@tangle-network/agent-eval 0.145.6 → 0.145.8

package/dist/benchmarks/index.d.ts

@@ -1,8 +1,8 @@
-import { s as RunSplitTag } from "../run-record-CKiihE6f.js";
-import { a as CampaignResult } from "../types-BxLccGMf.js";
-import { a as BenchmarkFamily, c as BenchmarkSource, i as BenchmarkEvaluation, l as BenchmarkTaskKind, n as BenchmarkAdapter, o as BenchmarkResponder, r as BenchmarkDatasetItem, s as BenchmarkScenario, t as BENCHMARK_SPLIT_SEED, u as deterministicSplit } from "../types-BRsxjg7z.js";
-import { xt as CampaignStorage } from "../provenance-8L-_4xiL.js";
-import "../index-DF3ynqmJ.js";
+import { s as RunSplitTag } from "../run-record-DVV82Gwh.js";
+import { a as CampaignResult } from "../types-BjsNDR49.js";
+import { a as BenchmarkFamily, c as BenchmarkSource, i as BenchmarkEvaluation, l as BenchmarkTaskKind, n as BenchmarkAdapter, o as BenchmarkResponder, r as BenchmarkDatasetItem, s as BenchmarkScenario, t as BENCHMARK_SPLIT_SEED, u as deterministicSplit } from "../types-C4bSVIr7.js";
+import { Tt as CampaignStorage } from "../provenance-AACfvbUR.js";
+import "../index-CM-e2LiV.js";
 //#region src/benchmarks/calibration.d.ts
 interface BenchmarkMetricCalibrationOptions<TPayload = unknown, TArtifact = string> {
   adapter: BenchmarkAdapter<BenchmarkDatasetItem<TPayload>, TPayload, TArtifact>;

package/dist/benchmarks/index.js

@@ -1,7 +1,7 @@
 import { t as __exportAll } from "../rolldown-runtime-8H4AJuhK.js";
-import { K as runCampaign } from "../llm-judge-DuYa4SEA.js";
+import { K as runCampaign } from "../llm-judge-DhiJqSjB.js";
 import { x as fsCampaignStorage } from "../external-optimizer-subprocess-BhKYK0Jv.js";
-import "../campaign-jTOvqnse.js";
+import "../campaign-Dg2B4W6d.js";
 import { join } from "node:path";
 //#region src/benchmarks/calibration.ts
 async function calibrateBenchmarkMetric(options) {

package/dist/campaign/index.d.ts

@@ -1,11 +1,11 @@
 import { c as CostLedgerHandle, w as PendingCostCallView } from "../cost-ledger-DbQdN3nO.js";
-import { _ as ProposalFinding, v as ProposalFindingOrigin, x as makeProposalFinding } from "../types-DLQx4mKU.js";
+import { _ as ProposalFinding, v as ProposalFindingOrigin, x as makeProposalFinding } from "../types-vXyshMwx.js";
 import { a as ExternalOptimizerEndpointFormat, c as ExternalOptimizerModelBudget, d as ExternalOptimizerModelCallResult, f as ExternalOptimizerModelExecutionObservation, g as ExternalTextCandidate, h as ExternalOptimizerRunnerCommand, i as ExternalOptimizerChatRequest, l as ExternalOptimizerModelCall, n as DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, o as ExternalOptimizerEvaluationObservation, p as ExternalOptimizerProcessLimits, r as ExternalOptimizerCallbackLimits, s as ExternalOptimizerEvaluationRefusalReason, t as DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, u as ExternalOptimizerModelCallRequest, v as resolveExternalOptimizerCallbackLimits, y as resolveExternalOptimizerProcessLimits } from "../external-optimizer-contracts-lixrOZdX.js";
-import { A as LabeledScenarioWrite, B as ScoredSurfaceOutcome, C as JudgeDimension, D as LabeledScenarioSampleArgs, E as LabeledScenarioRecord, F as ProposeContext, G as labelTrustRank, H as SurfaceProposer, I as ProposedCandidate, L as RedactionStatus, M as OptimizerConfig, N as ParetoParent, O as LabeledScenarioSource, P as ProposalTrackContext, R as Scenario, S as JudgeConfig, T as LabelTrust, U as TraceSpan, V as SessionScript, W as isProposedCandidate, _ as GateDecision, a as CampaignResult, b as GenerationRecord, c as CampaignTraceWriter, d as DispatchContext, f as DispatchFn, g as GateContribution, h as GateContext, i as CampaignCostMeter, j as MutableSurface, k as LabeledScenarioStore, l as CodeSurface, m as GateCheckStatus, n as CampaignArtifactWriter, o as CampaignScenarioIdentity, p as Gate, r as CampaignCellResult, s as CampaignTokenUsage, t as CampaignAggregates, u as ComponentSurface, v as GateResult, w as JudgeScore, x as JudgeAggregate, y as GenerationCandidate, z as ScenarioAggregate } from "../types-BxLccGMf.js";
-import { $ as OptimizationTokenUsage, At as llmJudge, C as RunOptimizationOptions, Ct as fsCampaignStorage, D as runEval, Dt as openAutoPr, E as RunEvalOptions, Et as OpenAutoPrResult, G as OptimizationMethodComparison, H as CompareOptimizationMethodsOptions, J as OptimizationMethodProvenance, K as OptimizationMethodInput, Ot as LlmJudgeDimension, Q as OptimizationPackageSource, S as PremeasuredOptimizationBaseline, St as createRunCostLedger, T as runOptimization, Tt as OpenAutoPrOptions, U as ComparisonCost, W as OptimizationMethod, X as OptimizationMethodRunOptions, Y as OptimizationMethodResult, Z as OptimizationMethodScore, _ as provenanceSpansPath, _t as ExternalOptimizerObservationArtifact, a as LoopProvenanceBackend, at as RunCampaignOptions, b as RunImprovementLoopResult, bt as readExternalOptimizerObservationArtifact, c as LoopProvenanceOptimizationMethod, ct as CampaignRunPlanCell, d as campaignMeasurementDigest, dt as GepaCandidatePopulationArtifact, et as combineComparisonCosts, f as canonicalDigest, ft as GepaCandidatePopulationCandidate, g as provenanceRecordPath, gt as ExternalOptimizerExecutionSummary, h as loopProvenanceSpans, ht as readGepaCandidatePopulationArtifact, i as LoopProvenanceArgsFromResult, it as CampaignCellFailureReceipt, kt as LlmJudgeOptions, l as LoopProvenanceRecord, lt as PlanCampaignRunOptions, m as loopProvenanceArgsFromResult, mt as GepaCandidateSelectionScore, n as EmitLoopProvenanceArgs, nt as costFromLedgerSummary, o as LoopProvenanceCandidate, ot as runCampaign, p as emitLoopProvenance, pt as GepaCandidatePopulationSummary, q as OptimizationMethodPairwise, r as EmitLoopProvenanceResult, rt as optimizationTokenUsageFromSummary, s as LoopProvenanceEvidence, st as CampaignRunPlan, t as BuildLoopProvenanceArgs, tt as compareOptimizationMethods, u as buildLoopProvenanceRecord, ut as planCampaignRun, v as verifyLoopProvenanceRecord, vt as ExternalOptimizerObservationSummary, w as RunOptimizationResult, wt as inMemoryCampaignStorage, x as runImprovementLoop, xt as CampaignStorage, y as RunImprovementLoopOptions, yt as ExternalOptimizerSubmittedCandidate } from "../provenance-8L-_4xiL.js";
-import { A as campaignScenarioIdentity, C as ExternalTextOptimizerContext, D as decodeExternalTextCandidate, E as ExternalTextEvaluationResponse, I as ReferenceEquivalenceJudgeOptions, M as campaignSplitDigestFromIdentities, O as assertCampaignDesign, R as ReferenceEquivalenceScenario, S as ExternalTextOptimizationMethodConfig, T as ExternalOptimizationExample, _ as DefaultProductionGateOptions, a as GepaAdaptiveEngineRun, b as composeGate, c as GepaOptimizationMethodConfig, d as gepaOptimizationMethod, f as OpenAICompatibleOptimizerModel, g as DefaultProductionGateCheck, h as heldOutGate, i as skillOptOptimizationMethod, j as campaignSplitDigest, k as assertCampaignSplitIdentity, l as GepaOptimizationRecipe, m as HeldOutGateOptions, n as SkillOptRunnerCommand, o as GepaEngineOptions, p as OptimizerModelBudget, r as SkillOptTrainerConfig, s as GepaEngineRun, t as SkillOptOptimizationMethodConfig, u as GepaRunnerCommand, v as DefaultProductionRewardHackingOptions, w as ExternalTextOptimizerResult, x as externalTextOptimizationMethod, y as defaultProductionGate, z as createReferenceEquivalenceJudge } from "../skillopt-optimization-method-CFDSzD-7.js";
-import { $ as SearchPlannedEvent, $n as CrossSurfacePairEvidence, $t as RunProfileMatrixResult, A as SearchAccountingAudit, An as CrossSurfaceAdditionRejectionReason, At as ProfileMatrixSegmentResult, B as SearchCostAccounting, Bn as CrossSurfaceComponentEvidence, Bt as ScoreboardRow, C as surfaceHash, Cn as discoverEvalFixtures, Ct as tangleTracesRoot, D as FileSearchLedger, Dn as analyzeCrossSurfaceInteractions, Dt as ProfileMatrixCoverage, E as acquireSingleRunLock, En as planEvalFixtureRun, Et as FinalizedProfileMatrixResult, F as SearchCandidateRegisteredEvent, Fn as CrossSurfaceCandidateComparison, Ft as runProfileMatrixSegment, G as SearchLedgerEvent, Gn as CrossSurfaceIneligibilityReason, Gt as renderScoreboardMarkdown, H as SearchLedger, Hn as CrossSurfaceDistribution, Ht as UserStory, I as SearchCandidateSlot, In as CrossSurfaceCandidateEvidence, It as PlaybackContext, J as SearchLedgerTrustedHeadMode, Jn as CrossSurfaceInteractionPath, Jt as userStoryScoreboard, K as SearchLedgerHash, Kn as CrossSurfaceInteractionAwareSelection, Kt as scoreUserStory, L as SearchCandidateSlotClosedEvent, Ln as CrossSurfaceCandidateOutcome, Lt as PlaybackDriver, M as SearchAttemptAccounting, Mn as CrossSurfaceBestSingleSelection, Mt as SegmentedProfileMatrixError, N as SearchCandidateDecidedEvent, Nn as CrossSurfaceBootstrapPolicy, Nt as createProfileMatrixPlan, O as OpenSearchLedgerOptions, On as AnalyzeCrossSurfaceInteractionsInput, Ot as ProfileMatrixPlan, P as SearchCandidateLineage, Pn as CrossSurfaceCandidate, Pt as finalizeProfileMatrix, Q as SearchPlan, Qn as CrossSurfacePairCompatibility, Qt as RunProfileMatrixOptions, R as SearchCandidateSurface, Rn as CrossSurfaceCandidateSummary, Rt as PlaybackStep, S as surfaceContentHash, Sn as PlanEvalFixtureRunOptions, St as resolveRunDir, T as SingleRunLockOptions, Tn as loadEvalFixtureScenarios, Tt as FinalizeProfileMatrixOptions, U as SearchLedgerAppendResult, Un as CrossSurfaceEligibility, Ut as UserStoryVerdict, V as SearchFailureReason, Vn as CrossSurfaceCompositionStep, Vt as ScoreboardSummary, W as SearchLedgerEntry, Wn as CrossSurfaceEvidenceBreakdown, Wt as makePlaybackDispatch, X as SearchOperationKind, Xn as CrossSurfaceInteractionTask, Xt as ProfileMatrixError, Y as SearchModelIdentity, Yn as CrossSurfaceInteractionReport, Yt as ProfileDispatchFn, Z as SearchOperationRecordedEvent, Zn as CrossSurfaceNaiveStackSelection, Zt as ProfileSummary, _ as assertCodeSurfaceIdentity, _n as EvalFixtureLoadOptions, _t as compareRankKeys, a as WorktreeAdapterError, an as LabeledScenarioStoreError, ar as CrossSurfaceSelections, at as SearchSurfaceKind, b as componentSurfaceIdentityMaterial, bn as EvalFixtureValidationMode, bt as scoreDiscrimination, c as verifyCodeSurface, cn as RolloutCall, cr as TraceAnalystArtifact, ct as SearchTokenAccounting, d as PhoenixEvaluationResultLike, dn as classifyUngroundedLiterals, dr as traceAnalystQualityJudge, dt as SearchLedgerConflictError, en as ScenarioRollup, er as CrossSurfacePairIncompatibilityReason, et as SearchPlannedOperation, f as PhoenixEvaluatorLike, fn as rolloutArgumentDiff, ft as SearchLedgerError, g as isTransientTransportFailure, gn as EvalFixtureFile, gt as campaignMeanComposite, h as TransientFailureOptions, hn as EvalFixture, ht as campaignBreakdown, i as WorktreeAdapter, in as FsLabeledScenarioStoreOptions, ir as CrossSurfaceSelectionPolicy, it as SearchSurfaceEvidence, j as SearchArtifactRef, jn as CrossSurfaceAttemptCompleteness, jt as RunProfileMatrixSegmentOptions, k as SEARCH_LEDGER_SCHEMA, kn as CrossSurfaceAdditionDecision, kt as ProfileMatrixRow, l as AutoevalsScoreLike, ln as ScoredRollout, lr as TraceAnalystScenario, lt as openSearchLedger, m as phoenixEvaluatorJudge, mn as neutralizationGate, mt as CampaignBreakdown, n as GitWorktreeAdapterOptions, nn as neutralizeText, nr as CrossSurfaceRankedSingle, nt as SearchSourceRef, o as gitWorktreeAdapter, on as RolloutArgumentDiff, or as CrossSurfaceTaskRow, ot as SearchTaskAttemptedEvent, p as autoevalsScorerJudge, pn as NeutralizationGateOptions, pt as SearchLedgerIntegrityError, q as SearchLedgerReplay, qn as CrossSurfaceInteractionEffect, qt as scoreboardSummary, r as Worktree, rn as FsLabeledScenarioStore, rr as CrossSurfaceRelativeCost, rt as SearchSurfaceEffect, s as resolveWorktreePath, sn as RolloutArgumentDiffOptions, sr as BuildTraceAnalystSurfaceDispatchOptions, st as SearchTaskOutcome, t as CodeSurfaceVerification, tn as runProfileMatrix, tr as CrossSurfacePairwiseEntry, tt as SearchPlannedTask, u as AutoevalsScorerLike, un as UngroundedLiteralReport, ur as buildTraceAnalystSurfaceDispatch, ut as validateSearchLedgerEvent, v as assertComponentSurface, vn as EvalFixtureRunPlan, vt as DiscriminationScore, w as SingleRunLock, wn as loadEvalFixture, wt as CreateProfileMatrixPlanOptions, x as renderSurfaceDiff, xn as LoadEvalFixtureScenariosOptions, xt as selectDiscriminative, y as codeSurfaceIdentityMaterial, yn as EvalFixtureScenario, yt as ScenarioSignal, z as SearchCompletedEvent, zn as CrossSurfaceComponent, zt as ScoreboardRenderOptions } from "../index-DF3ynqmJ.js";
+import { A as LabeledScenarioWrite, B as ScoredSurfaceOutcome, C as JudgeDimension, D as LabeledScenarioSampleArgs, E as LabeledScenarioRecord, F as ProposeContext, G as labelTrustRank, H as SurfaceProposer, I as ProposedCandidate, L as RedactionStatus, M as OptimizerConfig, N as ParetoParent, O as LabeledScenarioSource, P as ProposalTrackContext, R as Scenario, S as JudgeConfig, T as LabelTrust, U as TraceSpan, V as SessionScript, W as isProposedCandidate, _ as GateDecision, a as CampaignResult, b as GenerationRecord, c as CampaignTraceWriter, d as DispatchContext, f as DispatchFn, g as GateContribution, h as GateContext, i as CampaignCostMeter, j as MutableSurface, k as LabeledScenarioStore, l as CodeSurface, m as GateCheckStatus, n as CampaignArtifactWriter, o as CampaignScenarioIdentity, p as Gate, r as CampaignCellResult, s as CampaignTokenUsage, t as CampaignAggregates, u as ComponentSurface, v as GateResult, w as JudgeScore, x as JudgeAggregate, y as GenerationCandidate, z as ScenarioAggregate } from "../types-BjsNDR49.js";
+import { $ as OptimizationTokenUsage, At as OpenAutoPrResult, C as RunOptimizationOptions, Ct as PlanCampaignRunOptions, D as runEval, Dt as fsCampaignStorage, E as RunEvalOptions, Et as createRunCostLedger, G as OptimizationMethodComparison, H as CompareOptimizationMethodsOptions, J as OptimizationMethodProvenance, K as OptimizationMethodInput, Mt as LlmJudgeDimension, Nt as LlmJudgeOptions, Ot as inMemoryCampaignStorage, Pt as llmJudge, Q as OptimizationPackageSource, S as PremeasuredOptimizationBaseline, St as CampaignRunPlanCell, T as runOptimization, Tt as CampaignStorage, U as ComparisonCost, W as OptimizationMethod, X as OptimizationMethodRunOptions, Y as OptimizationMethodResult, Z as OptimizationMethodScore, _ as provenanceSpansPath, _t as cellCachePath, a as LoopProvenanceBackend, at as GepaCandidatePopulationCandidate, b as RunImprovementLoopResult, bt as runCampaign, c as LoopProvenanceOptimizationMethod, ct as readGepaCandidatePopulationArtifact, d as campaignMeasurementDigest, dt as ExternalOptimizerObservationSummary, et as combineComparisonCosts, f as canonicalDigest, ft as ExternalOptimizerSubmittedCandidate, g as provenanceRecordPath, gt as readCachedCell, h as loopProvenanceSpans, ht as CacheRead, i as LoopProvenanceArgsFromResult, it as GepaCandidatePopulationArtifact, jt as openAutoPr, kt as OpenAutoPrOptions, l as LoopProvenanceRecord, lt as ExternalOptimizerExecutionSummary, m as loopProvenanceArgsFromResult, mt as CacheIssueReason, n as EmitLoopProvenanceArgs, nt as costFromLedgerSummary, o as LoopProvenanceCandidate, ot as GepaCandidatePopulationSummary, p as emitLoopProvenance, pt as readExternalOptimizerObservationArtifact, q as OptimizationMethodPairwise, r as EmitLoopProvenanceResult, rt as optimizationTokenUsageFromSummary, s as LoopProvenanceEvidence, st as GepaCandidateSelectionScore, t as BuildLoopProvenanceArgs, tt as compareOptimizationMethods, u as buildLoopProvenanceRecord, ut as ExternalOptimizerObservationArtifact, v as verifyLoopProvenanceRecord, vt as CampaignCellFailureReceipt, w as RunOptimizationResult, wt as planCampaignRun, x as runImprovementLoop, xt as CampaignRunPlan, y as RunImprovementLoopOptions, yt as RunCampaignOptions } from "../provenance-AACfvbUR.js";
+import { A as campaignScenarioIdentity, C as ExternalTextOptimizerContext, D as decodeExternalTextCandidate, E as ExternalTextEvaluationResponse, I as ReferenceEquivalenceJudgeOptions, M as campaignSplitDigestFromIdentities, O as assertCampaignDesign, R as ReferenceEquivalenceScenario, S as ExternalTextOptimizationMethodConfig, T as ExternalOptimizationExample, _ as DefaultProductionGateOptions, a as GepaAdaptiveEngineRun, b as composeGate, c as GepaOptimizationMethodConfig, d as gepaOptimizationMethod, f as OpenAICompatibleOptimizerModel, g as DefaultProductionGateCheck, h as heldOutGate, i as skillOptOptimizationMethod, j as campaignSplitDigest, k as assertCampaignSplitIdentity, l as GepaOptimizationRecipe, m as HeldOutGateOptions, n as SkillOptRunnerCommand, o as GepaEngineOptions, p as OptimizerModelBudget, r as SkillOptTrainerConfig, s as GepaEngineRun, t as SkillOptOptimizationMethodConfig, u as GepaRunnerCommand, v as DefaultProductionRewardHackingOptions, w as ExternalTextOptimizerResult, x as externalTextOptimizationMethod, y as defaultProductionGate, z as createReferenceEquivalenceJudge } from "../skillopt-optimization-method-DO6OFgcz.js";
+import { $ as SearchPlannedEvent, $n as CrossSurfacePairEvidence, $t as RunProfileMatrixResult, A as SearchAccountingAudit, An as CrossSurfaceAdditionRejectionReason, At as ProfileMatrixSegmentResult, B as SearchCostAccounting, Bn as CrossSurfaceComponentEvidence, Bt as ScoreboardRow, C as surfaceHash, Cn as discoverEvalFixtures, Ct as tangleTracesRoot, D as FileSearchLedger, Dn as analyzeCrossSurfaceInteractions, Dt as ProfileMatrixCoverage, E as acquireSingleRunLock, En as planEvalFixtureRun, Et as FinalizedProfileMatrixResult, F as SearchCandidateRegisteredEvent, Fn as CrossSurfaceCandidateComparison, Ft as runProfileMatrixSegment, G as SearchLedgerEvent, Gn as CrossSurfaceIneligibilityReason, Gt as renderScoreboardMarkdown, H as SearchLedger, Hn as CrossSurfaceDistribution, Ht as UserStory, I as SearchCandidateSlot, In as CrossSurfaceCandidateEvidence, It as PlaybackContext, J as SearchLedgerTrustedHeadMode, Jn as CrossSurfaceInteractionPath, Jt as userStoryScoreboard, K as SearchLedgerHash, Kn as CrossSurfaceInteractionAwareSelection, Kt as scoreUserStory, L as SearchCandidateSlotClosedEvent, Ln as CrossSurfaceCandidateOutcome, Lt as PlaybackDriver, M as SearchAttemptAccounting, Mn as CrossSurfaceBestSingleSelection, Mt as SegmentedProfileMatrixError, N as SearchCandidateDecidedEvent, Nn as CrossSurfaceBootstrapPolicy, Nt as createProfileMatrixPlan, O as OpenSearchLedgerOptions, On as AnalyzeCrossSurfaceInteractionsInput, Ot as ProfileMatrixPlan, P as SearchCandidateLineage, Pn as CrossSurfaceCandidate, Pt as finalizeProfileMatrix, Q as SearchPlan, Qn as CrossSurfacePairCompatibility, Qt as RunProfileMatrixOptions, R as SearchCandidateSurface, Rn as CrossSurfaceCandidateSummary, Rt as PlaybackStep, S as surfaceContentHash, Sn as PlanEvalFixtureRunOptions, St as resolveRunDir, T as SingleRunLockOptions, Tn as loadEvalFixtureScenarios, Tt as FinalizeProfileMatrixOptions, U as SearchLedgerAppendResult, Un as CrossSurfaceEligibility, Ut as UserStoryVerdict, V as SearchFailureReason, Vn as CrossSurfaceCompositionStep, Vt as ScoreboardSummary, W as SearchLedgerEntry, Wn as CrossSurfaceEvidenceBreakdown, Wt as makePlaybackDispatch, X as SearchOperationKind, Xn as CrossSurfaceInteractionTask, Xt as ProfileMatrixError, Y as SearchModelIdentity, Yn as CrossSurfaceInteractionReport, Yt as ProfileDispatchFn, Z as SearchOperationRecordedEvent, Zn as CrossSurfaceNaiveStackSelection, Zt as ProfileSummary, _ as assertCodeSurfaceIdentity, _n as EvalFixtureLoadOptions, _t as compareRankKeys, a as WorktreeAdapterError, an as LabeledScenarioStoreError, ar as CrossSurfaceSelections, at as SearchSurfaceKind, b as componentSurfaceIdentityMaterial, bn as EvalFixtureValidationMode, bt as scoreDiscrimination, c as verifyCodeSurface, cn as RolloutCall, cr as TraceAnalystArtifact, ct as SearchTokenAccounting, d as PhoenixEvaluationResultLike, dn as classifyUngroundedLiterals, dr as traceAnalystQualityJudge, dt as SearchLedgerConflictError, en as ScenarioRollup, er as CrossSurfacePairIncompatibilityReason, et as SearchPlannedOperation, f as PhoenixEvaluatorLike, fn as rolloutArgumentDiff, ft as SearchLedgerError, g as isTransientTransportFailure, gn as EvalFixtureFile, gt as campaignMeanComposite, h as TransientFailureOptions, hn as EvalFixture, ht as campaignBreakdown, i as WorktreeAdapter, in as FsLabeledScenarioStoreOptions, ir as CrossSurfaceSelectionPolicy, it as SearchSurfaceEvidence, j as SearchArtifactRef, jn as CrossSurfaceAttemptCompleteness, jt as RunProfileMatrixSegmentOptions, k as SEARCH_LEDGER_SCHEMA, kn as CrossSurfaceAdditionDecision, kt as ProfileMatrixRow, l as AutoevalsScoreLike, ln as ScoredRollout, lr as TraceAnalystScenario, lt as openSearchLedger, m as phoenixEvaluatorJudge, mn as neutralizationGate, mt as CampaignBreakdown, n as GitWorktreeAdapterOptions, nn as neutralizeText, nr as CrossSurfaceRankedSingle, nt as SearchSourceRef, o as gitWorktreeAdapter, on as RolloutArgumentDiff, or as CrossSurfaceTaskRow, ot as SearchTaskAttemptedEvent, p as autoevalsScorerJudge, pn as NeutralizationGateOptions, pt as SearchLedgerIntegrityError, q as SearchLedgerReplay, qn as CrossSurfaceInteractionEffect, qt as scoreboardSummary, r as Worktree, rn as FsLabeledScenarioStore, rr as CrossSurfaceRelativeCost, rt as SearchSurfaceEffect, s as resolveWorktreePath, sn as RolloutArgumentDiffOptions, sr as BuildTraceAnalystSurfaceDispatchOptions, st as SearchTaskOutcome, t as CodeSurfaceVerification, tn as runProfileMatrix, tr as CrossSurfacePairwiseEntry, tt as SearchPlannedTask, u as AutoevalsScorerLike, un as UngroundedLiteralReport, ur as buildTraceAnalystSurfaceDispatch, ut as validateSearchLedgerEvent, v as assertComponentSurface, vn as EvalFixtureRunPlan, vt as DiscriminationScore, w as SingleRunLock, wn as loadEvalFixture, wt as CreateProfileMatrixPlanOptions, x as renderSurfaceDiff, xn as LoadEvalFixtureScenariosOptions, xt as selectDiscriminative, y as codeSurfaceIdentityMaterial, yn as EvalFixtureScenario, yt as ScenarioSignal, z as SearchCompletedEvent, zn as CrossSurfaceComponent, zt as ScoreboardRenderOptions } from "../index-CM-e2LiV.js";
 import { c as powerPreflight, o as PowerPreflight, s as PowerPreflightOptions } from "../pareto-BqNW3LJR.js";
-import { a as ObjectiveSource, c as PromotionPolicy, d as paretoSignificanceGate, i as EvidenceVector, l as buildEvidenceVector, n as AxisVerdict, o as ParetoSignificanceGateOptions, r as BuildEvidenceVectorOptions, s as PromotionObjective, t as AxisEvidence, u as paretoPolicy } from "../promotion-policy-CsMZJOB-.js";
-import { a as detectScale, c as pairHoldout, d as SequentialDecision, f as SequentialObservation, g as sequentialPairedGate, h as sequentialDecide, i as PairedHoldout, l as SequentialDecideFn, m as SequentialPairedGateOptions, n as HeldoutSignificance, o as dimensionRegressions, p as SequentialPairedGate, r as HeldoutSignificanceOptions, s as heldoutSignificance, t as DimensionRegression, u as SequentialDecideOptions } from "../statistical-heldout-C4De2tRI.js";
-export { type AnalyzeCrossSurfaceInteractionsInput, type AutoevalsScoreLike, type AutoevalsScorerLike, type AxisEvidence, type AxisVerdict, type BuildEvidenceVectorOptions, type BuildLoopProvenanceArgs, type BuildTraceAnalystSurfaceDispatchOptions, type CampaignAggregates, type CampaignArtifactWriter, type CampaignBreakdown, type CampaignCellFailureReceipt, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignRunPlan, type CampaignRunPlanCell, type CampaignScenarioIdentity, type CampaignStorage, type CampaignTokenUsage, type CampaignTraceWriter, type CodeSurface, type CodeSurfaceVerification, type CompareOptimizationMethodsOptions, type ComparisonCost, type ComponentSurface, type CostLedgerHandle, type CreateProfileMatrixPlanOptions, type CrossSurfaceAdditionDecision, type CrossSurfaceAdditionRejectionReason, type CrossSurfaceAttemptCompleteness, type CrossSurfaceBestSingleSelection, type CrossSurfaceBootstrapPolicy, type CrossSurfaceCandidate, type CrossSurfaceCandidateComparison, type CrossSurfaceCandidateEvidence, type CrossSurfaceCandidateOutcome, type CrossSurfaceCandidateSummary, type CrossSurfaceComponent, type CrossSurfaceComponentEvidence, type CrossSurfaceCompositionStep, type CrossSurfaceDistribution, type CrossSurfaceEligibility, type CrossSurfaceEvidenceBreakdown, type CrossSurfaceIneligibilityReason, type CrossSurfaceInteractionAwareSelection, type CrossSurfaceInteractionEffect, type CrossSurfaceInteractionPath, type CrossSurfaceInteractionReport, type CrossSurfaceInteractionTask, type CrossSurfaceNaiveStackSelection, type CrossSurfacePairCompatibility, type CrossSurfacePairEvidence, type CrossSurfacePairIncompatibilityReason, type CrossSurfacePairwiseEntry, type CrossSurfaceRankedSingle, type CrossSurfaceRelativeCost, type CrossSurfaceSelectionPolicy, type CrossSurfaceSelections, type CrossSurfaceTaskRow, DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, type DefaultProductionGateCheck, type DefaultProductionGateOptions, type DefaultProductionRewardHackingOptions, type DimensionRegression, type DiscriminationScore, type DispatchContext, type DispatchFn, type EmitLoopProvenanceArgs, type EmitLoopProvenanceResult, type EvalFixture, type EvalFixtureFile, type EvalFixtureLoadOptions, type EvalFixtureRunPlan, type EvalFixtureScenario, type EvalFixtureValidationMode, type EvidenceVector, type ExternalOptimizationExample, type ExternalOptimizerCallbackLimits, type ExternalOptimizerChatRequest, type ExternalOptimizerEndpointFormat, type ExternalOptimizerEvaluationObservation, type ExternalOptimizerEvaluationRefusalReason, type ExternalOptimizerExecutionSummary, type ExternalOptimizerModelBudget, type ExternalOptimizerModelCall, type ExternalOptimizerModelCallRequest, type ExternalOptimizerModelCallResult, type ExternalOptimizerModelExecutionObservation, type ExternalOptimizerObservationArtifact, type ExternalOptimizerObservationSummary, type ExternalOptimizerProcessLimits, type ExternalOptimizerRunnerCommand, type ExternalOptimizerSubmittedCandidate, type ExternalTextCandidate, type ExternalTextEvaluationResponse, type ExternalTextOptimizationMethodConfig, type ExternalTextOptimizerContext, type ExternalTextOptimizerResult, FileSearchLedger, type FinalizeProfileMatrixOptions, type FinalizedProfileMatrixResult, FsLabeledScenarioStore, type FsLabeledScenarioStoreOptions, type Gate, type GateCheckStatus, type GateContext, type GateContribution, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaAdaptiveEngineRun, type GepaCandidatePopulationArtifact, type GepaCandidatePopulationCandidate, type GepaCandidatePopulationSummary, type GepaCandidateSelectionScore, type GepaEngineOptions, type GepaEngineRun, type GepaOptimizationMethodConfig, type GepaOptimizationRecipe, type GepaRunnerCommand, type GitWorktreeAdapterOptions, type HeldOutGateOptions, type HeldoutSignificance, type HeldoutSignificanceOptions, type JudgeAggregate, type JudgeConfig, type JudgeDimension, type JudgeScore, type LabelTrust, type LabeledScenarioRecord, type LabeledScenarioSampleArgs, type LabeledScenarioSource, type LabeledScenarioStore, LabeledScenarioStoreError, type LabeledScenarioWrite, type LlmJudgeDimension, type LlmJudgeOptions, type LoadEvalFixtureScenariosOptions, type LoopProvenanceArgsFromResult, type LoopProvenanceBackend, type LoopProvenanceCandidate, type LoopProvenanceEvidence, type LoopProvenanceOptimizationMethod, type LoopProvenanceRecord, type MutableSurface, type NeutralizationGateOptions, type ObjectiveSource, type OpenAICompatibleOptimizerModel, type OpenAutoPrOptions, type OpenAutoPrResult, type OpenSearchLedgerOptions, type OptimizationMethod, type OptimizationMethodComparison, type OptimizationMethodInput, type OptimizationMethodPairwise, type OptimizationMethodProvenance, type OptimizationMethodResult, type OptimizationMethodRunOptions, type OptimizationMethodScore, type OptimizationPackageSource, type OptimizationTokenUsage, type OptimizerConfig, type OptimizerModelBudget, type PairedHoldout, type ParetoParent, type ParetoSignificanceGateOptions, type PendingCostCallView, type PhoenixEvaluationResultLike, type PhoenixEvaluatorLike, type PlanCampaignRunOptions, type PlanEvalFixtureRunOptions, type PlaybackContext, type PlaybackDriver, type PlaybackStep, type PowerPreflight, type PowerPreflightOptions, type PremeasuredOptimizationBaseline, type ProfileDispatchFn, type ProfileMatrixCoverage, ProfileMatrixError, type ProfileMatrixPlan, type ProfileMatrixRow, type ProfileMatrixSegmentResult, type ProfileSummary, type PromotionObjective, type PromotionPolicy, type ProposalFinding, type ProposalFindingOrigin, type ProposalTrackContext, type ProposeContext, type ProposedCandidate, type RedactionStatus, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceScenario, type RolloutArgumentDiff, type RolloutArgumentDiffOptions, type RolloutCall, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunOptimizationOptions, type RunOptimizationResult, type RunProfileMatrixOptions, type RunProfileMatrixResult, type RunProfileMatrixSegmentOptions, SEARCH_LEDGER_SCHEMA, type Scenario, type ScenarioAggregate, type ScenarioRollup, type ScenarioSignal, type ScoreboardRenderOptions, type ScoreboardRow, type ScoreboardSummary, type ScoredRollout, type ScoredSurfaceOutcome, type SearchAccountingAudit, type SearchArtifactRef, type SearchAttemptAccounting, type SearchCandidateDecidedEvent, type SearchCandidateLineage, type SearchCandidateRegisteredEvent, type SearchCandidateSlot, type SearchCandidateSlotClosedEvent, type SearchCandidateSurface, type SearchCompletedEvent, type SearchCostAccounting, type SearchFailureReason, type SearchLedger, type SearchLedgerAppendResult, SearchLedgerConflictError, type SearchLedgerEntry, SearchLedgerError, type SearchLedgerEvent, type SearchLedgerHash, SearchLedgerIntegrityError, type SearchLedgerReplay, type SearchLedgerTrustedHeadMode, type SearchModelIdentity, type SearchOperationKind, type SearchOperationRecordedEvent, type SearchPlan, type SearchPlannedEvent, type SearchPlannedOperation, type SearchPlannedTask, type SearchSourceRef, type SearchSurfaceEffect, type SearchSurfaceEvidence, type SearchSurfaceKind, type SearchTaskAttemptedEvent, type SearchTaskOutcome, type SearchTokenAccounting, SegmentedProfileMatrixError, type SequentialDecideFn, type SequentialDecideOptions, type SequentialDecision, type SequentialObservation, type SequentialPairedGate, type SequentialPairedGateOptions, type SessionScript, type SingleRunLock, type SingleRunLockOptions, type SkillOptOptimizationMethodConfig, type SkillOptRunnerCommand, type SkillOptTrainerConfig, type SurfaceProposer, type TraceAnalystArtifact, type TraceAnalystScenario, type TraceSpan, type TransientFailureOptions, type UngroundedLiteralReport, type UserStory, type UserStoryVerdict, type Worktree, type WorktreeAdapter, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, autoevalsScorerJudge, buildEvidenceVector, buildLoopProvenanceRecord, buildTraceAnalystSurfaceDispatch, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, combineComparisonCosts, compareOptimizationMethods, compareRankKeys, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createProfileMatrixPlan, createReferenceEquivalenceJudge, createRunCostLedger, decodeExternalTextCandidate, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, finalizeProfileMatrix, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, makeProposalFinding, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, phoenixEvaluatorJudge, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, readExternalOptimizerObservationArtifact, readGepaCandidatePopulationArtifact, renderScoreboardMarkdown, renderSurfaceDiff, resolveExternalOptimizerCallbackLimits, resolveExternalOptimizerProcessLimits, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, runProfileMatrixSegment, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, traceAnalystQualityJudge, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };
+import { a as ObjectiveSource, c as PromotionPolicy, d as paretoSignificanceGate, i as EvidenceVector, l as buildEvidenceVector, n as AxisVerdict, o as ParetoSignificanceGateOptions, r as BuildEvidenceVectorOptions, s as PromotionObjective, t as AxisEvidence, u as paretoPolicy } from "../promotion-policy-u3wj6w3U.js";
+import { a as detectScale, c as pairHoldout, d as SequentialDecision, f as SequentialObservation, g as sequentialPairedGate, h as sequentialDecide, i as PairedHoldout, l as SequentialDecideFn, m as SequentialPairedGateOptions, n as HeldoutSignificance, o as dimensionRegressions, p as SequentialPairedGate, r as HeldoutSignificanceOptions, s as heldoutSignificance, t as DimensionRegression, u as SequentialDecideOptions } from "../statistical-heldout-CBGPSZ5X.js";
+export { type AnalyzeCrossSurfaceInteractionsInput, type AutoevalsScoreLike, type AutoevalsScorerLike, type AxisEvidence, type AxisVerdict, type BuildEvidenceVectorOptions, type BuildLoopProvenanceArgs, type BuildTraceAnalystSurfaceDispatchOptions, type CacheIssueReason, type CacheRead, type CampaignAggregates, type CampaignArtifactWriter, type CampaignBreakdown, type CampaignCellFailureReceipt, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignRunPlan, type CampaignRunPlanCell, type CampaignScenarioIdentity, type CampaignStorage, type CampaignTokenUsage, type CampaignTraceWriter, type CodeSurface, type CodeSurfaceVerification, type CompareOptimizationMethodsOptions, type ComparisonCost, type ComponentSurface, type CostLedgerHandle, type CreateProfileMatrixPlanOptions, type CrossSurfaceAdditionDecision, type CrossSurfaceAdditionRejectionReason, type CrossSurfaceAttemptCompleteness, type CrossSurfaceBestSingleSelection, type CrossSurfaceBootstrapPolicy, type CrossSurfaceCandidate, type CrossSurfaceCandidateComparison, type CrossSurfaceCandidateEvidence, type CrossSurfaceCandidateOutcome, type CrossSurfaceCandidateSummary, type CrossSurfaceComponent, type CrossSurfaceComponentEvidence, type CrossSurfaceCompositionStep, type CrossSurfaceDistribution, type CrossSurfaceEligibility, type CrossSurfaceEvidenceBreakdown, type CrossSurfaceIneligibilityReason, type CrossSurfaceInteractionAwareSelection, type CrossSurfaceInteractionEffect, type CrossSurfaceInteractionPath, type CrossSurfaceInteractionReport, type CrossSurfaceInteractionTask, type CrossSurfaceNaiveStackSelection, type CrossSurfacePairCompatibility, type CrossSurfacePairEvidence, type CrossSurfacePairIncompatibilityReason, type CrossSurfacePairwiseEntry, type CrossSurfaceRankedSingle, type CrossSurfaceRelativeCost, type CrossSurfaceSelectionPolicy, type CrossSurfaceSelections, type CrossSurfaceTaskRow, DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, type DefaultProductionGateCheck, type DefaultProductionGateOptions, type DefaultProductionRewardHackingOptions, type DimensionRegression, type DiscriminationScore, type DispatchContext, type DispatchFn, type EmitLoopProvenanceArgs, type EmitLoopProvenanceResult, type EvalFixture, type EvalFixtureFile, type EvalFixtureLoadOptions, type EvalFixtureRunPlan, type EvalFixtureScenario, type EvalFixtureValidationMode, type EvidenceVector, type ExternalOptimizationExample, type ExternalOptimizerCallbackLimits, type ExternalOptimizerChatRequest, type ExternalOptimizerEndpointFormat, type ExternalOptimizerEvaluationObservation, type ExternalOptimizerEvaluationRefusalReason, type ExternalOptimizerExecutionSummary, type ExternalOptimizerModelBudget, type ExternalOptimizerModelCall, type ExternalOptimizerModelCallRequest, type ExternalOptimizerModelCallResult, type ExternalOptimizerModelExecutionObservation, type ExternalOptimizerObservationArtifact, type ExternalOptimizerObservationSummary, type ExternalOptimizerProcessLimits, type ExternalOptimizerRunnerCommand, type ExternalOptimizerSubmittedCandidate, type ExternalTextCandidate, type ExternalTextEvaluationResponse, type ExternalTextOptimizationMethodConfig, type ExternalTextOptimizerContext, type ExternalTextOptimizerResult, FileSearchLedger, type FinalizeProfileMatrixOptions, type FinalizedProfileMatrixResult, FsLabeledScenarioStore, type FsLabeledScenarioStoreOptions, type Gate, type GateCheckStatus, type GateContext, type GateContribution, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaAdaptiveEngineRun, type GepaCandidatePopulationArtifact, type GepaCandidatePopulationCandidate, type GepaCandidatePopulationSummary, type GepaCandidateSelectionScore, type GepaEngineOptions, type GepaEngineRun, type GepaOptimizationMethodConfig, type GepaOptimizationRecipe, type GepaRunnerCommand, type GitWorktreeAdapterOptions, type HeldOutGateOptions, type HeldoutSignificance, type HeldoutSignificanceOptions, type JudgeAggregate, type JudgeConfig, type JudgeDimension, type JudgeScore, type LabelTrust, type LabeledScenarioRecord, type LabeledScenarioSampleArgs, type LabeledScenarioSource, type LabeledScenarioStore, LabeledScenarioStoreError, type LabeledScenarioWrite, type LlmJudgeDimension, type LlmJudgeOptions, type LoadEvalFixtureScenariosOptions, type LoopProvenanceArgsFromResult, type LoopProvenanceBackend, type LoopProvenanceCandidate, type LoopProvenanceEvidence, type LoopProvenanceOptimizationMethod, type LoopProvenanceRecord, type MutableSurface, type NeutralizationGateOptions, type ObjectiveSource, type OpenAICompatibleOptimizerModel, type OpenAutoPrOptions, type OpenAutoPrResult, type OpenSearchLedgerOptions, type OptimizationMethod, type OptimizationMethodComparison, type OptimizationMethodInput, type OptimizationMethodPairwise, type OptimizationMethodProvenance, type OptimizationMethodResult, type OptimizationMethodRunOptions, type OptimizationMethodScore, type OptimizationPackageSource, type OptimizationTokenUsage, type OptimizerConfig, type OptimizerModelBudget, type PairedHoldout, type ParetoParent, type ParetoSignificanceGateOptions, type PendingCostCallView, type PhoenixEvaluationResultLike, type PhoenixEvaluatorLike, type PlanCampaignRunOptions, type PlanEvalFixtureRunOptions, type PlaybackContext, type PlaybackDriver, type PlaybackStep, type PowerPreflight, type PowerPreflightOptions, type PremeasuredOptimizationBaseline, type ProfileDispatchFn, type ProfileMatrixCoverage, ProfileMatrixError, type ProfileMatrixPlan, type ProfileMatrixRow, type ProfileMatrixSegmentResult, type ProfileSummary, type PromotionObjective, type PromotionPolicy, type ProposalFinding, type ProposalFindingOrigin, type ProposalTrackContext, type ProposeContext, type ProposedCandidate, type RedactionStatus, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceScenario, type RolloutArgumentDiff, type RolloutArgumentDiffOptions, type RolloutCall, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunOptimizationOptions, type RunOptimizationResult, type RunProfileMatrixOptions, type RunProfileMatrixResult, type RunProfileMatrixSegmentOptions, SEARCH_LEDGER_SCHEMA, type Scenario, type ScenarioAggregate, type ScenarioRollup, type ScenarioSignal, type ScoreboardRenderOptions, type ScoreboardRow, type ScoreboardSummary, type ScoredRollout, type ScoredSurfaceOutcome, type SearchAccountingAudit, type SearchArtifactRef, type SearchAttemptAccounting, type SearchCandidateDecidedEvent, type SearchCandidateLineage, type SearchCandidateRegisteredEvent, type SearchCandidateSlot, type SearchCandidateSlotClosedEvent, type SearchCandidateSurface, type SearchCompletedEvent, type SearchCostAccounting, type SearchFailureReason, type SearchLedger, type SearchLedgerAppendResult, SearchLedgerConflictError, type SearchLedgerEntry, SearchLedgerError, type SearchLedgerEvent, type SearchLedgerHash, SearchLedgerIntegrityError, type SearchLedgerReplay, type SearchLedgerTrustedHeadMode, type SearchModelIdentity, type SearchOperationKind, type SearchOperationRecordedEvent, type SearchPlan, type SearchPlannedEvent, type SearchPlannedOperation, type SearchPlannedTask, type SearchSourceRef, type SearchSurfaceEffect, type SearchSurfaceEvidence, type SearchSurfaceKind, type SearchTaskAttemptedEvent, type SearchTaskOutcome, type SearchTokenAccounting, SegmentedProfileMatrixError, type SequentialDecideFn, type SequentialDecideOptions, type SequentialDecision, type SequentialObservation, type SequentialPairedGate, type SequentialPairedGateOptions, type SessionScript, type SingleRunLock, type SingleRunLockOptions, type SkillOptOptimizationMethodConfig, type SkillOptRunnerCommand, type SkillOptTrainerConfig, type SurfaceProposer, type TraceAnalystArtifact, type TraceAnalystScenario, type TraceSpan, type TransientFailureOptions, type UngroundedLiteralReport, type UserStory, type UserStoryVerdict, type Worktree, type WorktreeAdapter, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, autoevalsScorerJudge, buildEvidenceVector, buildLoopProvenanceRecord, buildTraceAnalystSurfaceDispatch, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, cellCachePath, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, combineComparisonCosts, compareOptimizationMethods, compareRankKeys, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createProfileMatrixPlan, createReferenceEquivalenceJudge, createRunCostLedger, decodeExternalTextCandidate, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, finalizeProfileMatrix, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, makeProposalFinding, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, phoenixEvaluatorJudge, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, readCachedCell, readExternalOptimizerObservationArtifact, readGepaCandidatePopulationArtifact, renderScoreboardMarkdown, renderSurfaceDiff, resolveExternalOptimizerCallbackLimits, resolveExternalOptimizerProcessLimits, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, runProfileMatrixSegment, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, traceAnalystQualityJudge, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };

package/dist/campaign/index.js

@@ -1,10 +1,10 @@
-import { $ as assertCampaignDesign, A as surfaceHash, C as optimizationTokenUsageFromSummary, D as componentSurfaceIdentityMaterial, E as codeSurfaceIdentityMaterial, G as runEval, J as resolveRunDir, K as runCampaign, M as campaignMeanComposite, N as compareRankKeys, O as renderSurfaceDiff, R as readGepaCandidatePopulationArtifact, S as costFromLedgerSummary, T as assertComponentSurface, Y as tangleTracesRoot, a as canonicalDigest, b as combineComparisonCosts, c as loopProvenanceSpans, d as verifyLoopProvenanceRecord, et as assertCampaignSplitIdentity, f as runImprovementLoop, h as labelTrustRank, i as campaignMeasurementDigest, j as campaignBreakdown, k as surfaceContentHash, l as provenanceRecordPath, m as isProposedCandidate, nt as campaignSplitDigest, o as emitLoopProvenance, p as runOptimization, q as planCampaignRun, r as buildLoopProvenanceRecord, rt as campaignSplitDigestFromIdentities, s as loopProvenanceArgsFromResult, t as llmJudge, tt as campaignScenarioIdentity, u as provenanceSpansPath, v as openAutoPr, w as assertCodeSurfaceIdentity, x as compareOptimizationMethods, z as defaultProductionGate } from "../llm-judge-DuYa4SEA.js";
+import { $ as assertCampaignDesign, A as surfaceHash, C as optimizationTokenUsageFromSummary, D as componentSurfaceIdentityMaterial, E as codeSurfaceIdentityMaterial, G as runEval, J as resolveRunDir, K as runCampaign, M as campaignMeanComposite, N as compareRankKeys, O as renderSurfaceDiff, R as readGepaCandidatePopulationArtifact, S as costFromLedgerSummary, T as assertComponentSurface, Y as tangleTracesRoot, a as canonicalDigest, at as cellCachePath, b as combineComparisonCosts, c as loopProvenanceSpans, d as verifyLoopProvenanceRecord, et as assertCampaignSplitIdentity, f as runImprovementLoop, h as labelTrustRank, i as campaignMeasurementDigest, it as readCachedCell, j as campaignBreakdown, k as surfaceContentHash, l as provenanceRecordPath, m as isProposedCandidate, nt as campaignSplitDigest, o as emitLoopProvenance, p as runOptimization, q as planCampaignRun, r as buildLoopProvenanceRecord, rt as campaignSplitDigestFromIdentities, s as loopProvenanceArgsFromResult, t as llmJudge, tt as campaignScenarioIdentity, u as provenanceSpansPath, v as openAutoPr, w as assertCodeSurfaceIdentity, x as compareOptimizationMethods, z as defaultProductionGate } from "../llm-judge-DhiJqSjB.js";
 import { E as SearchLedgerIntegrityError, S as inMemoryCampaignStorage, T as SearchLedgerError, _ as resolveExternalOptimizerCallbackLimits, b as createRunCostLedger, c as DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, l as DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, v as resolveExternalOptimizerProcessLimits, w as SearchLedgerConflictError, x as fsCampaignStorage } from "../external-optimizer-subprocess-BhKYK0Jv.js";
 import { a as heldoutSignificance, i as dimensionRegressions, o as pairHoldout, r as detectScale, t as powerPreflight } from "../power-preflight-DEw-uC7q.js";
 import { r as makeProposalFinding } from "../types-BI4fT3HN.js";
 import { n as acquireSingleRunLock } from "../external-optimizer-process-BRE56woM.js";
-import { a as externalTextOptimizationMethod, i as composeGate, n as gepaOptimizationMethod, o as decodeExternalTextCandidate, r as heldOutGate, s as readExternalOptimizerObservationArtifact, t as skillOptOptimizationMethod, u as createReferenceEquivalenceJudge } from "../skillopt-optimization-method-BTls2l-Z.js";
-import { A as neutralizationGate, C as scoreboardSummary, D as LabeledScenarioStoreError, E as FsLabeledScenarioStore, F as analyzeCrossSurfaceInteractions, I as buildTraceAnalystSurfaceDispatch, L as traceAnalystQualityJudge, M as loadEvalFixture, N as loadEvalFixtureScenarios, O as classifyUngroundedLiterals, P as planEvalFixtureRun, S as scoreUserStory, T as neutralizeText, _ as runProfileMatrixSegment, a as autoevalsScorerJudge, b as makePlaybackDispatch, c as FileSearchLedger, d as validateSearchLedgerEvent, f as scoreDiscrimination, g as finalizeProfileMatrix, h as createProfileMatrixPlan, i as verifyCodeSurface, j as discoverEvalFixtures, k as rolloutArgumentDiff, l as SEARCH_LEDGER_SCHEMA, m as SegmentedProfileMatrixError, n as gitWorktreeAdapter, o as phoenixEvaluatorJudge, p as selectDiscriminative, r as resolveWorktreePath, s as isTransientTransportFailure, t as WorktreeAdapterError, u as openSearchLedger, v as ProfileMatrixError, w as userStoryScoreboard, x as renderScoreboardMarkdown, y as runProfileMatrix } from "../campaign-jTOvqnse.js";
+import { a as externalTextOptimizationMethod, i as composeGate, n as gepaOptimizationMethod, o as decodeExternalTextCandidate, r as heldOutGate, s as readExternalOptimizerObservationArtifact, t as skillOptOptimizationMethod, u as createReferenceEquivalenceJudge } from "../skillopt-optimization-method-C6JSfYxT.js";
+import { A as neutralizationGate, C as scoreboardSummary, D as LabeledScenarioStoreError, E as FsLabeledScenarioStore, F as analyzeCrossSurfaceInteractions, I as buildTraceAnalystSurfaceDispatch, L as traceAnalystQualityJudge, M as loadEvalFixture, N as loadEvalFixtureScenarios, O as classifyUngroundedLiterals, P as planEvalFixtureRun, S as scoreUserStory, T as neutralizeText, _ as runProfileMatrixSegment, a as autoevalsScorerJudge, b as makePlaybackDispatch, c as FileSearchLedger, d as validateSearchLedgerEvent, f as scoreDiscrimination, g as finalizeProfileMatrix, h as createProfileMatrixPlan, i as verifyCodeSurface, j as discoverEvalFixtures, k as rolloutArgumentDiff, l as SEARCH_LEDGER_SCHEMA, m as SegmentedProfileMatrixError, n as gitWorktreeAdapter, o as phoenixEvaluatorJudge, p as selectDiscriminative, r as resolveWorktreePath, s as isTransientTransportFailure, t as WorktreeAdapterError, u as openSearchLedger, v as ProfileMatrixError, w as userStoryScoreboard, x as renderScoreboardMarkdown, y as runProfileMatrix } from "../campaign-Dg2B4W6d.js";
 import { n as paretoPolicy, r as paretoSignificanceGate, t as buildEvidenceVector } from "../promotion-policy-xzA40Evo.js";
 import { n as sequentialPairedGate, t as sequentialDecide } from "../sequential-C458DXNf.js";
-export { DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, FileSearchLedger, FsLabeledScenarioStore, LabeledScenarioStoreError, ProfileMatrixError, SEARCH_LEDGER_SCHEMA, SearchLedgerConflictError, SearchLedgerError, SearchLedgerIntegrityError, SegmentedProfileMatrixError, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, autoevalsScorerJudge, buildEvidenceVector, buildLoopProvenanceRecord, buildTraceAnalystSurfaceDispatch, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, combineComparisonCosts, compareOptimizationMethods, compareRankKeys, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createProfileMatrixPlan, createReferenceEquivalenceJudge, createRunCostLedger, decodeExternalTextCandidate, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, finalizeProfileMatrix, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, makeProposalFinding, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, phoenixEvaluatorJudge, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, readExternalOptimizerObservationArtifact, readGepaCandidatePopulationArtifact, renderScoreboardMarkdown, renderSurfaceDiff, resolveExternalOptimizerCallbackLimits, resolveExternalOptimizerProcessLimits, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, runProfileMatrixSegment, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, traceAnalystQualityJudge, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };
+export { DEFAULT_EXTERNAL_OPTIMIZER_CALLBACK_LIMITS, DEFAULT_EXTERNAL_OPTIMIZER_PROCESS_LIMITS, FileSearchLedger, FsLabeledScenarioStore, LabeledScenarioStoreError, ProfileMatrixError, SEARCH_LEDGER_SCHEMA, SearchLedgerConflictError, SearchLedgerError, SearchLedgerIntegrityError, SegmentedProfileMatrixError, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, autoevalsScorerJudge, buildEvidenceVector, buildLoopProvenanceRecord, buildTraceAnalystSurfaceDispatch, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, cellCachePath, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, combineComparisonCosts, compareOptimizationMethods, compareRankKeys, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createProfileMatrixPlan, createReferenceEquivalenceJudge, createRunCostLedger, decodeExternalTextCandidate, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, finalizeProfileMatrix, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, makeProposalFinding, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, phoenixEvaluatorJudge, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, readCachedCell, readExternalOptimizerObservationArtifact, readGepaCandidatePopulationArtifact, renderScoreboardMarkdown, renderSurfaceDiff, resolveExternalOptimizerCallbackLimits, resolveExternalOptimizerProcessLimits, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, runProfileMatrixSegment, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, traceAnalystQualityJudge, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };

package/dist/{campaign-jTOvqnse.js → campaign-Dg2B4W6d.js}

@@ -1,12 +1,12 @@
 import { s as ValidationError, t as AgentEvalError } from "./errors-Dngq5h35.js";
-import { c as agentProfileId, d as harnessAxisOf, i as verifyCompletion, l as agentProfileModelId, s as agentProfileHash, t as extractProducedState } from "./produced-state-D6j7qy1Q.js";
+import { c as agentProfileId, d as harnessAxisOf, i as verifyCompletion, l as agentProfileModelId, s as agentProfileHash, t as extractProducedState } from "./produced-state-BNyyud4g.js";
 import { buildAgentProfileCell } from "./profile-cell.js";
 import { t as comparePairedArms } from "./paired-arms-D-XRF_fy.js";
 import { r as pairedBootstrap } from "./paired-tests-BHIhYVdu.js";
-import { o as runTaskScore, r as modelHasSnapshot, s as validateRunRecord } from "./run-record-BvHPVS-i.js";
+import { c as validateRunRecord, i as modelHasSnapshot, n as UNKNOWN_MODEL, s as runTaskScore } from "./run-record-D2lDdSAz.js";
 import { n as contentHash, t as canonicalJson } from "./verdict-cache-mZf5FEiY.js";
-import { $ as assertCampaignDesign, J as resolveRunDir, K as runCampaign, Q as summarizeBackendIntegrity, Z as assertRealBackend, at as computeManifestHash, h as labelTrustRank, it as cellCachePath, k as surfaceContentHash, q as planCampaignRun, tt as campaignScenarioIdentity, w as assertCodeSurfaceIdentity } from "./llm-judge-DuYa4SEA.js";
-import { a as campaignCellCostProvenance, l as campaignCellToRunRecord, u as projectCampaignCellQuality } from "./reward-hacking-DNgjilrV.js";
+import { $ as assertCampaignDesign, J as resolveRunDir, K as runCampaign, Q as summarizeBackendIntegrity, Z as assertRealBackend, at as cellCachePath, h as labelTrustRank, k as surfaceContentHash, ot as computeManifestHash, q as planCampaignRun, tt as campaignScenarioIdentity, w as assertCodeSurfaceIdentity } from "./llm-judge-DhiJqSjB.js";
+import { a as campaignCellCostProvenance, l as campaignCellToRunRecord, u as projectCampaignCellQuality } from "./reward-hacking-DFo2FU5J.js";
 import { t as CostAccountingIncompleteError } from "./cost-ledger-BSe92yAV.js";
 import { t as FileLedgerJournal, u as mapConcurrent } from "./ledger-core-BmZt19oQ.js";
 import { r as canonicalString } from "./canonical-D-XsTQ6_.js";
@@ -14,7 +14,7 @@ import { C as SEARCH_LEDGER_FILE_CONTEXT, E as SearchLedgerIntegrityError, T as
 import { o as pairHoldout } from "./power-preflight-DEw-uC7q.js";
 import { a as scoreAnalystFindings } from "./benchmark-BhT16ep9.js";
 import "./external-optimizer-process-BRE56woM.js";
-import "./skillopt-optimization-method-BTls2l-Z.js";
+import "./skillopt-optimization-method-C6JSfYxT.js";
 import { createHash } from "node:crypto";
 import { closeSync, constants, existsSync, fstatSync, lstatSync, mkdirSync, mkdtempSync, openSync, readFileSync, readSync, readdirSync, readlinkSync, realpathSync, rmSync, statSync, writeFileSync } from "node:fs";
 import { z } from "zod";
@@ -1545,6 +1545,7 @@ function receiptModels(cell) {
 /** Resolve and validate every paid-call model used by one cell. */
 function recordModel(cell, profileId, declaredModel) {
 	const reported = receiptModels(cell);
+	if (reported.length === 0 && cell.error !== void 0) return UNKNOWN_MODEL;
 	if (modelHasSnapshot(declaredModel)) {
 		for (const model of reported) if (model !== declaredModel) throw new ProfileMatrixError(`profile '${profileId}' paid-call model '${model}' for cell '${cell.cellId}' does not match its declared exact model '${declaredModel}'`);
 		return declaredModel;
@@ -1565,7 +1566,7 @@ function buildRunRecord(args) {
 	const { cell, profile, profileHash, configHash, experimentId, splitTag, commitSha, matrixId } = args;
 	const profileId = agentProfileId(profile);
 	const model = recordModel(cell, profileId, agentProfileModelId(profile));
-	const record = campaignCellToRunRecord(cell, {
+	const record = campaignCellToRunRecord(durableCellForRecord(cell), {
 		runId: `${matrixId}:${profileId}:${cell.cellId}`,
 		experimentId,
 		candidateId: profileId,
@@ -1586,6 +1587,28 @@ function buildRunRecord(args) {
 	return record;
 }
 /**
+* Keep failed rows truthful at the RunRecord boundary.
+*
+* A cell with no settled call currently has a zero subtotal and an observed
+* zero from the ledger. That is not measured provider usage or cost, so mark
+* usage incomplete and cost uncaptured while retaining the known subtotal.
+*/
+function durableCellForRecord(cell) {
+	if (cell.error === void 0) return cell;
+	const hasSettledAgentCalls = receiptModels(cell).length > 0;
+	return {
+		...cell,
+		tokenUsage: cell.tokenUsage.tokensKnown === false || hasSettledAgentCalls ? cell.tokenUsage : {
+			...cell.tokenUsage,
+			tokensKnown: false
+		},
+		costProvenance: !hasSettledAgentCalls && cell.costProvenance.kind === "observed" && cell.costUsd === 0 ? {
+			kind: "uncaptured",
+			usd: null
+		} : cell.costProvenance
+	};
+}
+/**
 * Profile × scenario matrix runner: fan N agent profiles across M scenarios, project each cell to a validated `RunRecord` with real token usage, and enforce the backend-integrity guard before returning.
 */
 async function runProfileMatrix(opts) {
@@ -1695,13 +1718,14 @@ async function runProfileMatrix(opts) {
 				kind: "agent-interface-profile",
 				hash: profileHash
 			},
-			model: cellModel,
+			...cellModel && cellModel !== "unknown" ? { model: cellModel } : {},
 			...axis ? { harness: { id: axis.harness } } : {}
 		});
 		const sharedCellIdentity = modelHasSnapshot(declaredModel) ? await buildCellIdentity(declaredModel) : void 0;
 		const profileRecords = [];
 		for (const cell of campaign.cells) {
-			const agentProfileCell = sharedCellIdentity ?? await buildCellIdentity(recordModel(cell, profileId, declaredModel));
+			const cellModel = recordModel(cell, profileId, declaredModel);
+			const agentProfileCell = cellModel === "unknown" ? await buildCellIdentity() : sharedCellIdentity ?? await buildCellIdentity(cellModel);
 			const record = buildRunRecord({
 				cell,
 				profile,
@@ -1718,7 +1742,7 @@ async function runProfileMatrix(opts) {
 			if (validate) validateRunRecord(record);
 			profileRecords.push(record);
 		}
-		const recordedModels = [...new Set(profileRecords.map((record) => record.model))];
+		const recordedModels = [...new Set(profileRecords.map((record) => record.model).filter((model) => model !== UNKNOWN_MODEL))];
 		if (recordedModels.length > 1) throw new ProfileMatrixError(`profile '${profileId}' resolved to multiple model snapshots: ${recordedModels.join(", ")}`);
 		const costProvenance = campaign.aggregates.cost.costProvenance;
 		const totalCostUsd = costProvenance.kind === "uncaptured" ? null : costProvenance.usd;
@@ -1729,7 +1753,7 @@ async function runProfileMatrix(opts) {
 			summary: {
 				profileId,
 				profileHash,
-				model: recordedModels[0] ?? declaredModel,
+				model: recordedModels[0] ?? "unknown",
 				records: profileRecords.length,
 				meanComposite: meanOrNull(profileRecords.map(scoreOf).filter((score) => score !== void 0)),
 				totalCostUsd,
@@ -3658,4 +3682,4 @@ function resolveWorktreePath(surface, worktreeDir) {
 //#endregion
 export { neutralizationGate as A, scoreboardSummary as C, LabeledScenarioStoreError as D, FsLabeledScenarioStore as E, analyzeCrossSurfaceInteractions as F, buildTraceAnalystSurfaceDispatch as I, traceAnalystQualityJudge as L, loadEvalFixture as M, loadEvalFixtureScenarios as N, classifyUngroundedLiterals as O, planEvalFixtureRun as P, scoreUserStory as S, neutralizeText as T, runProfileMatrixSegment as _, autoevalsScorerJudge as a, makePlaybackDispatch as b, FileSearchLedger as c, validateSearchLedgerEvent as d, scoreDiscrimination as f, finalizeProfileMatrix as g, createProfileMatrixPlan as h, verifyCodeSurface as i, discoverEvalFixtures as j, rolloutArgumentDiff as k, SEARCH_LEDGER_SCHEMA as l, SegmentedProfileMatrixError as m, gitWorktreeAdapter as n, phoenixEvaluatorJudge as o, selectDiscriminative as p, resolveWorktreePath as r, isTransientTransportFailure as s, WorktreeAdapterError as t, openSearchLedger as u, ProfileMatrixError as v, userStoryScoreboard as w, renderScoreboardMarkdown as x, runProfileMatrix as y };
 
-//# sourceMappingURL=campaign-jTOvqnse.js.map
+//# sourceMappingURL=campaign-Dg2B4W6d.js.map