEEPENED TO #A64E15, A DEPARTURE FROM THE BOARD, and
   it is a contrast fix rather than a preference: the board's pair
   measures 4.23:1, which is short of 4.5:1 for a 12px glyph. It
   PRE-DATES v2 -- the same figure appears on master, and the pair is
   pinned in both themes, so nothing about the recolour caused it. It
   is corrected here because it was measured here. 4.87:1 now.

   The GREEN pair is untouched: #0F6B3F on #E3F4EA measures 5.76:1
   and passes as it stands. Flagged for the canvas -- the board
   should take the deeper amber. */
.mission .drive-moments button i{width:22px;height:22px;flex:none;border-radius:50%;display:grid;place-items:center;font-size:12px;font-weight:800;font-style:normal}
.mission .drive-moments button[data-verdict=good] i{background:var(--ok-bg);color:var(--ok-ink)}
.mission .drive-moments button[data-verdict=improve] i{background:#FFEBD9;color:#A64E15}
.mission .drive-moments button[aria-pressed=true]{border-color:var(--ui-accent);border-width:1.5px;background:var(--ui-accent-soft);color:var(--ui-ink)}
.mission .drive-moments button[aria-pressed=true] b{background:var(--cta);color:var(--on-cta)}
.mission .opened-drive .moment-feedback{padding:22px 30px 26px}
.mission .opened-drive .moment-feedback>p{min-height:0}
.mission .opened-drive .hud-metrics{bottom:36px}
.mission .opened-drive .hud-event{top:100px}
@media(max-width:760px){.mission .drive-breadcrumb{padding:18px 19px 10px;font-size:9px}
.mission .opened-drive .trip-detail-header{padding:5px 19px 18px}
.mission .opened-drive .trip-detail-header h3{font-size:25px}
.mission .drive-summary{font-size:10px}
.mission .drive-summary>span{padding:0 4px}
.mission .drive-score{padding:8px 10px}
.mission .drive-score strong{font-size:23px}
.mission .drive-score b{font-size:7px}
.mission .ai-drive-summary{padding:12px 19px}
.mission .ai-drive-summary p{font-size:var(--t-micro)}
.mission .drive-moments{padding:14px 16px 10px;gap:6px}
/* STACKED ON THE PHONE, because three columns of a badge + two lines
   + an icon do not fit 390px side by side. The verdict icon moves to
   the TOP ROW beside the badge rather than disappearing: the whole
   point of review point 2 is that the verdict is visible without a
   click, and that must hold at 390 too. */
.mission .drive-moments button{padding:9px 8px;display:grid;grid-template-columns:auto 1fr auto;gap:6px;font-size:9px;align-items:center}
.mission .drive-moments button b{grid-row:1;grid-column:1;width:20px;height:20px;font-size:10px}
.mission .drive-moments button i{grid-row:1;grid-column:3;width:18px;height:18px;font-size:10px}
.mission .drive-moments button>span{grid-row:2;grid-column:1/-1;font-size:10px}
.mission .drive-moments button small{font-size:9px;margin-top:2px}
.mission .opened-drive .moment-feedback{padding:19px}
.mission .opened-drive .hud-event{top:69px}
.mission .opened-drive .hud-metrics{bottom:27px}}
/* Full-drive terrain map keeps image and overlays in one coordinate system. */
.mission .opened-drive .aerial-map svg{max-height:none;width:100%;height:auto;aspect-ratio:1642/996}
.mission .opened-drive .hud-event{top:auto;bottom:36px;right:18px}
/* THE OPENED-DRIVE HUD PILLS ARE PINNED CREAM IN BOTH THEMES and
   that is deliberate: they sit ON the Manchester basemap, which is a
   daylight raster in both themes (dimmed in dark, never inverted --
   see app.css). A pill that went dark would disappear into the
   dimmed map. So the fill is pinned and THE INK IS PINNED WITH IT,
   which is the rule: #142b49 on #fff9f1 is 13.64:1 either way. */
.mission .hud-mode{background:#fff9f1f5;color:#142b49;border-color:#cbd8e1;box-shadow:0 3px 15px #142b4910}
.mission .hud-mode small{color:#5C7084}
.mission .hud-compass{background:#fff9f1f5;color:#142b49;border-color:#cbd8e1}
.mission .aerial-credit{background:#fff9f1e8;color:#4A5E72}
.mission .route-number{filter:drop-shadow(0 2px 3px #142b4926)}
@media(max-width:760px){.mission .opened-drive .hud-event{right:10px;bottom:27px;top:auto}
.mission .hud-metrics>div+div{display:none}
.mission .hud-event div{max-width:112px}
.mission .aerial-credit{font-size:6px}
.mission .opened-drive .aerial-map svg{min-height:0}
.mission .hud-metrics{max-width:90px}}
/* Road-context comparison: photographic scene with native annotations. */
.mission .demo{grid-template-columns:1.2fr 1fr;border-radius:20px;background:#142B49}
.mission .road-panel{padding:0;min-height:0;background:#112136;justify-content:flex-start}
.mission .camera-topbar{display:flex;justify-content:space-between;gap:12px;align-items:center;background:#0e1b2c;padding:17px 21px;font-size:9px;color:#b4c6d4;letter-spacing:.09em}
.mission .camera-topbar span:first-child{font-weight:700;color:#e5eff7;display:flex;align-items:center;gap:7px}
.mission .camera-topbar i{width:5px;height:5px;background:#ffb264;border-radius:50%}
.mission .camera-topbar span:last-child{letter-spacing:0}
.mission .camera-scene{position:relative;overflow:hidden;isolation:isolate}
.mission .camera-scene img{display:block;width:100%;height:auto;aspect-ratio:1672/941;object-fit:cover}
.mission .camera-scene:after{content:'';position:absolute;inset:65% 0 0;background:linear-gradient(transparent,#08122090);pointer-events:none;z-index:0}
/* ---- THE OVERLAY BOX, RETUNED TO THE V2 SCENES -------------------
   22 Oct 2026. These percentages frame a SUBJECT INSIDE A PICTURE,
   so they are only ever correct for one picture: the v2 pack's
   scenes are newly composed, not recolours of the old ones, and the
   subjects are in different places. The old values (31%/45% and
   34%/30%) framed empty road on the new art -- looked at, not
   inferred.

   READ OFF THE SHIPPED FILES at their full 1279x720:

     context-following  the blue lead car spans x 55-76%, y 39-69%
     context-pedestrian the walking figure spans x 72-82%, y 36-72%

   The boxes below sit just inside those, which is what a detection
   box looks like. They are approximate ON PURPOSE -- the panel is
   labelled "ILLUSTRATED SCENARIO" and the box is a depiction of a
   detector, not the output of one. Re-read them if the art changes
   again; nothing computes these.

   THE DEFAULT RULE IS THE "following" SCENE and .scenario-safe
   overrides it with the pedestrian, which is the arrangement that
   was already here -- the class is on the panel, not the box, and
   mission.js toggles it. Unchanged. */
.mission .context-focus{position:absolute;left:55%;top:41.5%;width:21%;height:30%;border:1px solid #ffb26480;filter:drop-shadow(0 1px 2px #0008);z-index:1}
.mission .scenario-safe .context-focus{left:72.5%;top:43%;width:8%;height:30%}
.mission .context-focus>span{position:absolute;left:-1px;bottom:calc(100% + 7px);background:#142b49ed;border:1px solid #ffb26480;color:#fff4e8;white-space:nowrap;font-size:9px;line-height:1;padding:7px 9px;border-radius:4px}
.mission .context-focus i{position:absolute;width:13px;height:13px;border-color:#ffb264;border-style:solid}
.mission .context-focus i:nth-of-type(1){top:-2px;left:-2px;border-width:3px 0 0 3px}
.mission .context-focus i:nth-of-type(2){top:-2px;right:-2px;border-width:3px 3px 0 0}
.mission .context-focus i:nth-of-type(3){bottom:-2px;left:-2px;border-width:0 0 3px 3px}
.mission .context-focus i:nth-of-type(4){bottom:-2px;right:-2px;border-width:0 3px 3px 0}
.mission .camera-caption{position:absolute;bottom:10px;left:13px;font-size:8px;color:#fff;z-index:2;letter-spacing:.02em;text-shadow:0 1px 4px #000}
.mission .camera-readout{display:flex;justify-content:space-between;align-items:center;gap:20px;padding:23px 25px 20px}
.mission .readout-label{display:block;font-size:8px;letter-spacing:.12em;color:#afc8d6;margin-bottom:7px}
.mission .camera-readout strong{font-family:Manrope,sans-serif;font-size:26px;letter-spacing:-.05em;color:#fff9f1}
.mission .event-symbol{display:block;width:115px;flex:none;color:#ffb264}
.mission .event-symbol svg{display:block;width:100%;height:auto}
.mission .evidence-strip{display:flex;justify-content:center;align-items:center;gap:12px;border-top:1px solid #ffffff12;padding:14px 12px;font-size:10px;color:#b6cadb;background:#13243A}
/* A LITERAL: this strip is #13243A in BOTH themes, so its symbols
   must be light in both. The grey-ink pass converted this one and
   put a themed grey on a pinned dark band at 2.76:1 in LIGHT mode --
   the mirror image of the fault that pass was fixing. 9.29:1 now. */
.mission .evidence-strip b{font-size:13px;color:#b6cadb;font-weight:400}
.mission .evidence-strip strong{color:#ffcb98;font-weight:500}
.mission .context-detail{padding:30px}
.mission .context-detail .tabs{gap:7px;margin-bottom:27px}
.mission .context-detail .tabs button{font-size:11px;padding:10px 12px}
.mission .context-detail h3{font-size:28px}
.mission .context-detail p{font-size:var(--t-small)}
.mission .context-detail .meaning{font-size:var(--t-small)}
@media(max-width:760px){.mission .demo{grid-template-columns:1fr}
.mission .road-panel{padding:0}
.mission .camera-topbar{padding:14px 18px;font-size:8px}
.mission .context-focus>span{font-size:7px;padding:6px 7px;bottom:calc(100% + 5px)}
.mission .camera-readout{padding:19px 21px}
.mission .camera-readout strong{font-size:24px}
.mission .event-symbol{width:95px}
.mission .evidence-strip{font-size:9px;gap:9px}
.mission .context-detail{padding:24px}
.mission .context-detail h3{font-size:26px}
.mission .camera-caption{font-size:7px}}
.mission .opened-drive .aerial-map svg{aspect-ratio:1536/1024}
.mission .aerial-credit{font-size:10px}
.mission .aerial-credit a{text-decoration:underline;text-underline-offset:2px}
@media(max-width:760px){.mission .aerial-credit{font-size:8px}
.mission .drive-summary{max-width:205px}}
/* Cartoon scenarios keep annotation clear of the illustrated action. */
.mission .context-focus,.mission .scenario-safe .context-focus{left:16px;top:16px;width:auto;height:auto;border:0;filter:none}
.mission .context-focus>span{position:static;display:block;background:#142b49ed;border-color:#ffffff30;font-size:10px;border-radius:20px;padding:8px 12px}
.mission .context-focus i{display:none}
.mission .camera-caption{display:none}
@media(max-width:760px){.mission .context-focus,.mission .scenario-safe .context-focus{left:12px;top:12px}
.mission .context-focus>span{font-size:8px;padding:7px 10px}}
/* Replay and feedback share the same event thresholds, forwards and backwards. */
.mission .route-number circle,.mission .route-number text,.mission .drive-moments button,.mission .drive-moments button b,.mission .hud-event>span,.mission .replay-dot{transition:fill .2s ease,background-color .2s ease,color .2s ease,border-color .2s ease}
.mission .route-number.is-passed circle{fill:#1f5fa8}
.mission .route-number.is-passed text{fill:#fff}
.mission .drive-moments button.is-passed b{background:var(--ui-accent);color:var(--ui-card)}
.mission .drive-moments button.is-passed[aria-pressed=true]{background:var(--ui-accent);border-color:var(--ui-accent);color:var(--ui-card)}
.mission .drive-moments button.is-passed[aria-pressed=true] small{color:#dae9f6}
.mission .drive-moments button.is-passed[aria-pressed=true] b{background:var(--ui-card);color:var(--ui-accent)}
.mission .hud-event.is-passed>span{background:#1f5fa8;color:#fff;border-color:#1f5fa8}
@media(prefers-reduced-motion:reduce){.mission .route-number circle,.mission .route-number text,.mission .drive-moments button,.mission .drive-moments button b,.mission .hud-event>span,.mission .replay-dot{transition:none}}
/* ===================================================================
   ADDED DURING THE PORT — not in the handoff's mission.css
   =================================================================== */

/* ===================================================================
   THE app.css BOUNDARY RESET — CLASS-NAME COLLISIONS
   ===================================================================

   SCOPING mission.css WAS NECESSARY AND IS NOT SUFFICIENT, and this
   block is the other half.

   Scoping stops THIS stylesheet reaching OUT. It does nothing about
   app.css reaching IN: app.css's selectors are unscoped, and the
   handoff happens to use seven class names that already mean
   something else here —

       brand  hero  hero-copy  portal-brand  portal-top
       principle  wrap

   — plus the bare element selectors `button`, `h1` and `main`.

   IT WAS EIGHT UNTIL 12 Oct 2026: `pill` was on that list, and both
   rules for it went when the PRE-LAUNCH chip came out of the nav.
   The collision is still real in app.css — a bare `.pill` there
   sets a cream fill and dark ink — there is simply nothing on this
   page for it to collide with now. The measurements below were taken
   while the chip was still there and are not re-run; they describe
   how the collisions were found, which has not changed.

   HOW THIS WAS FOUND, because it matters for how it is maintained.
   Not by reading the two files: by asking the BROWSER, through
   CSS.getMatchedStylesForNode over every element inside .mission, to
   list the rules that actually matched and report the ones whose
   selector does not mention .mission. That returned TWELVE rules at
   390px and TEN at 1280px. Two of them were visible in a screenshot
   and ten were not.

   THE TWO THAT WERE VISIBLE, both caught by comparing against the
   handoff's own rendering side by side:

     .hero-copy { order: 1 } at <=61rem put the hero ARTWORK ABOVE
       THE HEADLINE at 390px. The image came first and "Our mission"
       followed it — the page read in the wrong order on a phone, and
       the DOM was never touched. app.css line 3419; it is the
       homepage's phone reorder, which lifts its video card out of
       the hero using `order` on a `main` turned into a flex column.
       Our page reuses the class name and inherited the behaviour.

     button { width: 100%; min-width: 14rem } made every one of the
       eleven buttons 224px wide: the two scenario pills stacked into
       a column instead of sharing a row, and the circular replay
       play button became a 224x44 ellipse. Handled in the button
       reset above, which has its own note.

   SO EVERY COLLIDING NAME IS NEUTRALISED HERE, not just the two that
   showed. A per-symptom patch fixes what someone happened to look at
   and leaves the rest — which is exactly how the play button stayed
   an ellipse through the first pass of this port.

   THESE RULES RESET TO THE BROWSER DEFAULT. They do not set values of
   our own. The handoff's own scoped rules then apply on top, so what
   renders is the approved design rather than a reinterpretation of
   it. Where the handoff sets nothing, the browser default is what the
   delivered page had.

   .mission IS ON THE SAME ELEMENT AS NOTHING ELSE, so `.mission .x`
   (0,2,0) beats app.css's `.x` (0,1,0) and `.mission main` beats
   `main`. The media queries below repeat the breakpoints app.css uses
   (61rem) rather than the handoff's (760px) because they exist to
   cancel app.css, and a cancel that does not cover the same range
   leaves a gap. */

/* main: app.css makes it a flex column at <=61rem for the homepage's
   card reorder. Nothing on this page wants that, and a flex container
   is what makes `order` do anything at all. */
.mission main { display: block; }

/* .hero-copy: the order is the fault above. The background, colour
   and padding come with it in the same rule — this page draws its own
   purple ground on .hero, so a second one on the copy block would
   paint a panel inside a panel. min-width:0 at wide widths is
   harmless but is reset for consistency. */
.mission .hero-copy {
  order: 0;
  background: none;
  color: inherit;
  padding: 0;
  min-width: 0;
}

/* .hero: app.css gives it vertical padding for the homepage's hero.
   The handoff's .hero is the full purple panel and sets its own. */
.mission .hero { padding-block: 0; }

/* .wrap: 75rem with a clamped gutter in app.css, 1180px with 36px
   here. The handoff's own .mission .wrap rule already wins on all
   four properties, so this is belt-and-braces against app.css growing
   a fifth. */
.mission .wrap { width: 100%; }

/* .pill: REMOVED 12 Oct 2026 with the PRE-LAUNCH chip it styled.
   Guillaume: "remove the pre-launch that is mensioned everywhere".
   Nothing on this page carries class="pill" any more, so both the
   handoff's own .mission .pill rule and this override of app.css's
   cream badge are gone. app.css's bare .pill is left alone: it is
   documented there as the quote flow's "Planned for launch" pill and
   is not this page's to delete. If a pill ever returns to this page,
   it needs BOTH rules back — app.css sets a cream fill and dark ink
   that an outlined chip on the hero has to override. */

/* .principle: a font-weight and colour in app.css, meant for a line in
   the quote flow. Here it is the bordered callout under section 01. */
.mission .principle { font-weight: inherit; color: inherit; }

/* .portal-top and .portal-brand: app.css styles the HOMEPAGE's portal
   mock-up, which is a different mock-up of the same product.

   ONLY THE PROPERTIES THE HANDOFF DOES NOT SET ARE RESET HERE, and
   that distinction was got wrong on the first attempt. Resetting
   flex-wrap, gap and font-size as well looked harmless and was not:
   the handoff sets flex-wrap:wrap and gap:10px on .portal-top at its
   phone breakpoint, and font-size:18px on .portal-brand. Writing our
   own values for those overrode the APPROVED DESIGN rather than
   app.css, and the difference showed up as a 17px wordmark and a
   non-wrapping portal header at 390px — caught by diffing computed
   values against the handoff's own rendering, not by eye.

   So: app.css sets `margin`, `padding-inline` and `color` on
   .portal-brand and the handoff sets none of the three. Those are the
   three that are reset. Everything else is left for the handoff's own
   rules to decide.

   .portal-top needs exactly ONE property, AND IT HAS TO BE RANGED.
   The handoff sets display, align-items, justify-content and gap at
   every width, but adds flex-wrap:wrap only at its PHONE breakpoint
   (max-width:760px) — so above that breakpoint flex-wrap is unset and
   app.css's `wrap` was winning. Measured at 1280: built `wrap`,
   reference `nowrap`.

   AN UNRANGED RESET FIXED 768-1450 AND BROKE 390. `.mission
   .portal-top { flex-wrap: nowrap }` written plainly here sits AFTER
   the handoff's phone rule in source order and ties it on
   specificity, so it won at every width and the portal header stopped
   wrapping on a phone — trading a desktop fault for a phone one.
   Caught by re-running the computed-value diff at all five widths
   rather than only at the one just fixed.

   So the reset is confined to the range where the handoff says
   nothing, using the handoff's own breakpoint as the boundary. Below
   761px this rule does not exist and the handoff's wrap stands. */
@media (min-width: 761px) {
  .mission .portal-top { flex-wrap: nowrap; }
}
.mission .portal-brand {
  margin: 0;
  padding-inline: 0;
  color: inherit;
}

/* h1: app.css caps it at 18ch with its own clamp and margin. The
   handoff's h1 is the hero headline and sets its own size, weight and
   max-width; the margin and the measure are what leak. */
.mission h1 { margin: 0; max-width: none; }


/* .brandrest wraps the part of the wordmark that is NOT the orange arc.

   The handoff hardcoded the string "oRoute" directly after the
   .brandmark span, because the arc IS the letter C — DESIGN-BRIEF.md:
   "the orange C is the initial letter of the CoRoute wordmark,
   followed by white `oRoute`; do not duplicate it with another white
   C."

   This project does not hardcode the brand name anywhere (BRAND_NAME,
   platform.Config.BrandName), so the handler passes the name with its
   first letter removed as .BrandRest and the template renders that.
   The span exists so the two halves can be targeted separately if the
   kerning ever needs it; it inherits everything today. */
.mission .brandrest { display: inline; }

/* The handoff footer's second sentence, kept as page content. See the
   note beside it in our-mission.html. Sized and spaced to read as a
   quiet closing qualifier rather than as a fifth section heading. */
.mission .mission-prelaunch {
  max-width: 1180px;
  margin: 0 auto;
  padding: 26px 36px 44px;
  font-size: var(--t-cap);
  line-height: 1.6;
  color: var(--muted);
  text-align: center;
}
@media (max-width: 760px) {
  .mission .mission-prelaunch { padding: 22px 22px 34px; }
}

/* ===================================================================
   THE 11 OCT 2026 BOARD CHANGES (canvas v180, version 1791751196-14a4)
   ===================================================================

   Four pieces, all scoped under .mission like everything else in this
   file. Read the header comment before adding to this block: a single
   unscoped selector here reaches the C1/D5 banner and the shared
   footer, which sit OUTSIDE <div class="mission">.

     1. .headline-stat   — the £1,813 figure under section 01's h2
     2. .mission-cards   — the three promises, replacing .principle
     3. .how-lede /      — section 05. ITS STEPPER IS GONE TOO,
        .price-note        13 Oct 2026, canvas v204: the three steps
                           became one sentence under the h2 and a
                           green price row inside the drive card.
     4. .agent-steps     — section 03's vertical stepper. GONE
                           12-13 Oct 2026, canvas v203: the three
                           steps were cut to one paragraph.
                           .step-dot/.step-join WERE kept then
                           "because section 03 still uses them";
                           it no longer does, so they are deleted
                           now. Nothing renders .step*. (That
                           section is numbered 05 from 13 Oct
                           2026 — see the renumbering note below.)
     5. .map-fallback    — the no-JS basemap layer

   TWO PALETTE VALUES THE BOARD USES AND THIS FILE DID NOT HAVE:
   --panel (the card bed) and --paper (the card). Declared on .mission
   with the rest of the palette so they cannot leak onto app.css's
   tokens, which use similar names for different values.

   ⚠️ THEY MUST BE PER-THEME, AND THEY WERE NOT. V2, 22 Oct 2026.

   This was `.mission { --panel: #DCEFFA; --paper: #FFFFFF; }` -- one
   value each, both light, overriding app.css's PER-THEME definitions
   for everything inside .mission. The scoping was right; the pinning
   was not, and it is the single largest reason this page did not work
   in dark mode. Measured on the rendered page before this change:
   every card on the page drew light ink on a white bed --

     h3 "Lower the cost of getting started"    1.19:1
     p  "Price new drivers on how they..."     1.75:1
     h2 "Our mission"                          1.01:1

   -- which is invisible text, not merely low, on the page's
   centrepiece. A scoped token that does not follow the theme is a
   pin, and a pin needs the same justification as any other: here
   there was none, only a value copied once from a board that had no
   dark theme to copy.

   The values match app.css's --panel and --paper exactly in both
   themes, which is the point: these two names mean the same thing
   inside .mission as outside it, and only the leak had to be
   prevented. */
.mission {
  --panel: #DCEFFA;
  --paper: #FFFFFF;
}
@media (prefers-color-scheme: dark) {
  .mission:where(:root:not([data-theme="light"]) *) {
    --panel: #1E3552;
    --paper: #182B43;
  }
}
.mission:where(:root[data-theme="dark"] *) {
  --panel: #1E3552;
  --paper: #182B43;
}

/* ---- 1. THE SOURCED STAT ------------------------------------------

   £1,813 is the only figure this page carries and the ONLY number
   being added by this round. Its attribution is laid out as part of
   the block rather than as a footnote, because an average premium
   with no source is the page's one unsubstantiated claim — and
   because CLAUDE.md's no-new-claim rule is what makes the source the
   price of the number being here at all.

   align-items:baseline, not center: the 64px figure and the 17px
   caption share a baseline on the board, which is what makes the
   caption read as a label on the number rather than as a line of body
   copy beside it. flex-wrap lets the caption drop below the figure on
   a narrow column instead of squeezing the digits.

   font-variant-numeric:tabular-nums is from the board. It matters
   less for one static figure than it would for a column of them, and
   it is kept so that a future second figure lines up with this one.

   NO <figure> MARGIN. Browsers default figure to 1em side margins,
   which would indent the whole block relative to the h2 above it.
   app.css has no figure rule, so this is the browser default being
   taken back, not a collision. */
/* ---- MOVED UNDER THE RIGHT COLUMN, 11 Oct 2026 -----------------

   Guillaume: "I would put that under the text on the right". It was
   under the heading in the LEFT column, where it made that column
   499px tall against the right column's 221px and left the body copy
   stranded at the top of a half-empty box.

   It now closes the right column instead, so the section reads
   heading -> argument -> the number that evidences the argument.
   That is also the better reading order: the figure now lands after
   "insurers price you on people like you", which is the sentence it
   actually supports.

   THE SEPARATOR IS A HAIRLINE, NOT A GAP. Sitting directly under a
   paragraph, the figure needs something to stop it reading as a
   continuation of the sentence above; 32px of air alone was not
   enough at 64px type. */
.mission .headline-stat {
  margin: 34px 0 0;
  padding-top: 30px;
  border-top: 1px solid var(--line, #e0e8ee);
  display: flex;
  align-items: center;
  gap: 20px;
  flex-wrap: wrap;
}
.mission .headline-stat strong {
  font-family: Manrope, sans-serif;
  font-size: 56px;
  line-height: 1;
  font-weight: 900;
  letter-spacing: -.03em;
  color: var(--strong-ink);
  font-variant-numeric: tabular-nums;
}
/* THE CAPTION SITS ON THE NUMBER'S BASELINE, NOT ITS OPTICAL CENTRE.
   align-items:baseline on the row lines the caption's FIRST line up
   with the bottom of a 56px numeral, so a three-line caption hangs
   well below it and reads as unaligned. Pulling it down by its own
   first-line offset puts the block's centre against the figure's.

   22ch forced three lines; 30ch gives two at this size. */
.mission .headline-stat > span {
  font-size: var(--t-small);
  line-height: 1.45;
  font-weight: 600;
  max-width: 30ch;
  color: var(--ink);
  align-self: center;
  position: relative;
  top: 4px;
}
/* THE SOURCE LINE sits on its own row under the figure and caption,
   at 13px — present and legible, not shouted. flex-basis:100% is what
   drops it to a new row regardless of how the two wrapped above it.

   max-width is a READING MEASURE, not a cap on the figure: the credit
   is now a full sentence (publisher, quarter, the WTW compilation and
   the five-cheapest-quotes basis) and a 13px line running the whole
   half-column width would be hard to read. It must stay visible and
   adjacent to the number — never a tooltip, never behind a toggle. */
.mission .headline-stat figcaption {
  flex: 0 0 100%;
  margin-top: 10px;
  max-width: 54ch;
  font-size: var(--t-cap);
  line-height: 1.55;
  color: var(--muted);
}
/* cite defaults to italic in every browser; the board sets it upright. */
.mission .headline-stat cite { font-style: normal; }
/* THE CREDIT IS A LINK BECAUSE THE PUBLISHER ASKS FOR ONE (the Q3
   report: "Please use this as credit for the data: …/price-index").
   `.mission a{color:inherit}` at the top of this file would render it
   as plain text indistinguishable from the sentence around it, so the
   underline is restored explicitly — an unmarked link is not a credit
   anyone can follow. Colour stays inherited so the line keeps reading
   as a quiet attribution rather than as a call to action. */
.mission .headline-stat figcaption a {
  color: inherit;
  text-decoration: underline;
  text-underline-offset: 2px;
}

/* The second paragraph of section 01 is the one that states the aim,
   so it takes the full ink colour while the first stays at --muted.
   The board draws exactly this contrast. */
.mission .body-copy .lead-out { color: var(--ink); }

/* ---- SECTION 01's RIGHT-HAND COLUMN, BIGGER AND MORE SALIENT ------

   Built 12 Oct 2026 from Mission.dc.html, canvas v1791836956-6698.
   The board's own comment states the brief: "01 BODY, 12 Oct
   (Guillaume: 'bigger and more salient'). Same words. The first line
   is larger; the belief is pulled out as the section's punchline
   (big, orange rule) with the aim as its follow-up line."

   SAME WORDS. Not one word of this copy changed — only its size,
   weight and arrangement. The old markup was two paragraphs, the
   second a .lead-out carrying the belief and the aim in one run of
   text with the belief bolded. The board splits them: the belief
   becomes a 31px punchline behind an orange rule, and the aim a
   quieter 19px line under it inside the same bordered block.

   THE BOARD'S VALUES, READ OFF ITS DECLARATIONS RATHER THAN BY EYE:

     para 1      23px / 1.5   colour #291449       "your age, your
                                                    postcode, your
                                                    car." at 700
     punchline   31px / 1.2   weight 800, #4B1FAE
                              6px solid #FFB264 left border,
                              24px left padding
     aim         19px / 1.55  --ink-soft

   COLOUR NOTE. The board writes the body ink as #291449 and this
   stylesheet's --ink is #291B56 — a 13-unit difference in green and
   blue, invisible side by side and consistent across every other
   section of this page. The token is used rather than the literal,
   so section 01 does not become the one place with a hand-typed ink
   that drifts from the rest when --ink is next adjusted. The
   punchline's #4B1FAE IS --purple exactly, and the rule's #FFB264 IS
   --orange exactly, so both are tokens.

   THE 34em CAP ON #why .body-copy STILL APPLIES and is left alone;
   at 23px that is ~780px, and the board's column is 792px. */
.mission #why .why-lede {
  font-size: 23px;
  line-height: 1.5;
  color: var(--ink);
}
.mission #why .why-lede strong { font-weight: 700; color: var(--ink); }

/* The punchline block. The border is the board's 6px orange rule and
   the 24px padding is its own; the 4px block padding keeps the rule
   from ending flush with the cap-height of the text, which is how
   the board draws it. */
.mission #why .why-punchline {
  margin-top: 32px;
  border-left: 6px solid var(--orange);
  padding: 4px 0 4px 24px;
}
.mission #why .why-belief {
  margin: 0 0 12px;
  font-size: 31px;
  line-height: 1.2;
  font-weight: 800;
  letter-spacing: -0.02em;
  color: var(--strong-ink);
}
.mission #why .why-aim {
  margin: 0;
  font-size: 19px;
  line-height: 1.55;
  color: var(--muted);
}

/* PHONE, the board's own smaller set: para 19, punchline 25, aim 17.
   48rem is an existing breakpoint in this stylesheet, not a new one. */
@media (max-width: 48rem) {
  .mission #why .why-lede   { font-size: 19px; }
  .mission #why .why-belief { font-size: 25px; }
  .mission #why .why-aim    { font-size: 17px; }
  /* The rule and its inset come in proportionally, so the block does
     not eat a fifth of a 390px column. */
  .mission #why .why-punchline { padding-left: 18px; border-left-width: 5px; }
}
/* "how *you* drive" — the board emphasises the pronoun, and an <em>
   inside body copy that is already --muted needs the weight to read. */
.mission .body-copy em { font-style: italic; }

/* ---- 2. THE MISSION CARDS -----------------------------------------

   REPLACED .principle, which was one sentence between two hairlines.
   The three promises are now scannable, which is the whole point of
   the redesign: a visitor who reads nothing else should still come
   away with what the company is for.

   The .principle rules above are left in place deliberately — see the
   note with them. Nothing renders them today.

   38px/900 ON THE HEADING with NO full stop, both from the board and
   both load-bearing: "Our mission" is a label, and a full stop makes
   a label read as a truncated sentence. The orange dot before it is
   the same device as .eyebrow's dot, at 12px rather than 8px.

   The h2 is flex so the dot sits on the text's optical centre rather
   than on its baseline. flex:none on the dot stops it being squeezed
   to nothing when the heading wraps. */
.mission .mission-cards {
  margin-top: 44px;
  background: var(--panel);
  border-radius: 24px;
  padding: 32px;
}
.mission .mission-cards-title {
  margin: 0 0 8px;
  display: flex;
  align-items: center;
  gap: 12px;
  font-family: Manrope, sans-serif;
  font-size: 38px;
  line-height: 1.1;
  font-weight: 900;
  letter-spacing: -.02em;
  color: var(--ink);
}
.mission .mission-cards-title .dot {
  flex: none;
  width: 12px;
  height: 12px;
  margin-right: 0;
}
/* 22px/700, and the word "Transparent" in purple — it is the key word
   of the whole page, which is why it also leads the first card. */
.mission .mission-cards-lede {
  margin: 0 0 24px;
  max-width: 52ch;
  font-size: 22px;
  line-height: 1.35;
  font-weight: 700;
  color: var(--ink);
}
.mission .mission-cards-lede strong {
  color: var(--strong-ink);
  font-weight: inherit;
}
/* A <ul>, because three promises are a list and a screen reader
   should say so. list-style and padding are the browser defaults
   being taken back. */
.mission .mission-card-grid {
  list-style: none;
  margin: 0;
  padding: 0;
  display: grid;
  grid-template-columns: repeat(3, minmax(0, 1fr));
  gap: 18px;
}
.mission .mission-card {
  background: var(--paper);
  border-radius: 16px;
  padding: 24px 22px;
  box-shadow: 0 6px 18px rgba(20, 43, 73, .06);
}
/* The icon tile. currentColor on the stroke is what lets the SVG take
   its colour from this rule rather than carrying a fill attribute,
   which keeps the markup free of presentation. */
/* THE TILE AND ITS GLYPH ARE --brand / --on-brand, A PAIR.

   They were var(--purple) and var(--paper), which resolve to the
   SAME VALUE in dark mode (#182B43) -- a glyph invisible against its
   own tile. Found by eye in the board comparison, not by the
   contrast walker: an SVG stroke is not a text node, so nothing that
   measures text could see it. That is the second thing in this job
   the walker could not catch and looking could.

   --brand/--on-brand are a declared pair and are correct together in
   both themes: 13.64:1 light (cream on navy), 9.72:1 dark (navy on
   light blue). */
.mission .mission-card-icon {
  width: 52px;
  height: 52px;
  border-radius: 14px;
  background: var(--brand);
  color: var(--on-brand);
  display: grid;
  place-items: center;
  margin-bottom: 16px;
}
.mission .mission-card-icon svg { width: 28px; height: 28px; display: block; }
.mission .mission-card h3 {
  margin: 0 0 8px;
  font-family: Manrope, sans-serif;
  font-size: 21px;
  line-height: 1.25;
  font-weight: 800;
  letter-spacing: -.01em;
  color: var(--ink);
}
.mission .mission-card p {
  margin: 0;
  font-size: 15px;
  line-height: 1.55;
  color: var(--muted);
}

/* ---- 3 & 4. THE STEPPERS ARE BOTH GONE ----------------------------

   THE COVERAGE SECTION'S VERTICAL STEPPER WENT 12-13 Oct 2026
   (canvas v203) and THE FEEDBACK SECTION'S HORIZONTAL ONE WENT
   13 Oct 2026 (canvas v204, live version 1791881544-d4b1).
   Those two were numbered 04 and 03 at the time and are 05 and 04
   now — the retention section took 02 on 13 Oct 2026, and this
   file names sections by their CURRENT number everywhere a rule
   depends on it — Guillaume: "remove those 1, 2, 3
   stages... everything should be part of the board above and
   condensed".

   SO .steps-across, .step, .step-dot AND .step-join ARE ALL DELETED
   from this file, along with the handoff-era `.mission .steps` /
   `.mission .step` grid and its phone overrides. NOTHING renders any
   of that markup now: the feedback section's three ideas moved INTO
   the drive review card (see .how-lede and .price-note below) and
   the coverage section's into one sentence beside its chat card.

   THE PREVIOUS ROUND KEPT .step-dot AND .step-join ALIVE ON PURPOSE,
   with a comment here saying so "because section 03 still uses
   them" — the feedback section, numbered 05 since 13 Oct 2026.
   That is now false, and a block of rules for markup nothing
   renders is the kind of thing a later reader restores markup to
   match — which is exactly what the 12-13 Oct note warned about.
   Deleted rather than left dormant, same reasoning as the coverage
   section's.

   BOARD LEADS, PAGE FOLLOWS: if a stepper should return anywhere on
   this page, it goes on Mission.dc.html first and these rules come
   back with it. */

/* ---- 3a. SECTION 05'S LEDE ---------------------------------------

   The board (canvas v204) puts ONE sentence under the h2, carrying
   what used to be step 1 ("See the full picture."):

     h2            margin-bottom 12px    <- the h2's OWN margin
     .how-lede     19px / 1.5, muted, max-width 72ch
                   margin-bottom 28px    <- the gap before the card

   The 12px and the 28px are two different elements' margins, not one
   gap each side of a rule, which is why the h2 is given a
   margin-bottom here at all — `.mission h2` sets none, and the old
   layout relied on .how-head's block spacing.

   SCOPED TO #progress's .how-head, not to .mission h2, because every
   other h2 on this page still wants no bottom margin.

   .how-head CARRIES max-width:680px and the lede must NOT inherit
   it: the board gives the lede 72ch, which is wider. 72ch is the
   board's own unit and is kept rather than converted to px, so it
   tracks the font the way the board does. The h2 keeps 680px.

   COLOUR IS --muted (#665d76), this file's token for the board's
   --ink-soft. The board's palette name does not exist here; app.css
   has an --ink-soft with a different value, and mission.css must not
   read app.css's tokens — see the header comment. #665d76 on the
   --cream ground measures 5.6:1. */
.mission #progress .how-head { max-width: none; }
.mission #progress .how-head h2 { max-width: 680px; margin-bottom: 12px; }
.mission .how-lede {
  margin: 0 0 28px;
  font-size: 19px;
  line-height: 1.5;
  color: var(--muted);
  max-width: 72ch;
}
/* THE CARD'S OWN margin-top:40px WOULD WIN THIS GAP, not add to it.
   Adjacent vertical margins collapse to the LARGER of the two, so the
   lede's 28px below and .driver-portal's 40px above gave 40px, and
   measuring the lede's own margin-bottom read a correct 28px while
   the page drew 40. The board draws 28px between the two.

   Zeroed on the CARD rather than raising the lede's margin, so the
   28px stays where the board puts it — on the lede — and the figure
   in this file matches the figure on the board. Scoped to #progress:
   .driver-portal is section 05's only card, but the 40px rule is
   written unscoped and section 03's cards must keep it. */
.mission #progress .how-lede + .driver-portal,
.mission #progress .how-head + .driver-portal { margin-top: 0; }

/* ---- 3b. THE PRICE ROW INSIDE THE DRIVE CARD ----------------------

   Carries what used to be step 3, "Better driving, fairer price.",
   and sits BETWEEN the detail block (.moment-feedback) and the card
   footer (.portal-bottom). Board geometry, to the value:

     panel      #EAF6EE, radius 12px, padding 14px 18px
     margin     0 24px 24px   (24px side margins, 24px clear of the
                               footer rule)
     coin       34px circle, #0F6B3F, white "£", 16px/800
     gap        14px between coin and text
     text       14px / 1.5

   THE CAVEAT STAYS IN THIS ROW. The spec is explicit about it, and
   the board's own note says "The price caveat ... stays next to the
   price line. Do not drop it." So the caveat is an inline <span>,
   NOT a second block: it wraps with the sentence it qualifies and
   cannot be read as a separate footnote. See the template comment
   for the one place the page says MORE than the board — the
   regulatory hedge — and why it may not be "fixed" back.

   ====================================================================
   PINNED LIGHT, INK AND FILL TOGETHER. THIS IS THE POINT.
   ====================================================================

   #EAF6EE and #0F6B3F are literals from a board that defines no dark
   theme (grep prefers-color-scheme over Mission.dc.html returns 0),
   and this whole file is a light design for the same reason. But
   /our-mission loads app.css FIRST, and app.css's dark block
   redefines --ink, --ink-soft and friends at :root.

   A pinned-light PANEL whose TEXT follows a dark theme is the fault
   that put "How it works" at 1.10:1 on the homepage explainer (see
   .explainer in app.css). So every colour in this row is a literal:
   the fill, the coin, the coin's white, the body ink and the
   caveat's ink. None of them can drift when a theme changes.

   The two inks are written out rather than var()'d, because the
   tokens they shadow are declared on .mission and would be correct
   today — but a literal is what makes this row immune to the next
   palette edit, and immunity is the property being bought.

   V2 RECOLOURED BOTH LITERALS, and that is not a contradiction of
   the paragraph above. "Immune to the next palette edit" means the
   row cannot be changed BY ACCIDENT, through a token moving under
   it; it was never a claim that the hexes are permanent. They were
   #291B56 and #665d76, an ink purple and a purple-cast grey. They
   are now the navy and the blue-cast grey at the same lightness --
   #665d76 is hue 262 and #55697D is hue 210 at an identical 0.41
   luminance, so the row looks the same and is no longer the one
   purple left in the stylesheet.

   Measured on #EAF6EE: #142B49 is 12.85:1, #55697D is 5.11:1,
   #fff on #0F6B3F is 6.0:1 -- all pass 4.5:1, in BOTH themes,
   because none of them move. */
/* --ok-bg, not #EAF6EE. MEASURED 22 Oct: the pale green was pinned in
   both themes while the text inside it followed the theme, so in dark
   mode this read #E5ECF3 on #EAF6EE -- 1.07:1, which is invisible, not
   merely low. Same fault as the premium chart: a surface that does not
   theme under ink that does. */
.mission .price-note {
  display: flex;
  align-items: center;
  gap: 14px;
  margin: 0 24px 24px;
  padding: 14px 18px;
  background: var(--ok-bg);
  border-radius: 12px;
}
.mission .price-note-coin {
  flex: none;
  display: grid;
  place-items: center;
  width: 34px;
  height: 34px;
  border-radius: 50%;
  background: var(--good-solid);
  color: #fff;
  font-family: Manrope, sans-serif;
  font-size: 16px;
  font-weight: 800;
  line-height: 1;
}
.mission .price-note p {
  margin: 0;
  font-size: 14px;
  line-height: 1.5;
  color: var(--ok-ink);
}
.mission .price-note p b { font-weight: 700; }
/* A LITERAL, AND IT WAS BRIEFLY NOT. This row pins its fill AND
   every ink on it -- see the long note above, which explains why
   immunity is the property being bought. A pass that converted the
   page's grey inks to var(--ui-soft) caught this one too, which put
   a themed ink back onto a pinned fill: 1.75:1 in dark mode, the
   exact fault the note warns about. Restored. */
.mission .price-note p > span { color: var(--ok-ink); opacity: .8; }

/* ---- 5. THE NO-JS BASEMAP -----------------------------------------

   Only ever rendered inside <noscript>, so a visitor with script
   never has this element in the DOM at all. It is absolutely
   positioned over the same box as the <svg> and sits BEHIND it
   (z-index 0 against the svg's own stacking) so the route, markers
   and vehicle stay visible on top.

   .aerial-map already carries isolation:isolate, which gives this a
   stacking context to be contained by. */
.mission .trip-map .map-fallback {
  position: absolute;
  inset: 0;
  width: 100%;
  height: 100%;
  object-fit: cover;
  z-index: 0;
}
.mission .trip-map svg { position: relative; z-index: 1; }

/* ---- PHONE -------------------------------------------------------

   Both steppers go to one column, and the horizontal joins must GO
   rather than shrink: a 2px rule reaching off the right edge of a
   single-column item points at a column that is not there. The
   vertical stepper keeps its join, because stacking is already its
   orientation.

   The stat drops to 48px so "£1,813" cannot collide with the
   24px-gutter at 320px. Measured, not guessed — see the report. */
@media (max-width: 760px) {
  .mission .headline-stat { margin-top: 32px; gap: 14px; }
  .mission .headline-stat strong { font-size: 48px; }
  .mission .headline-stat > span { font-size: 16px; max-width: none; }

  .mission .mission-cards { margin-top: 32px; padding: 22px; border-radius: 18px; }
  .mission .mission-cards-title { font-size: 30px; gap: 10px; }
  .mission .mission-cards-lede { font-size: 19px; margin-bottom: 18px; }
  .mission .mission-card-grid { grid-template-columns: 1fr; gap: 14px; }
  .mission .mission-card { padding: 20px 18px; }
  .mission .mission-card-icon { width: 44px; height: 44px; border-radius: 12px; margin-bottom: 12px; }
  .mission .mission-card-icon svg { width: 24px; height: 24px; }
  .mission .mission-card h3 { font-size: 19px; }

  /* SECTION 05's lede and price row at phone width. The stepper's
     four overrides that used to sit here are gone with it.

     THE PRICE ROW STAYS A FLEX ROW. The coin is flex:none at 34px,
     so the text column is 390 - 2*24 side margins - 2*18 padding -
     34 coin - 14 gap = 242px at 390, and the sentence wraps inside
     it. It is NOT stacked: the spec requires the caveat in the same
     row as the price line, and wrap-safe is what the measurement
     showed. Side margins come in to 16px so the text column is not
     squeezed below a comfortable measure. */
  .mission .how-lede { font-size: 17px; margin-bottom: 22px; }
  .mission .price-note { margin: 0 16px 20px; padding: 13px 14px; gap: 12px; }

}

/* ===================================================================
   SECTION ORDER CHANGED 13 Oct 2026 — board Mission.dc.html, live
   canvas version 1791886398-62fc (v213). BOARD LEADS, PAGE FOLLOWS.

   The AI-guide band (.coverage, beige #f0e7db) moved from LAST
   numbered section to directly after 02. The order is now:

     01 Why it costs so much      #why       (no band)
     02 After your first year     #renewal   (no band, .retention)
     03 Clarity before a claim    #cover     BEIGE  .coverage
     04 Context matters           #approach  NAVY   .context
     05 Feedback with a purpose   #progress  (no band)

   NO RULE IN THIS FILE CHANGED FOR THE MOVE, and that is the point
   worth recording. The bands are not positioned by source order or
   by margins — each is a full-width background on the section itself,
   so moving the <section> in the template moved the band with it.
   Only the comments naming sections by number changed.

   THE BRIEF ASKED FOR TWO MARGINS TO BE REMOVED. NEITHER EXISTED.
   It asked to drop .coverage's top margin and a 56px top margin on
   .context. MEASURED on the running page at 1280 before the move:
   every band already computed margin-top: 0px / margin-bottom: 0px,
   and every seam between consecutive bands already measured 0px.
   Band spacing comes entirely from each section's own padding
   (.section is 105px 0; .retention overrides to 56px 0 64px) — which
   is quote-layout R7, spacing from the container and not from a
   component margin. The "56px" in the brief is .retention's
   PADDING-top, documented immediately below: padding, not margin, and
   on a different section. The board agrees there is no margin — it
   draws both bands as a colour div wrapping `margin: 0 auto;
   padding: 56px 40px`. So nothing was removed. Changing .retention's
   padding to satisfy the brief's number would have re-spaced section
   02, which nobody asked for.

   THE BEIGE/NAVY SEAM IS NEW AND IT IS DELIBERATE. Two full-width
   bands now meet directly, beige #f0e7db above navy --navy, with 0px
   between them and 105px of padding inside each. Checked in the
   board-vs-page image: the board draws the same hard edge.
   =================================================================== */

/* ===================================================================
   02 / AFTER YOUR FIRST YEAR — the retention section
   ADDED 13 Oct 2026, from Mission.dc.html, live canvas version
   1791884504-d785. Board geometry, to the value:

     outer       max-width 1200px, padding 56px 40px 64px
     grid        minmax(0,1fr) 680px, gap 48px, align-items center
     card        --paper, radius 20px, padding 28px
     card lift   0 1px 0 rgba(41,20,73,.04), 0 14px 32px rgba(75,31,174,.10)
     "Our aim"   24px / 1.1 / 900, --brand, 10px orange dot, gap 10px
     aim flow    1fr auto 1fr auto 1fr, gap 8px, align-items stretch
     boxes       radius 14px, padding 18px 16px, text-wrap balance
     stage       15px / 800 / uppercase / .06em / opacity .85
     phrase      19px / 1.25 / 800
     arrows      30x30, stroke-width 3.5, round caps and joins
     caveat      12px / 1.5, 14px above a 1px --line rule, 20px clear

   THE OUTER PADDING IS NOT WRITTEN HERE. The page's .section carries
   105px 0 and .wrap carries the 1200px cap and the 40px gutters, so
   the board's 56px/40px/64px is EXPRESSED THROUGH THE EXISTING SHELL
   rather than duplicated: .retention overrides only the vertical
   figures. Writing a second max-width:1200px here would be a silent
   fork of the page's one container rule, which is how two sections
   end up a pixel apart at 1240px for no traceable reason.

   =====================================================================
   EVERY COLOUR IN THE CARD IS A LITERAL. THIS IS DELIBERATE.
   =====================================================================

   Same reasoning as .price-note above, and the same fault it was
   written against. Mission.dc.html defines NO dark theme (grep
   prefers-color-scheme over it returns 0) and this whole file is a
   light design. But /our-mission loads app.css FIRST, and app.css's
   dark block redefines --paper to #291449, --panel to #382455 and
   --ink-soft to #C9B8E8 at :root.

   mission.css re-declares --panel and --paper on .mission, so inside
   the page they would hold their light values TODAY by specificity.
   That is a coincidence of two files agreeing, not a guarantee: the
   .mission declaration is one edit away from being removed as
   "redundant", and a pinned-light PANEL whose TEXT follows a dark
   theme is exactly what put "How it works" at 1.10:1 on the homepage
   explainer. So the card's fill, both box beds, all four text inks,
   the green and its white, and the arrow stroke are written out.
   None of them can drift when a theme changes.

   The literals ARE this file's own palette, written out rather than
   var()'d:
     #FFFFFF  --paper      the card
     #F3ECFD  --panel      boxes 1-2
     #4B1FAE  --purple     their text, and "Our aim" (the board's
                           --brand #6624C9 is a different purple this
                           file does not have; --purple is the page's
                           own and measures BETTER — see below)
     #0F6B3F  (literal)    box 3, the same green as .price-note's coin
     #291B56  --ink        the bold aim line
     #665d76  --muted      the lede and the caveat
     #1B1033  (literal)    the arrows, the board's own stroke

   MEASURED, in BOTH SCHEMES, because none of these move (4.5:1 is
   the bar for text, 3:1 for the arrows, which are non-text):

     stage label, boxes 1-2   #4B1FAE at .85 on #F3ECFD   6.20:1
     phrase, boxes 1-2        #4B1FAE on #F3ECFD          8.65:1
     stage label, box 3       #FFFFFF at .85 on #0F6B3F   5.25:1
     phrase, box 3            #FFFFFF on #0F6B3F          6.57:1
     "Our aim"                #4B1FAE on #FFFFFF          9.97:1
     caveat                   #665d76 on #FFFFFF          6.19:1
     lede (on the page)       #665d76 on #FFF9F1          5.92:1
     aim line (on the page)   #291B56 on #FFF9F1         14.49:1
     arrows                   #1B1033 on #FFFFFF         17.97:1

   THE 85% OPACITY IS APPLIED AS A COLOUR, NOT AS opacity. The board
   writes opacity:.85 on the stage label, which would also fade the
   box's own text-wrap and any future focus ring drawn inside it, and
   which a compositing contrast check has to blend by hand. The
   blended values are written out instead — #643EBA on #F3ECFD and
   #DBE9E2 on #0F6B3F — so the rendered colour is the colour in the
   file and the measurement above is読 directly off it.
   =================================================================== */
.mission .retention { padding: 56px 0 64px; }
.mission .retention-grid {
  display: grid;
  grid-template-columns: minmax(0, 1fr) 680px;
  gap: 48px;
  align-items: center;
}
/* ---- THE AWKWARD BAND BETWEEN PHONE AND 1280 ---------------------

   THE BOARD'S 680px IS A FIXED TRACK AND IT DOES NOT FIT BELOW
   ~1120px. Found by measuring every width from 390 to 1440, not by
   looking at 390 and 1280 and assuming the middle interpolates:

     width   text column   card    overflow   h2 lines
      760    697 (1-col)   697        0          1     <- phone rules
      761      0 px        680       +3 px       5     <- BROKEN
      900     85 px        680        0          5     <- unreadable
     1000    185 px        680        0          4
     1100    285 px        680        0          3
     1280    380 px        680        0          2

   At 761 the phone rules stop, the 680px track takes the space, and
   minmax(0,1fr) correctly gives the text column ZERO — so the
   heading rendered in an 85px-wide gutter as a five-line ladder, and
   the card itself hung 3px past the viewport. 761 is one pixel above
   a breakpoint, which is exactly the width nobody opens.

   SO THE TRACK IS NOT FIXED UNTIL IT FITS. Up to 1199px the second
   track becomes minmax(0, 1fr) too, giving two even columns, and the
   gap steps down to 32px. The fixed 680px track returns at 1200px,
   where the text column first gets a usable measure: .wrap is
   min(1180, 100vw) less 36px of gutter each side, so at 1200 the
   content box is 1108px and the text column is 1108 - 680 - 48 =
   380px, which holds the 42px heading on two lines. At 1121 the same
   sum gave the text 306px and the heading went to THREE lines, so
   the boundary is set where the heading is right rather than where
   the card merely fits.

   THE BOUNDS ARE PIXELS, NOT rem, AND THAT IS DELIBERATE. The first
   attempt used min-width:47.6rem, which is 761.6px — so 761px fell
   in the one-pixel gap between the phone block's max-width:760px and
   this one, and the broken layout above survived at exactly the
   width the measurement had found it at. 760.01px closes it against
   the phone block's own figure, in the same unit.

   WHY NOT STACK ALL THE WAY TO 1120. Two even columns at 900px give
   the text 400px and the card 400px, which reads as a pair; one
   column there would leave a 828px-wide card with a 19px lede
   stretched across it, which is worse. The stack is for the phone,
   where the card genuinely has no room beside anything. */
@media (min-width: 760.01px) and (max-width: 1199px) {
  .mission .retention-grid {
    grid-template-columns: minmax(0, 1fr) minmax(0, 1fr);
    gap: 32px;
  }
  /* The desktop <br>s put "pay off at renewal." on its own line,
     which at a 400px column leaves the first line short and the
     second one wrapping anyway. Let both wrap naturally here — with
     the ::after space, for the reason given in the phone block at
     the foot of this file: a hidden <br> takes the only whitespace
     between the two words with it. */
  .mission .retention-copy h2 br,
  .mission .retention-aim br { display: inline; }
  .mission .retention-copy h2 br::after,
  .mission .retention-aim br::after { content: " "; }
  .mission .retention-copy h2 { font-size: 36px; }
}
.mission .retention-copy h2 {
  margin: 16px 0 24px;
  font-size: 42px;
  line-height: 1.12;
  font-weight: 800;
  letter-spacing: -.03em;
}
/* ---- A STRAIGHT RULE, NOT THE PAGE'S SWOOSH ----------------------

   THIS SECTION'S HEADING DOES NOT USE .underlined, AND THE BOARD IS
   WHY. .underlined is this page's hand-drawn device: a 5px bar with
   transform:rotate(-2deg) and border-radius:50%, so it reads as a
   marker stroke swept under the words. Sections 01 and 05 use it and
   should keep it.

   The board draws THIS heading with
     border-bottom: 5px solid var(--mission-orange); padding-bottom: 2px
   — a level rule, tight under the text. Same colour, same 5px, and a
   deliberately different gesture.

   FOUND BY LOOKING, NOT BY MEASURING. The first build reused
   .underlined because every other underlined heading on the page
   has it, and nothing caught it: the string was right, the colour
   was right, the thickness was right, nothing overflowed and the
   contrast was unchanged. It took the board and the page side by
   side to see that one rule was slanted and the other level. That is
   the whole argument for the image check.

   A SEPARATE CLASS RATHER THAN AN OVERRIDE, because unsetting
   .underlined's pseudo-element would leave a rule that exists only
   as the absence of another rule — and the next person adding a
   swoosh variant would have to discover this by reading it. */
.mission .rule-under {
  border-bottom: 5px solid var(--orange);
  padding-bottom: 2px;
}
/* THE RULE MUST NOT BE DRAWN TWICE.

   An inline element's border-bottom is drawn on EVERY line box it
   occupies. At 1280 "pay off at renewal." is one line and the rule
   is one bar, as the board draws it. At 393 the heading rewraps to
   "A good year should pay / off at renewal." — the span straddles
   the break, and the page drew TWO orange bars: a short one under
   "pay" at the end of line 1 and a second under "off at renewal."
   on line 2. Seen in the phone render; no measurement would have
   reported it, because both fragments have the right colour, the
   right 5px and the right offset.

   white-space:nowrap KEEPS THE SPAN WHOLE, so the rule is always
   one bar under one run of words, and the line breaks in front of
   it instead. .underlined does exactly this for the same reason
   (mission.css line 551) — the swoosh has the same problem — and
   this is that rule's one borrowed property.

   AND IT IS RELEASED AT THE NARROWEST WIDTHS, below, the way
   .underlined is: "pay off at renewal." is 19 characters and fits
   345px at 32px type, but a 320px screen with 200% text would be
   forced to overflow rather than wrap, and an unbreakable run that
   cannot fit is worse than a doubled rule. */
.mission .rule-under { white-space: nowrap; }
.mission .retention-lede {
  margin: 0 0 14px;
  font-size: 20px;
  line-height: 1.5;
  color:var(--ui-soft);
  text-wrap: pretty;
}
.mission .retention-aim {
  margin: 0;
  font-size: 22px;
  line-height: 1.4;
  font-weight: 800;
  /* --strong-ink, not a literal: this sentence sits on the PAGE, so
     it follows the theme. #142B49 on the dark page was 1.19:1 --
     invisible, on the one line in this section that states the aim. */
  color: var(--strong-ink);
  text-wrap: pretty;
}
/* THE HEDGED LINE UNDER THE AIM. Added 13 Oct 2026 with the green
   box's "Monitoring could become optional"; see the template comment
   for why both modals are load-bearing.

   COLOUR IS A LITERAL, like every other colour in this section, and
   #665d76 is this file's documented translation of the board's
   --ink-soft (see the .price-note and .aim-card headers above). The
   board's own token resolves to #56407A, which is NOT what the page
   uses anywhere in this section: .retention-lede and .aim-caveat
   both already render --ink-soft as #665d76, and matching them is
   what makes the two paragraphs in this column read as one voice.
   Measured 5.92:1 on the section's #FFF9F1 — same as the lede.

   NO text-wrap HERE, deliberately. The lede and the aim line both
   carry text-wrap:pretty/balance because they are short and their
   ragged edge is visible; this line is 78 characters and wraps to
   two full lines at 472px, where balancing moves the break without
   improving it. The board sets none either. */
.mission .retention-proved {
  margin: 12px 0 0;
  font-size: 18px;
  line-height: 1.5;
  color:var(--ui-soft);
}
/* THE CARD. 680px comes from the grid column, not from a width here,
   so the card and its column cannot disagree. */
.mission .aim-card {
  /* --paper, not a literal: this card holds the aim diagram and its
     caveat line, and a pinned white one put --muted on white at
     1.75:1 in dark mode. --paper is the page's card surface and is
     already per-theme (#FFFFFF / #182B43); --muted on it measures
     5.42:1 light and 8.17:1 dark. */
  background: var(--paper);
  border-radius: 20px;
  padding: 28px;
  box-shadow: 0 1px 0 rgba(20, 43, 73, .04), 0 14px 32px rgba(20, 43, 73, .10);
}
/* flex, not block, so the 10px dot sits on the text's optical centre
   rather than its baseline — the same device .mission-cards-title
   uses, at 10px rather than 12px per the board. */
.mission .aim-title {
  margin: 0 0 18px;
  display: flex;
  align-items: center;
  gap: 10px;
  font-family: Manrope, sans-serif;
  font-size: 24px;
  line-height: 1.1;
  font-weight: 900;
  letter-spacing: -.02em;
  /* --ink: this heading sits on .aim-card, whose fill now follows
     the theme, and the literal was 1:1 on the dark card. */
  color: var(--ink);
}
.mission .aim-title .dot {
  flex: none;
  width: 10px;
  height: 10px;
  margin-right: 0;
}
/* AN <ol>, AND THE ARROWS ARE INSIDE ITS ITEMS — BUT THE GRID STILL
   HAS THE BOARD'S FIVE TRACKS. See the template comment: an <ol> may
   only contain <li>, so each arrow sits at the end of the step it
   follows and the list has three children where the board's div has
   five.

   THE THREE BOXES MUST COME OUT EQUAL, AND MEASURING THE BOARD IS
   WHAT SETTLED IT. The board's 1fr auto 1fr auto 1fr gives three
   boxes of 177.3px each with two 30px arrow tracks between them:

     board, 1280, measured off its own render
       box 1  177.3px    arrow 1  30px at x=773.3
       box 2  177.3px    arrow 2  30px at x=996.7
       box 3  177.3px

   A FIRST ATTEMPT USED repeat(3, 1fr) ON THE ITEMS and let each
   arrow eat 38px out of its own item. That measured 164.7 / 164.7 /
   202.7 — the third box 38px wider than the other two, because it
   has no arrow to pay for. Equal-width beds are the whole visual
   idea of a three-stage flow, so that was a SHAPE difference, not a
   rounding one, and no text check would have seen it: every string
   was present and nothing overflowed.

   THE FIX IS display:contents ON THE ITEMS. The <li> stops
   generating a box, so its box and its arrow become direct grid
   items of .aim-flow, and the board's own five-track template
   applies unchanged. The list is still a list to a screen reader —
   display:contents does not remove list semantics in any browser
   that supports it, and the fallback below covers the ones that do
   not.

   align-items:stretch is the board's, and it is what makes all three
   beds the same height when a phrase wraps to two lines and another
   does not. */
.mission .aim-flow {
  list-style: none;
  margin: 0;
  padding: 0;
  display: grid;
  grid-template-columns: 1fr auto 1fr auto 1fr;
  gap: 8px;
  align-items: stretch;
}
.mission .aim-step { display: contents; }
.mission .aim-box {
  min-width: 0;
  border-radius: 14px;
  padding: 18px 16px;
  text-wrap: balance;
  /* --panel, for the same reason as .trust-band: a pinned pale box
     with themed ink inside it. The reward box below overrides this
     with its own green, which is a semantic fill and stays. */
  background: var(--panel);
}
/* LINE-HEIGHT 1.3 IS NOT DECORATION, IT IS THE BOX HEIGHT.

   The board gives this label no line-height, so it renders at
   `normal` — 20px at 15px/system-ui. The page inherits 1.65 from its
   own prose rule, which made the label 24.8px and every box 114.3px
   against the board's 109.5px. A 4.8px error in all three beds,
   invisible to every text check and to the overflow checker, and
   visible as soon as the two cards are put side by side.

   1.3 (= 19.5px) rather than `normal`, because `normal` is
   font-dependent: the board is system-ui and this page is DM Sans,
   so copying the keyword would copy a DIFFERENT number. A ratio is
   what makes the measurement reproducible.

   THE 109.5px ABOVE IS HISTORY, NOT THE CURRENT HEIGHT. It was the
   board's box height when all three beds held a label and a phrase.
   Since 13 Oct the green bed carries a third line ("Monitoring could
   become optional"), and .aim-flow is align-items:stretch, so all
   three measure 140px at 1280 — level, because the other two stretch
   to the tallest. What this rule still pins is the LABEL at 19.5px;
   that is the number to check if the beds ever disagree with the
   board again. */
.mission .aim-stage {
  display: block;
  font-size: 15px;
  line-height: 1.3;
  font-weight: 800;
  letter-spacing: .06em;
  text-transform: uppercase;
  /* --ui-accent, not a literal: .aim-box's fill follows the theme
     now, and #1F5FA8 on the dark panel was 1.93:1. The token is the
     per-theme pair -- #1F5FA8 light (5.45:1 on --panel) and #A9DCFA
     dark (8.48:1). The reward box overrides it below. */
  color: var(--ui-accent);
}
/* FONT-WEIGHT 700, NOT 800, AND THE BOARD IS WHY. Its markup is a
   bare <b style="display:block;margin-top:6px;font-size:19px;
   line-height:1.25"> with NO font-weight, so it renders at the UA
   default for <b>, which is `bold` = 700. The brief quotes the box
   phrases as "19px/800/1.25" from reading the surrounding inline
   styles; measured off the board's own render they are 700. Pinned
   to the measurement, not to the brief — and written out rather
   than left to the UA default, so a future `b { font-weight: 800 }`
   anywhere in app.css cannot move it. */
.mission .aim-box b {
  display: block;
  margin-top: 6px;
  font-size: 19px;
  line-height: 1.25;
  font-weight: 700;
  /* --ink follows the box's fill, which is now --panel. #142B49 on
     the dark panel measured 1.15:1 -- the two phrases that carry
     this diagram's whole argument. 12.08:1 / 10.45:1 now. */
  color: var(--ink);
}
/* BOX 3 IS THE GREEN ONE. Same #0F6B3F as .price-note's coin, so the
   two places this page connects driving to price read as one idea. */
.mission .aim-step-reward .aim-box { background: #0F6B3F; }
.mission .aim-step-reward .aim-box b { color: #FFFFFF; }
/* #FFFFFF at 85% over #0F6B3F, resolved. 5.25:1. */
.mission .aim-step-reward .aim-stage { color: #DBE9E2; }
/* "MONITORING COULD BECOME OPTIONAL" — THE THIRD LINE IN BOX 3.

   Added 13 Oct 2026 from the board. It only ever appears in the
   green box, so it is scoped under .aim-step-reward rather than
   given a width-independent rule of its own: a bare .aim-optional
   would invite a second use in a purple box, where #FFFFFF text and
   a white hairline are both wrong.

   PURE #FFFFFF, NOT THE STAGE LABEL'S #DBE9E2. The label above is
   deliberately dimmed to sit behind the phrase; this line is
   content, not a label, and reads at full strength. 6.57:1 on
   #0F6B3F, the same as the phrase it follows.

   THE HAIRLINE IS rgba(255,255,255,.3), FROM THE BOARD, and it is an
   alpha rather than the resolved #5B9579 on purpose: it is drawn
   over a fill this file pins (#0F6B3F), so the composite is already
   deterministic, and an alpha keeps the rule correct if the green is
   ever retuned. Composites to rgb(87, 151, 121) over the fill and
   measures 1.92:1 against it — a divider, not
   text, so 4.5:1 does not apply (WCAG 1.4.11 exempts purely
   decorative rules); it separates two lines that are each 6.57:1 and
   it is visible without cutting the box in half.

   8px ABOVE AND BELOW, which is the board's figure and is tighter
   than the 6px gap between the stage label and the phrase on
   purpose: the first two lines are one unit and this is a second
   thought joined to it, so the rule needs to read as a join rather
   than as a gap. */
.mission .aim-step-reward .aim-optional {
  display: block;
  margin-top: 8px;
  padding-top: 8px;
  border-top: 1px solid rgba(255, 255, 255, .3);
  font-size: 14px;
  line-height: 1.35;
  font-weight: 600;
  color: #FFFFFF;
}
/* currentColor on the stroke, set here: the markup carries no colour
   so the arrow follows this rule rather than an attribute. flex:none
   stops the 30px square being squeezed when a phrase wraps. */
.mission .aim-arrow {
  flex: none;
  align-self: center;
  width: 30px;
  height: 30px;
  color: #0E1B2C;
}
.mission .aim-caveat {
  margin: 20px 0 0;
  padding-top: 14px;
  border-top: 1px solid var(--line);
  font-size: 12px;
  line-height: 1.5;
  /* --muted, not --ui-soft, and the difference matters: this line is
     inside .aim-box, which is a PAGE surface (--paper) rather than a
     product mock-up. --ui-soft is the mock-ups' quiet ink and was
     2.03:1 here. */
  color: var(--muted);
}

/* ---- PHONE — 02 / AFTER YOUR FIRST YEAR, AND THE STRIP ------------

   THIS BLOCK IS AT THE END OF THE FILE ON PURPOSE, and the reason is
   a bug this round actually had. It was first written inside the
   existing @media (max-width:760px) block ABOVE the new section's own
   desktop rules. Media queries do not raise specificity: both rules
   are `.mission .retention-grid`, so the LATER one wins, and the
   desktop grid-template-columns beat the phone one at every width.
   Measured at 390 with the block in the wrong place: the grid still
   resolved to `0px 680px` and the card still hung 442px off the
   screen — while .aim-arrow's rotation DID apply, because that
   property has no desktop counterpart to lose to. A half-applied
   media query is worse than none: it looks like it is working.

   Same class of fault as the explainer's `margin-inline:auto` being
   overridden by a base rule written after its media query (see
   app.css). Any further phone rule for this section belongs here,
   below the desktop rules it overrides, not in the block above. */
@media (max-width: 760px) {
  /* ---- 02 / AFTER YOUR FIRST YEAR, AT PHONE WIDTH ----------------

     THE BOARD DOES NOT SETTLE THIS. There is no MissionPhone board —
     Mission.dc.html is the only board for this page, drawn at 1280 —
     so every figure below is a decision taken here, and the brief's
     own instruction for the phone is the whole of the specification:
     "text on top, card below; the three boxes stack vertically with
     the arrows rotated 90 degrees to point down."

     WHAT IT LOOKS LIKE WITHOUT THESE RULES, measured at 390 before
     they existed, because "it stacks sensibly" was not safe to
     assume: .retention-grid resolved to `0px 680px`. The minmax(0,
     1fr) text column collapsed to NOTHING and the 680px card kept
     its fixed track, so the card sat at x=72 with width 680 in a
     390px viewport — 442px of it outside the screen. The document
     reported ZERO horizontal overflow throughout, because the
     overflow was clipped rather than scrolled. So the single most
     visible fault on the phone was invisible to the overflow
     checker, which is the fault class this file keeps meeting.

     ONE COLUMN, and the 680px track must go with it: a fixed track
     in a single-column grid is still 680px wide. */
  .mission .retention { padding: 40px 0 44px; }
  .mission .retention-grid { grid-template-columns: 1fr; gap: 24px; }
  .mission .retention-copy h2 { font-size: 32px; margin: 12px 0 18px; }
  .mission .retention-lede { font-size: 17px; margin-bottom: 12px; }
  /* THE TWO <br>s IN THE LEFT COLUMN ARE DESKTOP LINE BREAKS. At 42px
     over a 472px column they are deliberate ("A good year should /
     pay off at renewal."); at 32px over 345px they break the line in
     the wrong place and leave a two-word orphan.

     display:none WAS WRONG AND THE PHONE RENDER IS WHAT SHOWED IT.
     A <br> is the ONLY whitespace between the words either side of
     it — the template has no space before or after the tag — so
     hiding it rendered "A good year shouldpay off at renewal." and
     "We want good drivers to stay,and be rewarded for it." Both
     measured perfectly: right font size, right colour, right line
     count, no overflow. They were simply missing a space, which only
     reading the picture catches.

     content:" " ON A ::after INSTEAD. The <br> stays in the flow as
     an inline box that no longer breaks the line (display:inline
     makes it inert in every engine) and the pseudo-element supplies
     the space the line break used to imply. The desktop breaks are
     untouched, so the board keeps its wording and its line endings. */
  .mission .retention-copy h2 br,
  .mission .retention-aim br { display: inline; }
  .mission .retention-copy h2 br::after,
  .mission .retention-aim br::after { content: " "; }
  .mission .retention-aim { font-size: 19px; }
  .mission .aim-card { padding: 20px; border-radius: 18px; }
  .mission .aim-title { font-size: 21px; margin-bottom: 14px; }

  /* THE FLOW STACKS, AND THE ARROWS TURN TO POINT DOWN.

     ONE COLUMN, FIVE ROWS. .aim-step is display:contents at every
     width, so the grid's children are box/arrow/box/arrow/box
     regardless — the phone only has to change the TRACK AXIS from
     five columns to one. grid-auto-flow is already row; replacing
     grid-template-columns with a single 1fr is the whole change, and
     the five items then fall into five rows in source order, which
     is exactly Year 1 -> arrow -> Renewal -> arrow -> Rewarded.

     rotate(90deg) ON THE ARROW, not a different glyph: the <path> is
     a right-pointing arrow and a quarter turn clockwise makes it
     point down. justify-self:center puts it on the column's centre
     line so it reads as a join between the two boxes rather than as
     an icon beside them. 10px of air above and below, so the chain
     reads as three beds joined by two arrows rather than five
     equally-spaced things.

     text-wrap:balance STAYS. The boxes are now 342px wide instead of
     177px, so no phrase wraps at all and balance is a no-op — kept
     because it costs nothing and is correct again the moment a
     phrase gets longer or the gutter narrows. */
  .mission .aim-flow { grid-template-columns: 1fr; gap: 10px; }
  .mission .aim-arrow {
    align-self: center;
    justify-self: center;
    transform: rotate(90deg);
  }
  .mission .aim-box { padding: 16px 16px; }
  .mission .aim-caveat { margin-top: 16px; padding-top: 12px; }

  /* ---- THE PRINCIPLE STRIP: FIVE ITEMS, ONE COLUMN ---------------

     A DECISION, NOT A BOARD FIGURE, and the reason is in the render
     rather than in the grid. At four items the 1fr 1fr above was a
     clean 2x2. At five it measured THREE rows with the last holding
     a single item:

       row 1  y=849   01 Recognise the individual | 02 Reward a good year
       row 2  y=909   03 Understand the context   | 04 Help drivers improve
       row 3  y=969   05 Make cover clear         |  (empty)

     Nothing overflowed — the fault is visual. Looked at on a 390
     device render, the orphan reads as a mistake: the strip sits
     directly under the hero, it is the page's table of contents, and
     a lone cell in the bottom-left makes the fifth principle look
     like an afterthought that did not fit rather than one of five
     equal items. A 2-up grid with an odd count has nowhere good to
     put the remainder.

     SO: ONE COLUMN, FIVE ROWS. It is taller — 5 rows of 19px plus
     four 10px gaps against the 2-up's 3 rows — and that is the cost
     of an even rhythm where every item is presented the same way.
     The alternative considered and rejected was leaving the orphan:
     it is not broken, but this strip is the first thing under the
     hero and "not broken" is a low bar for the most prominent
     element on the phone.

     ROW GAP 10px, NOT 14px. Five rows at 14px added 56px of air to a
     strip whose whole job is to be scanned in one glance; 10px keeps
     the list tight enough to read as one block. The column gap in
     the 1fr 1fr rule above no longer applies and is harmless.

     IF A SIXTH PRINCIPLE IS EVER ADDED, this stays one column — the
     reason above is about even presentation, not about oddness. */
  .mission .mission-strip .wrap { grid-template-columns: 1fr; gap: 10px; }
}

/* The same release .underlined gets at this width, and for the same
   reason: below here an unbreakable 19-character run at 32px can be
   wider than the column once the user's text size is raised, and an
   overflow is worse than a rule drawn on two line boxes. */
@media (max-width: 24.5rem) {
  .mission .rule-under { white-space: normal; }
}

/* ---- "CoRoute AI" with an orange C and AI ------------------------
   Guillaume, 16 Oct 2026: match the homepage's Contextual AI
   heading, which draws the C and the AI in #FFAA66.

   TWO ORANGES, NOT ONE, AND THAT IS THE WHOLE NOTE. The three places
   this appears do not share a background — measured on the rendered
   page rather than read off the stylesheet:

     chat-header      #291B56  dark
     ai-label         #FFFFFF  light  (the card, not the #F1E5FC
                                       sparkle tile beside it)
     ai-drive-summary #F7F2FC  light

   #FFAA66 scores 8.08:1 on the dark header but only 1.88:1 and
   1.70:1 on the two light ones. Using the homepage's orange
   everywhere would make the C and the AI all but invisible in two of
   the three — the opposite of the emphasis being asked for.

   So the DARK context keeps the homepage colour exactly, and the
   light contexts take a deeper orange of the same hue family:
   #9A4A00 scores 6.26:1 and 5.68:1. Both clear 4.5:1.

   THE FIGURES ABOVE ARE THE COMPUTED BACKGROUND OF THE TEXT, read
   off the rendered page with getComputedStyle and walked up to the
   first non-transparent ancestor. An earlier draft of this comment
   named #F1E5FC for the ai-label, which is the sparkle's own tile
   and not what the letters sit on.

   The default below is the light-ground one, because two of the
   three sit on light; the header overrides it. */
/* THE DEFAULT GROUND IS NOW PER-THEME, SO THIS IS TOO.

   IT IS --accent-text AND NOT --logo, and the difference is the
   point. --logo is #E8762A, the LOGOTYPE orange: 1.4.11 exempts a
   brand MARK from the contrast requirement, which is why the ring
   may be drawn in it. These are LETTERS -- the words "C...AI" in a
   sentence -- and 2.98:1 is not a legible word. --accent-text is the
   same family pushed to #9A4A00 in light (6.26:1 on --ui-card) and
   #FFB264 in dark (8.06:1).

   A first pass used --logo here and traded a dark-mode failure for a
   light-mode one at 2.98:1. Caught by running the contrast walker in
   BOTH themes after the change, which is the only reason it did not
   ship. */
/* --logo IS OVERRIDDEN FOR THE CARD, NOT WRITTEN ONTO THE C.
   23 Oct 2026, with the single-variable change in app.css. The
   reasoning above is unchanged and still correct: these are LETTERS
   in a readable word, so they need text contrast, and --logo's
   #E8762A is 2.98:1 on this card.

   What changed is WHERE that decision is expressed. Setting --logo
   for the card moves the C, the "o" and the "AI" together; writing
   a hex on .brandai-c moved only the C, which is precisely how the
   C and the "o" ended up different colours in four places.

   AND THE C IS NO LONGER A LETTER HERE. Later the same day the
   three typed Cs on this page became .brand-c rings (app.css), so
   this override now colours a BORDER as well as the "o" and the
   "AI" -- which is the point of expressing it as --logo rather than
   as a colour on text. The contrast reasoning above still applies:
   the ring sits in a readable word, so it is held to text contrast.
   .brandai-c is kept as the ring's hook at these three sites. */
.mission .brandai { --logo: var(--accent-text); }
.mission .brandai-ai { color: var(--logo); }

/* The dark chat header takes the homepage orange unchanged -- same
   mechanism, a --logo override scoped to the header. */
.mission .chat-header .brandai { --logo: #FFAA66; }
