Prompts / Eval

EvalRunCard

Eval-run summary — status, prompt, model, dataset, progress, headline metrics.

Preview

summarise-pr v2.1 · golden-100

completed

model: claude-sonnet-4-5dataset: golden-100100 / 100 rows3m

accuracy

0.062

0.842

latency p95

65ms

175ms

cost

0.0016$

0.0124$

summarise-pr v2.2 · golden-100

running

model: claude-sonnet-4-5dataset: golden-10042 / 100 rows

Installation

pnpm add nyxis-ui

Usage

import { EvalRunCard, type EvalRun } from 'nyxis-ui';

const run: EvalRun = {
  id: '1',
  name: 'summarise-pr v2.1 · golden-100',
  status: 'completed',
  modelId: 'claude-sonnet-4-5',
  datasetName: 'golden-100',
  totalRows: 100,
  processedRows: 100,
  durationMs: 184_000,
  metrics: [
    /* ... */
  ],
};

<EvalRunCard run={run} onSelect={(id) => router.push(`/runs/${id}`)} />;

Status pills

Status	Tone	Indicator
`queued`	muted	clock
`running`	primary	spinner + progress bar
`completed`	success	check
`failed`	destructive	warning + error banner
`cancelled`	muted	ban icon

Anatomy

Status icon + run name + status pill.
Meta row: model, dataset, rows processed, duration.
Progress bar visible while running.
Error banner visible while failed (renders error field).
Inline metric strip composes <MetricCard> for the first 6 metrics.