bridgebench 3.1.0-alpha.1 → 3.1.0-alpha.2

package/dist/{reports-4CejmOHf.d.cts → reports-B8TCJtPr.d.ts}

@@ -1,4 +1,4 @@
-import { B as BenchmarkCategory } from './tasks-CpaCJ6JE.cjs';
+import { B as BenchmarkCategory } from './tasks-BmhWuMBD.js';
 import { z } from 'zod';
 
 interface ResponseSummary {
@@ -57,7 +57,7 @@ interface ArenaEventDataMap {
         matchId: string;
         taskId: string;
         winnerModelId: string | null;
-        outcome: 'judged' | 'forfeit' | 'no-contest';
+        outcome: 'judged' | 'forfeit' | 'no-contest' | 'speed-decided';
         costUsd: number;
         eloAfter: Record<string, number>;
         completed: number;
@@ -133,9 +133,13 @@ declare const ModelCompletionSchema: z.ZodObject<{
     latencyMs: z.ZodNumber;
     finishReason: z.ZodString;
     attempts: z.ZodOptional<z.ZodNumber>;
+    /** Time from request start to the first non-empty content delta (speed arena metric). */
+    ttftMs: z.ZodOptional<z.ZodNumber>;
+    /** Request start to stream completion for the successful attempt (speed arena metric). */
+    totalMs: z.ZodOptional<z.ZodNumber>;
 }, "strip", z.ZodTypeAny, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -143,9 +147,11 @@ declare const ModelCompletionSchema: z.ZodObject<{
     finishReason: string;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -153,6 +159,8 @@ declare const ModelCompletionSchema: z.ZodObject<{
     finishReason: string;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }>;
 type ModelCompletion = z.infer<typeof ModelCompletionSchema>;
 interface ChatRequest {
@@ -218,12 +226,14 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     latencyMs: z.ZodNumber;
     finishReason: z.ZodString;
     attempts: z.ZodOptional<z.ZodNumber>;
+    ttftMs: z.ZodOptional<z.ZodNumber>;
+    totalMs: z.ZodOptional<z.ZodNumber>;
 } & {
     modelId: z.ZodString;
     success: z.ZodLiteral<true>;
 }, "strip", z.ZodTypeAny, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -233,9 +243,11 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     success: true;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -245,6 +257,8 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     success: true;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }>;
 type CompetitorSuccess = z.infer<typeof CompetitorSuccessSchema>;
 declare const CompetitorFailureSchema: z.ZodObject<{
@@ -280,14 +294,14 @@ declare const JudgeVerdictSchema: z.ZodObject<{
         completeness: z.ZodString;
     }, "strip", z.ZodTypeAny, {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     }, {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     }>;
     violations: z.ZodArray<z.ZodString, "many">;
 }, "strip", z.ZodTypeAny, {
@@ -296,9 +310,9 @@ declare const JudgeVerdictSchema: z.ZodObject<{
     rationale: string;
     criteria: {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     };
     violations: string[];
 }, {
@@ -307,9 +321,9 @@ declare const JudgeVerdictSchema: z.ZodObject<{
     rationale: string;
     criteria: {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     };
     violations: string[];
 }>;
@@ -382,6 +396,22 @@ interface EloState {
     ratings: Record<string, number>;
     points: Record<string, number>;
 }
+/** Per-competitor latency measurements the speed arena decides matches by. */
+interface SpeedMetric {
+    /** Time to first token: ms from request start to the first non-empty content delta. */
+    ttftMs: number;
+    /** Output tokens per second over the generation window: outputTokens / max(1e-3, (totalMs - ttftMs)/1000). */
+    tps: number;
+    /** Total wall-clock ms from request start to stream completion — the value that decides the winner. */
+    totalMs: number;
+    outputTokens: number;
+}
+/** Present only on 'speed-decided' matches; `a` is modelA's measurement, `b` is modelB's. */
+interface SpeedMetrics {
+    a: SpeedMetric;
+    b: SpeedMetric;
+}
+declare const SpeedMetricsSchema: z.ZodType<SpeedMetrics>;
 interface MatchResult {
     methodologyVersion: string;
     runId: string;
@@ -398,7 +428,8 @@ interface MatchResult {
         category?: BenchmarkCategory;
         cluster: string;
         publicHash: string;
-        privateHash: string;
+        /** Null for public-only packs (e.g. the speed arena), which have no hidden half. */
+        privateHash: string | null;
     };
     competitors: {
         modelA: string;
@@ -406,9 +437,11 @@ interface MatchResult {
         responseA: CompetitorResponse;
         responseB: CompetitorResponse;
     };
-    outcome: 'judged' | 'forfeit' | 'no-contest';
+    outcome: 'judged' | 'forfeit' | 'no-contest' | 'speed-decided';
     winnerModelId: string | null;
     panel: PanelDecision | null;
+    /** Present only when outcome is 'speed-decided'; null (or absent) otherwise. */
+    speedMetrics?: SpeedMetrics | null;
     eloBefore: Record<string, number>;
     eloAfter: Record<string, number>;
     pointAwarded: boolean;
@@ -422,6 +455,15 @@ declare function competitorOutputTokens(response: CompetitorResponse): number;
 declare function competitorReasoningTokens(response: CompetitorResponse): number | undefined;
 declare function competitorContent(response: CompetitorResponse): string;
 
+/** Per-model latency aggregates surfaced by the speed arena, derived from decided matches. */
+interface SpeedStats {
+    /** Number of speed-decided matches this model participated in. */
+    samples: number;
+    medianTtftMs: number;
+    avgTtftMs: number;
+    medianTps: number;
+    avgTps: number;
+}
 interface LeaderboardEntry {
     rank: number;
     modelId: string;
@@ -439,6 +481,8 @@ interface LeaderboardEntry {
         wins: number;
         losses: number;
     }>>;
+    /** Present only for the speed category; omitted for judged categories so their output is unchanged. */
+    speed?: SpeedStats;
 }
 interface ArenaSnapshot {
     version: '0.2.0';
@@ -451,4 +495,4 @@ interface ArenaSnapshot {
     matches: MatchResult[];
 }
 
-export { type ArenaEvent as A, competitorOutputTokens as B, type ChatRequest as C, competitorReasoningTokens as D, type EloState as E, JUDGE_VERDICT_TRANSPORT_SCHEMA as J, type LeaderboardEntry as L, type MatchResult as M, type OpenRouterGateway as O, type PanelDecision as P, type ScheduledMatch as S, type ArenaEventDataMap as a, type ArenaEventInput as b, type ArenaEventSink as c, type ArenaEventType as d, type ArenaExecutionOptions as e, type ArenaRunConfig as f, type ArenaRunResult as g, type ArenaSnapshot as h, type CompetitorFailure as i, type CompetitorResponse as j, CompetitorResponseSchema as k, type CompetitorSuccess as l, type JudgeVerdict as m, JudgeVerdictSchema as n, type JudgeVote as o, JudgeVoteSchema as p, MatchResultSchema as q, type ModelCompletion as r, ModelCompletionSchema as s, type ModelRegistryEntry as t, type ModelRequestPolicy as u, type ModelRole as v, ModelRoleSchema as w, PanelDecisionSchema as x, competitorContent as y, competitorCost as z };
+export { type ArenaEvent as A, SpeedMetricsSchema as B, type ChatRequest as C, type SpeedStats as D, type EloState as E, competitorContent as F, competitorCost as G, competitorOutputTokens as H, competitorReasoningTokens as I, JUDGE_VERDICT_TRANSPORT_SCHEMA as J, type LeaderboardEntry as L, type MatchResult as M, type OpenRouterGateway as O, type PanelDecision as P, type ScheduledMatch as S, type ArenaEventDataMap as a, type ArenaEventInput as b, type ArenaEventSink as c, type ArenaEventType as d, type ArenaExecutionOptions as e, type ArenaRunConfig as f, type ArenaRunResult as g, type ArenaSnapshot as h, type CompetitorFailure as i, type CompetitorResponse as j, CompetitorResponseSchema as k, type CompetitorSuccess as l, type JudgeVerdict as m, JudgeVerdictSchema as n, type JudgeVote as o, JudgeVoteSchema as p, MatchResultSchema as q, type ModelCompletion as r, ModelCompletionSchema as s, type ModelRegistryEntry as t, type ModelRequestPolicy as u, type ModelRole as v, ModelRoleSchema as w, PanelDecisionSchema as x, type SpeedMetric as y, type SpeedMetrics as z };

package/dist/{reports-s2CTnGN8.d.ts → reports-DPpOoOux.d.cts}

@@ -1,4 +1,4 @@
-import { B as BenchmarkCategory } from './tasks-CpaCJ6JE.js';
+import { B as BenchmarkCategory } from './tasks-BmhWuMBD.cjs';
 import { z } from 'zod';
 
 interface ResponseSummary {
@@ -57,7 +57,7 @@ interface ArenaEventDataMap {
         matchId: string;
         taskId: string;
         winnerModelId: string | null;
-        outcome: 'judged' | 'forfeit' | 'no-contest';
+        outcome: 'judged' | 'forfeit' | 'no-contest' | 'speed-decided';
         costUsd: number;
         eloAfter: Record<string, number>;
         completed: number;
@@ -133,9 +133,13 @@ declare const ModelCompletionSchema: z.ZodObject<{
     latencyMs: z.ZodNumber;
     finishReason: z.ZodString;
     attempts: z.ZodOptional<z.ZodNumber>;
+    /** Time from request start to the first non-empty content delta (speed arena metric). */
+    ttftMs: z.ZodOptional<z.ZodNumber>;
+    /** Request start to stream completion for the successful attempt (speed arena metric). */
+    totalMs: z.ZodOptional<z.ZodNumber>;
 }, "strip", z.ZodTypeAny, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -143,9 +147,11 @@ declare const ModelCompletionSchema: z.ZodObject<{
     finishReason: string;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -153,6 +159,8 @@ declare const ModelCompletionSchema: z.ZodObject<{
     finishReason: string;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }>;
 type ModelCompletion = z.infer<typeof ModelCompletionSchema>;
 interface ChatRequest {
@@ -218,12 +226,14 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     latencyMs: z.ZodNumber;
     finishReason: z.ZodString;
     attempts: z.ZodOptional<z.ZodNumber>;
+    ttftMs: z.ZodOptional<z.ZodNumber>;
+    totalMs: z.ZodOptional<z.ZodNumber>;
 } & {
     modelId: z.ZodString;
     success: z.ZodLiteral<true>;
 }, "strip", z.ZodTypeAny, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -233,9 +243,11 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     success: true;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }, {
-    generationId: string;
     content: string;
+    generationId: string;
     inputTokens: number;
     outputTokens: number;
     costUsd: number;
@@ -245,6 +257,8 @@ declare const CompetitorSuccessSchema: z.ZodObject<{
     success: true;
     reasoningTokens?: number | undefined;
     attempts?: number | undefined;
+    ttftMs?: number | undefined;
+    totalMs?: number | undefined;
 }>;
 type CompetitorSuccess = z.infer<typeof CompetitorSuccessSchema>;
 declare const CompetitorFailureSchema: z.ZodObject<{
@@ -280,14 +294,14 @@ declare const JudgeVerdictSchema: z.ZodObject<{
         completeness: z.ZodString;
     }, "strip", z.ZodTypeAny, {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     }, {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     }>;
     violations: z.ZodArray<z.ZodString, "many">;
 }, "strip", z.ZodTypeAny, {
@@ -296,9 +310,9 @@ declare const JudgeVerdictSchema: z.ZodObject<{
     rationale: string;
     criteria: {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     };
     violations: string[];
 }, {
@@ -307,9 +321,9 @@ declare const JudgeVerdictSchema: z.ZodObject<{
     rationale: string;
     criteria: {
         correctness: string;
-        grounding: string;
         constraintHandling: string;
         completeness: string;
+        grounding: string;
     };
     violations: string[];
 }>;
@@ -382,6 +396,22 @@ interface EloState {
     ratings: Record<string, number>;
     points: Record<string, number>;
 }
+/** Per-competitor latency measurements the speed arena decides matches by. */
+interface SpeedMetric {
+    /** Time to first token: ms from request start to the first non-empty content delta. */
+    ttftMs: number;
+    /** Output tokens per second over the generation window: outputTokens / max(1e-3, (totalMs - ttftMs)/1000). */
+    tps: number;
+    /** Total wall-clock ms from request start to stream completion — the value that decides the winner. */
+    totalMs: number;
+    outputTokens: number;
+}
+/** Present only on 'speed-decided' matches; `a` is modelA's measurement, `b` is modelB's. */
+interface SpeedMetrics {
+    a: SpeedMetric;
+    b: SpeedMetric;
+}
+declare const SpeedMetricsSchema: z.ZodType<SpeedMetrics>;
 interface MatchResult {
     methodologyVersion: string;
     runId: string;
@@ -398,7 +428,8 @@ interface MatchResult {
         category?: BenchmarkCategory;
         cluster: string;
         publicHash: string;
-        privateHash: string;
+        /** Null for public-only packs (e.g. the speed arena), which have no hidden half. */
+        privateHash: string | null;
     };
     competitors: {
         modelA: string;
@@ -406,9 +437,11 @@ interface MatchResult {
         responseA: CompetitorResponse;
         responseB: CompetitorResponse;
     };
-    outcome: 'judged' | 'forfeit' | 'no-contest';
+    outcome: 'judged' | 'forfeit' | 'no-contest' | 'speed-decided';
     winnerModelId: string | null;
     panel: PanelDecision | null;
+    /** Present only when outcome is 'speed-decided'; null (or absent) otherwise. */
+    speedMetrics?: SpeedMetrics | null;
     eloBefore: Record<string, number>;
     eloAfter: Record<string, number>;
     pointAwarded: boolean;
@@ -422,6 +455,15 @@ declare function competitorOutputTokens(response: CompetitorResponse): number;
 declare function competitorReasoningTokens(response: CompetitorResponse): number | undefined;
 declare function competitorContent(response: CompetitorResponse): string;
 
+/** Per-model latency aggregates surfaced by the speed arena, derived from decided matches. */
+interface SpeedStats {
+    /** Number of speed-decided matches this model participated in. */
+    samples: number;
+    medianTtftMs: number;
+    avgTtftMs: number;
+    medianTps: number;
+    avgTps: number;
+}
 interface LeaderboardEntry {
     rank: number;
     modelId: string;
@@ -439,6 +481,8 @@ interface LeaderboardEntry {
         wins: number;
         losses: number;
     }>>;
+    /** Present only for the speed category; omitted for judged categories so their output is unchanged. */
+    speed?: SpeedStats;
 }
 interface ArenaSnapshot {
     version: '0.2.0';
@@ -451,4 +495,4 @@ interface ArenaSnapshot {
     matches: MatchResult[];
 }
 
-export { type ArenaEvent as A, competitorOutputTokens as B, type ChatRequest as C, competitorReasoningTokens as D, type EloState as E, JUDGE_VERDICT_TRANSPORT_SCHEMA as J, type LeaderboardEntry as L, type MatchResult as M, type OpenRouterGateway as O, type PanelDecision as P, type ScheduledMatch as S, type ArenaEventDataMap as a, type ArenaEventInput as b, type ArenaEventSink as c, type ArenaEventType as d, type ArenaExecutionOptions as e, type ArenaRunConfig as f, type ArenaRunResult as g, type ArenaSnapshot as h, type CompetitorFailure as i, type CompetitorResponse as j, CompetitorResponseSchema as k, type CompetitorSuccess as l, type JudgeVerdict as m, JudgeVerdictSchema as n, type JudgeVote as o, JudgeVoteSchema as p, MatchResultSchema as q, type ModelCompletion as r, ModelCompletionSchema as s, type ModelRegistryEntry as t, type ModelRequestPolicy as u, type ModelRole as v, ModelRoleSchema as w, PanelDecisionSchema as x, competitorContent as y, competitorCost as z };
+export { type ArenaEvent as A, SpeedMetricsSchema as B, type ChatRequest as C, type SpeedStats as D, type EloState as E, competitorContent as F, competitorCost as G, competitorOutputTokens as H, competitorReasoningTokens as I, JUDGE_VERDICT_TRANSPORT_SCHEMA as J, type LeaderboardEntry as L, type MatchResult as M, type OpenRouterGateway as O, type PanelDecision as P, type ScheduledMatch as S, type ArenaEventDataMap as a, type ArenaEventInput as b, type ArenaEventSink as c, type ArenaEventType as d, type ArenaExecutionOptions as e, type ArenaRunConfig as f, type ArenaRunResult as g, type ArenaSnapshot as h, type CompetitorFailure as i, type CompetitorResponse as j, CompetitorResponseSchema as k, type CompetitorSuccess as l, type JudgeVerdict as m, JudgeVerdictSchema as n, type JudgeVote as o, JudgeVoteSchema as p, MatchResultSchema as q, type ModelCompletion as r, ModelCompletionSchema as s, type ModelRegistryEntry as t, type ModelRequestPolicy as u, type ModelRole as v, ModelRoleSchema as w, PanelDecisionSchema as x, type SpeedMetric as y, type SpeedMetrics as z };

package/dist/{tasks-CpaCJ6JE.d.cts → tasks-BmhWuMBD.d.cts}

@@ -1,9 +1,9 @@
 import { z } from 'zod';
 
 declare const METHODOLOGY_VERSION = "arena-v0.3.0";
-declare const BenchmarkCategorySchema: z.ZodEnum<["reasoning", "hallucination"]>;
+declare const BenchmarkCategorySchema: z.ZodEnum<["reasoning", "hallucination", "security", "bullshit", "refactoring", "debugging", "generation", "speed"]>;
 type BenchmarkCategory = z.infer<typeof BenchmarkCategorySchema>;
-declare const CATEGORIES: ["reasoning", "hallucination"];
+declare const CATEGORIES: ["reasoning", "hallucination", "security", "bullshit", "refactoring", "debugging", "generation", "speed"];
 declare const CATEGORY_CLUSTERS: Record<BenchmarkCategory, readonly string[]>;
 declare const CATEGORY_META: Record<BenchmarkCategory, {
     label: string;
@@ -16,21 +16,21 @@ declare const TaskArtifactSchema: z.ZodObject<{
     label: z.ZodString;
     content: z.ZodString;
 }, "strict", z.ZodTypeAny, {
-    type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
     id: string;
-    content: string;
+    type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
     label: string;
+    content: string;
 }, {
-    type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
     id: string;
-    content: string;
+    type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
     label: string;
+    content: string;
 }>;
 type TaskArtifact = z.infer<typeof TaskArtifactSchema>;
 declare const TaskPublicSchema: z.ZodObject<{
     id: z.ZodString;
     version: z.ZodString;
-    category: z.ZodEnum<["reasoning", "hallucination"]>;
+    category: z.ZodEnum<["reasoning", "hallucination", "security", "bullshit", "refactoring", "debugging", "generation", "speed"]>;
     cluster: z.ZodString;
     difficulty: z.ZodEnum<["hard", "expert"]>;
     title: z.ZodString;
@@ -42,47 +42,47 @@ declare const TaskPublicSchema: z.ZodObject<{
         label: z.ZodString;
         content: z.ZodString;
     }, "strict", z.ZodTypeAny, {
-        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         id: string;
-        content: string;
+        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         label: string;
+        content: string;
     }, {
-        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         id: string;
-        content: string;
+        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         label: string;
+        content: string;
     }>, "many">;
     tags: z.ZodDefault<z.ZodArray<z.ZodString, "many">>;
 }, "strict", z.ZodTypeAny, {
     id: string;
     version: string;
-    category: "reasoning" | "hallucination";
+    category: "reasoning" | "hallucination" | "security" | "bullshit" | "refactoring" | "debugging" | "generation" | "speed";
     cluster: string;
     difficulty: "hard" | "expert";
     title: string;
     summary: string;
     prompt: string;
     artifacts: {
-        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         id: string;
-        content: string;
+        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         label: string;
+        content: string;
     }[];
     tags: string[];
 }, {
     id: string;
     version: string;
-    category: "reasoning" | "hallucination";
+    category: "reasoning" | "hallucination" | "security" | "bullshit" | "refactoring" | "debugging" | "generation" | "speed";
     cluster: string;
     difficulty: "hard" | "expert";
     title: string;
     summary: string;
     prompt: string;
     artifacts: {
-        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         id: string;
-        content: string;
+        type: "code" | "log" | "config" | "spec" | "diff" | "table" | "note";
         label: string;
+        content: string;
     }[];
     tags?: string[] | undefined;
 }>;
@@ -100,14 +100,14 @@ declare const TaskPrivateSchema: z.ZodObject<{
         completeness: z.ZodString;
     }, "strict", z.ZodTypeAny, {
         correctness: string;
+        evidenceGrounding: string;
         constraintHandling: string;
         completeness: string;
-        evidenceGrounding: string;
     }, {
         correctness: string;
+        evidenceGrounding: string;
         constraintHandling: string;
         completeness: string;
-        evidenceGrounding: string;
     }>;
 }, "strict", z.ZodTypeAny, {
     id: string;
@@ -117,9 +117,9 @@ declare const TaskPrivateSchema: z.ZodObject<{
     disqualifyingErrors: string[];
     rubric: {
         correctness: string;
+        evidenceGrounding: string;
         constraintHandling: string;
         completeness: string;
-        evidenceGrounding: string;
     };
 }, {
     id: string;
@@ -128,9 +128,9 @@ declare const TaskPrivateSchema: z.ZodObject<{
     requiredEvidence: string[];
     rubric: {
         correctness: string;
+        evidenceGrounding: string;
         constraintHandling: string;
         completeness: string;
-        evidenceGrounding: string;
     };
     disqualifyingErrors?: string[] | undefined;
 }>;
@@ -147,5 +147,7 @@ type CompleteArenaTask = ArenaTask & {
     private: TaskPrivate;
     privateHash: string;
 };
+/** Narrow a task to one carrying its hidden reference (present for every judged category). */
+declare function isCompleteArenaTask(task: ArenaTask): task is CompleteArenaTask;
 
-export { type ArenaTask as A, type BenchmarkCategory as B, CATEGORIES as C, METHODOLOGY_VERSION as M, type TaskArtifact as T, BenchmarkCategorySchema as a, CATEGORY_CLUSTERS as b, CATEGORY_META as c, type CompleteArenaTask as d, TaskArtifactSchema as e, type TaskPrivate as f, TaskPrivateSchema as g, type TaskPublic as h, TaskPublicSchema as i };
+export { type ArenaTask as A, type BenchmarkCategory as B, CATEGORIES as C, METHODOLOGY_VERSION as M, type TaskArtifact as T, BenchmarkCategorySchema as a, CATEGORY_CLUSTERS as b, CATEGORY_META as c, type CompleteArenaTask as d, TaskArtifactSchema as e, type TaskPrivate as f, TaskPrivateSchema as g, type TaskPublic as h, TaskPublicSchema as i, isCompleteArenaTask as j };