Design Systems · Evaluation
Roast by AI
Compliance Index score
59.3/100 · F
↓ 7.8 pts since last week (67.1 → 59.3)
Per-category breakdown
Each category is scored independently. The cohort average shows how this site compares to the 30-site cohort in each dimension.
Evaluation
Roast by AI scores 59.3/F on the Designesy Compliance Index — 16 of 40 checks passed, 3 failed, 16 warned, 2 skipped. The cohort mean is 62.2/D across 30 scored sites. Roast by AI sits 5% below the cohort mean.
Strongest category: Identity at 100.0%. Weakest scored category: Takt at 25.0%. The gap between best and worst is the finding — most systems are strong in one dimension and absent in another.
Since last week's score, Roast by AI dropped by 7.8 points — moving from 67.1 to 59.3.
The engine detected 46 custom properties at :root — a modest token layer.
F-grade means Roast by AI scores below 60 — the contract's pass threshold. The failing checks are not subjective: they are missing CSS primitives that the contract requires at :root.
This evaluation is deterministic — no LLM, no human judgment, no survey. The same 40-check engine that scored Roast by AI scores every site on the leaderboard. Re-score roastbyai.com live to see if anything has changed since 2026-08-03.
Scored 2026-08-03 against contract v0.4.0 (40 checks, 14 weighted categories). Re-scored weekly via GitHub Action.
Compared to Design Systems peers
Roast by AI ranks #8 of 14 in the Design Systems category.
Cohort context
This site
59.3/100
Cohort mean
62.2/100
Cohort median
62.6/100
Delta from mean
-2.9
Why this site is scored
Roast competitor — leaderboard model reference