Implements the direction agreed from the identity board: Route C ("Cohort")
with the paper ground from C1, the Schibsted Grotesk / Literata pairing from
C2, and the iris accent from C3. Dark theme is in scope from the start rather
than retrofitted.
The audit found three things wrong beyond taste:
* No brand asset set. og:image was absent entirely, so every link shared into
a class WhatsApp group rendered as a bare grey card. apple-touch-icon pointed
at an SVG, which iOS ignores, and the manifest shipped no PNGs, so Android
installs had no icon. The header mark and the favicon had also drifted into
two different logos.
* No colour discipline. --primary and --trend-down were the same coral, so the
main CTA and "below average" shared a hue. 58 distinct hex values were spread
across component CSS, and the chart palette was still Chart.js's stock demo
colours.
* A dark theme that was declared but never built — themeColor announced a dark
variant with no dark styling behind it.
What changed:
Colour now has exactly three jobs that never borrow each other's hues: brand
(iris) for interactive and identity, status (teal/amber) for above/below a
comparison point, and phase for categories. Teal/amber rather than green/red
keeps the above/below signal readable for every form of colour blindness.
Every chromatic literal in component CSS is now a token, and the JS-painted
surfaces (Chart.js, Leaflet) read the tokens through lib/theme so they follow
the theme instead of ignoring it.
The mark is the five-bar cohort spread — the same object as the distribution
strip inside a school row, built from opacity steps so it inverts cleanly.
components/Logo.tsx is the single source; the favicon, apple-icon and share
card all derive from its geometry.
globals.css drops 123 dead global classes left over from the vanilla-JS app
(only the btn family, .skip-link and .main were still referenced), along with
the noise overlay. It also gains prefers-reduced-motion support, which was
missing entirely, and a type scale so the 54 ad-hoc font sizes have somewhere
to converge.
Verified: tsc clean, 159 unit tests pass, production build succeeds and
prerenders /icon.svg, /apple-icon and /opengraph-image. Three e2e journeys
added for the asset set, the themeColor/background match, and the dark theme
actually repainting — all silent failures that nothing on the page reveals.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Applies the compare screen's chart-truthfulness rules (spec §8.1) to the
school detail page's time-series charts.
PerformanceChart (Results Over Time, both phases):
- Fill every academic year between the first and last data point via the
shared fillAcademicYears helper, so cancelled/unpublished years (2019/20,
2020/21, and — for KS2 — 2021/22) render as real gaps instead of
compressed time. Each series and the England overlay map onto this filled
axis with null for missing years; spanGaps:false so school lines never
bridge a gap.
- Replace the primary-only COVID note with a distinct, honest gap caption:
KS2 names the cancelled tests plus the unpublished 2021/22 school-level
year; KS4 names the unpublished 2019/20–2020/21 GCSE grading years.
- Mobile x-axis switches to autoSkip so the longer (gap-honest) axis stays
readable; the broken line still marks a missing year even when its tick
label is skipped.
AdmissionsTrendChart:
- Same gap-honest axis + spanGaps:false so a missing admissions year is a
real gap, not compressed time.
Point 12 (definite canvas heights): desktop is already a definite 280px;
fix the secondary detail's mobile .chartContainer, which fixed the outer
box at 220px and double-constrained PerformanceChart's own 220px canvas +
chip strip (clipping the chips onto the plot) — now height:auto to match
the primary view.
SatsChart is out of scope (single-year per-subject CSS bars — no year axis,
no canvas height to constrain).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0146VHeLAWjDVE2B5uU67jCB
A first-choice offer rate sitting on the y-max ceiling (100%) was drawn
flush against the top of the plot area, clipping the point marker. Add 8px
top layout padding and clip:false so ceiling values render in full, without
introducing a misleading >100% tick.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The hand-rolled SVG sparkline used px font sizes inside a 520-wide viewBox
that stretched to the full card width, so labels ballooned ~4x and collided —
and with 10+ years of real data the per-point labels and year ticks
overlapped badly, while the oversized chart stretched the "this year" view.
Replace it with a Chart.js line chart (AdmissionsTrendChart) in a fixed
200px wrapper, matching PerformanceChart: responsive px fonts, auto-skipping
x ticks, auto-scaled y-axis clamped to 0-100, emphasised latest point.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
