# Agent OPS Evals

Eval runs comparing two prompt versions and models with score charts and a results matrix.

- id: `agent-ops-evals`
- Page: https://agentic-ui-kit.com/components/agent-ops-evals
- Live preview: https://agentic-ui-kit.com/live/fixture/agent-ops-evals?piece=agent-ops-evals

## Install

```bash
agentic-ui add agent-ops-evals
npx shadcn@4.17.0 add @agentic-ui/agent-ops-evals
```

## Metadata

- **Level:** page
- **Category:** page
- **Installable in:** Radix UI (registry variant new-york), Base UI (registry variant base-vega)
- **Capabilities:** eval-runs-list, new-run-dialog, run-progress, failed-run-retry, baseline-vs-candidate-chart, score-histogram, results-matrix, case-detail-dialog, pass-rate-and-cost
- **Intents:** evaluate-ai, compare-over-time, track-status
- **Suits domains:** ai-agents, devtools, analytics, saas
- **Interaction:** select, dialog, chart, grid, retry
- **Screen role:** overview
- **Density:** medium
- **Dark mode:** yes
- **Use it for:** eval runs comparing two prompt versions and models with score charts and a results matrix, evaluation dashboard where a new run is started on a dataset and compared with the previous one
- **Avoid when:** model training dashboards, A/B tests of web pages
- **Preserve when adapting:** single-h1-per-page, deterministic-seed-data, link-integrity
- **Replace when adapting:** brand-copy, seed-data, model-price-list
- **Adaptation effort:** low
- **Coupling:** medium
- **Quality breakdown:** visual 9/10, responsive 9/10, accessibility 9/10, code 9/10
- **Uses primitives:** button, card, badge, input, label, dialog, tabs, select, progress
- **npm dependencies:** next, lucide-react, next-themes, recharts, @tanstack/react-table
- **Motion:** profile none, library css, reduced-motion fallback yes
- **Used in:** Agent Ops Console (blueprint)
- **npm dependencies (registry):** lucide-react@0.577.0
- **Registry dependencies:** ai-eval-score-chart, ai-evals-results-table, lib-primitives