@tangle-network/agent-eval 0.116.0 → 0.117.1

package/dist/index.d.ts

@@ -1,52 +1,55 @@
-export { A as ActionExecutionPolicy, a as ActionPolicyDecision, C as ControlRunToRunRecordOptions, L as LlmJsonCall, b as LlmReviewerConfig, P as ProposeFn, c as ProposeInput, d as ProposeOutput, e as ProposeReviewConfig, f as ProposeReviewControlAction, g as ProposeReviewControlConfig, h as ProposeReviewControlResult, i as ProposeReviewControlState, j as ProposeReviewReport, k as ProposeReviewShot, R as Review, l as ReviewFn, m as ReviewInput, n as ReviewMemoryEntry, o as ReviewMemoryStore, p as RunEvidenceMetadata, V as Verification, q as VerifyFn, r as controlFailureClassFromVerification, s as controlRunToRunRecord, t as createLlmReviewer, u as evaluateActionPolicy, v as inMemoryReviewStore, w as jsonlReviewStore, x as runProposeReview, y as runProposeReviewAsControlLoop, z as scoreFromEvals } from './control-DbcDxouY.js';
-import { R as RunRecord, a as RunSplitTag } from './run-record-CZmcpWPo.js';
-export { g as AGENT_PROFILE_KINDS, h as AgentInterfaceProfileLike, A as AgentProfileCell, e as AgentProfileCellInput, i as AgentProfileCellSchemaVersion, j as AgentProfileCellValidationError, k as AgentProfileDimensionValue, l as AgentProfileHarness, f as AgentProfileJson, m as AgentProfileKind, n as AgentProfileSource, o as AgentProfileSourceInput, J as JudgeScoresRecord, b as RunCostProvenance, d as RunJudgeMetadata, p as RunOutcome, q as RunRecordValidationError, c as RunTokenUsage, r as agentProfileCellHashMaterial, s as agentProfileCellKey, t as assertRunAgentProfileCell, u as buildAgentInterfaceProfileCell, v as buildAgentProfileCell, w as groupRunsByAgentProfileCell, x as isRunRecord, y as modelHasSnapshot, z as parseRunRecordSafe, B as requireAgentProfileCell, C as resolveRunCostProvenance, D as roundTripRunRecord, E as toAgentProfileJson, F as validateAgentProfileCell, G as validateRunRecord, H as verifyAgentProfileCell } from './run-record-CZmcpWPo.js';
-import { B as BehavioralMetrics, A as RunScore, a as RunTrace, E as RunScoreWeights } from './semantic-concept-judge-CKjePUMh.js';
-export { G as ConceptComplexity, H as ConceptFinding, J as ConceptSpec, L as ConceptWeightStrategy, C as CreateAnalystAiConfig, M as DEFAULT_COMPLEXITY_WEIGHTS, N as DEFAULT_RUN_SCORE_WEIGHTS, D as DEFAULT_TRACE_ANALYST_KINDS, c as DiffPolicy, F as FAILURE_MODE_KIND_SPEC, g as FindingSubject, h as FindingSubjectKind, j as FindingsDiff, k as FindingsStore, I as IMPROVEMENT_KIND_SPEC, l as KNOWLEDGE_GAP_KIND_SPEC, m as KNOWLEDGE_POISONING_KIND_SPEC, P as PersistedFinding, R as RunCritic, O as RunCriticOptions, Q as SEMANTIC_CONCEPT_JUDGE_VERSION, n as SKILL_USAGE_ANALYST, b as SemanticConceptJudgeInput, S as SemanticConceptJudgeOptions, T as SemanticConceptJudgeResult, o as SkillUsageAnalyst, U as SuboptimalCode, V as SuboptimalSignal, W as aggregateRunScore, X as clamp01, Y as computeTraceMetrics, t as createAnalystAi, Z as createSemanticConceptJudge, u as defaultIsMaterial, v as diffFindings, _ as runSemanticConceptJudge } from './semantic-concept-judge-CKjePUMh.js';
-import { L as LlmClientOptions, m as ChatRequest, q as CreateChatClientOpts } from './policy-edit-Clb2v6Oa.js';
-export { A as Analyst, b as AnalystContext, h as AnalystCost, d as AnalystFinding, j as AnalystInputKind, k as AnalystRequirements, g as AnalystRunEvent, f as AnalystRunInputs, e as AnalystRunResult, c as AnalystRunSummary, i as AnalystSeverity, l as ChatCallOpts, C as ChatClient, n as ChatResponse, o as ChatTransport, p as CliBridgeTransportOpts, D as DirectProviderTransportOpts, E as EvidenceRef, F as FindingToPolicyEditOptions, a5 as LlmCallError, a6 as LlmCallRequest, a7 as LlmCallResult, a8 as LlmClient, a9 as LlmMessage, aa as LlmRouteAssertionError, a as LlmRouteRequirements, ab as LlmUsage, M as MockTransportOpts, r as POLICY_EDIT_AXES, s as POLICY_EDIT_CANDIDATE_RECORD_SCHEMA, t as POLICY_EDIT_TARGET_SURFACES, u as PolicyEdit, v as PolicyEditAdmission, w as PolicyEditAdmissionOptions, x as PolicyEditAxis, P as PolicyEditCandidateRecord, y as PolicyEditChange, z as PolicyEditExpectedGain, B as PolicyEditGainDirection, G as PolicyEditGainUnit, H as PolicyEditInit, I as PolicyEditRisk, J as PolicyEditSchemaVersion, K as PolicyEditSource, N as PolicyEditTarget, O as PolicyEditTargetSurface, Q as PolicyEditValidationError, R as RouterTransportOpts, S as SandboxSdkTransportOpts, T as admitPolicyEdit, U as applyPolicyEditToSurface, ac as assertLlmRoute, ad as backoffMs, ae as callLlm, af as callLlmJson, V as computeFindingId, W as computePolicyEditId, X as createChatClient, Y as isPolicyEdit, ag as isTransientLlmError, Z as makeFinding, _ as makePolicyEdit, $ as makePolicyEditCandidateRecord, a0 as policyEditFromFinding, a1 as policyEditsFromFindings, ah as probeLlm, a2 as scorePolicyEditReadiness, ai as stripFencedJson, a3 as validatePolicyEdit, a4 as validatePolicyEditCandidateRecord } from './policy-edit-Clb2v6Oa.js';
-export { a as AnalystHooks, A as AnalystRegistry, b as AnalystRegistryOptions, B as BudgetPolicy, D as DefaultAnalystRegistryOptions, R as RegistryRunOpts, c as buildDefaultAnalystRegistry } from './default-registry-DDfv22MQ.js';
-export { C as CreateTraceAnalystKindOpts, a as RawAnalystFinding, c as TraceAnalystGolden, T as TraceAnalystKindSpec, d as createTraceAnalystKind, r as renderPriorFindings } from './kind-factory-DWOvXjR_.js';
+export { A as ActionExecutionPolicy, a as ActionPolicyDecision, C as ControlRunToRunRecordOptions, L as LlmJsonCall, b as LlmReviewerConfig, P as ProposeFn, c as ProposeInput, d as ProposeOutput, e as ProposeReviewConfig, f as ProposeReviewControlAction, g as ProposeReviewControlConfig, h as ProposeReviewControlResult, i as ProposeReviewControlState, j as ProposeReviewReport, k as ProposeReviewShot, R as Review, l as ReviewFn, m as ReviewInput, n as ReviewMemoryEntry, o as ReviewMemoryStore, p as RunEvidenceMetadata, V as Verification, q as VerifyFn, r as controlFailureClassFromVerification, s as controlRunToRunRecord, t as createLlmReviewer, u as evaluateActionPolicy, v as inMemoryReviewStore, w as jsonlReviewStore, x as runProposeReview, y as runProposeReviewAsControlLoop, z as scoreFromEvals } from './control-6vuGfmDH.js';
+import { R as RunRecord, a as RunSplitTag } from './run-record-BDH49H2E.js';
+export { g as AGENT_PROFILE_KINDS, h as AgentInterfaceProfileLike, A as AgentProfileCell, e as AgentProfileCellInput, i as AgentProfileCellSchemaVersion, j as AgentProfileCellValidationError, k as AgentProfileDimensionValue, l as AgentProfileHarness, f as AgentProfileJson, m as AgentProfileKind, n as AgentProfileSource, o as AgentProfileSourceInput, J as JudgeScoresRecord, b as RunCostProvenance, d as RunJudgeMetadata, p as RunOutcome, q as RunRecordValidationError, c as RunTokenUsage, r as agentProfileCellHashMaterial, s as agentProfileCellKey, t as assertRunAgentProfileCell, u as buildAgentInterfaceProfileCell, v as buildAgentProfileCell, w as groupRunsByAgentProfileCell, x as isRunRecord, y as modelHasSnapshot, z as parseRunRecordSafe, B as requireAgentProfileCell, C as resolveRunCostProvenance, D as roundTripRunRecord, E as toAgentProfileJson, F as validateAgentProfileCell, G as validateRunRecord, H as verifyAgentProfileCell } from './run-record-BDH49H2E.js';
+import { B as BehavioralMetrics, A as RunScore, a as RunTrace, E as RunScoreWeights } from './semantic-concept-judge-CXnPEJbf.js';
+export { G as BehavioralTokenSequence, H as ConceptComplexity, J as ConceptFinding, L as ConceptSpec, M as ConceptWeightStrategy, C as CreateAnalystAiConfig, N as DEFAULT_COMPLEXITY_WEIGHTS, O as DEFAULT_RUN_SCORE_WEIGHTS, D as DEFAULT_TRACE_ANALYST_KINDS, c as DiffPolicy, F as FAILURE_MODE_KIND_SPEC, g as FindingSubject, h as FindingSubjectKind, j as FindingsDiff, k as FindingsStore, I as IMPROVEMENT_KIND_SPEC, l as KNOWLEDGE_GAP_KIND_SPEC, m as KNOWLEDGE_POISONING_KIND_SPEC, P as PersistedFinding, R as RunCritic, Q as RunCriticOptions, T as SEMANTIC_CONCEPT_JUDGE_VERSION, n as SKILL_USAGE_ANALYST, b as SemanticConceptJudgeInput, S as SemanticConceptJudgeOptions, U as SemanticConceptJudgeResult, o as SkillUsageAnalyst, V as SuboptimalCode, W as SuboptimalSignal, X as aggregateRunScore, Y as clamp01, Z as computeTraceMetrics, t as createAnalystAi, _ as createSemanticConceptJudge, u as defaultIsMaterial, v as diffFindings, $ as runSemanticConceptJudge } from './semantic-concept-judge-CXnPEJbf.js';
+import { l as ChatRequest, p as CreateChatClientOpts } from './policy-edit-wG9uFEFm.js';
+export { A as Analyst, a as AnalystContext, g as AnalystCost, c as AnalystFinding, i as AnalystInputKind, j as AnalystRequirements, f as AnalystRunEvent, e as AnalystRunInputs, d as AnalystRunResult, b as AnalystRunSummary, h as AnalystSeverity, k as ChatCallOpts, C as ChatClient, m as ChatResponse, n as ChatTransport, o as CliBridgeTransportOpts, D as DirectProviderTransportOpts, E as EvidenceRef, F as FindingToPolicyEditOptions, M as MockTransportOpts, q as POLICY_EDIT_AXES, r as POLICY_EDIT_CANDIDATE_RECORD_SCHEMA, s as POLICY_EDIT_TARGET_SURFACES, t as PolicyEdit, u as PolicyEditAdmission, v as PolicyEditAdmissionOptions, w as PolicyEditAxis, P as PolicyEditCandidateRecord, x as PolicyEditChange, y as PolicyEditExpectedGain, z as PolicyEditGainDirection, B as PolicyEditGainUnit, G as PolicyEditInit, H as PolicyEditRisk, I as PolicyEditSchemaVersion, J as PolicyEditSource, K as PolicyEditTarget, L as PolicyEditTargetSurface, N as PolicyEditValidationError, R as RouterTransportOpts, S as SandboxSdkTransportOpts, O as admitPolicyEdit, Q as applyPolicyEditToSurface, T as computeFindingId, U as computePolicyEditId, V as createChatClient, W as isPolicyEdit, X as makeFinding, Y as makePolicyEdit, Z as makePolicyEditCandidateRecord, _ as policyEditFromFinding, $ as policyEditsFromFindings, a0 as scorePolicyEditReadiness, a1 as validatePolicyEdit, a2 as validatePolicyEditCandidateRecord } from './policy-edit-wG9uFEFm.js';
+export { a as AnalystHooks, A as AnalystRegistry, b as AnalystRegistryOptions, B as BudgetPolicy, D as DefaultAnalystRegistryOptions, R as RegistryRunOpts, c as buildDefaultAnalystRegistry } from './default-registry-DaK8b3fv.js';
+export { C as CreateTraceAnalystKindOpts, a as RawAnalystFinding, c as TraceAnalystGolden, T as TraceAnalystKindSpec, d as createTraceAnalystKind, r as renderPriorFindings } from './kind-factory-ClZmO25A.js';
 import { TCloud } from '@tangle-network/tcloud';
-import { B as BenchmarkRunnerConfig, S as Scenario, c as BenchmarkReport, P as ProductClientConfig, C as CheckResult, T as TestResult, d as PersonaConfig, D as DriverResult, e as DriverState, b as JudgeFn, f as CollectedArtifacts, g as ScenarioResult, h as TurnMetrics, i as ScenarioFile, j as CompletionCriterion } from './types-C7DGg5ex.js';
-export { A as ArtifactCheck, k as ArtifactResult, E as EvalResult, F as FeedbackPattern, l as JudgeConfig, a as JudgeInput, m as JudgeRubric, J as JudgeScore, n as PersonaRigor, R as RouteMap, o as RubricDimension, p as Turn, q as TurnResult } from './types-C7DGg5ex.js';
-import { C as ControlSeverity, a as ControlEvalResult, b as FeedbackLabel, F as FeedbackTrajectoryStore, c as FeedbackTrajectory } from './feedback-trajectory-pDcz1lQ1.js';
-export { d as ControlActionFailureMode, e as ControlActionOutcome, f as ControlBudget, g as ControlContext, h as ControlDecision, i as ControlRunResult, j as ControlRuntimeConfig, k as ControlRuntimeError, l as ControlStep, m as ControlStopPolicies, n as FeedbackArtifactType, o as FeedbackAttempt, p as FeedbackLabelKind, q as FeedbackLabelSource, r as FeedbackOptimizerRow, s as FeedbackOutcome, t as FeedbackReplayAdapter, u as FeedbackReplayResult, v as FeedbackSeverity, w as FeedbackSplitPolicy, x as FeedbackTask, y as FeedbackTrajectoryFilter, z as FileSystemFeedbackTrajectoryStore, I as InMemoryFeedbackTrajectoryStore, P as PreferenceMemoryEntry, A as ProposedSideEffect, S as StopDecision, B as allCriticalPassed, D as assignFeedbackSplit, E as controlRunToFeedbackTrajectory, G as createFeedbackTrajectory, H as feedbackTrajectoriesToDatasetScenarios, J as feedbackTrajectoriesToOptimizerRows, K as feedbackTrajectoryToDatasetScenario, L as feedbackTrajectoryToOptimizerRow, M as objectiveEval, N as parseFeedbackTrajectoriesJsonl, O as renderPreferenceMemoryMarkdown, Q as replayFeedbackTrajectories, R as replayFeedbackTrajectory, T as runAgentControlLoop, U as serializeFeedbackTrajectoriesJsonl, V as stopOnNoProgress, W as stopOnRepeatedAction, X as subjectiveEval, Y as summarizePreferenceMemory, Z as withAssignedFeedbackSplit } from './feedback-trajectory-pDcz1lQ1.js';
-import { F as FailureClass, T as ToolSpan, h as BudgetSpec, B as BudgetLedgerEntry, R as Run, L as LlmSpan, S as Span } from './schema-SGWcK9wa.js';
-export { A as Artifact, E as EventKind, i as FAILURE_CLASSES, G as GenericSpan, J as JudgeSpan, M as Message, c as RetrievalSpan, g as RunLayer, f as RunStatus, d as SandboxSpan, j as SpanBase, b as SpanKind, k as SpanStatus, l as TRACE_SCHEMA_VERSION, e as TraceEvent, m as isJudgeSpan, n as isLlmSpan, o as isRetrievalSpan, p as isSandboxSpan, q as isToolSpan } from './schema-SGWcK9wa.js';
+import { B as BenchmarkRunnerConfig, S as Scenario, c as BenchmarkReport, P as ProductClientConfig, C as CheckResult, T as TestResult, d as PersonaConfig, D as DriverResult, e as DriverState, b as JudgeFn, f as CollectedArtifacts, g as ScenarioResult, h as TurnMetrics, i as ScenarioFile, j as CompletionCriterion } from './types-BkfcQnxV.js';
+export { A as ArtifactCheck, k as ArtifactResult, E as EvalResult, F as FeedbackPattern, l as JudgeConfig, a as JudgeInput, m as JudgeRubric, J as JudgeScore, n as PersonaRigor, R as RouteMap, o as RubricDimension, p as Turn, q as TurnResult } from './types-BkfcQnxV.js';
+import { C as ControlSeverity, a as ControlEvalResult, b as FeedbackLabel, F as FeedbackTrajectoryStore, c as FeedbackTrajectory } from './feedback-trajectory-BUnM58xL.js';
+export { d as ControlActionFailureMode, e as ControlActionOutcome, f as ControlBudget, g as ControlContext, h as ControlDecision, i as ControlRunResult, j as ControlRuntimeConfig, k as ControlRuntimeError, l as ControlStep, m as ControlStopPolicies, n as FeedbackArtifactType, o as FeedbackAttempt, p as FeedbackLabelKind, q as FeedbackLabelSource, r as FeedbackOptimizerRow, s as FeedbackOutcome, t as FeedbackReplayAdapter, u as FeedbackReplayResult, v as FeedbackSeverity, w as FeedbackSplitPolicy, x as FeedbackTask, y as FeedbackTrajectoryFilter, z as FileSystemFeedbackTrajectoryStore, I as InMemoryFeedbackTrajectoryStore, P as PreferenceMemoryEntry, A as ProposedSideEffect, S as StopDecision, B as allCriticalPassed, D as assignFeedbackSplit, E as controlRunToFeedbackTrajectory, G as createFeedbackTrajectory, H as feedbackTrajectoriesToDatasetScenarios, J as feedbackTrajectoriesToOptimizerRows, K as feedbackTrajectoryToDatasetScenario, L as feedbackTrajectoryToOptimizerRow, M as objectiveEval, N as parseFeedbackTrajectoriesJsonl, O as renderPreferenceMemoryMarkdown, Q as replayFeedbackTrajectories, R as replayFeedbackTrajectory, T as runAgentControlLoop, U as serializeFeedbackTrajectoriesJsonl, V as stopOnNoProgress, W as stopOnRepeatedAction, X as subjectiveEval, Y as summarizePreferenceMemory, Z as withAssignedFeedbackSplit } from './feedback-trajectory-BUnM58xL.js';
+import { F as FailureClass, T as ToolSpan, h as BudgetSpec, B as BudgetLedgerEntry, R as Run, L as LlmSpan, S as Span } from './schema-B3Q3l9Z_.js';
+export { A as Artifact, E as EventKind, i as FAILURE_CLASSES, G as GenericSpan, J as JudgeSpan, M as Message, c as RetrievalSpan, g as RunLayer, f as RunStatus, d as SandboxSpan, j as SpanBase, b as SpanKind, k as SpanStatus, l as TRACE_SCHEMA_VERSION, e as TraceEvent, m as isJudgeSpan, n as isLlmSpan, o as isRetrievalSpan, p as isSandboxSpan, q as isToolSpan } from './schema-B3Q3l9Z_.js';
+import { C as CostLedger, c as CostReceiptInput, M as MaximumCharge, e as ChannelRollup } from './cost-ledger-DWy3XdJc.js';
+export { f as CostAccountingIncompleteError, g as CostCallConflictError, h as CostCeilingReachedError, a as CostChannel, i as CostLedgerEntry, j as CostLedgerFilter, k as CostLedgerOptions, l as CostLedgerPersistence, m as CostLedgerPersistenceError, b as CostLedgerSummary, d as CostReceipt, n as CostReceiptCaptureError, o as CostReservationExceededError, p as CostResult, q as CostUsage, P as PaidCallResult, R as RunPaidCallInput, r as costForUsage, s as modelPriceKey } from './cost-ledger-DWy3XdJc.js';
 import { A as AgentEvalError, J as JudgeError, a as ConfigError } from './errors-oeQrLqXC.js';
 export { b as AgentEvalErrorCode, C as CaptureIntegrityError, N as NotFoundError, R as ReplayError, V as ValidationError, c as VerificationError } from './errors-oeQrLqXC.js';
-import { c as CorrectnessChecker } from './pre-registration--vU0mMtD.js';
-export { A as ArtifactCheckArtifact, e as ArtifactEventLike, f as ArtifactValidator, g as BackendIntegrityError, B as BackendIntegrityReport, h as ComparePairedArmsOptions, C as CompletionRequirement, a as CompletionVerdict, H as HypothesisManifest, i as HypothesisResult, j as LlmCorrectnessCheckerOpts, L as LlmJudgeDimension, d as LlmJudgeOptions, M as MatchedPair, k as PairArmsOptions, m as PairArmsResult, n as PairedArmRow, P as PairedArmsComparison, o as PairedCorrectness, p as PairedMetricDelta, q as ProducedProposal, b as ProducedState, r as ProposalEventLike, s as RequirementCheck, R as RuntimeEventLike, t as SatisfiedBy, S as SignedManifest, u as SignedManifestAlgo, T as TaskGold, v as ToolCallEventLike, V as ValidationContext, w as ValidationIssue, x as ValidationResult, y as assertRealBackend, z as byteLengthRange, D as canonicalize, E as comparePairedArms, F as completionVerdict, G as composeValidators, I as containsAll, J as createLlmCorrectnessChecker, K as createTokenRecallChecker, N as evaluateHypothesis, O as extractProducedState, Q as hashJson, U as jsonHasKeys, l as llmJudge, W as pairArms, X as parseCorrectnessResponse, Y as regexMatch, Z as signManifest, _ as summarizeBackendIntegrity, $ as verifyCompletion, a0 as verifyManifest } from './pre-registration--vU0mMtD.js';
-import { T as TraceEmitter } from './emitter-BRchAAAx.js';
-export { R as RunCompleteHook, a as RunCompleteHookContext, S as SpanHandle, b as TraceEmitterOptions, l as llmSpanFromProvider } from './emitter-BRchAAAx.js';
-import { h as ReleaseConfidenceThresholds, f as ReleaseConfidenceScorecard } from './release-report-CamNDe90.js';
-export { A as ActionableSideInfo, o as AsiSeverity, B as BootstrapOptions, a as BootstrapResult, J as JudgeReplayGateArgs, R as ReleaseConfidenceAxis, b as ReleaseConfidenceAxisName, c as ReleaseConfidenceInput, d as ReleaseConfidenceIssue, e as ReleaseConfidenceMetrics, g as ReleaseConfidenceStatus, i as ReleaseTraceEvidence, j as RenderReleaseReportOptions, V as Verdict, k as assertReleaseConfidence, l as bootstrapCi, m as evaluateReleaseConfidence, n as judgeReplayGate, r as renderReleaseReport } from './release-report-CamNDe90.js';
-export { c as CliffsMagnitude, d as CorpusAgreementOptions, e as CorpusAgreementPerDimension, C as CorpusAgreementReport, f as CorpusScoreRecord, g as EProcess, h as EProcessOptions, E as EProcessState, i as EProcessStep, M as McNemarResult, P as PairedBootstrapOptions, a as PairedBootstrapResult, j as PairedSignTestResult, k as ProportionInterval, R as RiskDifferenceResult, S as SignTestAlternative, W as WeightedCompositeInput, l as WeightedCompositeResult, b as benjaminiHochberg, m as bonferroni, n as cliffsDelta, o as cohensD, q as confidenceInterval, r as corpusInterRaterAgreement, s as corpusInterRaterAgreementFromJudgeScores, t as eProcess, u as holm, v as interRaterReliability, x as interpretCliffs, y as mannWhitneyU, z as mcnemar, A as mcnemarPower, B as mcnemarRequiredN, D as mulberry32, F as normalizeScores, p as pairedBootstrap, G as pairedMde, H as pairedRiskDifference, I as pairedSignTest, J as pairedTTest, K as partialCredit, L as passAtK, N as pearsonR, O as ranks, Q as requiredSampleSize, T as spearmanR, U as weightedComposite, V as weightedMean, w as wilcoxonSignedRank, X as wilson } from './statistics-oUbOJe-S.js';
+import { c as CorrectnessChecker } from './pre-registration-BWQhJ3vz.js';
+export { A as ArtifactCheckArtifact, e as ArtifactEventLike, f as ArtifactValidator, g as BackendIntegrityError, B as BackendIntegrityReport, h as ComparePairedArmsOptions, C as CompletionRequirement, a as CompletionVerdict, H as HypothesisManifest, i as HypothesisResult, j as LlmCorrectnessCheckerOpts, L as LlmJudgeDimension, d as LlmJudgeOptions, M as MatchedPair, k as PairArmsOptions, m as PairArmsResult, n as PairedArmRow, P as PairedArmsComparison, o as PairedCorrectness, p as PairedMetricDelta, q as ProducedProposal, b as ProducedState, r as ProposalEventLike, s as RequirementCheck, R as RuntimeEventLike, t as SatisfiedBy, S as SignedManifest, u as SignedManifestAlgo, T as TaskGold, v as ToolCallEventLike, V as ValidationContext, w as ValidationIssue, x as ValidationResult, y as assertRealBackend, z as byteLengthRange, D as canonicalize, E as comparePairedArms, F as completionVerdict, G as composeValidators, I as containsAll, J as createLlmCorrectnessChecker, K as createTokenRecallChecker, N as evaluateHypothesis, O as extractProducedState, Q as hashJson, U as jsonHasKeys, l as llmJudge, W as pairArms, X as parseCorrectnessResponse, Y as regexMatch, Z as signManifest, _ as summarizeBackendIntegrity, $ as verifyCompletion, a0 as verifyManifest } from './pre-registration-BWQhJ3vz.js';
+import { L as LlmCallMetadata, e as LlmUsage, a as LlmClientOptions } from './llm-client-qoDd18Qz.js';
+export { f as LlmCallError, c as LlmCallRequest, d as LlmCallResult, g as LlmClient, h as LlmMessage, i as LlmResponseError, j as LlmRouteAssertionError, b as LlmRouteRequirements, k as assertLlmRoute, l as backoffMs, m as callLlm, n as callLlmJson, o as costReceiptFromLlm, p as costReceiptFromLlmError, q as isTransientLlmError, r as maximumChargeForLlmRequest, s as probeLlm, t as stripFencedJson } from './llm-client-qoDd18Qz.js';
+import { T as TraceEmitter } from './emitter-CjD7vUwv.js';
+export { R as RunCompleteHook, a as RunCompleteHookContext, S as SpanHandle, b as TraceEmitterOptions, l as llmSpanFromProvider } from './emitter-CjD7vUwv.js';
+import { h as ReleaseConfidenceThresholds, f as ReleaseConfidenceScorecard } from './release-report-C8G2i5Xi.js';
+export { A as ActionableSideInfo, o as AsiSeverity, B as BootstrapOptions, a as BootstrapResult, J as JudgeReplayGateArgs, R as ReleaseConfidenceAxis, b as ReleaseConfidenceAxisName, c as ReleaseConfidenceInput, d as ReleaseConfidenceIssue, e as ReleaseConfidenceMetrics, g as ReleaseConfidenceStatus, i as ReleaseTraceEvidence, j as RenderReleaseReportOptions, V as Verdict, k as assertReleaseConfidence, l as bootstrapCi, m as evaluateReleaseConfidence, n as judgeReplayGate, r as renderReleaseReport } from './release-report-C8G2i5Xi.js';
+export { c as CliffsMagnitude, d as CorpusAgreementOptions, e as CorpusAgreementPerDimension, C as CorpusAgreementReport, f as CorpusScoreRecord, g as EProcess, h as EProcessOptions, E as EProcessState, i as EProcessStep, M as McNemarResult, P as PairedBootstrapOptions, a as PairedBootstrapResult, j as PairedSignTestResult, k as ProportionInterval, R as RiskDifferenceResult, S as SignTestAlternative, W as WeightedCompositeInput, l as WeightedCompositeResult, b as benjaminiHochberg, m as bonferroni, n as cliffsDelta, o as cohensD, q as confidenceInterval, r as corpusInterRaterAgreement, s as corpusInterRaterAgreementFromJudgeScores, t as eProcess, u as holm, v as interRaterReliability, x as interpretCliffs, y as mannWhitneyU, z as mcnemar, A as mcnemarPower, B as mcnemarRequiredN, D as mulberry32, F as normalizeScores, p as pairedBootstrap, G as pairedMde, H as pairedRiskDifference, I as pairedSignTest, J as pairedTTest, K as partialCredit, L as passAtK, N as pearsonR, O as ranks, Q as requiredSampleSize, T as spearmanR, U as weightedComposite, V as weightedMean, w as wilcoxonSignedRank, X as wilson } from './statistics-KUnG73jH.js';
 import { OtelExporter, OtelExportConfig } from './traces.js';
-export { CaptureFetchContext, CaptureFetchOptions, DEFAULT_REDACTION_RULES, ExportableSpan, ExtractedUsage, FlattenOtlpOptions, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_COST_ATTR_KEYS, LLM_COST_USD, LLM_INPUT_TOKENS, LLM_INPUT_TOKEN_ATTR_KEYS, LLM_MODEL_ATTR_KEYS, LLM_MODEL_NAME, LLM_OUTPUT_TOKENS, LLM_OUTPUT_TOKEN_ATTR_KEYS, OPENINFERENCE_SPAN_KIND, OTEL_AGENT_EVAL_SCOPE, OtlpExport, OtlpFileTraceStore, OtlpFileTraceStoreOptions, OtlpFlatLine, OtlpResourceSpans, OtlpSpan, OtlpToRunRecordsOptions, OtlpTraceRunRecord, ProjectedOtlpSpan, REDACTION_VERSION, RedactionReport, RedactionRule, ReplayCache, ReplayCacheEntry, ReplayCacheMissError, ReplayCacheStats, ReplayFetchOptions, SPAN_KIND_ATTR_KEYS, SpanNotFoundError, TOOL_NAME, TOOL_NAME_ATTR_KEYS, TRACE_ANALYST_ACTOR_DESCRIPTION, TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, TRACE_ANALYST_SUBAGENT_DESCRIPTION, TraceAggregate, TraceAnalystHookOptions, TraceFileMissingError, TraceInsightContext, TraceInsightFinding, TraceInsightPanelRole, TraceInsightPromptInput, TraceInsightQualityGate, TraceInsightQuestion, TraceInsightReadiness, TraceInsightSuite, TraceInsightTask, TraceNotFoundError, TraceStoreSource, TraceStoreToOtlpOptions, TracesToOtlpResult, asNumber, asString, buildTraceAnalystTools, buildTraceInsightContext, buildTraceInsightPrompt, captureFetchToRawSink, convertTraceStoresToOtlp, createOtelExporter, createOtelTracingStore, createReplayFetch, defaultTraceInsightPanel, describeTraceInsightScope, domainEvidencePattern, exportRunAsOtlp, extractOtlpAttributes, extractUsage, extractUsageFromResponse, extractUsageFromSse, firstNumberAttr, firstStringAttr, flattenOtlpExportToNdjson, inferDomainKeywords, inferOtlpKind, iterateRawCalls, otelRunCompleteHook, otlpToRunRecords, otlpToTraceRunRecords, planTraceInsightQuestions, projectOtlpFlatLine, readOtlpStatus, redactString, redactValue, scoreTraceInsightReadiness, stringField, tokenizeDomainWords, traceAnalystFunctionGroup, traceAnalystOnRunComplete, traceSpanKindToOpenInferenceKind } from './traces.js';
-import { a as AnalyzeTracesInput, A as AnalyzeTracesOptions, b as AnalyzeTracesResult } from './analyst-CFBc14Wc.js';
-export { c as AnalyzeTracesTurnSnapshot, d as analyzeTraces } from './analyst-CFBc14Wc.js';
-import { a as TraceAnalystSpan } from './store-9cAScOcb.js';
-export { D as DEFAULT_TRACE_ANALYST_BUDGETS, b as DatasetOverview, E as ErrorCluster, F as FileSystemRawProviderSink, c as FileSystemRawProviderSinkOptions, I as InMemoryRawProviderSink, d as InMemoryRawProviderSinkOptions, N as NoopRawProviderSink, P as ProviderRedactor, Q as QueryTracesPage, e as RawProviderDirection, f as RawProviderEvent, R as RawProviderSink, g as RawProviderSinkFilter, S as SearchSpanResult, h as SearchTraceResult, i as SpanMatchRecord, j as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, T as TraceAnalysisStore, k as TraceAnalystByteBudgets, l as TraceAnalystFilters, m as TraceAnalystSpanKind, n as TraceAnalystSpanStatus, o as TraceAnalystTraceSummary, V as ViewSpansResult, p as ViewTraceOversized, q as ViewTraceResult, r as defaultProviderRedactor, s as providerFromBaseUrl } from './store-9cAScOcb.js';
-import { b as JudgeConfig, J as JudgeScore, S as Scenario$1, g as Gate } from './types-Ca_63YSD.js';
-import { A as AnalyzeRunsOptions } from './analyze-runs-0rz_m29H.js';
-import { q as Objective, s as ParetoResult, h as GepaProposerConstraints, a as RunImprovementLoopResult } from './gepa-CQelRtuC.js';
-export { t as DEFAULT_RED_TEAM_CORPUS, D as Direction, e as RedTeamCase, u as RedTeamCategory, v as RedTeamFinding, w as RedTeamPayload, x as RedTeamReport, y as crowdingDistance, z as dominates, A as paretoFrontier, B as paretoFrontierWithCrowding, E as redTeamDataset, F as redTeamReport, H as scalarScore, I as scoreRedTeamOutput, J as toolNamesForRun } from './gepa-CQelRtuC.js';
-import { S as SandboxDriver, H as HarnessConfig, a as SandboxHarnessResult } from './test-graded-scenario-mzYBKspu.js';
-export { D as DockerSandboxDriver, c as SandboxHarness, d as SandboxResult, e as SubprocessSandboxDriver, f as SubprocessSandboxDriverOptions, g as TestGradedRunOptions, b as TestGradedRunResult, T as TestGradedScenario, h as TestOutputParser, i as composeParsers, j as jestTestParser, p as pytestTestParser, r as runTestGradedScenario, v as vitestTestParser } from './test-graded-scenario-mzYBKspu.js';
-export { b as RunIntegrityError, R as RunIntegrityExpectations, c as RunIntegrityIssue, d as RunIntegrityIssueCode, a as RunIntegrityReport, e as assertRunCaptured, t as throwIfRunIncomplete } from './integrity-C6PZ73iC.js';
-export { a as aggregateLlm, b as argHash, g as groupBy, j as judgeSpans, l as llmSpans, r as runFailureClass, c as runsForScenario, t as toolSpans } from './query-Ck190MOd.js';
-import { T as TraceStore, R as RunFilter } from './store-BsVi7ncX.js';
-export { E as EventFilter, F as FileSystemTraceStore, a as FileSystemTraceStoreOptions, I as InMemoryTraceStore, S as SpanFilter } from './store-BsVi7ncX.js';
-export { D as DEFAULT_FAILURE_RULES, b as FailureClassification, c as FailureContext, d as FailureRule, e as classifyFailure } from './failure-cluster-C48PiReX.js';
-export { P as ProjectRuntimeTrajectoryEvidenceOptions, a as RuntimeTrajectoryEvidenceProjection, b as RuntimeTrajectoryEvidenceSummary, c as RuntimeTrajectoryHookEvent, R as RuntimeTrajectoryRecord, d as RuntimeTrajectoryRunRecord, p as parseRuntimeTrajectoryHookEvent, e as projectRuntimeTrajectoryEvidence } from './runtime-trajectory-CC0jx9ql.js';
-import { a as BaselineReport, b as Trajectory, T as TrajectoryStep } from './baseline-DsNteOgR.js';
-export { B as BaselineOptions, M as MetricSamples, d as MetricVerdict, e as ToolStats, f as ToolUseMetrics, g as ToolUseOptions, h as buildTrajectory, i as compareToBaseline, c as computeToolUseMetrics, j as iqr, w as welchsTTest } from './baseline-DsNteOgR.js';
+export { CaptureFetchContext, CaptureFetchOptions, DEFAULT_REDACTION_RULES, ExportableSpan, ExtractedUsage, FlattenOtlpOptions, INPUT_VALUE, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_COST_ATTR_KEYS, LLM_COST_USD, LLM_INPUT_TOKENS, LLM_INPUT_TOKEN_ATTR_KEYS, LLM_MODEL_ATTR_KEYS, LLM_MODEL_NAME, LLM_OUTPUT_TOKENS, LLM_OUTPUT_TOKEN_ATTR_KEYS, OPENINFERENCE_SPAN_KIND, OTEL_AGENT_EVAL_SCOPE, OUTPUT_VALUE, OtlpExport, OtlpFileTraceStore, OtlpFileTraceStoreOptions, OtlpFlatLine, OtlpResourceSpans, OtlpSpan, OtlpToRunRecordsOptions, OtlpTraceRunRecord, ProjectedOtlpSpan, REDACTION_VERSION, RedactionReport, RedactionRule, ReplayCache, ReplayCacheEntry, ReplayCacheMissError, ReplayCacheStats, ReplayFetchOptions, SPAN_KIND_ATTR_KEYS, SpanNotFoundError, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS, TRACE_ANALYST_ACTOR_DESCRIPTION, TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, TRACE_ANALYST_SUBAGENT_DESCRIPTION, ToolSpanOtlpInput, TraceAggregate, TraceAnalystHookOptions, TraceFileMissingError, TraceInsightContext, TraceInsightFinding, TraceInsightPanelRole, TraceInsightPromptInput, TraceInsightQualityGate, TraceInsightQuestion, TraceInsightReadiness, TraceInsightSuite, TraceInsightTask, TraceNotFoundError, TraceStoreSource, TraceStoreToOtlpOptions, TracesToOtlpResult, applyToolSpanOtlpAttributes, asNumber, asString, buildTraceAnalystTools, buildTraceInsightContext, buildTraceInsightPrompt, captureFetchToRawSink, convertTraceStoresToOtlp, createOtelExporter, createOtelTracingStore, createReplayFetch, defaultTraceInsightPanel, describeTraceInsightScope, domainEvidencePattern, exportRunAsOtlp, extractOtlpAttributes, extractUsage, extractUsageFromResponse, extractUsageFromSse, firstNumberAttr, firstStringAttr, flattenOtlpExportToNdjson, inferDomainKeywords, inferOtlpKind, iterateRawCalls, otelRunCompleteHook, otlpToRunRecords, otlpToTraceRunRecords, planTraceInsightQuestions, projectOtlpFlatLine, readOtlpStatus, redactString, redactValue, scoreTraceInsightReadiness, stringField, tokenizeDomainWords, traceAnalystFunctionGroup, traceAnalystOnRunComplete, traceSpanKindToOpenInferenceKind } from './traces.js';
+import { a as AnalyzeTracesInput, A as AnalyzeTracesOptions, b as AnalyzeTracesResult } from './analyst-C8HHvfJp.js';
+export { c as AnalyzeTracesTurnSnapshot, d as analyzeTraces } from './analyst-C8HHvfJp.js';
+import { a as TraceAnalystSpan } from './store-C1YxJDEK.js';
+export { D as DEFAULT_TRACE_ANALYST_BUDGETS, b as DatasetOverview, E as ErrorCluster, Q as QueryTracesPage, S as SearchSpanResult, c as SearchTraceResult, d as SpanMatchRecord, e as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, T as TraceAnalysisStore, f as TraceAnalystByteBudgets, g as TraceAnalystFilters, h as TraceAnalystSpanKind, i as TraceAnalystSpanStatus, j as TraceAnalystTraceSummary, V as ViewSpansResult, k as ViewTraceOversized, l as ViewTraceResult } from './store-C1YxJDEK.js';
+import { b as JudgeConfig, J as JudgeScore, S as Scenario$1, G as Gate } from './types-BSw1rOUB.js';
+import { A as AnalyzeRunsOptions } from './analyze-runs--2x39HZ7.js';
+import { z as Objective, A as ParetoResult, p as GepaProposerConstraints, a as RunImprovementLoopResult } from './gepa-eESocoDi.js';
+export { B as DEFAULT_RED_TEAM_CORPUS, D as Direction, b as REFERENCE_EQUIVALENCE_INPUT_LIMITS, c as REFERENCE_EQUIVALENCE_JUDGE_VERSION, n as RedTeamCase, E as RedTeamCategory, F as RedTeamFinding, H as RedTeamPayload, I as RedTeamReport, d as ReferenceEquivalenceJudgeInput, e as ReferenceEquivalenceJudgeOptions, f as ReferenceEquivalenceJudgeResult, g as ReferenceEquivalenceScenario, j as createReferenceEquivalenceJudge, J as crowdingDistance, K as dominates, L as paretoFrontier, M as paretoFrontierWithCrowding, N as redTeamDataset, Q as redTeamReport, m as runReferenceEquivalenceJudge, S as scalarScore, T as scoreRedTeamOutput, U as toolNamesForRun } from './gepa-eESocoDi.js';
+import { S as SandboxDriver, H as HarnessConfig, a as SandboxHarnessResult } from './test-graded-scenario-B0ybnPY7.js';
+export { D as DockerSandboxDriver, c as SandboxHarness, d as SandboxResult, e as SubprocessSandboxDriver, f as SubprocessSandboxDriverOptions, g as TestGradedRunOptions, b as TestGradedRunResult, T as TestGradedScenario, h as TestOutputParser, i as composeParsers, j as jestTestParser, p as pytestTestParser, r as runTestGradedScenario, v as vitestTestParser } from './test-graded-scenario-B0ybnPY7.js';
+export { b as RunIntegrityError, R as RunIntegrityExpectations, c as RunIntegrityIssue, d as RunIntegrityIssueCode, a as RunIntegrityReport, e as assertRunCaptured, t as throwIfRunIncomplete } from './integrity-DqlBiLyK.js';
+export { a as aggregateLlm, b as argHash, g as groupBy, h as hasCapturedToolArgs, j as judgeSpans, l as llmSpans, r as runFailureClass, c as runsForScenario, t as toolSpans } from './query-CF7PG61p.js';
+export { F as FileSystemRawProviderSink, a as FileSystemRawProviderSinkOptions, I as InMemoryRawProviderSink, b as InMemoryRawProviderSinkOptions, N as NoopRawProviderSink, P as ProviderRedactor, c as RawProviderDirection, d as RawProviderEvent, R as RawProviderSink, e as RawProviderSinkFilter, f as defaultProviderRedactor, p as providerFromBaseUrl } from './raw-provider-sink-C46HDghv.js';
+import { T as TraceStore, R as RunFilter } from './store-DGqD0Pyo.js';
+export { E as EventFilter, F as FileSystemTraceStore, a as FileSystemTraceStoreOptions, I as InMemoryTraceStore, S as SpanFilter } from './store-DGqD0Pyo.js';
+export { D as DEFAULT_FAILURE_RULES, b as FailureClassification, c as FailureContext, d as FailureRule, e as classifyFailure } from './failure-cluster-DOAcSJ87.js';
+export { P as ProjectRuntimeTrajectoryEvidenceOptions, a as RuntimeTrajectoryEvidenceProjection, b as RuntimeTrajectoryEvidenceSummary, c as RuntimeTrajectoryHookEvent, R as RuntimeTrajectoryRecord, d as RuntimeTrajectoryRunRecord, p as parseRuntimeTrajectoryHookEvent, e as projectRuntimeTrajectoryEvidence } from './runtime-trajectory-DGBIUt4B.js';
+import { a as BaselineReport, b as Trajectory, T as TrajectoryStep } from './baseline-DKq3gJpP.js';
+export { B as BaselineOptions, M as MetricSamples, d as MetricVerdict, e as ToolStats, f as ToolUseMetrics, g as ToolUseOptions, h as buildTrajectory, i as compareToBaseline, c as computeToolUseMetrics, j as iqr, w as welchsTTest } from './baseline-DKq3gJpP.js';
 import { HarnessType, AgentProfile } from '@tangle-network/agent-interface';
 export { AgentProfile, HarnessType } from '@tangle-network/agent-interface';
-import { b as ChannelRollup, C as CostLedger } from './cost-ledger-DuSqlw5B.js';
-export { a as CostChannel, c as CostLedgerEntry, d as CostLedgerSummary, e as CostResult, f as CostUsage, g as costForUsage, m as modelPriceKey } from './cost-ledger-DuSqlw5B.js';
 export { S as SeriesConvergenceOptions, a as SeriesConvergenceResult, b as analyzeSeries } from './series-convergence-D5OWMBg6.js';
 import { D as DefaultVerdict } from './verdict-C9MlYujm.js';
 import { D as DatasetScenario, c as Dataset } from './dataset-NENEzRgk.js';
@@ -54,14 +57,14 @@ export { d as DatasetDifficulty, b as DatasetManifest, e as DatasetProvenance, a
 export { a as CalibrationResult, c as CandidateScore, C as ContinuousAgreement, d as ContinuousAgreementOptions, b as ContinuousCalibrationResult, G as GoldenItem, P as PositionalBiasResult, S as SelfPreferenceResult, V as VerbosityBiasResult, e as calibrateJudge, f as calibrateJudgeContinuous, g as continuousAgreement, p as positionalBias, s as selfPreference, v as verbosityBias } from './judge-calibration-7C-IDmKr.js';
 import { L as Layer, S as Severity, b as LayerResult, c as VerifyContext } from './multi-layer-verifier-BsqKuLyN.js';
 export { F as Finding, d as LayerStatus, M as MultiLayerVerifier, a as VerificationReport, V as VerifyOptions, g as gradeSemanticStatus } from './multi-layer-verifier-BsqKuLyN.js';
-export { B as BENCHMARK_SPLIT_SEED, a as BenchmarkAdapter, b as BenchmarkDatasetItem, c as BenchmarkEvaluation, d as BenchmarkFamily, e as BenchmarkResponder, f as BenchmarkScenario, g as BenchmarkSource, h as BenchmarkTaskKind, i as benchmarkDeterministicSplit, j as benchmarks } from './index-DbCXJfZ1.js';
-export { C as CallbackResearcher, d as CallbackResearcherOptions, e as CampaignFactoryParams, f as CampaignIntegrityPolicy, g as CampaignRunContext, h as CampaignRunOutcome, i as CampaignRunner, j as CampaignScenario, k as CampaignVariant, c as EvalCampaignOptions, b as EvalCampaignResult, E as ExperimentPlan, a as ExperimentResult, l as FailedRun, F as FailureMode, N as NoopResearcher, R as Researcher, S as SteeringChange, r as runEvalCampaign } from './researcher-Dwbo_Fxx.js';
-export { G as GainDistributionBin, a as GainDistributionFigureSpec, b as GainDistributionOptions, m as GateDecision, n as GateEvidence, H as HeldOutGate, o as HeldOutGateConfig, q as HeldOutGateRejectionCode, P as ParetoFigureSpec, c as ParetoPoint, R as RESEARCH_REPORT_HARD_PAIR_FLOOR, d as ResearchReport, e as ResearchReportCandidate, f as ResearchReportDecision, g as ResearchReportMethodology, h as ResearchReportOptions, i as ResearchReportRecommendation, S as SummaryTable, j as SummaryTableOptions, k as SummaryTableRow, l as gainHistogram, p as paretoChart, r as researchReport, s as summaryTable } from './summary-report-DTNgQycC.js';
+export { B as BENCHMARK_SPLIT_SEED, a as BenchmarkAdapter, b as BenchmarkDatasetItem, c as BenchmarkEvaluation, d as BenchmarkFamily, e as BenchmarkResponder, f as BenchmarkScenario, g as BenchmarkSource, h as BenchmarkTaskKind, i as benchmarkDeterministicSplit, j as benchmarks } from './index-PdX4VnPA.js';
+export { C as CallbackResearcher, d as CallbackResearcherOptions, e as CampaignFactoryParams, f as CampaignIntegrityPolicy, g as CampaignRunContext, h as CampaignRunOutcome, i as CampaignRunner, j as CampaignScenario, k as CampaignVariant, c as EvalCampaignOptions, b as EvalCampaignResult, E as ExperimentPlan, a as ExperimentResult, l as FailedRun, F as FailureMode, N as NoopResearcher, R as Researcher, S as SteeringChange, r as runEvalCampaign } from './researcher-C8XyxQsu.js';
+export { G as GainDistributionBin, a as GainDistributionFigureSpec, b as GainDistributionOptions, m as GateDecision, n as GateEvidence, H as HeldOutGate, o as HeldOutGateConfig, q as HeldOutGateRejectionCode, P as ParetoFigureSpec, c as ParetoPoint, R as RESEARCH_REPORT_HARD_PAIR_FLOOR, d as ResearchReport, e as ResearchReportCandidate, f as ResearchReportDecision, g as ResearchReportMethodology, h as ResearchReportOptions, i as ResearchReportRecommendation, S as SummaryTable, j as SummaryTableOptions, k as SummaryTableRow, l as gainHistogram, p as paretoChart, r as researchReport, s as summaryTable } from './summary-report-C5bKFfm-.js';
 export { I as InterimReleaseConfidence, a as InterimReleaseConfidenceInput, P as PairedEvalueOptions, b as PairedEvalueSequence, c as PairedEvalueStep, S as SequentialDecision, e as evaluateInterimReleaseConfidence, p as pairedEvalueSequence } from './sequential-5iSVfzl2.js';
 import '@ax-llm/ax';
 import 'zod';
-import './insight-report-oMVxDTxl.js';
-import './storage-Dw_f7WMt.js';
+import './insight-report-DY4nDW9Q.js';
+import './storage-DrX3v_5B.js';
 
 /**
  * Automated pull-request transports for the production loop.
@@ -568,6 +571,10 @@ interface AgentDriverConfig {
     driverModel?: string;
     /** System prompt context for the driver LLM to understand the product */
     productContext?: string;
+    /** Shared account for driver-model calls. */
+    costLedger?: CostLedger;
+    /** Exact provider attempt count, required when costLedger has a cap. */
+    tcloudMaximumAttempts?: number;
 }
 /**
  * AgentDriver — meta-agent that plays a persona against the real product.
@@ -581,6 +588,8 @@ declare class AgentDriver {
     private client;
     private driverModel;
     private productContext;
+    private costLedger;
+    private tcloudMaximumAttempts?;
     constructor(tc: TCloud, config: AgentDriverConfig);
     /**
      * Run a persona through the product.
@@ -645,6 +654,12 @@ interface DecideNextUserTurnOpts {
     productContext?: string;
     /** Driver LLM model. Defaults to claude-sonnet-4-6. */
     model?: string;
+    /** Shared account for the paid driver-model call. */
+    costLedger?: CostLedger;
+    /** Attribution tags merged into the paid-call receipt. */
+    costTags?: Record<string, string>;
+    /** Exact provider attempt count, required when costLedger has a cap. */
+    tcloudMaximumAttempts?: number;
 }
 /**
  * Decide the simulated user's next turn — the reactive, adversarial
@@ -662,6 +677,13 @@ interface ExecutorConfig {
     model?: string;
     /** Judges to run after execution */
     judges: JudgeFn[];
+    /** Shared ledger for paid built-in judges. */
+    costLedger?: CostLedger;
+    costPhase?: string;
+    costTags?: Record<string, string>;
+    /** Exact maximum provider attempts configured on the supplied TCloud client. */
+    tcloudMaximumAttempts?: number;
+    signal?: AbortSignal;
     /** Regex patterns for detecting tool/API calls in responses */
     toolCallPatterns?: RegExp[];
     /** Block delimiter pattern (default: :::type\n...\n:::) */
@@ -862,6 +884,10 @@ declare class CrossFamilyError extends Error {
  */
 declare function assertCrossFamily(models: string[], opts?: AssertCrossFamilyOptions): JudgeFamily[];
 
+type JudgeParseErrorOptions = {
+    cause?: unknown;
+    llmCall?: LlmCallMetadata;
+};
 /**
  * A judge's LLM response could not be parsed into scored dimensions.
  * Thrown instead of fabricating a `{ dimension: 'parse_error', score: 0 }`
@@ -874,9 +900,9 @@ declare class JudgeParseError extends JudgeError {
     readonly judgeName: string;
     /** The raw (truncated) model response that failed to parse. */
     readonly raw: string;
-    constructor(judgeName: string, raw: string, options?: {
-        cause?: unknown;
-    });
+    /** Paid-call metadata remains available even when the verdict is unusable. */
+    readonly llmCall?: LlmCallMetadata;
+    constructor(judgeName: string, raw: string, options?: JudgeParseErrorOptions);
 }
 /**
  * Create a domain expert judge with a configurable domain.
@@ -1725,9 +1751,9 @@ declare function summarizeHarnessResults(results: HarnessRunResult[]): HarnessVa
  *
  * Fail-loud: a judge that errored or returned malformed output is recorded in
  * `failedJudges` with `perDimension: null`, never folded into a zero. If EVERY
- * judge failed the reducer throws — a silent zero here would corrupt the gate's
- * number. A failed judge still burned tokens, so its `costUsd` is still summed.
+ * judge failed the reducer throws — a silent zero here would corrupt the result.
  */
+
 /** One judge's verdict. `perDimension: null` ⇒ that judge call failed (threw or
  *  returned malformed output) — recorded, never folded into a zero. */
 interface JudgeVerdict<D extends string = string> {
@@ -1737,8 +1763,10 @@ interface JudgeVerdict<D extends string = string> {
     perDimension: Record<D, number> | null;
     /** Optional one-line rationale; the first non-empty one becomes the aggregate's. */
     rationale?: string;
-    /** Optional reported cost — summed across ALL verdicts (failed included). */
+    /** Optional reported cost, committed by the caller's CostLedger. */
     costUsd?: number;
+    /** Optional provider usage, committed by the caller's CostLedger. */
+    usage?: LlmUsage;
     /** Optional per-dimension reasoning/evidence. Carried through to
      *  `EnsembleAggregate.verdicts` verbatim — never folded into the math. */
     detail?: Partial<Record<D, {
@@ -1758,8 +1786,6 @@ interface EnsembleAggregate<D extends string = string> {
     maxDisagreement: number;
     /** Models whose verdict was null (failed). */
     failedJudges: string[];
-    /** Sum of `costUsd` over ALL verdicts, failed included. */
-    costUsd: number;
     /** First non-empty survivor rationale, or `'llm-judge'`. */
     rationale: string;
     /** The input verdicts, verbatim — drill-down to raw scores, `detail`
@@ -1877,6 +1903,8 @@ interface EnsembleJudgeOptions<D extends string> {
     /** Judge model ids — one `scoreWith` call per entry. List a model twice to
      *  sample it twice (votes are suffix-keyed `model#2` so none overwrite). */
     models: string[];
+    /** Explicit scoring revision for closure state not visible in the static panel config. */
+    judgeVersion?: string;
     /**
      * Score the artifact with one model. Throw (or reject) on failure — the
      * panel records that model as a failed judge; it is never folded into a
@@ -1885,7 +1913,14 @@ interface EnsembleJudgeOptions<D extends string> {
     scoreWith: (model: string, input: {
         artifact: unknown;
         scenario?: unknown;
+        signal: AbortSignal;
     }) => Promise<JudgeVerdict<D>>;
+    /** Recover usage from a failed provider response when its error retains one. */
+    receiptFromError?: (error: Error, model: string) => CostReceiptInput | undefined;
+    /** Used by direct score calls; campaigns supply their run ledger in score(). */
+    costLedger?: CostLedger;
+    /** Required per model when the shared ledger has a dollar cap. */
+    maximumCharge?: MaximumCharge | ((model: string) => MaximumCharge);
     /**
      * Per-model retry policy, applied via `withJudgeRetry`. The panel's
      * `models` list drives the fan-out, so `retry.models` (the fallback
@@ -4871,10 +4906,14 @@ interface IntentMatchResult {
 interface IntentMatchOptions {
     model?: string;
     timeoutMs?: number;
+    maxTokens?: number;
     maxSourceChars?: number;
     maxPerFileChars?: number;
     maxHtmlChars?: number;
     llm?: LlmClientOptions;
+    costLedger?: CostLedger;
+    costPhase?: string;
+    signal?: AbortSignal;
 }
 /**
  * Run the intent-match judge. Soft-fails to available=false on error.
@@ -5314,8 +5353,14 @@ interface CreateDefaultReviewerOptions {
     model: string;
     /** Per-call timeout. Default 300s. */
     timeoutMs?: number;
+    /** Provider-enforced output limit. Default 4000. */
+    maxTokens?: number;
     /** LlmClient transport config (baseUrl, apiKey, authHeader, etc.). */
     llm?: LlmClientOptions;
+    /** Shared run spend account. */
+    costLedger?: CostLedger;
+    costPhase?: string;
+    signal?: AbortSignal;
     /**
      * Override the prompt builder. Default: `buildReviewerPrompt`.
      * Consumers with different reviewer voices pass their own.
@@ -6878,4 +6923,4 @@ type CachedJudge<TArtifact, TScenario extends Scenario$1 = Scenario$1> = JudgeCo
  */
 declare function cachedJudge<TArtifact, TScenario extends Scenario$1 = Scenario$1>(judge: JudgeConfig<TArtifact, TScenario>, store: VerdictCacheStore, options: CachedJudgeOptions): CachedJudge<TArtifact, TScenario>;
 
-export { ATTESTATION_ALGORITHM, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, type AgentProfileRuntimeReceipt, type AgreementResult, type AlignmentOp, AnalyzeTracesInput, AnalyzeTracesOptions, AnalyzeTracesResult, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, type BackendDescriptor, BaselineReport, BehaviorAssertion, BehavioralMetrics, BenchmarkReport, BenchmarkRunner, BenchmarkRunnerConfig, type BisectOptions, type BisectResult, type BisectStep, type BlendWeights, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, BudgetSpec, type BuildAgreementJudgeOptions, CODING_HARNESSES, type CachedJudge, type CachedJudgeOptions, CallExpectation, type CanaryAlert, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateComparison, type CandidateScenario, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, type CausalAttributionReport, type CellVerdict, ChannelRollup, ChatRequest, CheckResult, type ClusterBootstrapInterval, type ClusterSignFlipAlternative, type ClusterSignFlipResult, type ClusteredBinaryCluster, type ClusteredMatchedPair, type ClusteredPairedBinaryOptions, type ClusteredPairedBinaryResult, type ClusteredPairedBinaryStatistics, CollectedArtifacts, type CommandRunner, type CompareLabels, CompletionCriterion, ConfigError, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContractCheckResult, type ContractJudgeOptions, type ContractMetric, type ContractReport, type ContractRule, type ContractRuleKind, type ContractSpan, type ContractVerdict, type ContractViolation, ControlEvalResult, ControlSeverity, ConvergenceTracker, CorrectnessChecker, type CostEntry, CostLedger, type CostReport, type CostSummary, CostTracker, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, CreateChatClientOpts, type CreateDefaultReviewerOptions, type CreateExperimentInput, type CreateSandboxPoolOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, DEFAULT_AGENT_SLOS, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, type DataAcquisitionPlan, Dataset, DatasetScenario, type DecideNextUserTurnOpts, DefaultVerdict, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DescriptionLengthCandidate, type DescriptionLengthConfig, type DescriptionLengthDecision, type DescriptionLengthEvidence, DescriptionLengthGate, type DescriptionLengthRejectionCode, type DetectorEvent, type DetectorSeverity, type DetectorSignal, type DiffScorecardOptions, type DirEntry, type DiscoverPersonasOptions, type DiscoveredPersona, DriverResult, DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, ERROR_COUNT_PATTERNS, type EnsembleAggregate, type EnsembleJudgeOptions, type ErrorCountPattern, type ErrorStreakOptions, type EvalToolDef, EvalTraceStore, type EvolutionRound, type ExecutorConfig, type Expectation, type Experiment, type ExperimentProvenance, type ExperimentRep, type ExperimentStats, type ExperimentStore, ExperimentTracker, type ExperimentTrackerOptions, type ExperimentVerdict, type ExportedRewardModel, type ExtractOptions, type ExtractResult, type FactorContribution, type FactorialCell, FailureClass, FeedbackLabel, FeedbackTrajectory, FeedbackTrajectoryStore, type FieldAgreementSpec, type FieldDestination, type FileChange, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type GhCliClientOptions, type GoldScenario, type GoldSplit, type GoldenSeverity, type GoldenSpec, HARNESS_NATIVE_MODEL, type HarnessAdapter, HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessVariant, type HarnessVariantReport, type HeadroomClass, type HeadroomInput, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, HoldoutAuditor, type HttpGithubClientOptions, INTENT_MATCH_JUDGE_VERSION, type ImageData, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, JudgeError, type JudgeFamily, type JudgeFleetOptions, JudgeFn, JudgeParseError, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, JudgeRunner, type JudgeScoreInput, type JudgeVerdict, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type KnowledgeAcquisitionMode, type KnowledgeBundle, type KnowledgeFallbackPolicy, type KnowledgeFreshness, type KnowledgeImportance, type KnowledgeReadinessReport, type KnowledgeRecommendedAction, type KnowledgeRequirement, type KnowledgeRequirementCategory, type KnowledgeResponsibleSurface, type KnowledgeSensitivity, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, Layer, LayerResult, type LeaderboardOptions, type LeaderboardRow, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmClientOptions, LlmSpan, LockedJsonlAppender, MODEL_PRICING, type MakeEvalToolsConfig, type MatchResult, type MatcherResult, type MeasurementPolicy, type MergeOptions, MetricsCollector, type ModelCostRollup, type ModelPreflight, type ModelSeats, ModelsUnreachableError, type MuffledFinder, type MuffledFinding, type MultiToolchainLayerConfig, type Mutator, Mutex, type NoLeakOptions, type NoProgressOptions, Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, ParetoResult, type ParseStudentLabel, type PartitionHeldOutOptions, PersonaConfig, type Playbook, type PlaybookEntry, type PoolSlot, type PrReviewAuditCase, type PrReviewBenchmarkSummary, type PrReviewComment, type PrReviewMatchedFinding, type PrReviewOutcome, type PrReviewReferenceFinding, type PrReviewScore, type PrReviewScoreWeights, type PrReviewSeverity, type PrReviewSource, type PreflightModelsOptions, type PreflightOutcome, type ProductBenchmarkArm, type ProductBenchmarkArtifactPaths, type ProductBenchmarkBudgets, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkProfileRef, type ProductBenchmarkRecord, type ProductBenchmarkRepoRef, type ProductBenchmarkRunInput, type ProductBenchmarkScenario, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkSubstrateVersions, type ProductBenchmarkValidationReport, ProductClient, ProductClientConfig, type ProfileAxisSpec, type PromptHandle, PromptRegistry, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, type ProvenanceReader, type RecordRunsOptions, type ReferenceMatchResult, type ReferenceReplayAdapter, type ReferenceReplayAdapterFn, type ReferenceReplayAdapterLike, type ReferenceReplayAggregate, type ReferenceReplayCandidate, type ReferenceReplayCase, type ReferenceReplayCaseRun, type ReferenceReplayExecutionScenario, type ReferenceReplayItem, type ReferenceReplayMatch, type ReferenceReplayMatchStrategy, type ReferenceReplayMatcher, type ReferenceReplayPromotionDecision, type ReferenceReplayPromotionPolicy, type ReferenceReplayRun, type ReferenceReplayRunContext, type ReferenceReplayRunOptions, type ReferenceReplayRunStore, type ReferenceReplayScenario, type ReferenceReplayScenarioScore, type ReferenceReplayScore, type ReferenceReplayScoreOptions, type ReferenceReplaySplit, type ReferenceReplaySplitComparison, type ReferenceReplaySteeringRowsOptions, type ReflectionContext, type ReflectionProposal, ReleaseConfidenceScorecard, ReleaseConfidenceThresholds, type RenderStudentPrompt, type RepeatedActionOptions, type RepoRef, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RobustnessResult, type RoutedField, Run, type RunCommandInput, type RunCommandResult, type RunDistillationOptions, type RunDistillationResult, RunFilter, RunRecord, type RunRecordBackend, type RunRecordFilter, RunScore, RunScoreWeights, RunSplitTag, RunTrace, type RuntimeResolution, SandboxDriver, SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type ScanOptions, Scenario, type ScenarioCost, ScenarioFile, ScenarioRegistry, ScenarioResult, type ScoreKnowledgeReadinessOptions, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SeatName, type SeatPresetName, SeatUnsetError, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SerializedRegex, Severity, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, Span, type SpanPredicate, type SplitGoldOptions, type SteeringBundle, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type StreamingDetector, type SynthesisReason, type SynthesisTarget, type TaskHeadroom, TestResult, type TextMatcher, type ThresholdContract, TokenCounter, type TokenSpec, type ToolMatcher, ToolSpan, TraceAnalystSpan, type TraceContract, TraceContractBuilder, TraceEmitter, TraceStore, type TracedAnalystOptions, type TracedJudgeOptions, Trajectory, TrajectoryStep, type TreatmentClass, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, TurnMetrics, UI_FINDING_SEVERITIES, UI_LENSES, UNIVERSAL_FINDERS, type UiFinding, type UiFindingScreenshot, type UiFindingSeverity, type UiLens, type UserQuestion, type VerdictCacheStats, type VerdictCacheStore, VerifyContext, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, type WorkerDriverContext, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, acquisitionPlansForKnowledgeGaps, adversarialJudge, agentProfileHash, agentProfileId, agentProfileModelId, agentVisibleFields, aggregateJudgeVerdicts, aggregatePrReviewScore, analyzeAntiSlop, appendScorecard, assertCapabilityHeadroom, assertCrossFamily, assertModelsServed, assertNoHiddenLeak, assertProductBenchmarkRun, assertSingleBackend, assignHeldOutTag, attachCostToReport, attest, attributeCounterfactuals, bisect, blendHeldout, blockingKnowledgeEval, buildAgreementJudge, buildDriverSystemPrompt, buildProductBenchmarkManifest, buildReflectionPrompt, buildReviewerPrompt, buildWorkerDriverSystemPrompt, cachedJudge, canaryLeakView, canonicalJson, capabilityHeadroom, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, checkTraceContracts, classifyTreatment, clusteredPairedBinary, codeExecutionJudge, coherenceJudge, collectionPreserved, commentsForSource, commitBisect, compareReferenceReplay, compilerJudge, computeExperimentStats, contentHash, contractJudge, costReport, createAntiSlopJudge, createCustomJudge, createDefaultReviewer, createDomainExpertJudge, createIntentMatchJudge, createSandboxPool, crossTraceDiff, dataDescriptionBits, decideNextUserTurn, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultBlendWeights, defaultJudges, defaultParseStudentLabel, defaultReferenceReplayMatcher, defaultRenderStudentPrompt, deployGateLayer, diffScorecard, discoverPersonas, distillPlaybook, ensembleJudge, errorStreakDetector, estimateCost, estimateTokens, evaluateContract, evaluateOracles, evaluateTraceContract, executeScenario, expandProfileAxes, expectAgent, exportProductBenchmark, exportProductBenchmarkRuns, exportRewardModel, extractAssetUrls, extractErrorCount, fieldAgreement, fileContains, fileExists, fileExperimentStore, fileVerdictCache, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findProductBenchmarkArtifacts, findSkipCountsAsPass, flowLayer, fnv1a32, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, gateTreatmentApplied, gateTreatmentFromMetrics, gateTreatmentFromSpans, gateTreatmentFromToolSpans, ghCliClient, gitProvenanceReader, precision as goldenPrecision, gradeOnHidden, harnessAxisOf, hashContent, hashToUnit, hiddenGrade, htmlContainsElement, httpGithubClient, improvementVerdict, inMemoryExperimentStore, inMemoryReferenceReplayStore, inMemoryRunRecordBackend, inMemoryVerdictCache, isHiddenDestination, isModelPriced, isOtelConfigured, jsonShape, jsonlReferenceReplayStore, jsonlRunRecordBackend, judgeFamily, keyPreserved, knowledgeReadinessTracePayload, leaderboard, linterJudge, loadGoldScenarios, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, makeEvalTools, matchGoldens, matchSpan, mergeLayerResults, mergeSteeringBundle, modelDescriptionBits, multiToolchainLayer, noProgressDetector, notBlocked, observeAll, paraphraseRobustness, paraphraseRobustnessScenarios, parseGoldJsonl, parseReflectionResponse, partitionHeldOut, passOrthogonality, pixelDeltaRatio, politenessPrefixMutator, preflightModels, printDriverSummary, productBenchmarkIntegrityFailures, productBenchmarkMutableSurfaces, productBenchmarkRepoIdentity, productBenchmarkSplits, index as profile, promptBisect, proposeSynthesisTargets, readProductBenchmarkManifest, readProductBenchmarkRecords, recordRuns, recordRunsToScorecard, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatches, renderMarkdownReport, renderPlaybookMarkdown, renderSteeringText, repeatedActionDetector, replayScorerOverCorpus, replayTraceThroughJudge, resolveModelPricing, resolveSeat, routeFields, rowCount, rowWhere, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runDistillation, runE2EWorkflow, runExpectations, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runRecordToProductBenchmarkRecord, runReferenceReplay, runScore, runSelfPlay, scanForMuffledGates, scoreContinuity, scoreKnowledgeReadiness, scorePrReviewComments, scorePrReviewSource, scoreReferenceReplay, seatPresets, securityJudge, selectHarnessVariant, sentenceReorderMutator, splitGold, statusAdvanced, summarizeHarnessResults, summarizePrReviewBenchmark, testJudge, textInSnapshot, toLangfuseEnvelope, toOpenAiTool, toPrometheusText, traceContract, traceJudge, traceJudgeEnsemble, tracedAnalyzeTraces, typoMutator, urlContains, userQuestionsForKnowledgeGaps, validateProductBenchmarkManifest, validateProductBenchmarkRecord, validateProductBenchmarkRun, verifyAttestation, visualDiff, viteDeployRunner, weightedRecall, whitespaceCollapseMutator, withHeldoutBlend, withJudgeRetry, withOtelPipeline, wranglerDeployRunner };
+export { ATTESTATION_ALGORITHM, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, type AgentProfileRuntimeReceipt, type AgreementResult, type AlignmentOp, AnalyzeTracesInput, AnalyzeTracesOptions, AnalyzeTracesResult, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, type BackendDescriptor, BaselineReport, BehaviorAssertion, BehavioralMetrics, BenchmarkReport, BenchmarkRunner, BenchmarkRunnerConfig, type BisectOptions, type BisectResult, type BisectStep, type BlendWeights, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, BudgetSpec, type BuildAgreementJudgeOptions, CODING_HARNESSES, type CachedJudge, type CachedJudgeOptions, CallExpectation, type CanaryAlert, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateComparison, type CandidateScenario, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, type CausalAttributionReport, type CellVerdict, ChannelRollup, ChatRequest, CheckResult, type ClusterBootstrapInterval, type ClusterSignFlipAlternative, type ClusterSignFlipResult, type ClusteredBinaryCluster, type ClusteredMatchedPair, type ClusteredPairedBinaryOptions, type ClusteredPairedBinaryResult, type ClusteredPairedBinaryStatistics, CollectedArtifacts, type CommandRunner, type CompareLabels, CompletionCriterion, ConfigError, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContractCheckResult, type ContractJudgeOptions, type ContractMetric, type ContractReport, type ContractRule, type ContractRuleKind, type ContractSpan, type ContractVerdict, type ContractViolation, ControlEvalResult, ControlSeverity, ConvergenceTracker, CorrectnessChecker, type CostEntry, CostLedger, CostReceiptInput, type CostReport, type CostSummary, CostTracker, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, CreateChatClientOpts, type CreateDefaultReviewerOptions, type CreateExperimentInput, type CreateSandboxPoolOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, DEFAULT_AGENT_SLOS, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, type DataAcquisitionPlan, Dataset, DatasetScenario, type DecideNextUserTurnOpts, DefaultVerdict, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DescriptionLengthCandidate, type DescriptionLengthConfig, type DescriptionLengthDecision, type DescriptionLengthEvidence, DescriptionLengthGate, type DescriptionLengthRejectionCode, type DetectorEvent, type DetectorSeverity, type DetectorSignal, type DiffScorecardOptions, type DirEntry, type DiscoverPersonasOptions, type DiscoveredPersona, DriverResult, DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, ERROR_COUNT_PATTERNS, type EnsembleAggregate, type EnsembleJudgeOptions, type ErrorCountPattern, type ErrorStreakOptions, type EvalToolDef, EvalTraceStore, type EvolutionRound, type ExecutorConfig, type Expectation, type Experiment, type ExperimentProvenance, type ExperimentRep, type ExperimentStats, type ExperimentStore, ExperimentTracker, type ExperimentTrackerOptions, type ExperimentVerdict, type ExportedRewardModel, type ExtractOptions, type ExtractResult, type FactorContribution, type FactorialCell, FailureClass, FeedbackLabel, FeedbackTrajectory, FeedbackTrajectoryStore, type FieldAgreementSpec, type FieldDestination, type FileChange, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type GhCliClientOptions, type GoldScenario, type GoldSplit, type GoldenSeverity, type GoldenSpec, HARNESS_NATIVE_MODEL, type HarnessAdapter, HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessVariant, type HarnessVariantReport, type HeadroomClass, type HeadroomInput, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, HoldoutAuditor, type HttpGithubClientOptions, INTENT_MATCH_JUDGE_VERSION, type ImageData, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, JudgeError, type JudgeFamily, type JudgeFleetOptions, JudgeFn, JudgeParseError, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, JudgeRunner, type JudgeScoreInput, type JudgeVerdict, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type KnowledgeAcquisitionMode, type KnowledgeBundle, type KnowledgeFallbackPolicy, type KnowledgeFreshness, type KnowledgeImportance, type KnowledgeReadinessReport, type KnowledgeRecommendedAction, type KnowledgeRequirement, type KnowledgeRequirementCategory, type KnowledgeResponsibleSurface, type KnowledgeSensitivity, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, Layer, LayerResult, type LeaderboardOptions, type LeaderboardRow, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmCallMetadata, LlmClientOptions, LlmSpan, LlmUsage, LockedJsonlAppender, MODEL_PRICING, type MakeEvalToolsConfig, type MatchResult, type MatcherResult, MaximumCharge, type MeasurementPolicy, type MergeOptions, MetricsCollector, type ModelCostRollup, type ModelPreflight, type ModelSeats, ModelsUnreachableError, type MuffledFinder, type MuffledFinding, type MultiToolchainLayerConfig, type Mutator, Mutex, type NoLeakOptions, type NoProgressOptions, Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, ParetoResult, type ParseStudentLabel, type PartitionHeldOutOptions, PersonaConfig, type Playbook, type PlaybookEntry, type PoolSlot, type PrReviewAuditCase, type PrReviewBenchmarkSummary, type PrReviewComment, type PrReviewMatchedFinding, type PrReviewOutcome, type PrReviewReferenceFinding, type PrReviewScore, type PrReviewScoreWeights, type PrReviewSeverity, type PrReviewSource, type PreflightModelsOptions, type PreflightOutcome, type ProductBenchmarkArm, type ProductBenchmarkArtifactPaths, type ProductBenchmarkBudgets, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkProfileRef, type ProductBenchmarkRecord, type ProductBenchmarkRepoRef, type ProductBenchmarkRunInput, type ProductBenchmarkScenario, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkSubstrateVersions, type ProductBenchmarkValidationReport, ProductClient, ProductClientConfig, type ProfileAxisSpec, type PromptHandle, PromptRegistry, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, type ProvenanceReader, type RecordRunsOptions, type ReferenceMatchResult, type ReferenceReplayAdapter, type ReferenceReplayAdapterFn, type ReferenceReplayAdapterLike, type ReferenceReplayAggregate, type ReferenceReplayCandidate, type ReferenceReplayCase, type ReferenceReplayCaseRun, type ReferenceReplayExecutionScenario, type ReferenceReplayItem, type ReferenceReplayMatch, type ReferenceReplayMatchStrategy, type ReferenceReplayMatcher, type ReferenceReplayPromotionDecision, type ReferenceReplayPromotionPolicy, type ReferenceReplayRun, type ReferenceReplayRunContext, type ReferenceReplayRunOptions, type ReferenceReplayRunStore, type ReferenceReplayScenario, type ReferenceReplayScenarioScore, type ReferenceReplayScore, type ReferenceReplayScoreOptions, type ReferenceReplaySplit, type ReferenceReplaySplitComparison, type ReferenceReplaySteeringRowsOptions, type ReflectionContext, type ReflectionProposal, ReleaseConfidenceScorecard, ReleaseConfidenceThresholds, type RenderStudentPrompt, type RepeatedActionOptions, type RepoRef, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RobustnessResult, type RoutedField, Run, type RunCommandInput, type RunCommandResult, type RunDistillationOptions, type RunDistillationResult, RunFilter, RunRecord, type RunRecordBackend, type RunRecordFilter, RunScore, RunScoreWeights, RunSplitTag, RunTrace, type RuntimeResolution, SandboxDriver, SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type ScanOptions, Scenario, type ScenarioCost, ScenarioFile, ScenarioRegistry, ScenarioResult, type ScoreKnowledgeReadinessOptions, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SeatName, type SeatPresetName, SeatUnsetError, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SerializedRegex, Severity, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, Span, type SpanPredicate, type SplitGoldOptions, type SteeringBundle, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type StreamingDetector, type SynthesisReason, type SynthesisTarget, type TaskHeadroom, TestResult, type TextMatcher, type ThresholdContract, TokenCounter, type TokenSpec, type ToolMatcher, ToolSpan, TraceAnalystSpan, type TraceContract, TraceContractBuilder, TraceEmitter, TraceStore, type TracedAnalystOptions, type TracedJudgeOptions, Trajectory, TrajectoryStep, type TreatmentClass, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, TurnMetrics, UI_FINDING_SEVERITIES, UI_LENSES, UNIVERSAL_FINDERS, type UiFinding, type UiFindingScreenshot, type UiFindingSeverity, type UiLens, type UserQuestion, type VerdictCacheStats, type VerdictCacheStore, VerifyContext, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, type WorkerDriverContext, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, acquisitionPlansForKnowledgeGaps, adversarialJudge, agentProfileHash, agentProfileId, agentProfileModelId, agentVisibleFields, aggregateJudgeVerdicts, aggregatePrReviewScore, analyzeAntiSlop, appendScorecard, assertCapabilityHeadroom, assertCrossFamily, assertModelsServed, assertNoHiddenLeak, assertProductBenchmarkRun, assertSingleBackend, assignHeldOutTag, attachCostToReport, attest, attributeCounterfactuals, bisect, blendHeldout, blockingKnowledgeEval, buildAgreementJudge, buildDriverSystemPrompt, buildProductBenchmarkManifest, buildReflectionPrompt, buildReviewerPrompt, buildWorkerDriverSystemPrompt, cachedJudge, canaryLeakView, canonicalJson, capabilityHeadroom, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, checkTraceContracts, classifyTreatment, clusteredPairedBinary, codeExecutionJudge, coherenceJudge, collectionPreserved, commentsForSource, commitBisect, compareReferenceReplay, compilerJudge, computeExperimentStats, contentHash, contractJudge, costReport, createAntiSlopJudge, createCustomJudge, createDefaultReviewer, createDomainExpertJudge, createIntentMatchJudge, createSandboxPool, crossTraceDiff, dataDescriptionBits, decideNextUserTurn, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultBlendWeights, defaultJudges, defaultParseStudentLabel, defaultReferenceReplayMatcher, defaultRenderStudentPrompt, deployGateLayer, diffScorecard, discoverPersonas, distillPlaybook, ensembleJudge, errorStreakDetector, estimateCost, estimateTokens, evaluateContract, evaluateOracles, evaluateTraceContract, executeScenario, expandProfileAxes, expectAgent, exportProductBenchmark, exportProductBenchmarkRuns, exportRewardModel, extractAssetUrls, extractErrorCount, fieldAgreement, fileContains, fileExists, fileExperimentStore, fileVerdictCache, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findProductBenchmarkArtifacts, findSkipCountsAsPass, flowLayer, fnv1a32, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, gateTreatmentApplied, gateTreatmentFromMetrics, gateTreatmentFromSpans, gateTreatmentFromToolSpans, ghCliClient, gitProvenanceReader, precision as goldenPrecision, gradeOnHidden, harnessAxisOf, hashContent, hashToUnit, hiddenGrade, htmlContainsElement, httpGithubClient, improvementVerdict, inMemoryExperimentStore, inMemoryReferenceReplayStore, inMemoryRunRecordBackend, inMemoryVerdictCache, isHiddenDestination, isModelPriced, isOtelConfigured, jsonShape, jsonlReferenceReplayStore, jsonlRunRecordBackend, judgeFamily, keyPreserved, knowledgeReadinessTracePayload, leaderboard, linterJudge, loadGoldScenarios, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, makeEvalTools, matchGoldens, matchSpan, mergeLayerResults, mergeSteeringBundle, modelDescriptionBits, multiToolchainLayer, noProgressDetector, notBlocked, observeAll, paraphraseRobustness, paraphraseRobustnessScenarios, parseGoldJsonl, parseReflectionResponse, partitionHeldOut, passOrthogonality, pixelDeltaRatio, politenessPrefixMutator, preflightModels, printDriverSummary, productBenchmarkIntegrityFailures, productBenchmarkMutableSurfaces, productBenchmarkRepoIdentity, productBenchmarkSplits, index as profile, promptBisect, proposeSynthesisTargets, readProductBenchmarkManifest, readProductBenchmarkRecords, recordRuns, recordRunsToScorecard, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatches, renderMarkdownReport, renderPlaybookMarkdown, renderSteeringText, repeatedActionDetector, replayScorerOverCorpus, replayTraceThroughJudge, resolveModelPricing, resolveSeat, routeFields, rowCount, rowWhere, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runDistillation, runE2EWorkflow, runExpectations, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runRecordToProductBenchmarkRecord, runReferenceReplay, runScore, runSelfPlay, scanForMuffledGates, scoreContinuity, scoreKnowledgeReadiness, scorePrReviewComments, scorePrReviewSource, scoreReferenceReplay, seatPresets, securityJudge, selectHarnessVariant, sentenceReorderMutator, splitGold, statusAdvanced, summarizeHarnessResults, summarizePrReviewBenchmark, testJudge, textInSnapshot, toLangfuseEnvelope, toOpenAiTool, toPrometheusText, traceContract, traceJudge, traceJudgeEnsemble, tracedAnalyzeTraces, typoMutator, urlContains, userQuestionsForKnowledgeGaps, validateProductBenchmarkManifest, validateProductBenchmarkRecord, validateProductBenchmarkRun, verifyAttestation, visualDiff, viteDeployRunner, weightedRecall, whitespaceCollapseMutator, withHeldoutBlend, withJudgeRetry, withOtelPipeline, wranglerDeployRunner };