Eval card
Feedback paragraph card with a mono kicker. Surfaces evaluator prose alongside the dimension breakdown on /results.
Preview
What the grader noticed
You led with the read path, which is right for a fan-out heavy workload like this one. The choice of write-through over write-back is defensible at this scale.
Where the answer thinned out: bottleneck analysis at the queue layer. You named the queue but did not size it under burst.
Source
apps/docs/app/recipes/eval-card/recipe.tsxexport default function EvalCardRecipe() {
return (
<div className="bg-surface border-stroke rounded-xl border px-6 py-5">
<p className="text-ink-3 text-2xs mb-2 font-mono font-bold tracking-[0.06em] uppercase">
What the grader noticed
</p>
<div className="text-ink-2 space-y-3 text-sm leading-[1.6]">
<p>
You <strong className="text-ink-1 font-semibold">led with the read path</strong>, which is
right for a fan-out heavy workload like this one. The choice of write-through over
write-back is defensible at this scale.
</p>
<p>
Where the answer thinned out:{' '}
<strong className="text-ink-1 font-semibold">
bottleneck analysis at the queue layer
</strong>
. You named the queue but did not size it under burst.
</p>
</div>
</div>
);
}
export default function EvalCardRecipe() {
return (
<div className="bg-surface border-stroke rounded-xl border px-6 py-5">
<p className="text-ink-3 text-2xs mb-2 font-mono font-bold tracking-[0.06em] uppercase">
What the grader noticed
</p>
<div className="text-ink-2 space-y-3 text-sm leading-[1.6]">
<p>
You <strong className="text-ink-1 font-semibold">led with the read path</strong>, which is
right for a fan-out heavy workload like this one. The choice of write-through over
write-back is defensible at this scale.
</p>
<p>
Where the answer thinned out:{' '}
<strong className="text-ink-1 font-semibold">
bottleneck analysis at the queue layer
</strong>
. You named the queue but did not size it under burst.
</p>
</div>
</div>
);
}