@tangle-network/agent-eval 0.203.1 → 0.204.0

package/dist/index.d.ts

@@ -9,15 +9,15 @@ import { a as ContinuousCalibrationResult, d as calibrateJudgeContinuous, f as c
 import { a as CorpusAgreementPerDimension, c as corpusInterRaterAgreement, i as CorpusAgreementOptions, l as corpusInterRaterAgreementFromJudgeScores, n as SeriesConvergenceResult, o as CorpusAgreementReport, r as analyzeSeries, s as CorpusScoreRecord, t as SeriesConvergenceOptions, u as interRaterReliability } from "./series-convergence-DLy0bQMV.js";
 import { a as partialCredit, c as spearmanR, d as weightedMean, i as confidenceInterval, l as summarizeNumberSeries, n as WeightedCompositeInput, o as pearsonR, r as WeightedCompositeResult, s as ranks, t as SeriesDistribution, u as weightedComposite } from "./descriptive-B2iPaT9J.js";
 import { n as bonferroni, r as holm, t as benjaminiHochberg } from "./multiplicity-DIWHvysC.js";
-import { A as PairArmsOptions, Bt as RiskDifferenceResult, C as hashJson, Ct as verifyAttestation, D as ComparePairedArmsOptions, F as PairedCorrectness, Ft as EProcessStep, Gt as pairedRiskDifference, Ht as isBinaryOutcomeVector, I as PairedMetricDelta, It as eProcess, Jt as passAtK, Kt as pairedRiskDifferenceExact, L as comparePairedArms, Lt as ExactRiskDifferenceResult, M as PairRunRecordsResult, Mt as EProcess, N as PairedArmRow, Nt as EProcessOptions, O as MatchedPair, P as PairedArmsComparison, Pt as EProcessState, R as pairArms, Rt as McNemarResult, St as attest, Ut as mcnemar, Vt as ScoreRiskDifferenceResult, Wt as pairedBinaryScale, Yt as wilson, bt as AttestationVerification, et as Objective, j as PairArmsResult, k as MatchedRunRecordPair, nt as dominates, qt as pairedRiskDifferenceScore, rt as paretoFrontier, tt as ParetoResult, vt as ATTESTATION_ALGORITHM, w as manifestContentDigest, xt as AttestedReport, yt as AttestationProvenance, z as pairRunRecords, zt as ProportionInterval } from "./statistical-heldout-CJYAO-4a.js";
+import { A as PairArmsOptions, Bt as RiskDifferenceResult, C as hashJson, Ct as verifyAttestation, D as ComparePairedArmsOptions, F as PairedCorrectness, Ft as EProcessStep, Gt as pairedRiskDifference, Ht as isBinaryOutcomeVector, I as PairedMetricDelta, It as eProcess, Jt as passAtK, Kt as pairedRiskDifferenceExact, L as comparePairedArms, Lt as ExactRiskDifferenceResult, M as PairRunRecordsResult, Mt as EProcess, N as PairedArmRow, Nt as EProcessOptions, O as MatchedPair, P as PairedArmsComparison, Pt as EProcessState, R as pairArms, Rt as McNemarResult, St as attest, Ut as mcnemar, Vt as ScoreRiskDifferenceResult, Wt as pairedBinaryScale, Yt as wilson, bt as AttestationVerification, et as Objective, j as PairArmsResult, k as MatchedRunRecordPair, nt as dominates, qt as pairedRiskDifferenceScore, rt as paretoFrontier, tt as ParetoResult, vt as ATTESTATION_ALGORITHM, w as manifestContentDigest, xt as AttestedReport, yt as AttestationProvenance, z as pairRunRecords, zt as ProportionInterval } from "./statistical-heldout-Dq3pnmxd.js";
 import { _ as pairedSignTest, a as PairedPromotionDecision, c as BOOTSTRAP_GATE_MIN_N, d as PairedBootstrapResult, f as PairedSignTestResult, g as pairedDeltaTieFraction, h as pairedBootstrap, i as PairedMcNemarEvidence, l as DECISION_PAIRED_DELTA_STATISTIC, m as SignTestAlternative, n as PairedDecisionShape, o as PairedPromotionDecisionOptions, p as PairedTTestResult, r as PairedDecisionStatistic, s as decidePairedPromotion, t as PairedDecisionMethod, u as PairedBootstrapOptions, v as pairedTTest } from "./paired-promotion-decision-DPKz8ZnF.js";
 import { a as requiredPairedSampleSize, i as pairedMde, n as mcnemarPower, o as requiredSampleSize, r as mcnemarRequiredN, t as mulberry32 } from "./random-DGEyr3zG.js";
 import { C as RankTestMethod, D as WilcoxonSignedRankResult, E as WILCOXON_EXACT_MAX_N, O as mannWhitneyU, S as MannWhitneyResult, T as RankTestOptions, _ as bootstrapCi, a as ReleaseConfidenceIssue, b as MANN_WHITNEY_EXACT_MAX_STATES, f as evaluateReleaseConfidence, g as Verdict, i as ReleaseConfidenceInput, k as wilcoxonSignedRank, m as BootstrapResult, o as ReleaseConfidenceMetrics, p as BootstrapOptions, s as ReleaseConfidenceScorecard, t as ActionableSideInfo, u as ReleaseTraceEvidence, w as RankTestMethodRequest, x as MANN_WHITNEY_EXACT_MAX_WORK, y as DEFAULT_PERMUTATIONS } from "./release-confidence-CkoXFx9I.js";
 import { a as RunFilter, i as InMemoryTraceStore, n as FileSystemTraceStore, o as SpanFilter, s as TraceStore, t as EventFilter } from "./store-C4avFjIi.js";
 import { a as MintedRolloutLine } from "./schema-C-ziyJD7.js";
 import { a as mintRolloutRows, c as ScorePreference, i as RolloutSearchLineage, n as MintRolloutResult, t as MintRolloutOptions } from "./mint-C1W8w5wo.js";
-import { R as Scenario$1, S as JudgeConfig, _ as GateDecision, a as CampaignResult, w as JudgeScore$1 } from "./types-Duc6xOoW.js";
-import { Ar as DEFAULT_RED_TEAM_CORPUS, At as RequirementCheck, Br as CanaryKind, Ct as extractProducedState, Dt as LlmCorrectnessCheckerOpts, Et as CorrectnessChecker, Fr as redTeamDataset, Ft as verifyCompletion, Hr as CanaryReport, Ir as redTeamReport, Ji as llmJudge, Lr as scoreRedTeamOutput, Mr as RedTeamCategory, Mt as TaskGold, Nr as RedTeamFinding, Nt as completionVerdict, Ot as ProducedProposal, Pr as RedTeamReport, Pt as createLlmCorrectnessChecker, Qi as transientDispatchFailure, Rr as CanaryAlert, St as ToolCallEventLike, Tt as CompletionVerdict, Ur as runCanaries, Vr as CanaryOptions, Zi as quotaExhaustedUntil, _a as runCampaign, _t as assertRealBackend, bt as ProposalEventLike, ga as RunCampaignOptions, gt as BackendIntegrityReport, ha as CampaignCellRetryPolicy, ht as BackendIntegrityError, jr as RedTeamCase, jt as SatisfiedBy, kt as ProducedState, ma as CampaignCellFailureReceipt, qi as LlmJudgeOptions, vt as summarizeBackendIntegrity, wt as CompletionRequirement, xt as RuntimeEventLike, yt as ArtifactEventLike, zr as CanaryEvaluation } from "./index-Ce2HACcP.js";
+import { R as Scenario$1, S as JudgeConfig, _ as GateDecision, a as CampaignResult, w as JudgeScore$1 } from "./types-ZUy5HGY5.js";
+import { Ar as DEFAULT_RED_TEAM_CORPUS, At as RequirementCheck, Br as CanaryKind, Ct as extractProducedState, Dt as LlmCorrectnessCheckerOpts, Et as CorrectnessChecker, Fr as redTeamDataset, Ft as verifyCompletion, Hr as CanaryReport, Ir as redTeamReport, Ji as llmJudge, Lr as scoreRedTeamOutput, Mr as RedTeamCategory, Mt as TaskGold, Nr as RedTeamFinding, Nt as completionVerdict, Ot as ProducedProposal, Pr as RedTeamReport, Pt as createLlmCorrectnessChecker, Qi as transientDispatchFailure, Rr as CanaryAlert, St as ToolCallEventLike, Tt as CompletionVerdict, Ur as runCanaries, Vr as CanaryOptions, Zi as quotaExhaustedUntil, _a as runCampaign, _t as assertRealBackend, bt as ProposalEventLike, ga as RunCampaignOptions, gt as BackendIntegrityReport, ha as CampaignCellRetryPolicy, ht as BackendIntegrityError, jr as RedTeamCase, jt as SatisfiedBy, kt as ProducedState, ma as CampaignCellFailureReceipt, qi as LlmJudgeOptions, vt as summarizeBackendIntegrity, wt as CompletionRequirement, xt as RuntimeEventLike, yt as ArtifactEventLike, zr as CanaryEvaluation } from "./index-B8n12TR2.js";
 import { a as ProfileAxisSpec, c as expandProfileAxes, i as HarnessType, l as harnessAxisOf, n as CODING_HARNESSES, o as agentProfileHash, r as HARNESS_NATIVE_MODEL, s as agentProfileId, t as AgentProfile } from "./agent-profile-EpO0HURy.js";
 import { i as DatasetSplit, n as DatasetManifest, r as DatasetScenario, t as Dataset } from "./dataset-DQqhOCPt.js";
 import { C as defineEquivalenceCheck, S as buildEquivalenceRecord, _ as StrategyChecker, a as CheckerIdentity, b as VerificationStrategyProfile, c as EquivalenceCheckDefinition, d as EquivalenceCheckerInput, f as EquivalenceCheckerResult, g as EquivalenceRecord, h as EquivalenceProtocolError, i as equivalenceVerdict, l as EquivalenceCheckSpec, m as EquivalenceObligationStatus, n as VerdictCertification, o as CheckerOutcome, p as EquivalenceObligation, r as certificationEvidenceDigest, s as EquivalenceArm, t as DefaultVerdict, u as EquivalenceChecker, v as VERIFICATION_STRATEGIES, w as runEquivalenceCheck, x as VerificationStrategySource, y as VERIFICATION_STRATEGY_SOURCES } from "./verdict-E4eRNf7-.js";
@@ -27,11 +27,11 @@ import { a as FAILURE_BLAME, c as FailureContext, d as INFRA_FAILURE_BLAMES, f a
 import { d as InsightReport, i as analyzeRuns, t as AnalyzeRunsOptions } from "./analyze-runs-Q74o4Hfj.js";
 import { i as ExactAnalystRunEvent, l as ExactCapableAnalyst, o as ExactAnalystRunResult } from "./exact-types-BBMMYk3g.js";
 import { c as RegistryRunOpts, i as BudgetPolicy, n as AnalystRegistry, r as AnalystRegistryOptions, s as ExactRegistryRunOpts } from "./registry-CT-hMSkR.js";
-import { _ as SelfImproveMethodResult, a as DefinedAgentEval, c as SelfImproveMethodOptions, d as SelfImproveProposerOptions, f as SelfImproveProposerResult, g as SelfImproveMethodProvenance, h as selfImprove, i as DefineAgentEvalOptions, l as SelfImproveOptions, n as AgentEvalEvaluateOptions, o as defineAgentEval, p as SelfImproveResult, r as AgentEvalImproveOptions, s as SelfImproveBudget, t as AgentEvalAgent, u as SelfImproveProgressEvent } from "./define-agent-eval-BbEXdM3H.js";
-import { a as createTraceAnalyst, i as TraceAnalystDefinition, n as buildDefaultAnalystRegistry, r as CreateTraceAnalystOptions, t as DefaultAnalystRegistryOptions } from "./default-registry-CxEBVA6W.js";
-import { i as TraceAnalysisEngineResult, l as RawAnalystFinding, n as TraceAnalysisEngine, t as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS } from "./engine-CaSqEVlL.js";
-import { A as SemanticConceptJudgeInput, C as DspyRlmTraceEngineOptions, D as ConceptFinding, E as createChatTraceEngine, M as SemanticConceptJudgeResult, N as runSemanticConceptJudge, O as ConceptSpec, T as ChatTraceEngineOptions, _ as FindingSubject, a as FAILURE_MODE_KIND_SPEC, d as FindingsStore, f as PersistedFinding, j as SemanticConceptJudgeOptions, k as SEMANTIC_CONCEPT_JUDGE_VERSION, l as DiffPolicy, m as diffFindings, t as DEFAULT_TRACE_ANALYST_KINDS, u as FindingsDiff, v as FindingSubjectKind, w as createDspyRlmTraceEngine } from "./index-CIkxRsLS.js";
-import { i as BenchmarkEvaluation } from "./types-hH9APahd.js";
+import { _ as SelfImproveMethodResult, a as DefinedAgentEval, c as SelfImproveMethodOptions, d as SelfImproveProposerOptions, f as SelfImproveProposerResult, g as SelfImproveMethodProvenance, h as selfImprove, i as DefineAgentEvalOptions, l as SelfImproveOptions, n as AgentEvalEvaluateOptions, o as defineAgentEval, p as SelfImproveResult, r as AgentEvalImproveOptions, s as SelfImproveBudget, t as AgentEvalAgent, u as SelfImproveProgressEvent } from "./define-agent-eval-D8bPoBM5.js";
+import { a as createTraceAnalyst, i as TraceAnalystDefinition, n as buildDefaultAnalystRegistry, r as CreateTraceAnalystOptions, t as DefaultAnalystRegistryOptions } from "./default-registry-BHEfZHQd.js";
+import { i as TraceAnalysisEngineResult, l as RawAnalystFinding, n as TraceAnalysisEngine, t as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS } from "./engine-F_PvwesR.js";
+import { A as SemanticConceptJudgeInput, C as DspyRlmTraceEngineOptions, D as ConceptFinding, E as createChatTraceEngine, M as SemanticConceptJudgeResult, N as runSemanticConceptJudge, O as ConceptSpec, T as ChatTraceEngineOptions, _ as FindingSubject, a as FAILURE_MODE_KIND_SPEC, d as FindingsStore, f as PersistedFinding, j as SemanticConceptJudgeOptions, k as SEMANTIC_CONCEPT_JUDGE_VERSION, l as DiffPolicy, m as diffFindings, t as DEFAULT_TRACE_ANALYST_KINDS, u as FindingsDiff, v as FindingSubjectKind, w as createDspyRlmTraceEngine } from "./index-C9l_-23M.js";
+import { i as BenchmarkEvaluation } from "./types-DZQpUJse.js";
 import { a as SpanHandle, n as CaptureWrite, o as TraceEmitter, s as TraceEmitterOptions, t as CaptureErrorHandler } from "./emitter-HcCmFQl_.js";
 import { n as SandboxDriver } from "./sandbox-harness-BlSOu4LX.js";
 import { S as shareAllowed, b as redactForShare, d as ShareSafetyVerdict, f as assessShareSafety, i as RedactOptions, o as RedactionProfile, r as REDACTION_VERSION, s as RedactionReport, u as ShareSafetyStatus, x as redactText, y as redact } from "./redact-Cveoq6fJ.js";
@@ -40,7 +40,7 @@ import { n as TrajectoryStep, r as buildTrajectory, t as Trajectory } from "./tr
 import { a as runCounterfactual, i as CounterfactualRunner, n as CounterfactualMutation, r as CounterfactualResult, t as CounterfactualContext } from "./counterfactual-CPaWN99-.js";
 import { A as AnalystFindingDigest, B as ControlRunResult, C as createFeedbackTrajectory, D as renderPreferenceMemoryMarkdown, E as feedbackTrajectoryToOptimizerRow, F as ControlActionOutcome, G as StopDecision, H as ControlRuntimeError, I as ControlBudget, J as subjectiveEval, K as objectiveEval, L as ControlContext, M as AnalystRunDigest, N as analystFindingDigest, O as summarizePreferenceMemory, P as analystRunDigest, R as ControlDecision, S as controlRunToFeedbackTrajectory, T as feedbackTrajectoriesToOptimizerRows, U as ControlSeverity, V as ControlRuntimeConfig, W as ControlStep, _ as PreferenceMemoryEntry, a as FeedbackLabel, b as analystRunToReviewRequests, c as FeedbackOptimizerRow, d as FeedbackTask, f as FeedbackTrajectory, g as InMemoryFeedbackTrajectoryStore, h as FileSystemFeedbackTrajectoryStore, i as FeedbackAttempt, j as AnalystReviewDecision, k as withAssignedFeedbackSplit, l as FeedbackOutcome, m as FeedbackTrajectoryStore, n as AnalystReviewRequest, o as FeedbackLabelKind, p as FeedbackTrajectoryFilter, q as runAgentControlLoop, r as FeedbackArtifactType, s as FeedbackLabelSource, t as AnalystFeedbackTrajectoryOptions, u as FeedbackSplitPolicy, v as ProposedSideEffect, w as feedbackTrajectoriesToDatasetScenarios, x as assignFeedbackSplit, y as analystRunToFeedbackTrajectory, z as ControlEvalResult } from "./feedback-trajectory-C6NJxXEa.js";
 import { y as OUTPUT_VALUE } from "./attribute-vocabulary-CcXqHODT.js";
-import { C as inferDomainKeywords, H as captureFetchToRawSink, I as OtlpExport, J as extractUsageFromSse, K as extractUsage, M as AnalyzeTracesResult, N as analyzeTraces, O as scoreTraceInsightReadiness, P as OtlpFlatLine, R as OtlpSpan, S as domainEvidencePattern, W as ExtractedUsage, g as TraceInsightSuite, h as TraceInsightReadiness, k as tokenizeDomainWords, n as toolSpansToTraceAnalysisStore, o as otlpTextToTraceAnalysisStore, r as OtlpFileTraceStore, v as buildTraceInsightContext, x as describeTraceInsightScope, y as buildTraceInsightPrompt, z as exportRunAsOtlp } from "./store-tool-spans-fIE31bU8.js";
+import { A as scoreTraceInsightReadiness, B as OtlpSpan, C as describeTraceInsightScope, F as analyzeTraces, I as OtlpFlatLine, J as extractUsage, K as ExtractedUsage, P as AnalyzeTracesResult, R as OtlpExport, T as inferDomainKeywords, V as exportRunAsOtlp, W as captureFetchToRawSink, X as extractUsageFromSse, _ as TraceInsightReadiness, a as SpanRecordsToTraceAnalysisStoreOptions, b as buildTraceInsightContext, c as spanRecordsToTraceAnalysisStore, j as tokenizeDomainWords, n as toolSpansToTraceAnalysisStore, r as OtlpFileTraceStore, s as otlpTextToTraceAnalysisStore, v as TraceInsightSuite, w as domainEvidencePattern, x as buildTraceInsightPrompt } from "./store-tool-spans-BYIarVrG.js";
 import { a as RunIntegrityReport, o as assertRunCaptured, t as RunIntegrityError } from "./integrity-BWO6RPxv.js";
 import { c as judgeSpans, f as runsForScenario, i as argHash } from "./query-DBUQhehH.js";
 import { a as SteeringChange, c as CampaignRunContext, d as CampaignVariant, f as EvalCampaignOptions, h as runEvalCampaign, i as Researcher, l as CampaignRunOutcome, m as FailedRun, n as ExperimentResult, o as CampaignFactoryParams, p as EvalCampaignResult, r as FailureMode, s as CampaignIntegrityPolicy, t as ExperimentPlan, u as CampaignScenario } from "./researcher-zTpxNJBy.js";
@@ -3695,5 +3695,5 @@ export declare class PairwiseSteeringOptimizer {
   optimize(rows: SteeringOptimizationRow[], config?: SteeringOptimizerConfig): SteeringOptimizationResult;
 }
 //#endregion
-export { AGENT_PROFILE_KINDS, ATTESTATION_ALGORITHM, type ActionExecutionPolicy, type ActionPolicyDecision, type ActionableSideInfo, type ActiveLearningOptions, type AgentEvalAgent, AgentEvalError, type AgentEvalErrorCode, type AgentEvalEvaluateOptions, type AgentEvalImproveOptions, type AgentProfile, type AgentProfileCell, type AgentProfileCellInput, type AgentProfileDimensionValue, type AgentProfileJson, type AgentProfileSourceInput, type Analyst, type AnalystContext, type AnalystFeedbackTrajectoryOptions, type AnalystFinding, type AnalystFindingDigest, AnalystRegistry, type AnalystRegistryOptions, type AnalystReviewDecision, type AnalystReviewRequest, type AnalystRunDigest, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystSeverity, type AnalystUsageReceipt, type AnalyzeRunsOptions, type AnalyzeTracesResult, type ArgumentCheck, type Artifact, type ArtifactCheck, type ArtifactEventLike, type ArtifactResult, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertCrossFamilyServedOptions, type AssertServedModelOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, BOOTSTRAP_GATE_MIN_N, type BackendDescriptor, BackendIntegrityError, type BackendIntegrityReport, type BenchmarkEvaluation, type BlendWeights, type BootstrapOptions, type BootstrapResult, type BoundedProcessInput, type BoundedProcessResult, type BudgetLedgerEntry, type BudgetPolicy, type BudgetSpec, CODING_HARNESSES, type CalibrationResult, type CampaignCellFailureReceipt, type CampaignCellRetryPolicy, type CampaignFactoryParams, type CampaignIntegrityPolicy, type CampaignResult, type CampaignRunContext, type CampaignRunOutcome, type CampaignScenario, type CampaignVariant, type CanaryAlert, type CanaryEvaluation, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CandidateScore, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, type CaptureErrorHandler, type CaptureWrite, type CellVerdict, type ChannelRollup, type ChatCallOpts, type ChatClient, type ChatRequest, type ChatResponse, type ChatTraceEngineOptions, type ChatTransport, type CheckResult, type CheckerIdentity, type CheckerOutcome, type CliffsMagnitude, type CollectedArtifacts, type CommandRunner, type ComparePairedArmsOptions, type CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptFinding, type ConceptSpec, ConfigError, type ContinuousAgreement, type ContinuousAgreementOptions, type ContinuousCalibrationResult, type ContractCheckResult, type ContractLintFinding, type ContractRule, type ContractRuleExecution, type ContractRun, type ContractSpan, type ContractStatus, type ContractVerdict, type ControlActionOutcome, type ControlBudget, type ControlContext, type ControlDecision, type ControlEvalResult, type ControlRunResult, type ControlRuntimeConfig, type ControlRuntimeError, type ControlStep, type CorpusAgreementOptions, type CorpusAgreementPerDimension, type CorpusAgreementReport, type CorpusScoreRecord, type CorrectnessChecker, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, type CostChannel, type CostEntry, CostLedger, type CostLedgerFilter, type CostLedgerHandle, type CostLedgerOptions, type CostLedgerPersistence, CostLedgerPersistenceError, type CostLedgerSummary, type CostProvenance, type CostReceipt, CostReceiptCaptureError, type CostReceiptInput, CostReservationExceededError, type CostResult, type CostSummary, type CostUsage, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateChatClientOpts, type CreateTraceAnalystOptions, CrossFamilyError, type CustomTokenPricing, type CustomTransportOpts, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_FAILURE_REASON_RULES, DEFAULT_PERMUTATIONS, DEFAULT_RED_TEAM_CORPUS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, type DataAcquisitionPlan, type DatasetManifest, type DatasetOverview, type DatasetScenario, type DatasetSplit, type DefaultAnalystRegistryOptions, type DefaultVerdict, type DefineAgentEvalOptions, type DefinedAgentEval, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DetectorEvent, type DetectorSignal, type DiffPolicy, type DiffScorecardOptions, type JudgeScore as DimensionJudgeScore, type DirEntry, type DiscoverPersonasOptions, type DiscoveredPersona, type DriverState, type DspyRlmTraceEngineOptions, type EProcess, type EProcessOptions, type EProcessState, type EProcessStep, type EnsembleAggregate, type EnsembleJudgeOptions, type EquivalenceArm, type EquivalenceCheckDefinition, type EquivalenceCheckSpec, type EquivalenceChecker, type EquivalenceCheckerInput, type EquivalenceCheckerResult, type EquivalenceObligation, type EquivalenceObligationStatus, EquivalenceProtocolError, type EquivalenceRecord, type ErrorCluster, type ErrorCountPattern, type ErrorStreakOptions, type EvalCampaignOptions, type EvalCampaignResult, type EventFilter, type EvidenceRef, type ExactAnalystRunEvent, type ExactAnalystRunResult, type ExactCapableAnalyst, type ExactRegistryRunOpts, type ExactRiskDifferenceResult, type ExperimentPlan, type ExperimentRep, type ExperimentResult, type ExperimentStats, type ExtractOptions, type ExtractResult, type ExtractedUsage, FAILURE_BLAME, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, type FailedRun, type FailureBlame, type FailureClass, type FailureClassification, type FailureContext, type FailureMode, type FailureReasonRule, type FailureRule, type FeedbackArtifactType, type FeedbackAttempt, type FeedbackLabel, type FeedbackLabelKind, type FeedbackLabelSource, type FeedbackOptimizerRow, type FeedbackOutcome, type FeedbackSplitPolicy, type FeedbackTask, type FeedbackTrajectory, type FeedbackTrajectoryFilter, type FeedbackTrajectoryStore, type FieldDestination, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemTraceStore, type Finding, type FindingSubject, type FindingSubjectKind, type FindingToPolicyEditOptions, type FindingsDiff, FindingsStore, type GainDistributionBin, type GainDistributionFigureSpec, type GainDistributionOptions, type GateDecision, type GateEvidence, type GenericSpan, type GoldenItem, HARNESS_NATIVE_MODEL, type HarnessType, type HeadroomInput, HeldOutGate, type HeldOutGateConfig, type GateDecision$1 as HeldOutGateDecision, type HeldOutGateRejectionCode, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, INFRA_FAILURE_BLAMES, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryTraceStore, type InsightReport, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InterimReleaseConfidence, type InterimReleaseConfidenceInput, JudgeError, type JudgeFamily, type JudgeFn, type JudgeInput, type JudgeRetryOutcome, type JudgeRetryPolicy, type JudgeRubric, type JudgeScore$1 as JudgeScore, type JudgeScoreInput, type JudgeScoresRecord, type JudgeSpan, type JudgeVerdict, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type KnowledgeAcquisitionMode, type KnowledgeBundle, type KnowledgeFreshness, type KnowledgeImportance, type KnowledgeReadinessReport, type KnowledgeRequirement, type KnowledgeRequirementCategory, type KnowledgeSensitivity, type Layer, type LayerResult, type LayerStatus, type LeaderboardOptions, type LeaderboardRow, LlmCallError, type LlmCallMetadata, type LlmCallRequest, type LlmCallResult, type LlmChargeBounds, type LlmCorrectnessCheckerOpts, type LlmJudgeOptions, type LlmMessage, LlmResponseError, type LlmReviewerConfig, type LlmSpan, type LlmTokenLogprob, type LlmToolCall, type LlmToolChoice, type LlmToolDefinition, type LlmUsage, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, type MannWhitneyResult, type MatchedPair, type MatchedRunRecordPair, type MaximumCharge, type McNemarResult, type MintRolloutOptions, type MintRolloutResult, type MintedRolloutLine, type MockTransportOpts, type ModelEndpointCheck, type ModelEndpointRequest, type ModelSeats, ModelSubstitutionError, MultiLayerVerifier, type NoLeakOptions, NoopRawProviderSink, NotFoundError, OUTPUT_VALUE, type Objective, type OpenAiCompatibleTransportOpts, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrderMode, type OtlpExport, OtlpFileTraceStore, type OtlpFlatLine, type OtlpSpan, type PaidCallResult, type PairArmsOptions, type PairArmsResult, type PairRunRecordsResult, type PairedArmRow, type PairedArmsComparison, type PairedBootstrapOptions, type PairedBootstrapResult, type PairedCorrectness, type PairedDecisionMethod, type PairedDecisionShape, type PairedDecisionStatistic, type PairedDeltaTestOptions, type PairedDeltaTestResult, type PairedEvalueOptions, type PairedEvalueSequence, type PairedEvalueStep, type PairedMcNemarEvidence, type PairedMetricDelta, type PairedPromotionAlternative, type PairedPromotionAlternativeCell, type PairedPromotionCallRates, type PairedPromotionDecision, type PairedPromotionDecisionOptions, type PairedPromotionPowerCall, type PairedPromotionPowerOptions, type PairedPromotionPowerResult, type PairedSignTestResult, type PairedTTestResult, type PairedValueLaw, type ParetoPoint, type ParetoResult, type PartitionHeldOutOptions, type PendingCostCall, type PendingCostCallView, type PersistedFinding, type PersonaConfig, type PolicyEdit, type PolicyEditAdmission, type PolicyEditAdmissionOptions, type PolicyEditAxis, type PolicyEditCandidateRecord, type PolicyEditChange, type PolicyEditExpectedGain, type PolicyEditGainDirection, type PolicyEditGainUnit, type PolicyEditInit, type PolicyEditRisk, type PolicyEditSchemaVersion, type PolicyEditSource, type PolicyEditTarget, type PolicyEditTargetSurface, type PreferenceMemoryEntry, type ProducedProposal, type ProducedState, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkValidationReport, type ProductClientConfig, type Scenario as ProductScenario, type ProfileAxisSpec, type ProjectRuntimeTrajectoryEvidenceOptions, type PromptHandle, type ProportionInterval, type ProposalEventLike, type ProposalFinding, type ProposeFn, type ProposeInput, type ProposeOutput, type ProposeReviewConfig, type ProposeReviewControlAction, type ProposeReviewControlConfig, type ProposeReviewControlResult, type ProposeReviewControlState, type ProposeReviewReport, type ProposedSideEffect, type QueryTracesPage, REDACTION_VERSION, type RankTestMethod, type RankTestMethodRequest, type RankTestOptions, type RawAnalystFinding, type RawProviderEvent, type RawProviderSink, type RecordRunsOptions, type RedTeamCase, type RedTeamCategory, type RedTeamFinding, type RedTeamReport, type RedactOptions, type RedactionProfile, type RedactionReport, type ReferenceReplayCaseRun, type ReferenceReplayRun, type ReferenceReplaySplit, type ReflectionContext, type ReflectionProposal, type RegistryRunOpts, type ReleaseConfidenceInput, type ReleaseConfidenceIssue, type ReleaseConfidenceMetrics, type ReleaseConfidenceScorecard, type ReleaseTraceEvidence, type RepeatedActionOptions, type RequiredPairsForPairedPromotionOptions, type RequiredPairsForPairedPromotionResult, type RequirementCheck, type ResearchReport, type ResearchReportOptions, type Researcher, type RetryWrite, type Review, type ReviewFn, type ReviewInput, type ReviewMemoryEntry, type ReviewMemoryStore, type RiskDifferenceResult, type RolloutSearchLineage, type RouteMap, type RoutedField, type RubricDimension, type Run, type RunCampaignOptions, type RunCaptureReport, type RunCommandInput, type RunCommandResult, type RunCostProvenance, type RunFilter, RunIntegrityError, type RunIntegrityReport, type RunJudgeMetadata, type RunOutcome, type RunPaidCallInput, type RunRecord, type RunRecordBackend, RunRecordValidationError, type RunScore, type RunScoreWeights, type RunSearchCoordinates, type RunSplitTag, type RunStatus, type RunTaskFailure, type RunTerminalOutcome, type RunTokenUsage, type RunTraceRef, type RuntimeEventLike, type RuntimeTrajectoryEvidenceProjection, type RuntimeTrajectoryEvidenceSummary, type RuntimeTrajectoryHookEvent, type RuntimeTrajectoryRecord, type RuntimeTrajectoryRunRecord, SEMANTIC_CONCEPT_JUDGE_VERSION, type SandboxDriver, type SandboxSdkTransportOpts, type SatisfiedBy, type Scenario$1 as Scenario, type ScenarioCost, type ScenarioFile, type ScenarioResult, type ScorePreference, type ScoreRiskDifferenceResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type SearchSpanResult, type SearchTraceResult, type SelfImproveBudget, type SelfImproveMethodOptions, type SelfImproveMethodProvenance, type SelfImproveMethodResult, type SelfImproveOptions, type SelfImproveProgressEvent, type SelfImproveProposerOptions, type SelfImproveProposerResult, type SelfImproveResult, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SequentialDecision, type SeriesConvergenceOptions, type SeriesConvergenceResult, type SeriesDistribution, ServedCrossFamilyError, type ServedModelCheck, type ServedModelPolicy, type ServedModelVerdict, type Severity, type ShareSafetyStatus, type ShareSafetyVerdict, type SignTestAlternative, type SingleBackendDivergence, type SingleBackendReport, type Span, type SpanFilter, type SpanHandle, type SpanMatchRecord, type SpanPredicate, type SplitCoverage, type SteeringBundle, type SteeringChange, type SteeringOptimizationResult, type SteeringOptimizationRow, type StopDecision, type StrategyChecker, type StreamingDetector, type SummaryTable, type SummaryTableOptions, type SummaryTableRow, type SynthesisTarget, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, type TaskGold, type TaskHeadroom, type ToolCallEventLike, type ToolSpan, type ToolStats, type ToolUseMetrics, type ToolUseOptions, type TraceAnalysisEngine, type TraceAnalysisEngineResult, type TraceAnalysisStore, type TraceAnalystByteBudgets, type TraceAnalystDefinition, type TraceAnalystFilters, type TraceAnalystSpan, type TraceAnalystSpanKind, type TraceAnalystSpanStatus, type TraceAnalystTraceSummary, type TraceContract, type TraceContractSpec, TraceEmitter, type TraceEmitterOptions, type TraceEvent, type TraceInsightReadiness, type TraceInsightSuite, type TraceStore, type TrainSelectionTestOptions, type TrainSelectionTestPartition, type Trajectory, type TrajectoryStep, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, type Turn, type TurnResult, UNKNOWN_MODEL, type UserQuestion, VERIFICATION_STRATEGIES, VERIFICATION_STRATEGY_SOURCES, ValidationError, type VerbosityBiasResult, type Verdict, type VerdictCacheStore, type VerdictCertification, type Verification, type VerificationReport, type VerificationStrategyProfile, type VerificationStrategySource, type VerifyFn, type VerifyOptions, type ViewSpansResult, type ViewTraceOversized, type ViewTraceResult, type ViteDeployRunnerInput, WILCOXON_EXACT_MAX_N, type WeightedCompositeInput, type WeightedCompositeResult, type WilcoxonSignedRankResult, type WranglerDeployRunnerInput, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, analystFindingDigest, analystRunDigest, analystRunToFeedbackTrajectory, analystRunToReviewRequests, analyzeRuns, analyzeSeries, analyzeTraces, argHash, assertCrossFamily, assertCrossFamilyServed, assertRealBackend, assertRunCaptured, assertServedModel, assertServedModels, assessShareSafety, assignFeedbackSplit, attest, benjaminiHochberg, bonferroni, bootstrapCi, budgetBreachView, buildAgentProfileCell, buildDefaultAnalystRegistry, buildEquivalenceRecord, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, calibrateJudge, calibrateJudgeContinuous, captureFetchToRawSink, certificationEvidenceDigest, checkServedModel, classifyFailure, classifyFailureReason, comparePairedArms, completionVerdict, computeFindingId, computeToolUseMetrics, confidenceInterval, continuousAgreement, controlRunToFeedbackTrajectory, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, createBoundedTraceAnalysisStore, createChatClient, createChatTraceEngine, createDspyRlmTraceEngine, createFeedbackTrajectory, createLlmCorrectnessChecker, createTraceAnalyst, decidePairedPromotion, defineAgentEval, defineEquivalenceCheck, describeTraceInsightScope, diffFindings, domainEvidencePattern, dominates, eProcess, equivalenceVerdict, evaluateInterimReleaseConfidence, evaluateReleaseConfidence, expandProfileAxes, exportRunAsOtlp, extractProducedState, extractUsage, extractUsageFromSse, failureBlame, failureClusterView, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToOptimizerRow, gainHistogram, gradeSemanticStatus, groupRunsByAgentProfileCell, harnessAxisOf, hashJson, holm, inferDomainKeywords, interRaterReliability, iqr, isBinaryOutcomeVector, isJudgeSpan, isLlmSpan, isRunRecord, isToolSpan, isTransientLlmError, judgeAgreementView, judgeFamily, judgeSpans, llmJudge, makeFinding, makeProposalFinding, manifestContentDigest, mannWhitneyU, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, mintRolloutRows, modelHasSnapshot, modelPriceKey, mulberry32, objectiveEval, otlpTextToTraceAnalysisStore, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paretoChart, paretoFrontier, parseRunRecordSafe, partialCredit, passAtK, pearsonR, index_d_exports as profile, quotaExhaustedUntil, ranks, redTeamDataset, redTeamReport, redact, redactForShare, redactText, renderPreferenceMemoryMarkdown, requiredPairedSampleSize, requiredSampleSize, roundTripRunRecord, runAgentControlLoop, runCampaign, runCanaries, runCostFloorUsd, runCounterfactual, runEquivalenceCheck, runEvalCampaign, runSemanticConceptJudge, runTaskScore, runsForScenario, scoreRedTeamOutput, scoreTraceInsightReadiness, searchCellRunId, selfImprove, servedModelAcceptable, shareAllowed, spearmanR, stripFencedJson, subjectiveEval, summarizeBackendIntegrity, summarizeNumberSeries, summarizePreferenceMemory, summaryTable, toAgentProfileJson, tokenizeDomainWords, toolSpansToTraceAnalysisStore, toolWasteView, transientDispatchFailure, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, weightedComposite, weightedMean, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit };
+export { AGENT_PROFILE_KINDS, ATTESTATION_ALGORITHM, type ActionExecutionPolicy, type ActionPolicyDecision, type ActionableSideInfo, type ActiveLearningOptions, type AgentEvalAgent, AgentEvalError, type AgentEvalErrorCode, type AgentEvalEvaluateOptions, type AgentEvalImproveOptions, type AgentProfile, type AgentProfileCell, type AgentProfileCellInput, type AgentProfileDimensionValue, type AgentProfileJson, type AgentProfileSourceInput, type Analyst, type AnalystContext, type AnalystFeedbackTrajectoryOptions, type AnalystFinding, type AnalystFindingDigest, AnalystRegistry, type AnalystRegistryOptions, type AnalystReviewDecision, type AnalystReviewRequest, type AnalystRunDigest, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystSeverity, type AnalystUsageReceipt, type AnalyzeRunsOptions, type AnalyzeTracesResult, type ArgumentCheck, type Artifact, type ArtifactCheck, type ArtifactEventLike, type ArtifactResult, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertCrossFamilyServedOptions, type AssertServedModelOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, BOOTSTRAP_GATE_MIN_N, type BackendDescriptor, BackendIntegrityError, type BackendIntegrityReport, type BenchmarkEvaluation, type BlendWeights, type BootstrapOptions, type BootstrapResult, type BoundedProcessInput, type BoundedProcessResult, type BudgetLedgerEntry, type BudgetPolicy, type BudgetSpec, CODING_HARNESSES, type CalibrationResult, type CampaignCellFailureReceipt, type CampaignCellRetryPolicy, type CampaignFactoryParams, type CampaignIntegrityPolicy, type CampaignResult, type CampaignRunContext, type CampaignRunOutcome, type CampaignScenario, type CampaignVariant, type CanaryAlert, type CanaryEvaluation, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CandidateScore, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, type CaptureErrorHandler, type CaptureWrite, type CellVerdict, type ChannelRollup, type ChatCallOpts, type ChatClient, type ChatRequest, type ChatResponse, type ChatTraceEngineOptions, type ChatTransport, type CheckResult, type CheckerIdentity, type CheckerOutcome, type CliffsMagnitude, type CollectedArtifacts, type CommandRunner, type ComparePairedArmsOptions, type CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptFinding, type ConceptSpec, ConfigError, type ContinuousAgreement, type ContinuousAgreementOptions, type ContinuousCalibrationResult, type ContractCheckResult, type ContractLintFinding, type ContractRule, type ContractRuleExecution, type ContractRun, type ContractSpan, type ContractStatus, type ContractVerdict, type ControlActionOutcome, type ControlBudget, type ControlContext, type ControlDecision, type ControlEvalResult, type ControlRunResult, type ControlRuntimeConfig, type ControlRuntimeError, type ControlStep, type CorpusAgreementOptions, type CorpusAgreementPerDimension, type CorpusAgreementReport, type CorpusScoreRecord, type CorrectnessChecker, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, type CostChannel, type CostEntry, CostLedger, type CostLedgerFilter, type CostLedgerHandle, type CostLedgerOptions, type CostLedgerPersistence, CostLedgerPersistenceError, type CostLedgerSummary, type CostProvenance, type CostReceipt, CostReceiptCaptureError, type CostReceiptInput, CostReservationExceededError, type CostResult, type CostSummary, type CostUsage, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateChatClientOpts, type CreateTraceAnalystOptions, CrossFamilyError, type CustomTokenPricing, type CustomTransportOpts, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_FAILURE_REASON_RULES, DEFAULT_PERMUTATIONS, DEFAULT_RED_TEAM_CORPUS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, type DataAcquisitionPlan, type DatasetManifest, type DatasetOverview, type DatasetScenario, type DatasetSplit, type DefaultAnalystRegistryOptions, type DefaultVerdict, type DefineAgentEvalOptions, type DefinedAgentEval, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DetectorEvent, type DetectorSignal, type DiffPolicy, type DiffScorecardOptions, type JudgeScore as DimensionJudgeScore, type DirEntry, type DiscoverPersonasOptions, type DiscoveredPersona, type DriverState, type DspyRlmTraceEngineOptions, type EProcess, type EProcessOptions, type EProcessState, type EProcessStep, type EnsembleAggregate, type EnsembleJudgeOptions, type EquivalenceArm, type EquivalenceCheckDefinition, type EquivalenceCheckSpec, type EquivalenceChecker, type EquivalenceCheckerInput, type EquivalenceCheckerResult, type EquivalenceObligation, type EquivalenceObligationStatus, EquivalenceProtocolError, type EquivalenceRecord, type ErrorCluster, type ErrorCountPattern, type ErrorStreakOptions, type EvalCampaignOptions, type EvalCampaignResult, type EventFilter, type EvidenceRef, type ExactAnalystRunEvent, type ExactAnalystRunResult, type ExactCapableAnalyst, type ExactRegistryRunOpts, type ExactRiskDifferenceResult, type ExperimentPlan, type ExperimentRep, type ExperimentResult, type ExperimentStats, type ExtractOptions, type ExtractResult, type ExtractedUsage, FAILURE_BLAME, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, type FailedRun, type FailureBlame, type FailureClass, type FailureClassification, type FailureContext, type FailureMode, type FailureReasonRule, type FailureRule, type FeedbackArtifactType, type FeedbackAttempt, type FeedbackLabel, type FeedbackLabelKind, type FeedbackLabelSource, type FeedbackOptimizerRow, type FeedbackOutcome, type FeedbackSplitPolicy, type FeedbackTask, type FeedbackTrajectory, type FeedbackTrajectoryFilter, type FeedbackTrajectoryStore, type FieldDestination, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemTraceStore, type Finding, type FindingSubject, type FindingSubjectKind, type FindingToPolicyEditOptions, type FindingsDiff, FindingsStore, type GainDistributionBin, type GainDistributionFigureSpec, type GainDistributionOptions, type GateDecision, type GateEvidence, type GenericSpan, type GoldenItem, HARNESS_NATIVE_MODEL, type HarnessType, type HeadroomInput, HeldOutGate, type HeldOutGateConfig, type GateDecision$1 as HeldOutGateDecision, type HeldOutGateRejectionCode, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, INFRA_FAILURE_BLAMES, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryTraceStore, type InsightReport, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InterimReleaseConfidence, type InterimReleaseConfidenceInput, JudgeError, type JudgeFamily, type JudgeFn, type JudgeInput, type JudgeRetryOutcome, type JudgeRetryPolicy, type JudgeRubric, type JudgeScore$1 as JudgeScore, type JudgeScoreInput, type JudgeScoresRecord, type JudgeSpan, type JudgeVerdict, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type KnowledgeAcquisitionMode, type KnowledgeBundle, type KnowledgeFreshness, type KnowledgeImportance, type KnowledgeReadinessReport, type KnowledgeRequirement, type KnowledgeRequirementCategory, type KnowledgeSensitivity, type Layer, type LayerResult, type LayerStatus, type LeaderboardOptions, type LeaderboardRow, LlmCallError, type LlmCallMetadata, type LlmCallRequest, type LlmCallResult, type LlmChargeBounds, type LlmCorrectnessCheckerOpts, type LlmJudgeOptions, type LlmMessage, LlmResponseError, type LlmReviewerConfig, type LlmSpan, type LlmTokenLogprob, type LlmToolCall, type LlmToolChoice, type LlmToolDefinition, type LlmUsage, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, type MannWhitneyResult, type MatchedPair, type MatchedRunRecordPair, type MaximumCharge, type McNemarResult, type MintRolloutOptions, type MintRolloutResult, type MintedRolloutLine, type MockTransportOpts, type ModelEndpointCheck, type ModelEndpointRequest, type ModelSeats, ModelSubstitutionError, MultiLayerVerifier, type NoLeakOptions, NoopRawProviderSink, NotFoundError, OUTPUT_VALUE, type Objective, type OpenAiCompatibleTransportOpts, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrderMode, type OtlpExport, OtlpFileTraceStore, type OtlpFlatLine, type OtlpSpan, type PaidCallResult, type PairArmsOptions, type PairArmsResult, type PairRunRecordsResult, type PairedArmRow, type PairedArmsComparison, type PairedBootstrapOptions, type PairedBootstrapResult, type PairedCorrectness, type PairedDecisionMethod, type PairedDecisionShape, type PairedDecisionStatistic, type PairedDeltaTestOptions, type PairedDeltaTestResult, type PairedEvalueOptions, type PairedEvalueSequence, type PairedEvalueStep, type PairedMcNemarEvidence, type PairedMetricDelta, type PairedPromotionAlternative, type PairedPromotionAlternativeCell, type PairedPromotionCallRates, type PairedPromotionDecision, type PairedPromotionDecisionOptions, type PairedPromotionPowerCall, type PairedPromotionPowerOptions, type PairedPromotionPowerResult, type PairedSignTestResult, type PairedTTestResult, type PairedValueLaw, type ParetoPoint, type ParetoResult, type PartitionHeldOutOptions, type PendingCostCall, type PendingCostCallView, type PersistedFinding, type PersonaConfig, type PolicyEdit, type PolicyEditAdmission, type PolicyEditAdmissionOptions, type PolicyEditAxis, type PolicyEditCandidateRecord, type PolicyEditChange, type PolicyEditExpectedGain, type PolicyEditGainDirection, type PolicyEditGainUnit, type PolicyEditInit, type PolicyEditRisk, type PolicyEditSchemaVersion, type PolicyEditSource, type PolicyEditTarget, type PolicyEditTargetSurface, type PreferenceMemoryEntry, type ProducedProposal, type ProducedState, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkValidationReport, type ProductClientConfig, type Scenario as ProductScenario, type ProfileAxisSpec, type ProjectRuntimeTrajectoryEvidenceOptions, type PromptHandle, type ProportionInterval, type ProposalEventLike, type ProposalFinding, type ProposeFn, type ProposeInput, type ProposeOutput, type ProposeReviewConfig, type ProposeReviewControlAction, type ProposeReviewControlConfig, type ProposeReviewControlResult, type ProposeReviewControlState, type ProposeReviewReport, type ProposedSideEffect, type QueryTracesPage, REDACTION_VERSION, type RankTestMethod, type RankTestMethodRequest, type RankTestOptions, type RawAnalystFinding, type RawProviderEvent, type RawProviderSink, type RecordRunsOptions, type RedTeamCase, type RedTeamCategory, type RedTeamFinding, type RedTeamReport, type RedactOptions, type RedactionProfile, type RedactionReport, type ReferenceReplayCaseRun, type ReferenceReplayRun, type ReferenceReplaySplit, type ReflectionContext, type ReflectionProposal, type RegistryRunOpts, type ReleaseConfidenceInput, type ReleaseConfidenceIssue, type ReleaseConfidenceMetrics, type ReleaseConfidenceScorecard, type ReleaseTraceEvidence, type RepeatedActionOptions, type RequiredPairsForPairedPromotionOptions, type RequiredPairsForPairedPromotionResult, type RequirementCheck, type ResearchReport, type ResearchReportOptions, type Researcher, type RetryWrite, type Review, type ReviewFn, type ReviewInput, type ReviewMemoryEntry, type ReviewMemoryStore, type RiskDifferenceResult, type RolloutSearchLineage, type RouteMap, type RoutedField, type RubricDimension, type Run, type RunCampaignOptions, type RunCaptureReport, type RunCommandInput, type RunCommandResult, type RunCostProvenance, type RunFilter, RunIntegrityError, type RunIntegrityReport, type RunJudgeMetadata, type RunOutcome, type RunPaidCallInput, type RunRecord, type RunRecordBackend, RunRecordValidationError, type RunScore, type RunScoreWeights, type RunSearchCoordinates, type RunSplitTag, type RunStatus, type RunTaskFailure, type RunTerminalOutcome, type RunTokenUsage, type RunTraceRef, type RuntimeEventLike, type RuntimeTrajectoryEvidenceProjection, type RuntimeTrajectoryEvidenceSummary, type RuntimeTrajectoryHookEvent, type RuntimeTrajectoryRecord, type RuntimeTrajectoryRunRecord, SEMANTIC_CONCEPT_JUDGE_VERSION, type SandboxDriver, type SandboxSdkTransportOpts, type SatisfiedBy, type Scenario$1 as Scenario, type ScenarioCost, type ScenarioFile, type ScenarioResult, type ScorePreference, type ScoreRiskDifferenceResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type SearchSpanResult, type SearchTraceResult, type SelfImproveBudget, type SelfImproveMethodOptions, type SelfImproveMethodProvenance, type SelfImproveMethodResult, type SelfImproveOptions, type SelfImproveProgressEvent, type SelfImproveProposerOptions, type SelfImproveProposerResult, type SelfImproveResult, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SequentialDecision, type SeriesConvergenceOptions, type SeriesConvergenceResult, type SeriesDistribution, ServedCrossFamilyError, type ServedModelCheck, type ServedModelPolicy, type ServedModelVerdict, type Severity, type ShareSafetyStatus, type ShareSafetyVerdict, type SignTestAlternative, type SingleBackendDivergence, type SingleBackendReport, type Span, type SpanFilter, type SpanHandle, type SpanMatchRecord, type SpanPredicate, type SpanRecordsToTraceAnalysisStoreOptions, type SplitCoverage, type SteeringBundle, type SteeringChange, type SteeringOptimizationResult, type SteeringOptimizationRow, type StopDecision, type StrategyChecker, type StreamingDetector, type SummaryTable, type SummaryTableOptions, type SummaryTableRow, type SynthesisTarget, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, type TaskGold, type TaskHeadroom, type ToolCallEventLike, type ToolSpan, type ToolStats, type ToolUseMetrics, type ToolUseOptions, type TraceAnalysisEngine, type TraceAnalysisEngineResult, type TraceAnalysisStore, type TraceAnalystByteBudgets, type TraceAnalystDefinition, type TraceAnalystFilters, type TraceAnalystSpan, type TraceAnalystSpanKind, type TraceAnalystSpanStatus, type TraceAnalystTraceSummary, type TraceContract, type TraceContractSpec, TraceEmitter, type TraceEmitterOptions, type TraceEvent, type TraceInsightReadiness, type TraceInsightSuite, type TraceStore, type TrainSelectionTestOptions, type TrainSelectionTestPartition, type Trajectory, type TrajectoryStep, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, type Turn, type TurnResult, UNKNOWN_MODEL, type UserQuestion, VERIFICATION_STRATEGIES, VERIFICATION_STRATEGY_SOURCES, ValidationError, type VerbosityBiasResult, type Verdict, type VerdictCacheStore, type VerdictCertification, type Verification, type VerificationReport, type VerificationStrategyProfile, type VerificationStrategySource, type VerifyFn, type VerifyOptions, type ViewSpansResult, type ViewTraceOversized, type ViewTraceResult, type ViteDeployRunnerInput, WILCOXON_EXACT_MAX_N, type WeightedCompositeInput, type WeightedCompositeResult, type WilcoxonSignedRankResult, type WranglerDeployRunnerInput, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, analystFindingDigest, analystRunDigest, analystRunToFeedbackTrajectory, analystRunToReviewRequests, analyzeRuns, analyzeSeries, analyzeTraces, argHash, assertCrossFamily, assertCrossFamilyServed, assertRealBackend, assertRunCaptured, assertServedModel, assertServedModels, assessShareSafety, assignFeedbackSplit, attest, benjaminiHochberg, bonferroni, bootstrapCi, budgetBreachView, buildAgentProfileCell, buildDefaultAnalystRegistry, buildEquivalenceRecord, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, calibrateJudge, calibrateJudgeContinuous, captureFetchToRawSink, certificationEvidenceDigest, checkServedModel, classifyFailure, classifyFailureReason, comparePairedArms, completionVerdict, computeFindingId, computeToolUseMetrics, confidenceInterval, continuousAgreement, controlRunToFeedbackTrajectory, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, createBoundedTraceAnalysisStore, createChatClient, createChatTraceEngine, createDspyRlmTraceEngine, createFeedbackTrajectory, createLlmCorrectnessChecker, createTraceAnalyst, decidePairedPromotion, defineAgentEval, defineEquivalenceCheck, describeTraceInsightScope, diffFindings, domainEvidencePattern, dominates, eProcess, equivalenceVerdict, evaluateInterimReleaseConfidence, evaluateReleaseConfidence, expandProfileAxes, exportRunAsOtlp, extractProducedState, extractUsage, extractUsageFromSse, failureBlame, failureClusterView, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToOptimizerRow, gainHistogram, gradeSemanticStatus, groupRunsByAgentProfileCell, harnessAxisOf, hashJson, holm, inferDomainKeywords, interRaterReliability, iqr, isBinaryOutcomeVector, isJudgeSpan, isLlmSpan, isRunRecord, isToolSpan, isTransientLlmError, judgeAgreementView, judgeFamily, judgeSpans, llmJudge, makeFinding, makeProposalFinding, manifestContentDigest, mannWhitneyU, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, mintRolloutRows, modelHasSnapshot, modelPriceKey, mulberry32, objectiveEval, otlpTextToTraceAnalysisStore, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paretoChart, paretoFrontier, parseRunRecordSafe, partialCredit, passAtK, pearsonR, index_d_exports as profile, quotaExhaustedUntil, ranks, redTeamDataset, redTeamReport, redact, redactForShare, redactText, renderPreferenceMemoryMarkdown, requiredPairedSampleSize, requiredSampleSize, roundTripRunRecord, runAgentControlLoop, runCampaign, runCanaries, runCostFloorUsd, runCounterfactual, runEquivalenceCheck, runEvalCampaign, runSemanticConceptJudge, runTaskScore, runsForScenario, scoreRedTeamOutput, scoreTraceInsightReadiness, searchCellRunId, selfImprove, servedModelAcceptable, shareAllowed, spanRecordsToTraceAnalysisStore, spearmanR, stripFencedJson, subjectiveEval, summarizeBackendIntegrity, summarizeNumberSeries, summarizePreferenceMemory, summaryTable, toAgentProfileJson, tokenizeDomainWords, toolSpansToTraceAnalysisStore, toolWasteView, transientDispatchFailure, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, weightedComposite, weightedMean, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit };
 //# sourceMappingURL=index.d.ts.map

package/dist/index.js

@@ -1,7 +1,7 @@
 import { t as __exportAll } from "./rolldown-runtime-8H4AJuhK.js";
 import { i as JudgeError, n as CaptureIntegrityError, o as NotFoundError, r as ConfigError, s as ValidationError, t as AgentEvalError } from "./errors-Dngq5h35.js";
 import { a as hashCanonical, r as canonicalString } from "./canonical-BsZVgBhW.js";
-import { a as CODING_HARNESSES, c as agentProfileId, d as harnessAxisOf, i as verifyCompletion, l as agentProfileModelId, n as completionVerdict, o as HARNESS_NATIVE_MODEL, r as createLlmCorrectnessChecker, s as agentProfileHash, t as extractProducedState, u as expandProfileAxes } from "./produced-state-g9BPgu4D.js";
+import { a as CODING_HARNESSES, c as agentProfileId, d as harnessAxisOf, i as verifyCompletion, l as agentProfileModelId, n as completionVerdict, o as HARNESS_NATIVE_MODEL, r as createLlmCorrectnessChecker, s as agentProfileHash, t as extractProducedState, u as expandProfileAxes } from "./produced-state-DLToUXrD.js";
 import { c as groupRunsByAgentProfileCell, d as validateAgentProfileCell, f as verifyAgentProfileCell, h as manifestContentDigest, i as agentProfileCellKey, m as hashJson, r as agentProfileCellHashMaterial, s as buildAgentProfileCell, t as AGENT_PROFILE_KINDS, u as toAgentProfileJson } from "./agent-profile-cell-DcFW4vOc.js";
 import { t as mulberry32 } from "./random-Dn5fPWkt.js";
 import { a as spearmanR, c as weightedMean, i as ranks, n as partialCredit, o as summarizeNumberSeries, r as pearsonR, s as weightedComposite, t as confidenceInterval } from "./descriptive-D9LUFk1-.js";
@@ -23,24 +23,24 @@ import { r as observedSplitScore, t as isRealnessGated } from "./reward-nw2xZGZG
 import { a as parseRunRecordSafe, c as runTaskScore, i as modelHasSnapshot, l as searchCellRunId, n as UNKNOWN_MODEL, o as roundTripRunRecord, r as isRunRecord, s as runCostFloorUsd, t as RunRecordValidationError, u as validateRunRecord } from "./run-record-BRlM99rX.js";
 import { a as summaryTable, n as gainHistogram, r as paretoChart } from "./summary-report-DLzptLSA.js";
 import { n as contentHash, r as fileVerdictCache, t as canonicalJson } from "./verdict-cache-BQEsVy9n.js";
-import { J as redTeamDataset, X as scoreRedTeamOutput, Y as redTeamReport, Z as runCanaries, a as transientDispatchFailure, i as quotaExhaustedUntil, o as aggregateRunScore, q as DEFAULT_RED_TEAM_CORPUS, rt as runCampaign, s as clamp01, t as llmJudge } from "./llm-judge-EUMHOZOQ.js";
+import { J as redTeamDataset, X as scoreRedTeamOutput, Y as redTeamReport, Z as runCanaries, a as transientDispatchFailure, i as quotaExhaustedUntil, o as aggregateRunScore, q as DEFAULT_RED_TEAM_CORPUS, rt as runCampaign, s as clamp01, t as llmJudge } from "./llm-judge-DyZdrp6I.js";
 import { a as resolveModelPricing, i as isModelPriced, n as estimateCost, r as estimateTokens, t as MODEL_PRICING } from "./metrics-DOMpkoeb.js";
 import { a as CostLedgerPersistenceError, c as costForTokenPricing, i as CostLedger, l as costForUsage, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError, u as modelPriceKey } from "./cost-ledger-_IbkejCF.js";
 import { B as assertRealBackend, F as surfaceContentHash, T as parseReflectionResponse, V as summarizeBackendIntegrity, c as ATTESTATION_ALGORITHM, l as attest, u as verifyAttestation, w as buildReflectionPrompt, z as BackendIntegrityError } from "./campaign-evidence-Dt_rWPk2.js";
 import { d as redactForShare, f as redactText, n as REDACTION_VERSION, p as shareAllowed, r as assessShareSafety, u as redact } from "./redact-Dk1ZdJEm.js";
 import { D as paretoFrontier, E as dominates, d as pairedPromotionPower, f as requiredPairsForPairedPromotion } from "./search-kernel-Dpniloks.js";
 import { S as judgeFamily, _ as assertServedModels, b as CrossFamilyError, d as stripFencedJson, f as ModelSubstitutionError, h as assertServedModel, i as backoffMs, l as isTransientLlmError, m as assertCrossFamilyServed, o as costReceiptFromLlm, p as ServedCrossFamilyError, r as LlmResponseError, s as costReceiptFromLlmError, t as LlmCallError, u as maximumChargeForLlmRequest, v as checkServedModel, x as assertCrossFamily, y as servedModelAcceptable } from "./llm-client-DmCNzrc9.js";
-import { r as selfImprove, t as defineAgentEval } from "./define-agent-eval-Da8mHyc7.js";
+import { r as selfImprove, t as defineAgentEval } from "./define-agent-eval-B25Cc3ki.js";
 import { n as equivalenceVerdict, t as certificationEvidenceDigest } from "./verdict-DhPoXApk.js";
 import { t as TraceEmitter } from "./emitter-DbMXUIVP.js";
 import { t as buildTrajectory } from "./trajectory-D_7rLrvE.js";
 import { t as runCounterfactual } from "./counterfactual-DtMWvn7D.js";
-import { G as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, U as parseFindingSubject, h as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, m as DEFAULT_TRACE_ANALYST_BUDGETS, t as createTraceAnalyst, u as createBoundedTraceAnalysisStore } from "./kind-factory-CdPPHtGZ.js";
-import { _ as completedAnalystReviewQuality, b as validateAnalystReviewDecisions, c as FAILURE_MODE_KIND_SPEC, g as assertUniqueFindingIds, h as analystRunDigest, i as DEFAULT_TRACE_ANALYST_KINDS, m as analystFindingDigest, n as AnalystRegistry, t as buildDefaultAnalystRegistry, v as readAnalystReview, y as snapshotAnalystRun } from "./default-registry-wInsQWlQ.js";
+import { K as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, W as parseFindingSubject, g as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, h as DEFAULT_TRACE_ANALYST_BUDGETS, t as createTraceAnalyst, u as createBoundedTraceAnalysisStore } from "./kind-factory-Cn1lbsAd.js";
+import { _ as completedAnalystReviewQuality, b as validateAnalystReviewDecisions, c as FAILURE_MODE_KIND_SPEC, g as assertUniqueFindingIds, h as analystRunDigest, i as DEFAULT_TRACE_ANALYST_KINDS, m as analystFindingDigest, n as AnalystRegistry, t as buildDefaultAnalystRegistry, v as readAnalystReview, y as snapshotAnalystRun } from "./default-registry-BthquIoC.js";
 import { OUTPUT_VALUE } from "./trace-attributes.js";
 import { r as extractUsageFromSse, t as extractUsage } from "./extract-usage-BrnzRJ6H.js";
 import { n as InMemoryRawProviderSink, r as NoopRawProviderSink, t as FileSystemRawProviderSink } from "./raw-provider-sink-rB2DYxr6.js";
-import { S as exportRunAsOtlp, T as captureFetchToRawSink, _ as tokenizeDomainWords, d as inferDomainKeywords, g as scoreTraceInsightReadiness, l as describeTraceInsightScope, n as toolSpansToTraceAnalysisStore, o as buildTraceInsightContext, s as buildTraceInsightPrompt, u as domainEvidencePattern, v as analyzeTraces } from "./store-tool-spans-D5kOU0mr.js";
+import { S as exportRunAsOtlp, T as captureFetchToRawSink, _ as tokenizeDomainWords, d as inferDomainKeywords, g as scoreTraceInsightReadiness, l as describeTraceInsightScope, n as toolSpansToTraceAnalysisStore, o as buildTraceInsightContext, s as buildTraceInsightPrompt, u as domainEvidencePattern, v as analyzeTraces } from "./store-tool-spans-DKoVbE-w.js";
 import { n as assertRunCaptured, t as RunIntegrityError } from "./integrity-BvnVARVl.js";
 import { n as InMemoryTraceStore, t as FileSystemTraceStore } from "./store-BhNuLdJw.js";
 import { t as packageVersion$1 } from "./package-version-D7lQHt_-.js";
@@ -52,10 +52,10 @@ import { t as mintRolloutRows } from "./mint-CWG-s2mI.js";
 import { n as pairedEvalueSequence, t as evaluateInterimReleaseConfidence } from "./sequential-CzK5DarL.js";
 import { n as paidJsonChat } from "./chat-json-call-Digk-pfF.js";
 import { i as makeProposalFinding, n as computeFindingId, r as makeFinding } from "./types-CW5b90JH.js";
-import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-D0h-P2Mz.js";
-import { a as diffFindings, n as runSemanticConceptJudge, o as createChatTraceEngine, r as FindingsStore, t as SEMANTIC_CONCEPT_JUDGE_VERSION } from "./semantic-concept-judge-BgWxa4L_.js";
+import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-DCQKNQMn.js";
+import { a as diffFindings, n as runSemanticConceptJudge, o as createChatTraceEngine, r as FindingsStore, t as SEMANTIC_CONCEPT_JUDGE_VERSION } from "./semantic-concept-judge-BbiYuv7U.js";
 import { t as analyzeSeries } from "./series-convergence-CjO2QdRW.js";
-import { i as otlpTextToTraceAnalysisStore, n as OtlpFileTraceStore } from "./store-otlp-yd9lDziN.js";
+import { a as spanRecordsToTraceAnalysisStore, i as otlpTextToTraceAnalysisStore, n as OtlpFileTraceStore } from "./store-otlp-BA7G1Du-.js";
 import { n as evaluateReleaseConfidence, r as bootstrapCi } from "./release-confidence-CiFg1EPq.js";
 import { t as createChatClient } from "./chat-client-5LLl4Ctl.js";
 import { createHash } from "node:crypto";
@@ -9694,6 +9694,6 @@ function rankRows(rows, weights) {
 	})).sort((a, b) => b.mean - a.mean);
 }
 //#endregion
-export { AGENT_PROFILE_KINDS, ATTESTATION_ALGORITHM, AgentEvalError, AnalystRegistry, BOOTSTRAP_GATE_MIN_N, BackendIntegrityError, BudgetBreachError, BudgetGuard, CODING_HARNESSES, ConfigError, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, CostLedger, CostLedgerPersistenceError, CostReceiptCaptureError, CostReservationExceededError, CostTracker, CrossFamilyError, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_BOUNDED_PROCESS_TIMEOUT_MS, DEFAULT_FAILURE_REASON_RULES, DEFAULT_PERMUTATIONS, DEFAULT_RED_TEAM_CORPUS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, ERROR_COUNT_PATTERNS, EquivalenceProtocolError, FAILURE_BLAME, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemTraceStore, FindingsStore, HARNESS_NATIVE_MODEL, HeldOutGate, INFRA_FAILURE_BLAMES, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryTraceStore, JudgeError, LOCAL_COMMAND_RUNNER_DEFAULT_CAP_MS, LlmCallError, LlmResponseError, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, MODEL_PRICING, ModelSubstitutionError, MultiLayerVerifier, NoopRawProviderSink, NotFoundError, OUTPUT_VALUE, OtlpFileTraceStore, POLICY_EDIT_AXES, POLICY_EDIT_CANDIDATE_RECORD_SCHEMA, POLICY_EDIT_TARGET_SURFACES, PairwiseSteeringOptimizer, PolicyEditValidationError, ProductClient, PromptRegistry, REDACTION_VERSION, RunIntegrityError, RunRecordValidationError, SEMANTIC_CONCEPT_JUDGE_VERSION, ServedCrossFamilyError, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, TraceEmitter, UNKNOWN_MODEL, VERIFICATION_STRATEGIES, VERIFICATION_STRATEGY_SOURCES, ValidationError, WILCOXON_EXACT_MAX_N, acquisitionPlansForKnowledgeGaps, admitPolicyEdit, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, aggregateJudgeVerdicts, aggregateRunScore, analystFindingDigest, analystRunDigest, analystRunToFeedbackTrajectory, analystRunToReviewRequests, analyzeAntiSlop, analyzeRuns, analyzeSeries, analyzeTraces, applyPolicyEditToSurface, argHash, assertCapabilityHeadroom, assertCrossFamily, assertCrossFamilyServed, assertNoHiddenLeak, assertNoJudgeVerdict, assertProductBenchmarkRun, assertRealBackend, assertRunCaptured, assertServedModel, assertServedModels, assertSingleBackend, assessShareSafety, assignFeedbackSplit, attest, benjaminiHochberg, blendHeldout, blockingKnowledgeEval, bonferroni, bootstrapCi, budgetBreachView, buildAgentProfileCell, buildDefaultAnalystRegistry, buildEquivalenceRecord, buildReflectionPrompt, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, calibrateJudge, calibrateJudgeContinuous, canonicalJson, capabilityHeadroom, captureFetchToRawSink, certificationEvidenceDigest, checkCanaries, checkServedModel, checkTraceContracts, clamp01, classifyFailure, classifyFailureReason, cliffsDelta, cohensD, comparePairedArms, compileTraceContractSpec, completionVerdict, computeExperimentStats, computeFindingId, computePolicyEditId, computeToolUseMetrics, confidenceInterval, contentHash, continuousAgreement, controlRunToFeedbackTrajectory, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, createAntiSlopJudge, createBoundedTraceAnalysisStore, createChatClient, createChatTraceEngine, createDspyRlmTraceEngine, createFeedbackTrajectory, createLlmCorrectnessChecker, createLlmReviewer, createTraceAnalyst, decidePairedPromotion, defaultBlendWeights, defineAgentEval, defineEquivalenceCheck, deployGateLayer, describeTraceInsightScope, diffFindings, diffScorecard, discoverPersonas, domainEvidencePattern, dominates, eProcess, ensembleJudge, equivalenceVerdict, errorStreakDetector, estimateCost, estimateTokens, evaluateActionPolicy, evaluateInterimReleaseConfidence, evaluateOracles, evaluateReleaseConfidence, evaluateTraceContract, expandProfileAxes, explainTraceContract, exportProductBenchmark, exportProductBenchmarkRuns, exportRunAsOtlp, extractErrorCount, extractProducedState, extractUsage, extractUsageFromSse, failureBlame, failureClusterView, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToOptimizerRow, fileVerdictCache, formatScorecardDiff, gainHistogram, gateTreatmentApplied, gradeOnHidden, gradeSemanticStatus, groupRunsByAgentProfileCell, harnessAxisOf, hashContent, hashJson, hiddenGrade, holm, improvementVerdict, inMemoryReviewStore, inferDomainKeywords, interRaterReliability, interpretCliffs, iqr, isBinaryOutcomeVector, isJudgeSpan, isLlmSpan, isModelPriced, isPolicyEdit, isRunRecord, isToolSpan, isTransientLlmError, jsonShape, jsonlReviewStore, jsonlRunRecordBackend, judgeAgreementView, judgeFamily, judgeSpans, knowledgeReadinessTracePayload, leaderboard, lintTraceContractSpec, llmJudge, loadScorecard, localCommandRunner, makeFinding, makePolicyEdit, makePolicyEditCandidateRecord, makeProposalFinding, manifestContentDigest, mannWhitneyU, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, minimumPairsForPairedDeltaTest, mintRolloutRows, modelHasSnapshot, modelPriceKey, mulberry32, notBlocked, objectiveEval, observeAll, otlpTextToTraceAnalysisStore, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedCohensDz, pairedDeltaTest, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedPromotionPower, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paretoChart, paretoFrontier, parseReflectionResponse, parseRunRecordSafe, partialCredit, partitionHeldOut, partitionTrainSelectionTest, passAtK, pearsonR, policyEditFromFinding, policyEditsFromFindings, preflightModels, productBenchmarkRepoIdentity, profile_exports as profile, projectRuntimeTrajectoryEvidence, proposeSynthesisTargets, quotaExhaustedUntil, ranks, readProductBenchmarkManifest, recordRuns, recordRunsToScorecard, redTeamDataset, redTeamReport, redact, redactForShare, redactText, regexMatches, renderPreferenceMemoryMarkdown, repeatedActionDetector, requiredPairedSampleSize, requiredPairsForPairedPromotion, requiredSampleSize, resolveModelPricing, resolveSeat, roundTripRunRecord, routeFields, runAgentControlLoop, runBoundedProcess, runCampaign, runCanaries, runCostFloorUsd, runCounterfactual, runEquivalenceCheck, runEvalCampaign, runIntentMatchJudge, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runProposeReview, runProposeReviewAsControlLoop, runSemanticConceptJudge, runTaskScore, runsForScenario, scoreKnowledgeReadiness, scorePolicyEditReadiness, scoreRedTeamOutput, scoreTraceInsightReadiness, searchCellRunId, seatPresets, selfImprove, servedModelAcceptable, shareAllowed, spearmanR, stripFencedJson, subjectiveEval, summarizeBackendIntegrity, summarizeNumberSeries, summarizePreferenceMemory, summaryTable, textInSnapshot, toAgentProfileJson, tokenizeDomainWords, toolSpansToTraceAnalysisStore, toolWasteView, traceContract, transientDispatchFailure, urlContains, userQuestionsForKnowledgeGaps, validatePolicyEdit, validatePolicyEditCandidateRecord, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, viteDeployRunner, weightedComposite, weightedMean, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit, withHeldoutBlend, withJudgeRetry, wranglerDeployRunner };
+export { AGENT_PROFILE_KINDS, ATTESTATION_ALGORITHM, AgentEvalError, AnalystRegistry, BOOTSTRAP_GATE_MIN_N, BackendIntegrityError, BudgetBreachError, BudgetGuard, CODING_HARNESSES, ConfigError, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, CostLedger, CostLedgerPersistenceError, CostReceiptCaptureError, CostReservationExceededError, CostTracker, CrossFamilyError, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_BOUNDED_PROCESS_TIMEOUT_MS, DEFAULT_FAILURE_REASON_RULES, DEFAULT_PERMUTATIONS, DEFAULT_RED_TEAM_CORPUS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, ERROR_COUNT_PATTERNS, EquivalenceProtocolError, FAILURE_BLAME, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemTraceStore, FindingsStore, HARNESS_NATIVE_MODEL, HeldOutGate, INFRA_FAILURE_BLAMES, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryTraceStore, JudgeError, LOCAL_COMMAND_RUNNER_DEFAULT_CAP_MS, LlmCallError, LlmResponseError, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, MODEL_PRICING, ModelSubstitutionError, MultiLayerVerifier, NoopRawProviderSink, NotFoundError, OUTPUT_VALUE, OtlpFileTraceStore, POLICY_EDIT_AXES, POLICY_EDIT_CANDIDATE_RECORD_SCHEMA, POLICY_EDIT_TARGET_SURFACES, PairwiseSteeringOptimizer, PolicyEditValidationError, ProductClient, PromptRegistry, REDACTION_VERSION, RunIntegrityError, RunRecordValidationError, SEMANTIC_CONCEPT_JUDGE_VERSION, ServedCrossFamilyError, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, TraceEmitter, UNKNOWN_MODEL, VERIFICATION_STRATEGIES, VERIFICATION_STRATEGY_SOURCES, ValidationError, WILCOXON_EXACT_MAX_N, acquisitionPlansForKnowledgeGaps, admitPolicyEdit, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, aggregateJudgeVerdicts, aggregateRunScore, analystFindingDigest, analystRunDigest, analystRunToFeedbackTrajectory, analystRunToReviewRequests, analyzeAntiSlop, analyzeRuns, analyzeSeries, analyzeTraces, applyPolicyEditToSurface, argHash, assertCapabilityHeadroom, assertCrossFamily, assertCrossFamilyServed, assertNoHiddenLeak, assertNoJudgeVerdict, assertProductBenchmarkRun, assertRealBackend, assertRunCaptured, assertServedModel, assertServedModels, assertSingleBackend, assessShareSafety, assignFeedbackSplit, attest, benjaminiHochberg, blendHeldout, blockingKnowledgeEval, bonferroni, bootstrapCi, budgetBreachView, buildAgentProfileCell, buildDefaultAnalystRegistry, buildEquivalenceRecord, buildReflectionPrompt, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, calibrateJudge, calibrateJudgeContinuous, canonicalJson, capabilityHeadroom, captureFetchToRawSink, certificationEvidenceDigest, checkCanaries, checkServedModel, checkTraceContracts, clamp01, classifyFailure, classifyFailureReason, cliffsDelta, cohensD, comparePairedArms, compileTraceContractSpec, completionVerdict, computeExperimentStats, computeFindingId, computePolicyEditId, computeToolUseMetrics, confidenceInterval, contentHash, continuousAgreement, controlRunToFeedbackTrajectory, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, createAntiSlopJudge, createBoundedTraceAnalysisStore, createChatClient, createChatTraceEngine, createDspyRlmTraceEngine, createFeedbackTrajectory, createLlmCorrectnessChecker, createLlmReviewer, createTraceAnalyst, decidePairedPromotion, defaultBlendWeights, defineAgentEval, defineEquivalenceCheck, deployGateLayer, describeTraceInsightScope, diffFindings, diffScorecard, discoverPersonas, domainEvidencePattern, dominates, eProcess, ensembleJudge, equivalenceVerdict, errorStreakDetector, estimateCost, estimateTokens, evaluateActionPolicy, evaluateInterimReleaseConfidence, evaluateOracles, evaluateReleaseConfidence, evaluateTraceContract, expandProfileAxes, explainTraceContract, exportProductBenchmark, exportProductBenchmarkRuns, exportRunAsOtlp, extractErrorCount, extractProducedState, extractUsage, extractUsageFromSse, failureBlame, failureClusterView, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToOptimizerRow, fileVerdictCache, formatScorecardDiff, gainHistogram, gateTreatmentApplied, gradeOnHidden, gradeSemanticStatus, groupRunsByAgentProfileCell, harnessAxisOf, hashContent, hashJson, hiddenGrade, holm, improvementVerdict, inMemoryReviewStore, inferDomainKeywords, interRaterReliability, interpretCliffs, iqr, isBinaryOutcomeVector, isJudgeSpan, isLlmSpan, isModelPriced, isPolicyEdit, isRunRecord, isToolSpan, isTransientLlmError, jsonShape, jsonlReviewStore, jsonlRunRecordBackend, judgeAgreementView, judgeFamily, judgeSpans, knowledgeReadinessTracePayload, leaderboard, lintTraceContractSpec, llmJudge, loadScorecard, localCommandRunner, makeFinding, makePolicyEdit, makePolicyEditCandidateRecord, makeProposalFinding, manifestContentDigest, mannWhitneyU, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, minimumPairsForPairedDeltaTest, mintRolloutRows, modelHasSnapshot, modelPriceKey, mulberry32, notBlocked, objectiveEval, observeAll, otlpTextToTraceAnalysisStore, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedCohensDz, pairedDeltaTest, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedPromotionPower, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paretoChart, paretoFrontier, parseReflectionResponse, parseRunRecordSafe, partialCredit, partitionHeldOut, partitionTrainSelectionTest, passAtK, pearsonR, policyEditFromFinding, policyEditsFromFindings, preflightModels, productBenchmarkRepoIdentity, profile_exports as profile, projectRuntimeTrajectoryEvidence, proposeSynthesisTargets, quotaExhaustedUntil, ranks, readProductBenchmarkManifest, recordRuns, recordRunsToScorecard, redTeamDataset, redTeamReport, redact, redactForShare, redactText, regexMatches, renderPreferenceMemoryMarkdown, repeatedActionDetector, requiredPairedSampleSize, requiredPairsForPairedPromotion, requiredSampleSize, resolveModelPricing, resolveSeat, roundTripRunRecord, routeFields, runAgentControlLoop, runBoundedProcess, runCampaign, runCanaries, runCostFloorUsd, runCounterfactual, runEquivalenceCheck, runEvalCampaign, runIntentMatchJudge, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runProposeReview, runProposeReviewAsControlLoop, runSemanticConceptJudge, runTaskScore, runsForScenario, scoreKnowledgeReadiness, scorePolicyEditReadiness, scoreRedTeamOutput, scoreTraceInsightReadiness, searchCellRunId, seatPresets, selfImprove, servedModelAcceptable, shareAllowed, spanRecordsToTraceAnalysisStore, spearmanR, stripFencedJson, subjectiveEval, summarizeBackendIntegrity, summarizeNumberSeries, summarizePreferenceMemory, summaryTable, textInSnapshot, toAgentProfileJson, tokenizeDomainWords, toolSpansToTraceAnalysisStore, toolWasteView, traceContract, transientDispatchFailure, urlContains, userQuestionsForKnowledgeGaps, validatePolicyEdit, validatePolicyEditCandidateRecord, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, viteDeployRunner, weightedComposite, weightedMean, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit, withHeldoutBlend, withJudgeRetry, wranglerDeployRunner };
 
 //# sourceMappingURL=index.js.map

package/dist/jev.d.ts

@@ -1,6 +1,6 @@
 import { b as CustomTokenPricing, g as CostReceiptInput } from "./cost-ledger-DbQdN3nO.js";
 import { i as AnalystFinding, n as AnalystContext, t as Analyst } from "./types-6VEehmEg.js";
-import { R as Scenario, S as JudgeConfig, w as JudgeScore } from "./types-Duc6xOoW.js";
+import { R as Scenario, S as JudgeConfig, w as JudgeScore } from "./types-ZUy5HGY5.js";
 import { EvaluationAnalystOptions, EvaluationContext, EvaluationJudgeOptions, EvaluationResult, Evaluator, EvaluatorOptions, asAnalyst, asJudge, createEvaluator } from "./evaluation.js";
 import { _ as parseJevRecordedRequest, a as JevQuestion, c as JevRecordedQuestions, d as JevResponseError, f as JevResult, g as parseJevQuestions, h as jevUsage, i as JevJson, l as JevRecordedRequest, m as JevState, n as JevAnswerFor, o as JevQuestions, p as JevScoreCriteria, r as JevDescription, s as JevRecordedQuestion, t as JevAnswer, u as JevRequest, v as parseJevRequest, y as parseJevResult } from "./jev-protocol-BQLxwbqD.js";
 import { a as PreparedJevReview, c as prepareJevReview, i as JevReviewReport, n as JevReviewCheck, o as assessJevReview, r as JevReviewInput, s as jevReviewFindings, t as JevReviewAssessment } from "./jev-review-Cw06AKuF.js";

package/dist/{kind-factory-CdPPHtGZ.js → kind-factory-Cn1lbsAd.js}

@@ -1280,7 +1280,7 @@ const spanStatus = z.enum([
 	"ERROR",
 	"UNSET"
 ]);
-const traceSpan = z.object({
+const traceSpanSchema = z.object({
 	trace_id: identifier,
 	span_id: identifier,
 	parent_span_id: nullableIdentifier,
@@ -1352,12 +1352,12 @@ const oversizedTrace = z.object({
 }).strict();
 const traceView = z.object({
 	trace_id: identifier,
-	spans: z.array(traceSpan).optional(),
+	spans: z.array(traceSpanSchema).optional(),
 	oversized: oversizedTrace.optional()
 }).strict().refine((value) => value.spans === void 0 !== (value.oversized === void 0), { message: "exactly one of spans or oversized is required" });
 const spansView = z.object({
 	trace_id: identifier,
-	spans: z.array(traceSpan),
+	spans: z.array(traceSpanSchema),
 	missing_span_ids: z.array(identifier),
 	omitted_span_ids: z.array(identifier),
 	has_more: z.boolean(),
@@ -2088,6 +2088,6 @@ function truncateForContext(value, max) {
 	return `${value.slice(0, max - 3).trimEnd()}...`;
 }
 //#endregion
-export { spanEpochMillis as A, FINDING_SUBJECT_SYNTAX as B, TraceFileTooLargeError as C, extractOtlpAttributes as D, compareSpanTime as E, parseRawFinding as F, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS as G, findingSubjectGrammarPromptFor as H, parseTraceSpanEvidenceUri as I, snapshotExactExecutionPlan as J, resolveTraceAnalystLimits as K, coerceJson as L, RAW_FINDING_SCHEMA_PROMPT as M, RawAnalystFindingSchema as N, firstStringAttr$1 as O, evidenceRefsFromRawFinding as P, stripCodeFences as R, TraceFileMissingError as S, asString as T, parseFindingSubject as U, KIND_EXPECTED_SUBJECTS as V, renderFindingSubject as W, SpanNotFoundError as _, TRACE_ANALYST_TOOL_NAMESPACE as a, TraceAnalysisValidationError as b, traceAnalystFunctionGroup as c, compileSearchRegex as d, truncateForBudget as f, TRACE_ANALYSIS_LIMITS as g, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX as h, buildTraceToolsForGroup as i, stringField as j, projectOtlpFlatLine as k, bindSpanSourceReader as l, DEFAULT_TRACE_ANALYST_BUDGETS as m, renderPriorFindings as n, UNTRUSTED_TRACE_TEXT as o, validateInteger as p, snapshotExactExecutionComponentIdentity as q, runTraceAnalyst as r, buildTraceAnalysisToolDescriptors as s, createTraceAnalyst as t, createBoundedTraceAnalysisStore as u, TraceAnalysisLimitError as v, TraceNotFoundError as w, TraceFileMalformedError as x, TraceAnalysisStoreContractError as y, FINDING_SUBJECT_KINDS as z };
+export { projectOtlpFlatLine as A, FINDING_SUBJECT_KINDS as B, TraceFileMissingError as C, compareSpanTime as D, asString as E, evidenceRefsFromRawFinding as F, renderFindingSubject as G, KIND_EXPECTED_SUBJECTS as H, parseRawFinding as I, snapshotExactExecutionComponentIdentity as J, DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS as K, parseTraceSpanEvidenceUri as L, stringField as M, RAW_FINDING_SCHEMA_PROMPT as N, extractOtlpAttributes as O, RawAnalystFindingSchema as P, coerceJson as R, TraceFileMalformedError as S, TraceNotFoundError as T, findingSubjectGrammarPromptFor as U, FINDING_SUBJECT_SYNTAX as V, parseFindingSubject as W, snapshotExactExecutionPlan as Y, TRACE_ANALYSIS_LIMITS as _, TRACE_ANALYST_TOOL_NAMESPACE as a, TraceAnalysisStoreContractError as b, traceAnalystFunctionGroup as c, traceSpanSchema as d, compileSearchRegex as f, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX as g, DEFAULT_TRACE_ANALYST_BUDGETS as h, buildTraceToolsForGroup as i, spanEpochMillis as j, firstStringAttr$1 as k, bindSpanSourceReader as l, validateInteger as m, renderPriorFindings as n, UNTRUSTED_TRACE_TEXT as o, truncateForBudget as p, resolveTraceAnalystLimits as q, runTraceAnalyst as r, buildTraceAnalysisToolDescriptors as s, createTraceAnalyst as t, createBoundedTraceAnalysisStore as u, SpanNotFoundError as v, TraceFileTooLargeError as w, TraceAnalysisValidationError as x, TraceAnalysisLimitError as y, stripCodeFences as z };
 
-//# sourceMappingURL=kind-factory-CdPPHtGZ.js.map
+//# sourceMappingURL=kind-factory-Cn1lbsAd.js.map