The answer's conclusion under pressure: one AI argues for it, one against, for a few rounds, and a third says what held up.
Each side is named, and the rounds are shown in full rather than summarised, so a reader can see how a point broke instead of being told that it did. What nobody settled is said plainly, for a person to decide: a check that always reaches a verdict is a check pretending to know more than it does.
While running, the rounds arrive one at a time under a Streaming line that can stop it.
import { ChallengeReport } from '@redrob-labs/ui';
| Prop | Type | Notes |
|---|---|---|
claim |
React.ReactNode |
The conclusion under challenge, in the answer's words. |
sides |
{ for?: React.ReactNode; against?: React.ReactNode; judge?: React.ReactNode } |
|
rounds |
Array<{ for: React.ReactNode; against: React.ReactNode }> |
|
state |
'running' | 'done' |
|
shown |
number |
Rounds shown so far while running. |
of |
number |
|
took |
React.ReactNode |
|
verdict |
Array<{ kind: 'broke' | 'held' | 'changed' | string; text: React.ReactNode }> |
|
unsettled |
React.ReactNode |
|
applied |
boolean |
|
onApply |
() => void |
|
onRerun |
() => void |
|
onClose |
() => void |
|
onStop |
() => void |
|
title |
React.ReactNode |
|
label |
string |
|
claimLabel |
string |
|
forLabel |
string |
|
againstLabel |
string |
|
judgeLabel |
string |
|
roundLabel |
string |
|
verdictLabel |
string |
|
unsettledLabel |
string |
|
applyLabel |
string |
|
appliedLabel |
string |
|
rerunLabel |
string |
|
runningLabel |
string |
|
progressLabel |
string |
|
kindLabels |
Record<string, string> |
|
className |
string |