Is your feature request related to a problem?
v2 runs include score.overall and an ai_summary for identifying the weakest questions, but these are not displayed on the run detail page. This makes it difficult for users to quickly assess the overall performance of their runs.
Describe the solution you'd like
- Display
score.overall at the top of the run detail page, including overall_score, verdict, and breakdown (score, weight, delta, verdict)
- Show
ai_summary as a diagnostic note above the per-row table
- Ensure graceful degradation when
ai_summary is null or when overall is absent on v1 runs
Original issue
Is your feature request related to a problem?
v2 runs carry score.overall — a weighted rollup across the judge metrics plus a natural-language ai_summary pointing at the worst questions. Today the run detail page shows neither, so the reader has to eyeball three metrics and every row to work out how the run went.
Describe the solution you'd like
- Render
score.overall at the top of the run detail page: overall_score, verdict, and the per-metric breakdown (score, weight, delta, verdict)
- Render
ai_summary as a diagnostic note above the per-row table
- Degrade gracefully when
ai_summary is null (best-effort, never fails the run) or when overall is absent on v1 runs
Parent: #265
Is your feature request related to a problem?
v2 runs include
score.overalland anai_summaryfor identifying the weakest questions, but these are not displayed on the run detail page. This makes it difficult for users to quickly assess the overall performance of their runs.Describe the solution you'd like
score.overallat the top of the run detail page, includingoverall_score,verdict, andbreakdown(score, weight, delta, verdict)ai_summaryas a diagnostic note above the per-row tableai_summaryisnullor whenoverallis absent on v1 runsOriginal issue
Is your feature request related to a problem?
v2 runs carry
score.overall— a weighted rollup across the judge metrics plus a natural-languageai_summarypointing at the worst questions. Today the run detail page shows neither, so the reader has to eyeball three metrics and every row to work out how the run went.Describe the solution you'd like
score.overallat the top of the run detail page:overall_score,verdict, and the per-metricbreakdown(score, weight, delta, verdict)ai_summaryas a diagnostic note above the per-row tableai_summaryisnull(best-effort, never fails the run) or whenoverallis absent on v1 runsParent: #265