@tangle-network/agent-eval 0.60.0 → 0.61.0

package/dist/index.d.ts

@@ -1,38 +1,40 @@
-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-DjEgwWNo.js';
-import { R as RunRecord } from './run-record-etiCMsUq.js';
-export { e as AGENT_PROFILE_KINDS, A as AgentProfileCell, d as AgentProfileCellInput, f as AgentProfileCellSchemaVersion, g as AgentProfileCellValidationError, h as AgentProfileDimensionValue, i as AgentProfileHarness, j as AgentProfileJson, k as AgentProfileKind, l as AgentProfileSource, m as AgentProfileSourceInput, J as JudgeScoresRecord, c as RunJudgeMetadata, n as RunOutcome, o as RunRecordValidationError, a as RunSplitTag, b as RunTokenUsage, S as SandboxAgentProfileLike, p as agentProfileCellHashMaterial, q as agentProfileCellKey, r as assertRunAgentProfileCell, s as buildAgentProfileCell, t as buildSandboxAgentProfileCell, u as groupRunsByAgentProfileCell, v as isRunRecord, w as parseRunRecordSafe, x as requireAgentProfileCell, y as roundTripRunRecord, z as toAgentProfileJson, B as validateAgentProfileCell, C as validateRunRecord, D as verifyAgentProfileCell } from './run-record-etiCMsUq.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-Bf8owbuG.js';
+import { R as RunRecord } from './run-record-DgUVo5pw.js';
+export { e as AGENT_PROFILE_KINDS, A as AgentProfileCell, d as AgentProfileCellInput, f as AgentProfileCellSchemaVersion, g as AgentProfileCellValidationError, h as AgentProfileDimensionValue, i as AgentProfileHarness, j as AgentProfileJson, k as AgentProfileKind, l as AgentProfileSource, m as AgentProfileSourceInput, J as JudgeScoresRecord, c as RunJudgeMetadata, n as RunOutcome, o as RunRecordValidationError, a as RunSplitTag, b as RunTokenUsage, S as SandboxAgentProfileLike, p as agentProfileCellHashMaterial, q as agentProfileCellKey, r as assertRunAgentProfileCell, s as buildAgentProfileCell, t as buildSandboxAgentProfileCell, u as groupRunsByAgentProfileCell, v as isRunRecord, w as parseRunRecordSafe, x as requireAgentProfileCell, y as roundTripRunRecord, z as toAgentProfileJson, B as validateAgentProfileCell, C as validateRunRecord, D as verifyAgentProfileCell } from './run-record-DgUVo5pw.js';
 import { AxAIService, AxFunction } from '@ax-llm/ax';
-import { d as Severity, M as MultiLayerVerifier, e as VerifyOptions, L as Layer, f as LayerResult, g as VerifyContext } from './researcher-JP8EvnLv.js';
-export { C as CallbackResearcher, h as CallbackResearcherOptions, i as CampaignFactoryParams, j as CampaignIntegrityPolicy, k as CampaignRunContext, l as CampaignRunOutcome, m as CampaignRunner, n as CampaignScenario, o as CampaignVariant, c as EvalCampaignOptions, b as EvalCampaignResult, E as ExperimentPlan, a as ExperimentResult, p as FailedRun, F as FailureMode, q as Finding, s as LayerStatus, N as NoopResearcher, R as Researcher, S as SteeringChange, V as VerificationReport, t as gradeSemanticStatus, r as runEvalCampaign } from './researcher-JP8EvnLv.js';
+import { d as Severity, M as MultiLayerVerifier, e as VerifyOptions, L as Layer, f as LayerResult, g as VerifyContext } from './researcher-BaVsy0sW.js';
+export { C as CallbackResearcher, h as CallbackResearcherOptions, i as CampaignFactoryParams, j as CampaignIntegrityPolicy, k as CampaignRunContext, l as CampaignRunOutcome, m as CampaignRunner, n as CampaignScenario, o as CampaignVariant, c as EvalCampaignOptions, b as EvalCampaignResult, E as ExperimentPlan, a as ExperimentResult, p as FailedRun, F as FailureMode, q as Finding, s as LayerStatus, N as NoopResearcher, R as Researcher, S as SteeringChange, V as VerificationReport, t as gradeSemanticStatus, r as runEvalCampaign } from './researcher-BaVsy0sW.js';
 import { R as Run$1, S as Span, a as TraceEvent, A as Artifact$1, B as BudgetLedgerEntry, h as BudgetSpec, L as LlmSpan } from './schema-m0gsnbt3.js';
 export { E as EventKind, i as FAILURE_CLASSES, F as FailureClass, G as GenericSpan, J as JudgeSpan, M as Message, d as RetrievalSpan, g as RunLayer, f as RunStatus, e as SandboxSpan, j as SpanBase, c as SpanKind, k as SpanStatus, l as TRACE_SCHEMA_VERSION, T as ToolSpan, m as isJudgeSpan, n as isLlmSpan, o as isRetrievalSpan, p as isSandboxSpan, q as isToolSpan } from './schema-m0gsnbt3.js';
 import { T as TraceStore, R as RunFilter } from './store-CKUAgsJz.js';
 export { E as EventFilter, F as FileSystemTraceStore, a as FileSystemTraceStoreOptions, I as InMemoryTraceStore, S as SpanFilter } from './store-CKUAgsJz.js';
-import { L as LlmClientOptions } from './llm-client-BXVRUZyX.js';
-export { d as LlmCallError, b as LlmCallRequest, c as LlmCallResult, e as LlmClient, f as LlmMessage, g as LlmRouteAssertionError, a as LlmRouteRequirements, h as LlmUsage, i as assertLlmRoute, j as backoffMs, k as callLlm, l as callLlmJson, m as isTransientLlmError, p as probeLlm, s as stripFencedJson } from './llm-client-BXVRUZyX.js';
+import { L as LlmClientOptions } from './llm-client-DbjLfz-K.js';
+export { d as LlmCallError, b as LlmCallRequest, c as LlmCallResult, e as LlmClient, f as LlmMessage, g as LlmRouteAssertionError, a as LlmRouteRequirements, h as LlmUsage, i as assertLlmRoute, j as backoffMs, k as callLlm, l as callLlmJson, m as isTransientLlmError, p as probeLlm, s as stripFencedJson } from './llm-client-DbjLfz-K.js';
 import { AnalyzeTracesOptions, OtelExporter, OtelExportConfig, AnalyzeTracesInput, AnalyzeTracesResult } from './traces.js';
 export { AnalyzeTracesTurnSnapshot, CaptureFetchContext, CaptureFetchOptions, DEFAULT_REDACTION_RULES, ExportableSpan, FlattenOtlpOptions, OTEL_AGENT_EVAL_SCOPE, OtlpExport, OtlpFileTraceStore, OtlpFileTraceStoreOptions, OtlpFlatLine, OtlpResourceSpans, OtlpSpan, REDACTION_VERSION, RedactionReport, RedactionRule, ReplayCache, ReplayCacheEntry, ReplayCacheMissError, ReplayCacheStats, ReplayFetchOptions, SpanNotFoundError, TRACE_ANALYST_ACTOR_DESCRIPTION, TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, TRACE_ANALYST_SUBAGENT_DESCRIPTION, TraceAnalystHookOptions, TraceFileMissingError, TraceInsightContext, TraceInsightFinding, TraceInsightPanelRole, TraceInsightPromptInput, TraceInsightQualityGate, TraceInsightQuestion, TraceInsightReadiness, TraceInsightSuite, TraceInsightTask, TraceNotFoundError, analyzeTraces, buildTraceAnalystTools, buildTraceInsightContext, buildTraceInsightPrompt, captureFetchToRawSink, createOtelExporter, createOtelTracingStore, createReplayFetch, defaultTraceInsightPanel, describeTraceInsightScope, domainEvidencePattern, exportRunAsOtlp, flattenOtlpExportToNdjson, inferDomainKeywords, iterateRawCalls, otelRunCompleteHook, planTraceInsightQuestions, redactString, redactValue, scoreTraceInsightReadiness, tokenizeDomainWords, traceAnalystFunctionGroup, traceAnalystOnRunComplete } from './traces.js';
 import { T as TraceAnalysisStore } from './store-jzKpMl16.js';
 export { D as DEFAULT_TRACE_ANALYST_BUDGETS, a as DatasetOverview, Q as QueryTracesPage, S as SearchSpanResult, b as SearchTraceResult, c as SpanMatchRecord, d as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, e as TraceAnalystByteBudgets, f as TraceAnalystFilters, g as TraceAnalystSpan, h as TraceAnalystSpanKind, i as TraceAnalystSpanStatus, j as TraceAnalystTraceSummary, V as ViewSpansResult, k as ViewTraceOversized, l as ViewTraceResult } from './store-jzKpMl16.js';
 import { b as JudgeFn, J as JudgeInput, 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, f as CollectedArtifacts, g as ScenarioResult, h as TurnMetrics, i as ScenarioFile, j as CompletionCriterion } from './types-DhqpAi_z.js';
 export { A as ArtifactCheck, k as ArtifactResult, E as EvalResult, F as FeedbackPattern, l as JudgeConfig, m as JudgeRubric, a as JudgeScore, n as PersonaRigor, R as RouteMap, o as RubricDimension, p as Turn, q as TurnResult } from './types-DhqpAi_z.js';
-import { a as Analyst, b as AnalystSeverity, c as AnalystFinding, d as AnalystCost, e as AnalystContext } from './registry-DK9kqXvb.js';
-export { f as AnalystHooks, g as AnalystInputKind, A as AnalystRegistry, h as AnalystRegistryOptions, i as AnalystRequirements, j as AnalystRunEvent, k as AnalystRunInputs, l as AnalystRunResult, m as AnalystRunSummary, B as BudgetPolicy, C as ChatCallOpts, n as ChatClient, o as ChatRequest, p as ChatResponse, q as ChatTransport, r as CliBridgeTransportOpts, s as CreateChatClientOpts, D as DirectProviderTransportOpts, E as EvidenceRef, M as MockTransportOpts, R as RegistryRunOpts, t as RouterTransportOpts, S as SandboxSdkTransportOpts, u as computeFindingId, v as createChatClient, w as makeFinding } from './registry-DK9kqXvb.js';
+import { a as Analyst, b as AnalystSeverity, c as AnalystFinding, d as AnalystCost, e as AnalystContext } from './registry-qmbYT3Eo.js';
+export { f as AnalystHooks, g as AnalystInputKind, A as AnalystRegistry, h as AnalystRegistryOptions, i as AnalystRequirements, j as AnalystRunEvent, k as AnalystRunInputs, l as AnalystRunResult, m as AnalystRunSummary, B as BudgetPolicy, C as ChatCallOpts, n as ChatClient, o as ChatRequest, p as ChatResponse, q as ChatTransport, r as CliBridgeTransportOpts, s as CreateChatClientOpts, D as DirectProviderTransportOpts, E as EvidenceRef, M as MockTransportOpts, R as RegistryRunOpts, t as RouterTransportOpts, S as SandboxSdkTransportOpts, u as computeFindingId, v as createChatClient, w as makeFinding } from './registry-qmbYT3Eo.js';
 import { TCloud } from '@tangle-network/tcloud';
 import { z } from 'zod';
 export { c as ControlActionFailureMode, d as ControlActionOutcome, e as ControlBudget, f as ControlContext, g as ControlDecision, C as ControlEvalResult, a as ControlRunResult, h as ControlRuntimeConfig, i as ControlRuntimeError, j as ControlSeverity, b as ControlStep, k as ControlStopPolicies, S as StopDecision, l as allCriticalPassed, o as objectiveEval, r as runAgentControlLoop, s as stopOnNoProgress, m as stopOnRepeatedAction, n as subjectiveEval } from './control-runtime-DuFBYg7A.js';
-import { A as AgentEvalError } from './errors-mje_cKOs.js';
-export { a as AgentEvalErrorCode, C as CaptureIntegrityError, b as ConfigError, J as JudgeError, N as NotFoundError, R as ReplayError, V as ValidationError, c as VerificationError } from './errors-mje_cKOs.js';
-import { a as FeedbackLabel, F as FeedbackTrajectoryStore, b as FeedbackTrajectory } from './feedback-trajectory-DpUmE90J.js';
-export { c as FeedbackArtifactType, d as FeedbackAttempt, e as FeedbackLabelKind, f as FeedbackLabelSource, g as FeedbackOptimizerRow, h as FeedbackOutcome, i as FeedbackReplayAdapter, j as FeedbackReplayResult, k as FeedbackSeverity, l as FeedbackSplitPolicy, m as FeedbackTask, n as FeedbackTrajectoryFilter, o as FileSystemFeedbackTrajectoryStore, I as InMemoryFeedbackTrajectoryStore, P as PreferenceMemoryEntry, p as ProposedSideEffect, q as assignFeedbackSplit, r as controlRunToFeedbackTrajectory, s as createFeedbackTrajectory, t as feedbackTrajectoriesToDatasetScenarios, u as feedbackTrajectoriesToOptimizerRows, v as feedbackTrajectoryToDatasetScenario, w as feedbackTrajectoryToOptimizerRow, x as parseFeedbackTrajectoriesJsonl, y as renderPreferenceMemoryMarkdown, z as replayFeedbackTrajectories, A as replayFeedbackTrajectory, B as serializeFeedbackTrajectoriesJsonl, C as summarizePreferenceMemory, D as withAssignedFeedbackSplit } from './feedback-trajectory-DpUmE90J.js';
+import { A as AgentEvalError } from './errors-Dwqw-T_m.js';
+export { a as AgentEvalErrorCode, C as CaptureIntegrityError, b as ConfigError, J as JudgeError, N as NotFoundError, R as ReplayError, V as ValidationError, c as VerificationError } from './errors-Dwqw-T_m.js';
+import { a as FeedbackLabel, F as FeedbackTrajectoryStore, b as FeedbackTrajectory } from './feedback-trajectory-8hKC5EOb.js';
+export { c as FeedbackArtifactType, d as FeedbackAttempt, e as FeedbackLabelKind, f as FeedbackLabelSource, g as FeedbackOptimizerRow, h as FeedbackOutcome, i as FeedbackReplayAdapter, j as FeedbackReplayResult, k as FeedbackSeverity, l as FeedbackSplitPolicy, m as FeedbackTask, n as FeedbackTrajectoryFilter, o as FileSystemFeedbackTrajectoryStore, I as InMemoryFeedbackTrajectoryStore, P as PreferenceMemoryEntry, p as ProposedSideEffect, q as assignFeedbackSplit, r as controlRunToFeedbackTrajectory, s as createFeedbackTrajectory, t as feedbackTrajectoriesToDatasetScenarios, u as feedbackTrajectoriesToOptimizerRows, v as feedbackTrajectoryToDatasetScenario, w as feedbackTrajectoryToOptimizerRow, x as parseFeedbackTrajectoriesJsonl, y as renderPreferenceMemoryMarkdown, z as replayFeedbackTrajectories, A as replayFeedbackTrajectory, B as serializeFeedbackTrajectoriesJsonl, C as summarizePreferenceMemory, D as withAssignedFeedbackSplit } from './feedback-trajectory-8hKC5EOb.js';
+import { A as AgentProfile } from './agent-profile-9J9hxdm2.js';
+export { a as BackendIntegrityError, B as BackendIntegrityReport, b as agentProfileHash, c as assertRealBackend, s as summarizeBackendIntegrity } from './agent-profile-9J9hxdm2.js';
 export { DataAcquisitionPlan, KnowledgeAcquisitionMode, KnowledgeBundle, KnowledgeFallbackPolicy, KnowledgeFreshness, KnowledgeImportance, KnowledgeReadinessReport, KnowledgeRecommendedAction, KnowledgeRequirement, KnowledgeRequirementCategory, KnowledgeResponsibleSurface, KnowledgeSensitivity, ScoreKnowledgeReadinessOptions, UserQuestion, acquisitionPlansForKnowledgeGaps, blockingKnowledgeEval, knowledgeReadinessTracePayload, scoreKnowledgeReadiness, userQuestionsForKnowledgeGaps } from './knowledge/index.js';
-import { i as ReleaseConfidenceThresholds, g as ReleaseConfidenceScorecard } from './release-report-DmPjIce3.js';
-export { A as ActionableSideInfo, s as AsiSeverity, B as BootstrapOptions, a as BootstrapResult, C as CliffsMagnitude, t as CorpusAgreementOptions, u as CorpusAgreementPerDimension, v as CorpusAgreementReport, x as CorpusScoreRecord, J as JudgeReplayGateArgs, P as PairedBootstrapOptions, b as PairedBootstrapResult, R as ReleaseConfidenceAxis, c as ReleaseConfidenceAxisName, d as ReleaseConfidenceInput, e as ReleaseConfidenceIssue, f as ReleaseConfidenceMetrics, h as ReleaseConfidenceStatus, j as ReleaseTraceEvidence, k as RenderReleaseReportOptions, V as Verdict, W as WeightedCompositeInput, y as WeightedCompositeResult, l as assertReleaseConfidence, m as benjaminiHochberg, z as bonferroni, n as bootstrapCi, D as cliffsDelta, E as cohensD, F as confidenceInterval, G as corpusInterRaterAgreement, H as corpusInterRaterAgreementFromJudgeScores, o as evaluateReleaseConfidence, I as interRaterReliability, K as interpretCliffs, p as judgeReplayGate, L as mannWhitneyU, M as normalizeScores, q as pairedBootstrap, N as pairedMde, O as pairedTTest, Q as partialCredit, r as renderReleaseReport, S as requiredSampleSize, T as weightedComposite, U as weightedMean, w as wilcoxonSignedRank } from './release-report-DmPjIce3.js';
+import { i as ReleaseConfidenceThresholds, g as ReleaseConfidenceScorecard } from './release-report-DszkgvJ3.js';
+export { A as ActionableSideInfo, s as AsiSeverity, B as BootstrapOptions, a as BootstrapResult, C as CliffsMagnitude, t as CorpusAgreementOptions, u as CorpusAgreementPerDimension, v as CorpusAgreementReport, x as CorpusScoreRecord, J as JudgeReplayGateArgs, P as PairedBootstrapOptions, b as PairedBootstrapResult, R as ReleaseConfidenceAxis, c as ReleaseConfidenceAxisName, d as ReleaseConfidenceInput, e as ReleaseConfidenceIssue, f as ReleaseConfidenceMetrics, h as ReleaseConfidenceStatus, j as ReleaseTraceEvidence, k as RenderReleaseReportOptions, V as Verdict, W as WeightedCompositeInput, y as WeightedCompositeResult, l as assertReleaseConfidence, m as benjaminiHochberg, z as bonferroni, n as bootstrapCi, D as cliffsDelta, E as cohensD, F as confidenceInterval, G as corpusInterRaterAgreement, H as corpusInterRaterAgreementFromJudgeScores, o as evaluateReleaseConfidence, I as interRaterReliability, K as interpretCliffs, p as judgeReplayGate, L as mannWhitneyU, M as normalizeScores, q as pairedBootstrap, N as pairedMde, O as pairedTTest, Q as partialCredit, r as renderReleaseReport, S as requiredSampleSize, T as weightedComposite, U as weightedMean, w as wilcoxonSignedRank } from './release-report-DszkgvJ3.js';
 import { S as SandboxDriver, H as HarnessConfig, a as SandboxHarnessResult } from './test-graded-scenario-BdVaPyHT.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-BdVaPyHT.js';
 import { T as TraceEmitter } from './emitter-DEZwY14K.js';
 export { R as RunCompleteHook, a as RunCompleteHookContext, S as SpanHandle, b as TraceEmitterOptions, l as llmSpanFromProvider } from './emitter-DEZwY14K.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-CfXjSqEv.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-CJzrpUua.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-CqTxMwDw.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';
 export { D as DEFAULT_FAILURE_RULES, b as FailureClassification, c as FailureContext, d as FailureRule, e as classifyFailure } from './failure-cluster-CL7IVgkJ.js';
@@ -41,14 +43,14 @@ export { B as BaselineOptions, M as MetricSamples, b as MetricVerdict, T as Tool
 import { T as Trajectory, a as TrajectoryStep } from './trajectory-GEdXJCL5.js';
 export { b as buildTrajectory } from './trajectory-GEdXJCL5.js';
 export { D as DefaultVerdict } from './verdict-CeEgtjyI.js';
-import { a as DatasetScenario, b as Dataset } from './dataset-BlwAtYYf.js';
-export { d as DatasetDifficulty, c as DatasetManifest, e as DatasetProvenance, D as DatasetSplit, H as HoldoutLockedError, S as SliceOptions, h as hashScenarios } from './dataset-BlwAtYYf.js';
+import { a as DatasetScenario, b as Dataset } from './dataset-B2kL-fSM.js';
+export { d as DatasetDifficulty, c as DatasetManifest, e as DatasetProvenance, D as DatasetSplit, H as HoldoutLockedError, S as SliceOptions, h as hashScenarios } from './dataset-B2kL-fSM.js';
 export { b as CalibrationResult, c as CandidateScore, a as ContinuousAgreement, C as ContinuousAgreementOptions, d 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-DilmB3Ml.js';
-export { D as DEFAULT_RED_TEAM_CORPUS, R as RedTeamCase, a as RedTeamCategory, b as RedTeamFinding, c as RedTeamPayload, d as RedTeamReport, r as redTeamDataset, e as redTeamReport, s as scoreRedTeamOutput, t as toolNamesForRun } from './red-team-CrC5MZYd.js';
+export { D as DEFAULT_RED_TEAM_CORPUS, R as RedTeamCase, a as RedTeamCategory, b as RedTeamFinding, c as RedTeamPayload, d as RedTeamReport, r as redTeamDataset, e as redTeamReport, s as scoreRedTeamOutput, t as toolNamesForRun } from './red-team-DW9Ca_tj.js';
 import { a as PrmGrader } from './rubric-BOfxn4ja.js';
 export { EuRiskClass, GovernanceContext, GovernanceFinding, GovernanceReport, UseCaseSignals, classifyEuAiRisk, euAiActReport, nistAiRmfReport, renderMarkdown, soc2Report, summarize } from './governance/index.js';
-export { B as BENCHMARK_SPLIT_SEED, a as BenchmarkAdapter, b as BenchmarkDatasetItem, c as BenchmarkEvaluation, d as benchmarkDeterministicSplit, i as benchmarks } from './index-wlaiph9Y.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-DLxh4yWk.js';
+export { B as BENCHMARK_SPLIT_SEED, a as BenchmarkAdapter, b as BenchmarkDatasetItem, c as BenchmarkEvaluation, d as benchmarkDeterministicSplit, i as benchmarks } from './index-Bvk35ils.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-BQvXpvaR.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 './outcome-store-D6KWmYvj.js';
 
@@ -1121,76 +1123,6 @@ interface ExecutorConfig {
  */
 declare function executeScenario(tc: TCloud, scenario: Scenario, config: ExecutorConfig): Promise<ScenarioResult>;
 
-/**
- * Backend-integrity guard: distinguish "agent failed" from "eval ran against
- * a stub / unconfigured backend." Without this guard a canonical eval can
- * silently report `0/N passed` and look like an agent-quality problem when
- * the LLM was never actually called — the failure mode we just hit running
- * the 4-vertical parallel eval (legal-sandbox-stub returned hard-coded 33-104
- * char strings; gtm/creative defaulted to a cli-bridge that wasn't running).
- *
- * The shape:
- *
- *   const report = summarizeBackendIntegrity(records)
- *   assertRealBackend(records)   // throws BackendIntegrityError if 100% stub
- *
- * A record is "stub-mode" if its `tokenUsage.input === 0 && tokenUsage.output === 0`.
- * (`costUsd` alone is unreliable — some backends successfully call LLMs but
- *  don't propagate pricing, producing real tokens with $0 cost.)
- *
- * Verdicts:
- *   - `real`   — at least one record has nonzero token usage
- *   - `stub`   — every record is stub-mode (eval ran blind)
- *   - `mixed`  — some records real, some stub (partial backend failure;
- *                often the 429-cascade or auth-half-failed case)
- */
-
-interface BackendIntegrityReport {
-    /** Total records inspected. */
-    totalRecords: number;
-    /** Records with input=0 AND output=0 (a stub fingerprint). */
-    stubRecords: number;
-    /** Records with nonzero token usage (real LLM activity). */
-    realRecords: number;
-    /** Records where output>0 but costUsd=0 (real LLM, broken cost ledger). */
-    uncostedRecords: number;
-    /** Sum of input tokens across all records. */
-    totalInputTokens: number;
-    /** Sum of output tokens across all records. */
-    totalOutputTokens: number;
-    /** Sum of costUsd across all records. */
-    totalCostUsd: number;
-    /** Worst-case integrity verdict. */
-    verdict: 'real' | 'mixed' | 'stub';
-    /** Human-readable diagnosis suitable for terminal output. */
-    diagnosis: string;
-}
-/**
- * Error thrown when an integrity assertion fails. Caller can pattern-match
- * by `code === 'AGENT_EVAL_BACKEND_STUB'` to differentiate from other
- * errors.
- */
-declare class BackendIntegrityError extends AgentEvalError {
-    readonly report: BackendIntegrityReport;
-    constructor(message: string, report: BackendIntegrityReport);
-}
-/**
- * Inspect a batch of RunRecords and return an integrity report. Pure
- * function — no I/O, no logging. The caller decides what to do with the
- * verdict (print warning, throw, gate CI, etc.).
- */
-declare function summarizeBackendIntegrity(records: ReadonlyArray<RunRecord>): BackendIntegrityReport;
-/**
- * Throw BackendIntegrityError if the verdict is 'stub' — i.e. every record
- * shows zero LLM activity. Non-strict callers can pass `{ allowMixed: false }`
- * to also reject mixed verdicts (recommended for CI gates).
- *
- * Real backends pass through silently.
- */
-declare function assertRealBackend(records: ReadonlyArray<RunRecord>, opts?: {
-    allowMixed?: boolean;
-}): BackendIntegrityReport;
-
 /**
  * Single-backend guard: assert the agent and the rubric judge run through the
  * SAME backend config, so the judge can't silently re-route through a
@@ -2667,46 +2599,6 @@ declare class BudgetGuard {
     get state(): Record<keyof BudgetSpec, number>;
 }
 
-/**
- * @stable
- *
- * AgentProfile — the eval harness's unit of variation.
- *
- * A profile pins everything that changes agent behaviour for a benchmark
- * cell: the model, the active skills, the prompt version, the available
- * tools. Vary the profile — swap a model, add a skill — and re-run the suite
- * to benchmark the change. The scorecard keys a cell on
- * `(scenarioId, profileHash)`, so the model is not a separate axis: it lives
- * inside the profile, and two profiles with the same model but different
- * skills are different cells.
- *
- * `agentProfileHash` is the profile's behaviour identity. Two profiles that
- * produce the same agent behaviour share a hash (and a scorecard cell);
- * reordering `skills` or `tools` does not change it; the human-facing `id`
- * label does not affect it.
- */
-interface AgentProfile {
-    /** Human-facing label, e.g. `sonnet-legal-skills-v3`. Not part of the hash. */
-    id: string;
-    /** Model snapshot id this profile pins, e.g. `claude-sonnet-4-6@2025-04-15`. */
-    model: string;
-    /** Skill ids/versions active in this profile — the primary behaviour lever. */
-    skills?: string[];
-    /** Prompt version identifier. */
-    promptVersion?: string;
-    /** Tool ids available to the agent. */
-    tools?: string[];
-    /** Any other behaviour-bearing knobs that should fingerprint into the hash. */
-    metadata?: Record<string, string | number | boolean>;
-}
-/**
- * Deterministic behaviour identity of a profile — a sha256 over the
- * behaviour-bearing fields. `skills` and `tools` are order-insensitive; the
- * `id` label is excluded. Throws on a profile with no `model` — an unkeyable
- * profile must fail loud rather than collapse into a blank-model cell.
- */
-declare function agentProfileHash(profile: AgentProfile): string;
-
 /**
  * Cost tracker — token + USD accounting per scenario and per run.
  *
@@ -5601,4 +5493,4 @@ declare function traceJudge(judge: JudgeFn, judgeName: string, opts: TracedJudge
  */
 declare function traceJudgeEnsemble(judges: JudgeFn[], judgeNames: string[], opts: TracedJudgeOptions): JudgeFn;
 
-export { ANALYST_SEVERITIES, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, type AgentProfile, type AlignmentOp, Analyst, AnalystContext, AnalystCost, AnalystFinding, AnalystSeverity, AnalyzeTracesInput, AnalyzeTracesOptions, AnalyzeTracesResult, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, Artifact$1 as Artifact, type Artifact as ArtifactCheckArtifact, type ArtifactEventLike, type ArtifactValidator, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, type BackendDescriptor, BackendIntegrityError, type BackendIntegrityReport, BaselineReport, BehaviorAssertion, BenchmarkReport, BenchmarkRunner, BenchmarkRunnerConfig, type BisectOptions, type BisectResult, type BisectStep, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, BudgetSpec, CallExpectation, type CanaryAlert, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateScenario, type CausalAttributionReport, type CellVerdict, CheckResult, CollectedArtifacts, type CommandRunner, CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptComplexity, type ConceptFinding, type ConceptSpec, type ConceptWeightStrategy, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContractMetric, type ContractReport, ConvergenceTracker, type CorrectnessChecker, type CostEntry, type CostSummary, CostTracker, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateDefaultReviewerOptions, type CreateSandboxPoolOpts, type CreateTraceAnalystKindOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, D1ExperimentStore, type D1ExperimentStoreOptions, type D1Like, type D1PreparedStatementLike, DEFAULT_AGENT_SLOS, DEFAULT_COMPLEXITY_WEIGHTS, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_RUN_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, DEFAULT_TRACE_ANALYST_KINDS, Dataset, DatasetScenario, type DecideNextUserTurnOpts, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DiffPolicy, type DiffScorecardOptions, type DirEntry, type Direction, type DiscoverPersonasOptions, type DiscoveredPersona, DriverResult, DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, ERROR_COUNT_PATTERNS, type ErrorCountPattern, type EvolutionRound, type ExecutorConfig, type Expectation, type Experiment, type Run as ExperimentRun, type ExperimentStore, ExperimentTracker, type ExportedRewardModel, type ExtractOptions, type ExtractResult, FAILURE_MODE_KIND_SPEC, FINDING_SUBJECT_GRAMMAR_PROMPT, FINDING_SUBJECT_KINDS, type FactorContribution, type FactorialCell, FeedbackLabel, FeedbackTrajectory, FeedbackTrajectoryStore, type FileChange, FileSystemExperimentStore, type FileSystemExperimentStoreOptions, type FindingSubject, type FindingSubjectKind, FindingSubjectStringSchema, type FindingsDiff, FindingsStore, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type GhCliClientOptions, type GoldenSeverity, type GoldenSpec, type HarnessAdapter, HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessVariant, type HarnessVariantReport, HoldoutAuditor, type HostedJudgeConfig, type HostedJudgeDimension, type HostedJudgeRequest, type HostedJudgeResponse, type HostedRunCriticConfig, type HostedRunScoreRequest, type HostedRunScoreResponse, type HttpGithubClientOptions, type HypothesisManifest, type HypothesisResult, IMPROVEMENT_KIND_SPEC, INTENT_MATCH_JUDGE_VERSION, type ImageData, InMemoryExperimentStore, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, type JudgeAdapterOpts, type JudgeFamily, type JudgeFleetOptions, JudgeFn, JudgeInput, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, JudgeRunner, KIND_EXPECTED_SUBJECTS, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, Layer, LayerResult, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmClientOptions, type LlmCorrectnessCheckerOpts, LlmSpan, LockedJsonlAppender, MODEL_PRICING, type MatchResult, type MatcherResult, type MeasurementPolicy, type MergeOptions, MetricsCollector, type MuffledFinder, type MuffledFinding, MultiLayerVerifier, type MultiToolchainLayerConfig, type Mutator, Mutex, type Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, type ParetoResult, type PersistedFinding, 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 ProducedProposal, type ProducedState, ProductClient, ProductClientConfig, type PromptHandle, PromptRegistry, type ProposalEventLike, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, RAW_FINDING_SCHEMA_PROMPT, type RawAnalystFinding, RawAnalystFindingSchema, 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 RepoRef, type RequirementCheck, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RobustnessResult, Run$1 as Run, type RunCommandInput, type RunCommandResult, type RunConfig, RunCritic, type RunCriticAdapterOpts, type RunCriticOptions, type RunDiff, RunFilter, RunRecord, type RunScore, type RunScoreWeights, type RunTrace, type RuntimeEventLike, SEMANTIC_CONCEPT_JUDGE_VERSION, SandboxDriver, SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type SatisfiedBy, type ScanOptions, Scenario, type ScenarioCost, ScenarioFile, ScenarioRegistry, ScenarioResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SemanticConceptJudgeAdapterOpts, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SeriesConvergenceOptions, type SeriesConvergenceResult, Severity, type SignedManifest, type SignedManifestAlgo, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, Span, type SteeringBundle, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type SynthesisReason, type SynthesisTarget, type TaskGold, TestResult, type ThresholdContract, TokenCounter, type TokenSpec, type ToolCallEventLike, TraceAnalysisStore, type TraceAnalystAdapterOpts, type TraceAnalystGolden, type TraceAnalystKindSpec, TraceEmitter, TraceEvent, TraceStore, type TraceToolGroupName, type TracedAnalystOptions, type TracedJudgeOptions, Trajectory, TrajectoryStep, type TrialTrace, TurnMetrics, UNIVERSAL_FINDERS, type ValidationContext, type ValidationIssue, type ValidationResult, type VerifierAdapterOpts, VerifyContext, VerifyOptions, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, adversarialJudge, agentProfileHash, aggregatePrReviewScore, aggregateRunScore, analyzeAntiSlop, analyzeSeries, appendScorecard, assertCrossFamily, assertRealBackend, assertSingleBackend, attributeCounterfactuals, bisect, buildDriverSystemPrompt, buildReflectionPrompt, buildReviewerPrompt, buildTraceToolsForGroup, byteLengthRange, canaryLeakView, canonicalize, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, clamp01, codeExecutionJudge, coherenceJudge, collectionPreserved, commentsForSource, commitBisect, compareReferenceReplay, compilerJudge, composeValidators, containsAll, createAntiSlopJudge, createCustomJudge, createDefaultReviewer, createDomainExpertJudge, createIntentMatchJudge, createJudgeAdapter, createLlmCorrectnessChecker, createRunCriticAdapter, createSandboxPool, createSemanticConceptJudge, createSemanticConceptJudgeAdapter, createTraceAnalystAdapter, createTraceAnalystKind, createVerifierAdapter, crossTraceDiff, crowdingDistance, decideNextUserTurn, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultIsMaterial, defaultJudges, defaultReferenceReplayMatcher, deployGateLayer, diffFindings, diffScorecard, discoverPersonas, distillPlaybook, dominates, estimateCost, estimateTokens, evaluateContract, evaluateHypothesis, evaluateOracles, executeScenario, expectAgent, exportRewardModel, extractAssetUrls, extractErrorCount, extractProducedState, fileContains, fileExists, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findSkipCountsAsPass, flowLayer, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, ghCliClient, precision as goldenPrecision, hashContent, hashJson, htmlContainsElement, httpGithubClient, inMemoryReferenceReplayStore, isModelPriced, isOtelConfigured, jsonHasKeys, jsonShape, jsonlReferenceReplayStore, judgeFamily, keyPreserved, liftSeverity, linterJudge, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, matchGoldens, mergeLayerResults, mergeSteeringBundle, multiToolchainLayer, notBlocked, paraphraseRobustness, paraphraseRobustnessScenarios, paretoFrontier, paretoFrontierWithCrowding, parseCorrectnessResponse, parseFindingSubject, parseRawFinding, parseReflectionResponse, passOrthogonality, pixelDeltaRatio, politenessPrefixMutator, printDriverSummary, promptBisect, proposeAutomatedPullRequest, proposeSynthesisTargets, recordRuns, recordRunsToScorecard, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatch, regexMatches, renderFindingSubject, renderMarkdownReport, renderPlaybookMarkdown, renderPriorFindings, renderSteeringText, replayScorerOverCorpus, replayTraceThroughJudge, resetLockedAppendersForTesting, resolveModelPricing, rowCount, rowWhere, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runE2EWorkflow, runExpectations, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runReferenceReplay, runSelfPlay, runSemanticConceptJudge, scalarScore, scanForMuffledGates, scoreContinuity, scorePrReviewComments, scorePrReviewSource, scoreReferenceReplay, securityJudge, selectHarnessVariant, sentenceReorderMutator, signManifest, statusAdvanced, summarizeBackendIntegrity, summarizeHarnessResults, summarizePrReviewBenchmark, testJudge, textInSnapshot, toLangfuseEnvelope, toPrometheusText, traceJudge, traceJudgeEnsemble, tracedAnalyzeTraces, typoMutator, urlContains, verifyCompletion, verifyManifest, visualDiff, viteDeployRunner, weightedRecall, whitespaceCollapseMutator, withJudgeRetry, withOtelPipeline, wranglerDeployRunner };
+export { ANALYST_SEVERITIES, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, AgentProfile, type AlignmentOp, Analyst, AnalystContext, AnalystCost, AnalystFinding, AnalystSeverity, AnalyzeTracesInput, AnalyzeTracesOptions, AnalyzeTracesResult, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, Artifact$1 as Artifact, type Artifact as ArtifactCheckArtifact, type ArtifactEventLike, type ArtifactValidator, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, type BackendDescriptor, BaselineReport, BehaviorAssertion, BenchmarkReport, BenchmarkRunner, BenchmarkRunnerConfig, type BisectOptions, type BisectResult, type BisectStep, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, BudgetSpec, CallExpectation, type CanaryAlert, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateScenario, type CausalAttributionReport, type CellVerdict, CheckResult, CollectedArtifacts, type CommandRunner, CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptComplexity, type ConceptFinding, type ConceptSpec, type ConceptWeightStrategy, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContractMetric, type ContractReport, ConvergenceTracker, type CorrectnessChecker, type CostEntry, type CostSummary, CostTracker, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateDefaultReviewerOptions, type CreateSandboxPoolOpts, type CreateTraceAnalystKindOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, D1ExperimentStore, type D1ExperimentStoreOptions, type D1Like, type D1PreparedStatementLike, DEFAULT_AGENT_SLOS, DEFAULT_COMPLEXITY_WEIGHTS, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_RUN_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, DEFAULT_TRACE_ANALYST_KINDS, Dataset, DatasetScenario, type DecideNextUserTurnOpts, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DiffPolicy, type DiffScorecardOptions, type DirEntry, type Direction, type DiscoverPersonasOptions, type DiscoveredPersona, DriverResult, DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, ERROR_COUNT_PATTERNS, type ErrorCountPattern, type EvolutionRound, type ExecutorConfig, type Expectation, type Experiment, type Run as ExperimentRun, type ExperimentStore, ExperimentTracker, type ExportedRewardModel, type ExtractOptions, type ExtractResult, FAILURE_MODE_KIND_SPEC, FINDING_SUBJECT_GRAMMAR_PROMPT, FINDING_SUBJECT_KINDS, type FactorContribution, type FactorialCell, FeedbackLabel, FeedbackTrajectory, FeedbackTrajectoryStore, type FileChange, FileSystemExperimentStore, type FileSystemExperimentStoreOptions, type FindingSubject, type FindingSubjectKind, FindingSubjectStringSchema, type FindingsDiff, FindingsStore, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type GhCliClientOptions, type GoldenSeverity, type GoldenSpec, type HarnessAdapter, HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessVariant, type HarnessVariantReport, HoldoutAuditor, type HostedJudgeConfig, type HostedJudgeDimension, type HostedJudgeRequest, type HostedJudgeResponse, type HostedRunCriticConfig, type HostedRunScoreRequest, type HostedRunScoreResponse, type HttpGithubClientOptions, type HypothesisManifest, type HypothesisResult, IMPROVEMENT_KIND_SPEC, INTENT_MATCH_JUDGE_VERSION, type ImageData, InMemoryExperimentStore, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, type JudgeAdapterOpts, type JudgeFamily, type JudgeFleetOptions, JudgeFn, JudgeInput, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, JudgeRunner, KIND_EXPECTED_SUBJECTS, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, Layer, LayerResult, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmClientOptions, type LlmCorrectnessCheckerOpts, LlmSpan, LockedJsonlAppender, MODEL_PRICING, type MatchResult, type MatcherResult, type MeasurementPolicy, type MergeOptions, MetricsCollector, type MuffledFinder, type MuffledFinding, MultiLayerVerifier, type MultiToolchainLayerConfig, type Mutator, Mutex, type Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, type ParetoResult, type PersistedFinding, 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 ProducedProposal, type ProducedState, ProductClient, ProductClientConfig, type PromptHandle, PromptRegistry, type ProposalEventLike, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, RAW_FINDING_SCHEMA_PROMPT, type RawAnalystFinding, RawAnalystFindingSchema, 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 RepoRef, type RequirementCheck, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RobustnessResult, Run$1 as Run, type RunCommandInput, type RunCommandResult, type RunConfig, RunCritic, type RunCriticAdapterOpts, type RunCriticOptions, type RunDiff, RunFilter, RunRecord, type RunScore, type RunScoreWeights, type RunTrace, type RuntimeEventLike, SEMANTIC_CONCEPT_JUDGE_VERSION, SandboxDriver, SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type SatisfiedBy, type ScanOptions, Scenario, type ScenarioCost, ScenarioFile, ScenarioRegistry, ScenarioResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SemanticConceptJudgeAdapterOpts, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SeriesConvergenceOptions, type SeriesConvergenceResult, Severity, type SignedManifest, type SignedManifestAlgo, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, Span, type SteeringBundle, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type SynthesisReason, type SynthesisTarget, type TaskGold, TestResult, type ThresholdContract, TokenCounter, type TokenSpec, type ToolCallEventLike, TraceAnalysisStore, type TraceAnalystAdapterOpts, type TraceAnalystGolden, type TraceAnalystKindSpec, TraceEmitter, TraceEvent, TraceStore, type TraceToolGroupName, type TracedAnalystOptions, type TracedJudgeOptions, Trajectory, TrajectoryStep, type TrialTrace, TurnMetrics, UNIVERSAL_FINDERS, type ValidationContext, type ValidationIssue, type ValidationResult, type VerifierAdapterOpts, VerifyContext, VerifyOptions, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, adversarialJudge, aggregatePrReviewScore, aggregateRunScore, analyzeAntiSlop, analyzeSeries, appendScorecard, assertCrossFamily, assertSingleBackend, attributeCounterfactuals, bisect, buildDriverSystemPrompt, buildReflectionPrompt, buildReviewerPrompt, buildTraceToolsForGroup, byteLengthRange, canaryLeakView, canonicalize, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, clamp01, codeExecutionJudge, coherenceJudge, collectionPreserved, commentsForSource, commitBisect, compareReferenceReplay, compilerJudge, composeValidators, containsAll, createAntiSlopJudge, createCustomJudge, createDefaultReviewer, createDomainExpertJudge, createIntentMatchJudge, createJudgeAdapter, createLlmCorrectnessChecker, createRunCriticAdapter, createSandboxPool, createSemanticConceptJudge, createSemanticConceptJudgeAdapter, createTraceAnalystAdapter, createTraceAnalystKind, createVerifierAdapter, crossTraceDiff, crowdingDistance, decideNextUserTurn, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultIsMaterial, defaultJudges, defaultReferenceReplayMatcher, deployGateLayer, diffFindings, diffScorecard, discoverPersonas, distillPlaybook, dominates, estimateCost, estimateTokens, evaluateContract, evaluateHypothesis, evaluateOracles, executeScenario, expectAgent, exportRewardModel, extractAssetUrls, extractErrorCount, extractProducedState, fileContains, fileExists, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findSkipCountsAsPass, flowLayer, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, ghCliClient, precision as goldenPrecision, hashContent, hashJson, htmlContainsElement, httpGithubClient, inMemoryReferenceReplayStore, isModelPriced, isOtelConfigured, jsonHasKeys, jsonShape, jsonlReferenceReplayStore, judgeFamily, keyPreserved, liftSeverity, linterJudge, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, matchGoldens, mergeLayerResults, mergeSteeringBundle, multiToolchainLayer, notBlocked, paraphraseRobustness, paraphraseRobustnessScenarios, paretoFrontier, paretoFrontierWithCrowding, parseCorrectnessResponse, parseFindingSubject, parseRawFinding, parseReflectionResponse, passOrthogonality, pixelDeltaRatio, politenessPrefixMutator, printDriverSummary, promptBisect, proposeAutomatedPullRequest, proposeSynthesisTargets, recordRuns, recordRunsToScorecard, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatch, regexMatches, renderFindingSubject, renderMarkdownReport, renderPlaybookMarkdown, renderPriorFindings, renderSteeringText, replayScorerOverCorpus, replayTraceThroughJudge, resetLockedAppendersForTesting, resolveModelPricing, rowCount, rowWhere, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runE2EWorkflow, runExpectations, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runReferenceReplay, runSelfPlay, runSemanticConceptJudge, scalarScore, scanForMuffledGates, scoreContinuity, scorePrReviewComments, scorePrReviewSource, scoreReferenceReplay, securityJudge, selectHarnessVariant, sentenceReorderMutator, signManifest, statusAdvanced, summarizeHarnessResults, summarizePrReviewBenchmark, testJudge, textInSnapshot, toLangfuseEnvelope, toPrometheusText, traceJudge, traceJudgeEnsemble, tracedAnalyzeTraces, typoMutator, urlContains, verifyCompletion, verifyManifest, visualDiff, viteDeployRunner, weightedRecall, whitespaceCollapseMutator, withJudgeRetry, withOtelPipeline, wranglerDeployRunner };

package/dist/index.js

@@ -1,3 +1,6 @@
+import {
+  agentProfileHash
+} from "./chunk-PQV2TKC3.js";
 import {
   HoldoutAuditor,
   canaryLeakView,
@@ -21,7 +24,7 @@ import {
   scoreRedTeamOutput,
   summarizeBackendIntegrity,
   toolNamesForRun
-} from "./chunk-GBHRUAOF.js";
+} from "./chunk-GMXHLSLL.js";
 import {
   BENCHMARK_SPLIT_SEED,
   benchmarks_exports,
@@ -34,7 +37,7 @@ import {
   computeToolUseMetrics,
   iqr,
   welchsTTest
-} from "./chunk-QDOSODID.js";
+} from "./chunk-3B7Y5AUR.js";
 import {
   exportTrainingData,
   toNdjson
@@ -51,7 +54,7 @@ import {
   pytestTestParser,
   runTestGradedScenario,
   vitestTestParser
-} from "./chunk-YTMXBHFM.js";
+} from "./chunk-T375SUOZ.js";
 import {
   classifyEuAiRisk,
   euAiActReport,
@@ -77,7 +80,7 @@ import {
   runProposeReview,
   runProposeReviewAsControlLoop,
   scoreFromEvals
-} from "./chunk-J4DIMSRK.js";
+} from "./chunk-6EKXFFGQ.js";
 import {
   allCriticalPassed,
   objectiveEval,
@@ -92,10 +95,10 @@ import {
   evaluateReleaseConfidence,
   judgeReplayGate,
   renderReleaseReport
-} from "./chunk-AIXHUIHG.js";
+} from "./chunk-B26KI423.js";
 import {
   runEvalCampaign
-} from "./chunk-GM476SZU.js";
+} from "./chunk-AIWHLG7J.js";
 import {
   AGENT_PROFILE_KINDS,
   AgentProfileCellValidationError,
@@ -114,7 +117,7 @@ import {
   validateAgentProfileCell,
   validateRunRecord,
   verifyAgentProfileCell
-} from "./chunk-NCK5QLGT.js";
+} from "./chunk-F3SRAAZO.js";
 import {
   evaluateInterimReleaseConfidence,
   pairedEvalueSequence
@@ -125,7 +128,7 @@ import {
   paretoChart,
   researchReport,
   summaryTable
-} from "./chunk-OLIBRKRD.js";
+} from "./chunk-KX6F6NCG.js";
 import {
   benjaminiHochberg,
   bonferroni,
@@ -152,7 +155,7 @@ import {
   weightedComposite,
   weightedMean,
   wilcoxonSignedRank
-} from "./chunk-S3SDD56V.js";
+} from "./chunk-ITBRCT73.js";
 import {
   DEFAULT_TRACE_ANALYST_BUDGETS,
   FileSystemTraceStore,
@@ -189,7 +192,7 @@ import {
   tokenizeDomainWords,
   traceAnalystFunctionGroup,
   traceAnalystOnRunComplete
-} from "./chunk-PIEAE33T.js";
+} from "./chunk-Z4ZCBC7M.js";
 import {
   DEFAULT_REDACTION_RULES,
   REDACTION_VERSION,
@@ -219,7 +222,7 @@ import {
   RunIntegrityError,
   assertRunCaptured,
   throwIfRunIncomplete
-} from "./chunk-UBPIXOC4.js";
+} from "./chunk-SBCB6VZY.js";
 import {
   TraceEmitter,
   llmSpanFromProvider
@@ -242,7 +245,7 @@ import {
   isTransientLlmError,
   probeLlm,
   stripFencedJson
-} from "./chunk-VXNVVBZO.js";
+} from "./chunk-IHDHUN2X.js";
 import {
   FileSystemRawProviderSink,
   InMemoryRawProviderSink,
@@ -259,7 +262,7 @@ import {
   ReplayError,
   ValidationError,
   VerificationError
-} from "./chunk-QYJT52YW.js";
+} from "./chunk-3BFEG2F6.js";
 import "./chunk-PZ5AY32C.js";
 
 // src/run-score.ts
@@ -5953,22 +5956,6 @@ var BudgetGuard = class {
   }
 };
 
-// src/agent-profile.ts
-import { createHash as createHash2 } from "crypto";
-function agentProfileHash(profile) {
-  if (typeof profile.model !== "string" || profile.model.trim().length === 0) {
-    throw new ValidationError(`AgentProfile "${profile.id}" has no model \u2014 cannot hash`);
-  }
-  const behaviour = {
-    model: profile.model.trim(),
-    skills: [...profile.skills ?? []].sort(),
-    promptVersion: profile.promptVersion ?? null,
-    tools: [...profile.tools ?? []].sort(),
-    metadata: profile.metadata ?? {}
-  };
-  return createHash2("sha256").update(JSON.stringify(canonicalize(behaviour))).digest("hex");
-}
-
 // src/cost-tracker.ts
 var CostTracker = class {
   byScenario = /* @__PURE__ */ new Map();