# Designesy — /work/designesy-org Canonical page: https://www.designesy.org/work/designesy-org --- ## designesy.org · D to A The publisher scores itself, fixes the gaps, and publishes the grade. The same /api/score engine that grades every other site graded designesy.org. The result was a D — 9 failures, 2 warnings, 3 skips. The fixes were real: token discipline, focus-visible rings, reduced-motion tiering, Cadence typography rules, and the static halves of the Poise/Takt checks. The score is now A, verified on the same terms as everyone else’s. ## Summary designesy.org was scored by its own engine and got a D. The failures were concentrated in four areas: token discipline (raw hex, magic numbers), keyboard affordance (no focus-visible rings), reduced-motion handling (a kill switch rather than tiered), and Cadence typography (no text-wrap, no tabular-nums, no font-synthesis guards). One session resolved all 9 failures and 2 warnings, plus converted 3 SKIPs to PASSes by implementing the static contract-diff halves. The score moved from D 67.4 to A 96.3 — a 28.9-point gain in a single working session. ## Score delta ### Before · D · 67.4 12 pass · 9 fail · 2 warn · 3 skip ### After · A · 96.3 23 pass · 0 fail · 0 warn · 3 skip **+ 11** pass · **- 9** fail · **- 2** warn · skip unchanged (browser-only checks, honestly labeled). ## Checks that moved Every check that changed status between the before and after score. The fix that moved each one is documented in the corrections section. ### Token values match live site :root foundation **FAIL** → **PASS** ### focus-visible present on all interactive elements **FAIL** → **PASS** ### Reduced-motion tiering (Tier 1/2/3) not a kill switch **FAIL** → **PASS** ### No raw hex colors in component CSS **FAIL** → **PASS** ### Cadence typography contract-diff **SKIP** → **PASS** ### text-wrap: balance + pretty both present **WARN** → **PASS** ### tabular-nums: 8 instances across the live CSS **SKIP** → **PASS** ### Poise interaction rules match live /labs/poise **SKIP** → **PASS** ### Poise keyboard-path published **SKIP** → **PASS** ### Takt rules match CSS **SKIP** → **PASS** ## Corrections Replaced raw hex values with contract tokens across component CSS. Added focus-visible rings on every interactive element, tiered reduced-motion (Tier 1 remove, Tier 2 soften ≤200ms, Tier 3 keep — never a kill switch). Implemented the Cadence typography rules: text-wrap balance + pretty, tabular-nums in numeric contexts, font-synthesis guards, skip-ink on underlines. Shipped the static contract-diff halves of the Poise interaction, Poise keyboard-path, and Takt feel checks — converting 3 SKIPs to PASSes without a browser path. ## Lessons ## What is still open Three checks remain SKIP: v02 (responsive overflow at 4 viewports), v04 (sound toggle aria-pressed flip), and v21 (Core Web Vitals). All three need a browser path. The audit endpoint at /api/score/audit returns honest SKIPs with diagnostic strings until ENABLE_BROWSER_AUDIT=1 and PSI_API_KEY are set on the deployment. They are not hidden — they are unrun, and labeled as such. ## Sources - [Score designesy.org now — Run the live engine against the publisher](https://www.designesy.org/score?url=designesy.org) - [Design system contract v0.4.0 — The standard the score runs against](https://www.designesy.org/contracts/design-system) - [Why a public score — The argument this case study is evidence for](https://www.designesy.org/learn/why-we-built-a-public-design-score) - [Work — case studies — Index](https://www.designesy.org/work)