@tangle-network/agent-eval 0.144.4 → 0.144.5

package/dist/index.d.ts

@@ -2,7 +2,7 @@ import { a as JudgeError, c as ReplayError, i as ConfigError, l as ValidationErr
 import { C as verifyAgentProfileCell, S as validateAgentProfileCell, _ as buildAgentInterfaceProfileCell, a as AgentProfileCellSchemaVersion, b as requireAgentProfileCell, c as AgentProfileHarness, d as AgentProfileKind, f as AgentProfileSource, g as assertRunAgentProfileCell, h as agentProfileCellKey, i as AgentProfileCellInput, l as AgentProfileJson, m as agentProfileCellHashMaterial, n as AgentInterfaceProfileLike, o as AgentProfileCellValidationError, p as AgentProfileSourceInput, r as AgentProfileCell, s as AgentProfileDimensionValue, t as AGENT_PROFILE_KINDS, u as AgentProfileJsonObject, v as buildAgentProfileCell, x as toAgentProfileJson, y as groupRunsByAgentProfileCell } from "./agent-profile-cell-Cw0PVwDr.js";
 import { t as DefaultVerdict } from "./verdict-Dps8_okt.js";
 import { a as MultiLayerVerifier, c as VerifyContext, i as LayerStatus, l as VerifyOptions, n as Layer, o as Severity, r as LayerResult, s as VerificationReport, t as Finding, u as gradeSemanticStatus } from "./multi-layer-verifier-BHY1gWAc.js";
-import { $ as RunScore, B as ConceptSpec, E as FindingSubjectKind, G as SemanticConceptJudgeOptions, H as DEFAULT_COMPLEXITY_WEIGHTS, J as runSemanticConceptJudge, K as SemanticConceptJudgeResult, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, Q as DEFAULT_RUN_SCORE_WEIGHTS, R as ConceptComplexity, T as FindingSubject, U as SEMANTIC_CONCEPT_JUDGE_VERSION, V as ConceptWeightStrategy, W as SemanticConceptJudgeInput, X as RunCriticOptions, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, et as RunScoreWeights, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, nt as clamp01, p as CONTROL_INTEGRITY_ANALYST, q as createSemanticConceptJudge, t as SKILL_USAGE_ANALYST, tt as aggregateRunScore, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, x as diffFindings, y as PersistedFinding, z as ConceptFinding } from "./skill-usage-DtpLou9L.js";
+import { $ as RunScore, B as ConceptSpec, E as FindingSubjectKind, G as SemanticConceptJudgeOptions, H as DEFAULT_COMPLEXITY_WEIGHTS, J as runSemanticConceptJudge, K as SemanticConceptJudgeResult, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, Q as DEFAULT_RUN_SCORE_WEIGHTS, R as ConceptComplexity, T as FindingSubject, U as SEMANTIC_CONCEPT_JUDGE_VERSION, V as ConceptWeightStrategy, W as SemanticConceptJudgeInput, X as RunCriticOptions, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, et as RunScoreWeights, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, nt as clamp01, p as CONTROL_INTEGRITY_ANALYST, q as createSemanticConceptJudge, t as SKILL_USAGE_ANALYST, tt as aggregateRunScore, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, x as diffFindings, y as PersistedFinding, z as ConceptFinding } from "./skill-usage-DUvvudWR.js";
 import { C as PendingCostCall, D as costForUsage, E as costForTokenPricing, O as modelPriceKey, S as PaidCallResult, T as RunPaidCallInput, _ as CostReservationExceededError, a as CostChannel, b as CustomTokenPricing, c as CostLedgerHandle, d as CostLedgerPersistenceError, f as CostLedgerSummary, g as CostReceiptInput, h as CostReceiptCaptureError, i as CostCeilingReachedError, l as CostLedgerOptions, m as CostReceipt, n as CostAccountingIncompleteError, o as CostLedger, p as CostProvenance, r as CostCallConflictError, s as CostLedgerFilter, t as ChannelRollup, u as CostLedgerPersistence, v as CostResult, w as PendingCostCallView, x as MaximumCharge, y as CostUsage } from "./cost-ledger-FuQvHxPm.js";
 import { C as TraceEvent, D as isSandboxSpan, E as isRetrievalSpan, O as isToolSpan, S as ToolSpan, T as isLlmSpan, _ as Span, a as FAILURE_CLASSES, b as SpanStatus, c as JudgeSpan, d as RetrievalSpan, f as Run, g as SandboxSpan, h as RunStatus, i as EventKind, l as LlmSpan, n as BudgetLedgerEntry, o as FailureClass, p as RunLayer, r as BudgetSpec, s as GenericSpan, t as Artifact, u as Message, v as SpanBase, w as isJudgeSpan, x as TRACE_SCHEMA_VERSION, y as SpanKind } from "./schema-BtVldJ3T.js";
 import { a as RunRecord, c as RunTaskFailure, d as isRunRecord, f as modelHasSnapshot, g as validateRunRecord, h as runTaskScore, i as RunOutcome, l as RunTerminalOutcome, m as roundTripRunRecord, n as RunCostProvenance, o as RunRecordValidationError, p as parseRunRecordSafe, r as RunJudgeMetadata, s as RunSplitTag, t as JudgeScoresRecord, u as RunTokenUsage } from "./run-record-ooo9FWns.js";
@@ -11,27 +11,27 @@ import { $ as assertLlmRoute, A as ChatClient, B as SandboxSdkTransportOpts, C a
 import { a as RunFilter, i as InMemoryTraceStore, n as FileSystemTraceStore, o as SpanFilter, r as FileSystemTraceStoreOptions, s as TraceStore, t as EventFilter } from "./store-CT9YIIve.js";
 import { a as TraceEmitterOptions, i as TraceEmitter, n as RunCompleteHookContext, o as llmSpanFromProvider, r as SpanHandle, t as RunCompleteHook } from "./emitter-DGQGoLyj.js";
 import { a as RunIntegrityReport, i as RunIntegrityIssueCode, n as RunIntegrityExpectations, o as assertRunCaptured, r as RunIntegrityIssue, s as throwIfRunIncomplete, t as RunIntegrityError } from "./integrity-DRXobPEs.js";
-import { $ as traceAnalystOnRunComplete, A as stringField, At as OtlpSpanRoleInput, B as TraceInsightReadiness, Bt as exportRunAsOtlp, C as ProjectedOtlpSpan, Ct as createOtelTracingStore, D as inferOtlpKind, Dt as OtelExporter, E as firstStringAttr, Et as OtelExportConfig, F as TraceInsightFinding, Ft as traceSpanKindToOpenInferenceKind, G as defaultTraceInsightPanel, H as TraceInsightTask, I as TraceInsightPanelRole, It as OTEL_AGENT_EVAL_SCOPE, J as inferDomainKeywords, K as describeTraceInsightScope, L as TraceInsightPromptInput, Lt as OtlpExport, M as flattenOtlpExportToNdjson, Mt as applyToolSpanOtlpAttributes, N as OtlpFlatLine, Nt as classifyOtlpSpanRole, O as projectOtlpFlatLine, Ot as createOtelExporter, P as TraceInsightContext, Pt as isOtlpModelCall, Q as TraceAnalystHookOptions, R as TraceInsightQualityGate, Rt as OtlpResourceSpans, S as otlpToTraceRunRecords, St as redactValue, T as extractOtlpAttributes, Tt as ExportableSpan, U as buildTraceInsightContext, V as TraceInsightSuite, W as buildTraceInsightPrompt, X as scoreTraceInsightReadiness, Y as planTraceInsightQuestions, Z as tokenizeDomainWords, _ as OtlpTraceRunRecord, _t as DEFAULT_REDACTION_RULES, a as ReplayFetchOptions, at as TraceFileMissingError, b as otlpRowsToTraceRunRecords, bt as RedactionRule, c as ToolTraceMissingError, ct as AnalyzeTracesInput, d as OtlpFileTraceStoreOptions, dt as analyzeTraces, et as SpanNotFoundError, f as ToolSpansToTraceAnalysisStoreOptions, ft as createBoundedTraceAnalysisStore, g as OtlpToRunRecordsOptions, gt as convertTraceStoresToOtlp, h as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, ht as TracesToOtlpResult, i as ReplayCacheStats, it as TraceFileMalformedError, j as FlattenOtlpOptions, jt as ToolSpanOtlpInput, k as readOtlpStatus, kt as OtlpSpanRole, l as toolSpansToTraceAnalysisStore, lt as AnalyzeTracesOptions, m as TRACE_ANALYST_ACTOR_DESCRIPTION, mt as TraceStoreToOtlpOptions, n as ReplayCacheEntry, nt as TraceAnalysisStoreContractError, o as createReplayFetch, ot as TraceFileTooLargeError, p as otlpTextToTraceAnalysisStore, pt as TraceStoreSource, q as domainEvidencePattern, r as ReplayCacheMissError, rt as TraceAnalysisValidationError, s as iterateRawCalls, st as TraceNotFoundError, t as ReplayCache, tt as TraceAnalysisLimitError, u as OtlpFileTraceStore, ut as AnalyzeTracesResult, v as TraceAggregate, vt as REDACTION_VERSION, w as asString, wt as otelRunCompleteHook, x as otlpToRunRecords, xt as redactString, y as otlpRowsToRunRecords, yt as RedactionReport, z as TraceInsightQuestion, zt as OtlpSpan } from "./replay-DQ-55DC_.js";
+import { $ as traceAnalystOnRunComplete, A as stringField, At as OtlpSpanRoleInput, B as TraceInsightReadiness, Bt as exportRunAsOtlp, C as ProjectedOtlpSpan, Ct as createOtelTracingStore, D as inferOtlpKind, Dt as OtelExporter, E as firstStringAttr, Et as OtelExportConfig, F as TraceInsightFinding, Ft as traceSpanKindToOpenInferenceKind, G as defaultTraceInsightPanel, H as TraceInsightTask, I as TraceInsightPanelRole, It as OTEL_AGENT_EVAL_SCOPE, J as inferDomainKeywords, K as describeTraceInsightScope, L as TraceInsightPromptInput, Lt as OtlpExport, M as flattenOtlpExportToNdjson, Mt as applyToolSpanOtlpAttributes, N as OtlpFlatLine, Nt as classifyOtlpSpanRole, O as projectOtlpFlatLine, Ot as createOtelExporter, P as TraceInsightContext, Pt as isOtlpModelCall, Q as TraceAnalystHookOptions, R as TraceInsightQualityGate, Rt as OtlpResourceSpans, S as otlpToTraceRunRecords, St as redactValue, T as extractOtlpAttributes, Tt as ExportableSpan, U as buildTraceInsightContext, V as TraceInsightSuite, W as buildTraceInsightPrompt, X as scoreTraceInsightReadiness, Y as planTraceInsightQuestions, Z as tokenizeDomainWords, _ as OtlpTraceRunRecord, _t as DEFAULT_REDACTION_RULES, a as ReplayFetchOptions, at as TraceFileMissingError, b as otlpRowsToTraceRunRecords, bt as RedactionRule, c as ToolTraceMissingError, ct as AnalyzeTracesInput, d as OtlpFileTraceStoreOptions, dt as analyzeTraces, et as SpanNotFoundError, f as ToolSpansToTraceAnalysisStoreOptions, ft as createBoundedTraceAnalysisStore, g as OtlpToRunRecordsOptions, gt as convertTraceStoresToOtlp, h as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, ht as TracesToOtlpResult, i as ReplayCacheStats, it as TraceFileMalformedError, j as FlattenOtlpOptions, jt as ToolSpanOtlpInput, k as readOtlpStatus, kt as OtlpSpanRole, l as toolSpansToTraceAnalysisStore, lt as AnalyzeTracesOptions, m as TRACE_ANALYST_ACTOR_DESCRIPTION, mt as TraceStoreToOtlpOptions, n as ReplayCacheEntry, nt as TraceAnalysisStoreContractError, o as createReplayFetch, ot as TraceFileTooLargeError, p as otlpTextToTraceAnalysisStore, pt as TraceStoreSource, q as domainEvidencePattern, r as ReplayCacheMissError, rt as TraceAnalysisValidationError, s as iterateRawCalls, st as TraceNotFoundError, t as ReplayCache, tt as TraceAnalysisLimitError, u as OtlpFileTraceStore, ut as AnalyzeTracesResult, v as TraceAggregate, vt as REDACTION_VERSION, w as asString, wt as otelRunCompleteHook, x as otlpToRunRecords, xt as redactString, y as otlpRowsToRunRecords, yt as RedactionReport, z as TraceInsightQuestion, zt as OtlpSpan } from "./replay-B7S7Pdbw.js";
 import { C as TOOL_LATENCY_MS, D as asNumber, E as applyLlmSpanOtlpAttributes, O as contextInputTokens, S as TOOL_ARGS_CAPTURED, T as TOOL_NAME_ATTR_KEYS, _ as LlmSpanOtlpInput, a as LLM_CACHE_WRITE_TOKEN_ATTR_KEYS, b as RUN_COST_ATTR_KEYS, c as LLM_COST_USD, d as LLM_MODEL_ATTR_KEYS, f as LLM_MODEL_NAME, g as LLM_REASONING_TOKEN_ATTR_KEYS, h as LLM_REASONING_TOKENS, i as LLM_CACHE_WRITE_TOKENS, k as firstNumberAttr, l as LLM_INPUT_TOKENS, m as LLM_OUTPUT_TOKEN_ATTR_KEYS, n as LLM_CACHED_TOKENS, o as LLM_CONTEXT_TOKENS, p as LLM_OUTPUT_TOKENS, r as LLM_CACHED_TOKEN_ATTR_KEYS, s as LLM_COST_ATTR_KEYS, t as INPUT_VALUE, u as LLM_INPUT_TOKEN_ATTR_KEYS, v as OPENINFERENCE_SPAN_KIND, w as TOOL_NAME, x as SPAN_KIND_ATTR_KEYS, y as OUTPUT_VALUE } from "./attribute-vocabulary-DLJ6303h.js";
 import { a as judgeSpans, c as runsForScenario, i as hasCapturedToolArgs, l as toolSpans, n as argHash, o as llmSpans, r as groupBy, s as runFailureClass, t as aggregateLlm } from "./query-CJ_DX8vl.js";
-import { A as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, C as DEFAULT_TRACE_ANALYST_BUDGETS, D as SearchSpanResult, E as QueryTracesPage, F as TraceAnalystSpanStatus, I as TraceAnalystTraceSummary, L as ViewSpansResult, M as TraceAnalystFilters, N as TraceAnalystSpan, O as SearchTraceResult, P as TraceAnalystSpanKind, R as ViewTraceOversized, S as TraceAnalysisStoreContext, T as ErrorCluster, _ as makeFinding, a as AnalystInputKind, b as TRACE_ANALYSIS_LIMITS, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, j as TraceAnalystByteBudgets, k as SpanMatchRecord, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, w as DatasetOverview, x as TraceAnalysisStore, y as BoundedTraceAnalysisStoreOptions, z as ViewTraceResult } from "./types-y8jrxXWd.js";
-import { a as createTraceAnalyst, c as runTraceAnalyst, f as BehavioralMetrics, g as computeTraceMetrics, h as SuboptimalSignal, i as TraceAnalystDefinition, m as SuboptimalCode, n as buildDefaultAnalystRegistry, o as renderPriorFindings, p as BehavioralTokenSequence, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions } from "./default-registry-iXfu2trt.js";
-import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "./exact-types-BygCBR4L.js";
-import { A as ExactAnalystRunExecutionError, C as jsonHasKeys, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, S as containsAll, T as AnalystHooks, _ as ValidationContext, a as ProducedProposal, b as byteLengthRange, c as SatisfiedBy, d as createLlmCorrectnessChecker, f as createTokenRecallChecker, g as ArtifactValidator, h as Artifact$1, i as LlmCorrectnessCheckerOpts, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy, l as TaskGold, m as verifyCompletion, n as CompletionVerdict, o as ProducedState, p as parseCorrectnessResponse, r as CorrectnessChecker, s as RequirementCheck, t as CompletionRequirement, u as completionVerdict, v as ValidationIssue, w as regexMatch, x as composeValidators, y as ValidationResult } from "./completion-verifier-EJERfFwF.js";
-import { R as Scenario$1, S as JudgeConfig$1, w as JudgeScore$1 } from "./types-DcJxgsLy.js";
+import { A as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, C as DEFAULT_TRACE_ANALYST_BUDGETS, D as SearchSpanResult, E as QueryTracesPage, F as TraceAnalystSpanStatus, I as TraceAnalystTraceSummary, L as ViewSpansResult, M as TraceAnalystFilters, N as TraceAnalystSpan, O as SearchTraceResult, P as TraceAnalystSpanKind, R as ViewTraceOversized, S as TraceAnalysisStoreContext, T as ErrorCluster, _ as makeFinding, a as AnalystInputKind, b as TRACE_ANALYSIS_LIMITS, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, j as TraceAnalystByteBudgets, k as SpanMatchRecord, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, w as DatasetOverview, x as TraceAnalysisStore, y as BoundedTraceAnalysisStoreOptions, z as ViewTraceResult } from "./types-CZt1PBIk.js";
+import { a as createTraceAnalyst, c as runTraceAnalyst, f as BehavioralMetrics, g as computeTraceMetrics, h as SuboptimalSignal, i as TraceAnalystDefinition, m as SuboptimalCode, n as buildDefaultAnalystRegistry, o as renderPriorFindings, p as BehavioralTokenSequence, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions } from "./default-registry-BwbZ9N9v.js";
+import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "./exact-types-BQ7W90C4.js";
+import { A as ExactAnalystRunExecutionError, C as jsonHasKeys, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, S as containsAll, T as AnalystHooks, _ as ValidationContext, a as ProducedProposal, b as byteLengthRange, c as SatisfiedBy, d as createLlmCorrectnessChecker, f as createTokenRecallChecker, g as ArtifactValidator, h as Artifact$1, i as LlmCorrectnessCheckerOpts, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy, l as TaskGold, m as verifyCompletion, n as CompletionVerdict, o as ProducedState, p as parseCorrectnessResponse, r as CorrectnessChecker, s as RequirementCheck, t as CompletionRequirement, u as completionVerdict, v as ValidationIssue, w as regexMatch, x as composeValidators, y as ValidationResult } from "./completion-verifier-VvpHRu78.js";
+import { R as Scenario$1, S as JudgeConfig$1, w as JudgeScore$1 } from "./types-Dcoaqcsc.js";
 import { a as DatasetScenario, c as SliceOptions, i as DatasetProvenance, l as hashScenarios, n as DatasetDifficulty, o as DatasetSplit, r as DatasetManifest, s as HoldoutLockedError, t as Dataset } from "./dataset-v_Y5902-.js";
 import { $ as pairedCohensDz, A as WeightedCompositeInput, At as positionalBias, B as eProcess, C as RankTestMethod, Ct as GoldenItem, D as ScoreRiskDifferenceResult, Dt as calibrateJudge, E as RiskDifferenceResult, Et as VerbosityBiasResult, F as cliffsDelta, G as mannWhitneyU, H as interRaterReliability, I as cohensD, J as mcnemarRequiredN, K as mcnemar, L as confidenceInterval, M as WilcoxonSignedRankResult, Mt as verbosityBias, N as benjaminiHochberg, O as SignTestAlternative, Ot as calibrateJudgeContinuous, P as bonferroni, Q as pairedBootstrap, R as corpusInterRaterAgreement, S as ProportionInterval, St as ContinuousCalibrationResult, T as RankTestOptions, Tt as SelfPreferenceResult, U as interpretCliffs, V as holm, W as isBinaryOutcomeVector, X as normalizeScores, Y as mulberry32, Z as pairedBinaryScale, _ as McNemarResult, _t as wilson, a as CorpusAgreementReport, at as pairedSignTest, b as PairedSignTestResult, bt as ContinuousAgreement, c as DEFAULT_PERMUTATIONS, ct as passAtK, d as EProcessState, dt as requiredPairedSampleSize, et as pairedDeltaTieFraction, f as EProcessStep, ft as requiredSampleSize, g as MannWhitneyResult, gt as wilcoxonSignedRank, h as MANN_WHITNEY_EXACT_MAX_WORK, ht as weightedMean, i as CorpusAgreementPerDimension, it as pairedRiskDifferenceScore, j as WeightedCompositeResult, jt as selfPreference, k as WILCOXON_EXACT_MAX_N, kt as continuousAgreement, l as EProcess, lt as pearsonR, m as MANN_WHITNEY_EXACT_MAX_STATES, mt as weightedComposite, n as CliffsMagnitude, nt as pairedRiskDifference, o as CorpusScoreRecord, ot as pairedTTest, p as ExactRiskDifferenceResult, pt as spearmanR, q as mcnemarPower, r as CorpusAgreementOptions, rt as pairedRiskDifferenceExact, s as DECISION_PAIRED_DELTA_STATISTIC, st as partialCredit, t as BOOTSTRAP_GATE_MIN_N, tt as pairedMde, u as EProcessOptions, ut as ranks, v as PairedBootstrapOptions, vt as CalibrationResult, w as RankTestMethodRequest, wt as PositionalBiasResult, x as PairedTTestResult, xt as ContinuousAgreementOptions, y as PairedBootstrapResult, yt as CandidateScore, z as corpusInterRaterAgreementFromJudgeScores } from "./statistics-B4u_CiFd.js";
 import { A as PairedMcNemarEvidence, C as HeldOutGate, D as PairedDecisionMethod, E as SplitCoverage, M as PairedPromotionDecisionOptions, N as decidePairedPromotion, O as PairedDecisionShape, P as pairedDecisionShape, S as GateEvidence, T as HeldOutGateRejectionCode, _ as paretoChart, a as ParetoPoint, b as DeltaStatistic, c as ResearchReportCandidate, d as ResearchReportOptions, f as ResearchReportRecommendation, g as gainHistogram, h as SummaryTableRow, i as ParetoFigureSpec, j as PairedPromotionDecision, k as PairedDecisionStatistic, l as ResearchReportDecision, m as SummaryTableOptions, n as GainDistributionFigureSpec, o as RESEARCH_REPORT_HARD_PAIR_FLOOR, p as SummaryTable, r as GainDistributionOptions, s as ResearchReport, t as GainDistributionBin, u as ResearchReportMethodology, v as researchReport, w as HeldOutGateConfig, x as GateDecision, y as summaryTable } from "./summary-report-BOM6dfP7.js";
 import { a as FailureClassification, c as classifyFailure, i as DEFAULT_RULES, o as FailureContext, s as FailureRule } from "./failure-cluster-CqcvCcdR.js";
-import { C as MakeEvalToolsConfig, S as EvalToolDef, T as toOpenAiTool, _ as BuildTraceAnalysisToolsOptions, a as TraceAnalysisEngineRequest, b as buildTraceAnalysisToolDescriptors, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, i as TraceAnalysisEngine, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, v as TRACE_ANALYST_TOOL_NAMESPACE, w as makeEvalTools, x as traceAnalystFunctionGroup, y as TraceAnalysisToolDescriptor } from "./tool-groups-CMmsgTzj.js";
+import { C as MakeEvalToolsConfig, S as EvalToolDef, T as toOpenAiTool, _ as BuildTraceAnalysisToolsOptions, a as TraceAnalysisEngineRequest, b as buildTraceAnalysisToolDescriptors, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, i as TraceAnalysisEngine, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, v as TRACE_ANALYST_TOOL_NAMESPACE, w as makeEvalTools, x as traceAnalystFunctionGroup, y as TraceAnalysisToolDescriptor } from "./tool-groups-ByZiqpVk.js";
 import { A as isTrainableSplit, D as assertRolloutLine, E as assertMintedLines, T as assertMinted, b as RolloutSplit, h as RolloutLine, i as ChatToolCall, j as validateRolloutLine, k as isRolloutLine, n as ChatMessage, o as MintedRolloutLine, p as RolloutCapture, s as MintedRolloutOutcome, u as ROLLOUT_SCHEMA, w as ToolDef, x as RolloutStep, y as RolloutRole } from "./schema-Cef2cFmb.js";
 import { C as Unavailable, D as showMeasured, E as isUnavailable, S as SupervisorRunTreeGapCode, _ as SupervisorRunReport, b as SupervisorRunTree, f as SUPERVISOR_RUN_SCHEMA, g as SupervisorRunReader, h as SupervisorRunNodeRole, p as SourceLimits, r as Measured, v as SupervisorRunRollup, x as SupervisorRunTreeGap, y as SupervisorRunSources } from "./types-yLK8gXE9.js";
-import { B as BenchmarkSource, F as BenchmarkDatasetItem, H as deterministicSplit, I as BenchmarkEvaluation, L as BenchmarkFamily, N as BENCHMARK_SPLIT_SEED, P as BenchmarkAdapter, R as BenchmarkResponder, V as BenchmarkTaskKind, t as index_d_exports, z as BenchmarkScenario } from "./index-D_qTihaQ.js";
-import { $ as Objective, An as LlmJudgeDimension, At as CanarySeverity, Cn as REFERENCE_EQUIVALENCE_JUDGE_VERSION, Ct as scoreRedTeamOutput, Dn as ReferenceEquivalenceScenario, Dt as CanaryKind, En as ReferenceEquivalenceJudgeResult, Et as CanaryEvaluation, Mn as llmJudge, On as createReferenceEquivalenceJudge, Ot as CanaryOptions, Q as Direction, Sn as REFERENCE_EQUIVALENCE_INPUT_LIMITS, St as redTeamReport, Tn as ReferenceEquivalenceJudgeOptions, Tt as CanaryAlert, _t as RedTeamCategory, at as scalarScore, bt as RedTeamReport, et as ParetoResult, gt as RedTeamCase, ht as DEFAULT_RED_TEAM_CORPUS, it as paretoFrontierWithCrowding, jn as LlmJudgeOptions, jt as runCanaries, kn as runReferenceEquivalenceJudge, kt as CanaryReport, nt as dominates, rt as paretoFrontier, tt as crowdingDistance, vt as RedTeamFinding, wn as ReferenceEquivalenceJudgeInput, wt as toolNamesForRun, xt as redTeamDataset, yt as RedTeamPayload } from "./skillopt-optimization-method-CvSJGdm3.js";
-import { $r as pairRunRecords, $t as HarnessType, Bn as SignedManifest, Gn as signManifest, Gr as PairArmsResult, Gt as BackendIntegrityReport, Hn as canonicalize, Hr as MatchedPair, Jr as PairedArmsComparison, Jt as summarizeAgentReceiptIntegrity, Kn as verifyManifest, Kr as PairRunRecordsResult, Kt as assertRealAgentReceipts, Qr as pairArms, Qt as HARNESS_NATIVE_MODEL, Rn as HypothesisManifest, Un as evaluateHypothesis, Ur as MatchedRunRecordPair, Vn as SignedManifestAlgo, Vr as ComparePairedArmsOptions, Wn as hashJson, Wr as PairArmsOptions, Wt as BackendIntegrityError, Xr as PairedMetricDelta, Xt as AgentProfile, Yr as PairedCorrectness, Yt as summarizeBackendIntegrity, Zr as comparePairedArms, Zt as CODING_HARNESSES, an as harnessAxisOf, cn as RuntimeEventLike, en as ProfileAxisSpec, in as expandProfileAxes, ln as ToolCallEventLike, nn as agentProfileId, on as ArtifactEventLike, qr as PairedArmRow, qt as assertRealBackend, rn as agentProfileModelId, sn as ProposalEventLike, tn as agentProfileHash, un as extractProducedState, zn as HypothesisResult } from "./index-BuHs_OnD.js";
+import { B as BenchmarkSource, F as BenchmarkDatasetItem, H as deterministicSplit, I as BenchmarkEvaluation, L as BenchmarkFamily, N as BENCHMARK_SPLIT_SEED, P as BenchmarkAdapter, R as BenchmarkResponder, V as BenchmarkTaskKind, t as index_d_exports, z as BenchmarkScenario } from "./index-CsuAo2-J.js";
+import { $ as Objective, An as LlmJudgeDimension, At as CanarySeverity, Cn as REFERENCE_EQUIVALENCE_JUDGE_VERSION, Ct as scoreRedTeamOutput, Dn as ReferenceEquivalenceScenario, Dt as CanaryKind, En as ReferenceEquivalenceJudgeResult, Et as CanaryEvaluation, Mn as llmJudge, On as createReferenceEquivalenceJudge, Ot as CanaryOptions, Q as Direction, Sn as REFERENCE_EQUIVALENCE_INPUT_LIMITS, St as redTeamReport, Tn as ReferenceEquivalenceJudgeOptions, Tt as CanaryAlert, _t as RedTeamCategory, at as scalarScore, bt as RedTeamReport, et as ParetoResult, gt as RedTeamCase, ht as DEFAULT_RED_TEAM_CORPUS, it as paretoFrontierWithCrowding, jn as LlmJudgeOptions, jt as runCanaries, kn as runReferenceEquivalenceJudge, kt as CanaryReport, nt as dominates, rt as paretoFrontier, tt as crowdingDistance, vt as RedTeamFinding, wn as ReferenceEquivalenceJudgeInput, wt as toolNamesForRun, xt as redTeamDataset, yt as RedTeamPayload } from "./skillopt-optimization-method-CGz9ywhM.js";
+import { $r as pairRunRecords, $t as HarnessType, Bn as SignedManifest, Gn as signManifest, Gr as PairArmsResult, Gt as BackendIntegrityReport, Hn as canonicalize, Hr as MatchedPair, Jr as PairedArmsComparison, Jt as summarizeAgentReceiptIntegrity, Kn as verifyManifest, Kr as PairRunRecordsResult, Kt as assertRealAgentReceipts, Qr as pairArms, Qt as HARNESS_NATIVE_MODEL, Rn as HypothesisManifest, Un as evaluateHypothesis, Ur as MatchedRunRecordPair, Vn as SignedManifestAlgo, Vr as ComparePairedArmsOptions, Wn as hashJson, Wr as PairArmsOptions, Wt as BackendIntegrityError, Xr as PairedMetricDelta, Xt as AgentProfile, Yr as PairedCorrectness, Yt as summarizeBackendIntegrity, Zr as comparePairedArms, Zt as CODING_HARNESSES, an as harnessAxisOf, cn as RuntimeEventLike, en as ProfileAxisSpec, in as expandProfileAxes, ln as ToolCallEventLike, nn as agentProfileId, on as ArtifactEventLike, qr as PairedArmRow, qt as assertRealBackend, rn as agentProfileModelId, sn as ProposalEventLike, tn as agentProfileHash, un as extractProducedState, zn as HypothesisResult } from "./index-CpxZSlB7.js";
 import { _ as vitestTestParser, a as DockerSandboxDriver, c as SandboxHarness, d as SubprocessSandboxDriver, f as SubprocessSandboxDriverOptions, g as pytestTestParser, h as jestTestParser, i as runTestGradedScenario, l as SandboxHarnessResult, m as composeParsers, n as TestGradedRunResult, o as HarnessConfig, p as TestOutputParser, r as TestGradedScenario, s as SandboxDriver, t as TestGradedRunOptions, u as SandboxResult } from "./test-graded-scenario-D1TaI2va.js";
-import { $ as ControlRunResult, A as feedbackTrajectoryToOptimizerRow, B as AnalystReviewCounts, C as analystRunToReviewRequests, D as feedbackTrajectoriesToDatasetScenarios, E as createFeedbackTrajectory, F as serializeFeedbackTrajectoriesJsonl, G as analystFindingDigest, H as AnalystReviewQuality, I as summarizePreferenceMemory, J as ControlActionOutcome, K as analystRunDigest, L as withAssignedFeedbackSplit, M as renderPreferenceMemoryMarkdown, N as replayFeedbackTrajectories, O as feedbackTrajectoriesToOptimizerRows, P as replayFeedbackTrajectory, Q as ControlEvalResult, R as AnalystFindingDigest, S as analystRunToFeedbackTrajectory, T as controlRunToFeedbackTrajectory, U as AnalystReviewSource, V as AnalystReviewDecision, W as AnalystRunDigest, X as ControlContext, Y as ControlBudget, Z as ControlDecision, _ as FeedbackTrajectoryStore, a as FeedbackLabel, at as StopDecision, b as PreferenceMemoryEntry, c as FeedbackOptimizerRow, ct as runAgentControlLoop, d as FeedbackReplayResult, dt as subjectiveEval, et as ControlRuntimeConfig, f as FeedbackSeverity, g as FeedbackTrajectoryFilter, h as FeedbackTrajectory, i as FeedbackAttempt, it as ControlStopPolicies, j as parseFeedbackTrajectoriesJsonl, k as feedbackTrajectoryToDatasetScenario, l as FeedbackOutcome, lt as stopOnNoProgress, m as FeedbackTask, n as AnalystReviewRequest, nt as ControlSeverity, o as FeedbackLabelKind, ot as allCriticalPassed, p as FeedbackSplitPolicy, q as ControlActionFailureMode, r as FeedbackArtifactType, rt as ControlStep, s as FeedbackLabelSource, st as objectiveEval, t as AnalystFeedbackTrajectoryOptions, tt as ControlRuntimeError, u as FeedbackReplayAdapter, ut as stopOnRepeatedAction, v as FileSystemFeedbackTrajectoryStore, w as assignFeedbackSplit, x as ProposedSideEffect, y as InMemoryFeedbackTrajectoryStore, z as AnalystMissedIssue } from "./feedback-trajectory-CSIkRLQX.js";
-import { A as ActionExecutionPolicy, C as ReviewMemoryStore, D as inMemoryReviewStore, E as createLlmReviewer, M as evaluateActionPolicy, O as jsonlReviewStore, S as ReviewMemoryEntry, T as VerifyFn, _ as ProposeReviewReport, a as ProposeReviewControlAction, b as ReviewFn, c as ProposeReviewControlState, d as LlmJsonCall, f as LlmReviewerConfig, g as ProposeReviewConfig, h as ProposeOutput, i as scoreFromEvals, j as ActionPolicyDecision, k as runProposeReview, l as controlFailureClassFromVerification, m as ProposeInput, n as RunEvidenceMetadata, o as ProposeReviewControlConfig, p as ProposeFn, r as controlRunToRunRecord, s as ProposeReviewControlResult, t as ControlRunToRunRecordOptions, u as runProposeReviewAsControlLoop, v as ProposeReviewShot, w as Verification, x as ReviewInput, y as Review } from "./run-evidence-H1vRpIdT.js";
+import { $ as ControlRunResult, A as feedbackTrajectoryToOptimizerRow, B as AnalystReviewCounts, C as analystRunToReviewRequests, D as feedbackTrajectoriesToDatasetScenarios, E as createFeedbackTrajectory, F as serializeFeedbackTrajectoriesJsonl, G as analystFindingDigest, H as AnalystReviewQuality, I as summarizePreferenceMemory, J as ControlActionOutcome, K as analystRunDigest, L as withAssignedFeedbackSplit, M as renderPreferenceMemoryMarkdown, N as replayFeedbackTrajectories, O as feedbackTrajectoriesToOptimizerRows, P as replayFeedbackTrajectory, Q as ControlEvalResult, R as AnalystFindingDigest, S as analystRunToFeedbackTrajectory, T as controlRunToFeedbackTrajectory, U as AnalystReviewSource, V as AnalystReviewDecision, W as AnalystRunDigest, X as ControlContext, Y as ControlBudget, Z as ControlDecision, _ as FeedbackTrajectoryStore, a as FeedbackLabel, at as StopDecision, b as PreferenceMemoryEntry, c as FeedbackOptimizerRow, ct as runAgentControlLoop, d as FeedbackReplayResult, dt as subjectiveEval, et as ControlRuntimeConfig, f as FeedbackSeverity, g as FeedbackTrajectoryFilter, h as FeedbackTrajectory, i as FeedbackAttempt, it as ControlStopPolicies, j as parseFeedbackTrajectoriesJsonl, k as feedbackTrajectoryToDatasetScenario, l as FeedbackOutcome, lt as stopOnNoProgress, m as FeedbackTask, n as AnalystReviewRequest, nt as ControlSeverity, o as FeedbackLabelKind, ot as allCriticalPassed, p as FeedbackSplitPolicy, q as ControlActionFailureMode, r as FeedbackArtifactType, rt as ControlStep, s as FeedbackLabelSource, st as objectiveEval, t as AnalystFeedbackTrajectoryOptions, tt as ControlRuntimeError, u as FeedbackReplayAdapter, ut as stopOnRepeatedAction, v as FileSystemFeedbackTrajectoryStore, w as assignFeedbackSplit, x as ProposedSideEffect, y as InMemoryFeedbackTrajectoryStore, z as AnalystMissedIssue } from "./feedback-trajectory-WK7x4mhy.js";
+import { A as ActionExecutionPolicy, C as ReviewMemoryStore, D as inMemoryReviewStore, E as createLlmReviewer, M as evaluateActionPolicy, O as jsonlReviewStore, S as ReviewMemoryEntry, T as VerifyFn, _ as ProposeReviewReport, a as ProposeReviewControlAction, b as ReviewFn, c as ProposeReviewControlState, d as LlmJsonCall, f as LlmReviewerConfig, g as ProposeReviewConfig, h as ProposeOutput, i as scoreFromEvals, j as ActionPolicyDecision, k as runProposeReview, l as controlFailureClassFromVerification, m as ProposeInput, n as RunEvidenceMetadata, o as ProposeReviewControlConfig, p as ProposeFn, r as controlRunToRunRecord, s as ProposeReviewControlResult, t as ControlRunToRunRecordOptions, u as runProposeReviewAsControlLoop, v as ProposeReviewShot, w as Verification, x as ReviewInput, y as Review } from "./run-evidence-j5Ynww6L.js";
 import { _ as ReleaseConfidenceStatus, a as JudgeReplayGateArgs, b as assertReleaseConfidence, c as judgeReplayGate, d as ReleaseConfidenceAxis, f as ReleaseConfidenceAxisName, g as ReleaseConfidenceScorecard, h as ReleaseConfidenceMetrics, i as BootstrapResult, l as ActionableSideInfo, m as ReleaseConfidenceIssue, n as renderReleaseReport, o as Verdict, p as ReleaseConfidenceInput, r as BootstrapOptions, s as bootstrapCi, t as RenderReleaseReportOptions, u as AsiSeverity, v as ReleaseConfidenceThresholds, x as evaluateReleaseConfidence, y as ReleaseTraceEvidence } from "./release-report-BZRWdq_t.js";
 import { $ as ScoreOrigin, $t as toSftRows, Ct as HarborSubagentTrajectoryRef, Dt as relabelImportedSplit, Et as fromHarborTrajectory, Gt as SftRow, Ht as RewardRow, J as MintRolloutOptions, Ot as toHarborTrajectories, Q as unmintableReasons, St as HarborStepSource, Tt as HarborTrajectory, Wt as SftExportOptions, X as RolloutScrubber, Xt as toRewardRows, Y as MintRolloutResult, Yt as toJsonl, Z as mintRolloutRows, _t as HarborImageSource, at as trainingReward, bt as HarborObservationResult, dt as ATIF_SCHEMA_VERSION, et as ScorePreference, ft as FromHarborOptions, gt as HarborFinalMetrics, ht as HarborContentPart, it as scoreOrigin, kt as toHarborTrajectory, mt as HarborAgent, nt as observedScore, ot as trainingScore, pt as HARBOR_IMPORT_GAP, rt as observedSplitScore, tt as isRealnessGated, vt as HarborMetrics, wt as HarborToolCall, xt as HarborStep, yt as HarborObservation } from "./index-CNOCxBLh.js";
 import { C as SupervisorRunIntegrityIssue, D as SupervisorRunIntegritySeverity, E as SupervisorRunIntegrityReport, F as analyzeSupervisorRunSources, L as rollupSupervisorRuns, N as claudeCodeSupervisorRunReader, O as SupervisorRunSourceOnlyCheckCode, P as readClaudeCodeSupervisorRun, S as SupervisorRunIntegrityEvidence, T as SupervisorRunIntegrityOptions, a as supervisorRunRolloutLines, b as analyzeSupervisorRunIntegrity, c as renderSupervisorRunMarkdown, d as analyzeSupervisorRun, n as readRuntimeSupervisorRun, r as runtimeSupervisorRunReader, s as renderSupervisorRunHeadline, t as isRuntimeSupervisorRunDir, v as writeSupervisorRunReport, w as SupervisorRunIntegrityIssueCode, x as SUPERVISOR_RUN_INTEGRITY_SCHEMA } from "./index-BZUe-ODI.js";

package/dist/index.js

@@ -6,17 +6,17 @@ import { a as estimateTokens, i as estimateCost, n as MetricsCollector, o as isM
 import { a as CostLedgerPersistenceError, c as costForTokenPricing, i as CostLedger, l as costForUsage, n as CostCallConflictError, o as CostReceiptCaptureError, r as CostCeilingReachedError, s as CostReservationExceededError, t as CostAccountingIncompleteError, u as modelPriceKey } from "./cost-ledger-DMFxsLKr.js";
 import { a as providerFromBaseUrl, i as defaultProviderRedactor, n as InMemoryRawProviderSink, r as NoopRawProviderSink, t as FileSystemRawProviderSink } from "./raw-provider-sink-BQd7mzyT.js";
 import { a as assertLlmRoute, c as callLlmJson, d as isTransientLlmError, f as maximumChargeForLlmRequest, i as LlmRouteAssertionError, l as costReceiptFromLlm, m as stripFencedJson, n as LlmClient, o as backoffMs, p as probeLlm, r as LlmResponseError, s as callLlm, t as LlmCallError, u as costReceiptFromLlmError } from "./llm-client-D3EoChAU.js";
-import { C as createChatClient, S as computeTraceMetrics, _ as CONTROL_INTEGRITY_ANALYST, a as analystFindingDigest, c as completedAnalystReviewQuality, d as validateAnalystReviewDecisions, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, l as readAnalystReview, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, o as analystRunDigest, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, s as assertUniqueFindingIds, t as buildDefaultAnalystRegistry, u as snapshotAnalystRun, v as ControlIntegrityAnalyst, y as emitControlIntegrityFindings } from "./default-registry-SOyHB6qG.js";
+import { C as createChatClient, S as computeTraceMetrics, _ as CONTROL_INTEGRITY_ANALYST, a as analystFindingDigest, c as completedAnalystReviewQuality, d as validateAnalystReviewDecisions, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, l as readAnalystReview, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, o as analystRunDigest, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, s as assertUniqueFindingIds, t as buildDefaultAnalystRegistry, u as snapshotAnalystRun, v as ControlIntegrityAnalyst, y as emitControlIntegrityFindings } from "./default-registry-RLNNoeEP.js";
 import { INPUT_VALUE, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_CACHE_WRITE_TOKENS, LLM_CACHE_WRITE_TOKEN_ATTR_KEYS, LLM_CONTEXT_TOKENS, 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, LLM_REASONING_TOKENS, LLM_REASONING_TOKEN_ATTR_KEYS, OPENINFERENCE_SPAN_KIND, OUTPUT_VALUE, RUN_COST_ATTR_KEYS, SPAN_KIND_ATTR_KEYS, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS, applyLlmSpanOtlpAttributes, asNumber, contextInputTokens, firstNumberAttr } from "./trace-attributes.js";
-import { C as TraceNotFoundError, G as resolveTraceAnalystLimits, J as extractOtlpAttributes, K as asString, Q as readOtlpStatus, S as TraceFileTooLargeError, W as DEFAULT_TRACE_ANALYST_LIMITS, X as inferOtlpKind, Y as firstStringAttr, Z as projectOtlpFlatLine, _ as TraceAnalysisLimitError, b as TraceFileMalformedError, c as traceAnalystFunctionGroup, et as stringField, g as SpanNotFoundError, h as TRACE_ANALYSIS_LIMITS, i as runTraceAnalyst, it as traceSpanKindToOpenInferenceKind, l as createBoundedTraceAnalysisStore, m as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, n as renderPriorFindings, nt as classifyOtlpSpanRole, o as TRACE_ANALYST_TOOL_NAMESPACE, p as DEFAULT_TRACE_ANALYST_BUDGETS, r as renderUpstreamFindings, rt as isOtlpModelCall, s as buildTraceAnalysisToolDescriptors, t as createTraceAnalyst, tt as applyToolSpanOtlpAttributes, v as TraceAnalysisStoreContractError, x as TraceFileMissingError, y as TraceAnalysisValidationError } from "./kind-factory-Bvwe3pup.js";
-import { a as computeFindingId, o as makeFinding, s as makeProposalFinding } from "./usage-receipt-CgxMEBZq.js";
+import { C as TraceNotFoundError, G as resolveTraceAnalystLimits, J as extractOtlpAttributes, K as asString, Q as readOtlpStatus, S as TraceFileTooLargeError, W as DEFAULT_TRACE_ANALYST_LIMITS, X as inferOtlpKind, Y as firstStringAttr, Z as projectOtlpFlatLine, _ as TraceAnalysisLimitError, b as TraceFileMalformedError, c as traceAnalystFunctionGroup, et as stringField, g as SpanNotFoundError, h as TRACE_ANALYSIS_LIMITS, i as runTraceAnalyst, it as traceSpanKindToOpenInferenceKind, l as createBoundedTraceAnalysisStore, m as TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, n as renderPriorFindings, nt as classifyOtlpSpanRole, o as TRACE_ANALYST_TOOL_NAMESPACE, p as DEFAULT_TRACE_ANALYST_BUDGETS, r as renderUpstreamFindings, rt as isOtlpModelCall, s as buildTraceAnalysisToolDescriptors, t as createTraceAnalyst, tt as applyToolSpanOtlpAttributes, v as TraceAnalysisStoreContractError, x as TraceFileMissingError, y as TraceAnalysisValidationError } from "./kind-factory-BHIgPmzS.js";
+import { a as computeFindingId, o as makeFinding, s as makeProposalFinding } from "./usage-receipt-EVI8B8Xu.js";
 import { c as SUPERVISOR_RUN_INTEGRITY_SCHEMA, n as supervisorRunRolloutLines, t as analyzeSupervisorRunIntegrity } from "./integrity-DY6tIbl0.js";
 import { c as assertMintedLines, f as isRolloutLine, i as ROLLOUT_SCHEMA, m as validateRolloutLine, p as isTrainableSplit, s as assertMinted, u as assertRolloutLine } from "./schema-C6DW4ZHR.js";
 import { a as scoreOrigin, n as observedScore, o as trainingReward, r as observedSplitScore, s as trainingScore, t as isRealnessGated } from "./reward-nw2xZGZG.js";
 import { a as clamp01, i as aggregateRunScore, r as DEFAULT_RUN_SCORE_WEIGHTS } from "./proposal-findings-2GIUo1et.js";
-import { a as RunCritic, d as defaultIsMaterial, f as diffFindings, h as defineTraceAnalyst, i as runSemanticConceptJudge, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, p as LockedJsonlAppender, r as createSemanticConceptJudge, s as SkillUsageAnalyst, t as DEFAULT_COMPLEXITY_WEIGHTS, u as FindingsStore } from "./semantic-concept-judge-Do5aM9wP.js";
+import { a as RunCritic, d as defaultIsMaterial, f as diffFindings, h as defineTraceAnalyst, i as runSemanticConceptJudge, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, p as LockedJsonlAppender, r as createSemanticConceptJudge, s as SkillUsageAnalyst, t as DEFAULT_COMPLEXITY_WEIGHTS, u as FindingsStore } from "./semantic-concept-judge-DKRtp2sY.js";
 import { _ as cachedJudge, b as fileVerdictCache, v as canonicalJson, x as inMemoryVerdictCache, y as contentHash } from "./single-run-lock-D5iN0Xzb.js";
-import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-IRCG8kdi.js";
+import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-19FQEMBK.js";
 import { f as Mutex, p as mapConcurrent } from "./ledger-core-DXZIqu17.js";
 import { At as summarizeAgentReceiptIntegrity, B as surfaceContentHash, Ct as decidePairedPromotion, D as runCanaries, Dt as BackendIntegrityError, Et as pairedDeltaTest, G as DEFAULT_MUTATION_PRIMITIVES, K as buildReflectionPrompt, Mt as JudgeParseError, Ot as assertRealAgentReceipts, Tt as minimumPairsForPairedDeltaTest, _t as crowdingDistance, at as REFERENCE_EQUIVALENCE_JUDGE_VERSION, bt as paretoFrontierWithCrowding, ct as DEFAULT_RED_TEAM_CORPUS, dt as scoreRedTeamOutput, ft as toolNamesForRun, gt as llmJudge, ht as hashScenarios, it as REFERENCE_EQUIVALENCE_INPUT_LIMITS, jt as summarizeBackendIntegrity, kt as assertRealBackend, lt as redTeamDataset, mt as HoldoutLockedError, ot as createReferenceEquivalenceJudge, pt as Dataset, q as parseReflectionResponse, st as runReferenceEquivalenceJudge, ut as redTeamReport, vt as dominates, wt as pairedDecisionShape, xt as scalarScore, yt as paretoFrontier } from "./skillopt-optimization-method-C4FX42dy.js";
 import { $ as selfPreference, A as pairedRiskDifference, B as requiredSampleSize, C as mulberry32, D as pairedCohensDz, E as pairedBootstrap, F as partialCredit, G as wilson, H as weightedComposite, I as passAtK, J as normalCdf, K as studentTCdf, L as pearsonR, M as pairedRiskDifferenceScore, N as pairedSignTest, O as pairedDeltaTieFraction, P as pairedTTest, Q as positionalBias, R as ranks, S as mcnemarRequiredN, T as pairedBinaryScale, U as weightedMean, V as spearmanR, W as wilcoxonSignedRank, X as calibrateJudgeContinuous, Y as calibrateJudge, Z as continuousAgreement, _ as interpretCliffs, a as MANN_WHITNEY_EXACT_MAX_WORK, b as mcnemar, c as bonferroni, d as confidenceInterval, et as verbosityBias, f as corpusInterRaterAgreement, g as interRaterReliability, h as holm, i as MANN_WHITNEY_EXACT_MAX_STATES, j as pairedRiskDifferenceExact, k as pairedMde, l as cliffsDelta, m as eProcess, n as DECISION_PAIRED_DELTA_STATISTIC, o as WILCOXON_EXACT_MAX_N, p as corpusInterRaterAgreementFromJudgeScores, r as DEFAULT_PERMUTATIONS, s as benjaminiHochberg, t as BOOTSTRAP_GATE_MIN_N, u as cohensD, v as isBinaryOutcomeVector, w as normalizeScores, x as mcnemarPower, y as mannWhitneyU, z as requiredPairedSampleSize } from "./statistics-ByxzSiOM.js";
@@ -30,10 +30,10 @@ import { a as toHarborTrajectories, i as relabelImportedSplit, n as HARBOR_IMPOR
 import { t as buildTrajectory } from "./trajectory-D_7rLrvE.js";
 import { n as unmintableReasons, t as mintRolloutRows } from "./mint-DD-0oQTA.js";
 import { C as rollupSupervisorRuns, D as isUnavailable, E as SUPERVISOR_RUN_SCHEMA, O as showMeasured, b as readClaudeCodeSupervisorRun, c as writeSupervisorRunReport, d as readRuntimeSupervisorRun, f as runtimeSupervisorRunReader, h as renderSupervisorRunMarkdown, m as renderSupervisorRunHeadline, t as analyzeSupervisorRun, u as isRuntimeSupervisorRunDir, x as analyzeSupervisorRunSources, y as claudeCodeSupervisorRunReader } from "./supervisor-run-DiyQVczd.js";
-import { A as TRACE_ANALYST_ACTOR_DESCRIPTION, C as domainEvidencePattern, D as tokenizeDomainWords, E as scoreTraceInsightReadiness, O as traceAnalystOnRunComplete, S as describeTraceInsightScope, T as planTraceInsightQuestions, _ as otlpToTraceRunRecords, a as convertTraceStoresToOtlp, b as buildTraceInsightPrompt, c as otelRunCompleteHook, d as captureFetchToRawSink, f as ToolTraceMissingError, g as otlpToRunRecords, h as otlpRowsToTraceRunRecords, i as iterateRawCalls, j as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, k as analyzeTraces, l as OTEL_AGENT_EVAL_SCOPE, m as otlpRowsToRunRecords, n as ReplayCacheMissError, o as createOtelExporter, p as toolSpansToTraceAnalysisStore, r as createReplayFetch, s as createOtelTracingStore, t as ReplayCache, u as exportRunAsOtlp, v as flattenOtlpExportToNdjson, w as inferDomainKeywords, x as defaultTraceInsightPanel, y as buildTraceInsightContext } from "./replay-CqOsGjzU.js";
-import { i as otlpTextToTraceAnalysisStore, n as OtlpFileTraceStore } from "./store-otlp-D4I90_vR.js";
-import { n as extractUsageFromResponse, r as extractUsageFromSse, t as extractUsage } from "./extract-usage-7l1Xq5ti.js";
-import { B as agentProfileModelId, G as createLlmCorrectnessChecker, H as harnessAxisOf, I as CODING_HARNESSES, J as verifyCompletion, K as createTokenRecallChecker, L as HARNESS_NATIVE_MODEL, R as agentProfileHash, U as extractProducedState, V as expandProfileAxes, W as completionVerdict, q as parseCorrectnessResponse, z as agentProfileId } from "./campaign-DjGFyPxH.js";
+import { A as TRACE_ANALYST_ACTOR_DESCRIPTION, C as domainEvidencePattern, D as tokenizeDomainWords, E as scoreTraceInsightReadiness, O as traceAnalystOnRunComplete, S as describeTraceInsightScope, T as planTraceInsightQuestions, _ as otlpToTraceRunRecords, a as convertTraceStoresToOtlp, b as buildTraceInsightPrompt, c as otelRunCompleteHook, d as captureFetchToRawSink, f as ToolTraceMissingError, g as otlpToRunRecords, h as otlpRowsToTraceRunRecords, i as iterateRawCalls, j as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, k as analyzeTraces, l as OTEL_AGENT_EVAL_SCOPE, m as otlpRowsToRunRecords, n as ReplayCacheMissError, o as createOtelExporter, p as toolSpansToTraceAnalysisStore, r as createReplayFetch, s as createOtelTracingStore, t as ReplayCache, u as exportRunAsOtlp, v as flattenOtlpExportToNdjson, w as inferDomainKeywords, x as defaultTraceInsightPanel, y as buildTraceInsightContext } from "./replay-GW61ezMW.js";
+import { i as otlpTextToTraceAnalysisStore, n as OtlpFileTraceStore } from "./store-otlp-CKtTpRhv.js";
+import { n as extractUsageFromResponse, r as extractUsageFromSse, t as extractUsage } from "./extract-usage-BW27f3XW.js";
+import { B as agentProfileModelId, G as createLlmCorrectnessChecker, H as harnessAxisOf, I as CODING_HARNESSES, J as verifyCompletion, K as createTokenRecallChecker, L as HARNESS_NATIVE_MODEL, R as agentProfileHash, U as extractProducedState, V as expandProfileAxes, W as completionVerdict, q as parseCorrectnessResponse, z as agentProfileId } from "./campaign-ClpnD7Ug.js";
 import { n as iqr, r as welchsTTest, t as compareToBaseline } from "./baseline-C-GocmIW.js";
 import { a as judgeSpans, c as runsForScenario, i as hasCapturedToolArgs, l as toolSpans, n as argHash, o as llmSpans, r as groupBy, s as runFailureClass, t as aggregateLlm } from "./query-Di7eEQ79.js";
 import { a as checkBehavioralCanary, i as canaryLeakView, o as checkCanaries, r as HoldoutAuditor, s as runBehavioralCanaries, t as analyzeRuns } from "./analyze-runs-DWIvOAGk.js";
@@ -43,7 +43,7 @@ import { a as InMemoryTraceStore, i as FileSystemTraceStore, n as assertRunCaptu
 import { i as redactValue, n as REDACTION_VERSION, r as redactString, t as DEFAULT_REDACTION_RULES } from "./redact-7Aq1ukl-.js";
 import { n as DEFAULT_RULES, r as classifyFailure, t as computeToolUseMetrics } from "./tool-use-metrics-DEGMKycK.js";
 import { t as analyzeSeries } from "./series-convergence-CjO2QdRW.js";
-import { _ as deterministicSplit, g as BENCHMARK_SPLIT_SEED, t as benchmarks_exports } from "./benchmarks-CkG1bWFa.js";
+import { _ as deterministicSplit, g as BENCHMARK_SPLIT_SEED, t as benchmarks_exports } from "./benchmarks-CWbsj0t4.js";
 import { t as runEvalCampaign } from "./eval-campaign-lZcDIwQM.js";
 import { n as pairedEvalueSequence, t as evaluateInterimReleaseConfidence } from "./sequential-Br0mAPHA.js";
 import { accessSync, appendFileSync, constants, cpSync, existsSync, mkdirSync, promises, readFileSync, readdirSync, statSync, writeFileSync } from "node:fs";

package/dist/{kind-factory-Bvwe3pup.js → kind-factory-BHIgPmzS.js}

@@ -1,7 +1,7 @@
 import { a as LimitExceededError, c as ValidationError, n as CaptureIntegrityError, o as NotFoundError, t as AgentEvalError } from "./errors-D-LKuDhb.js";
 import { i as CostLedger } from "./cost-ledger-DMFxsLKr.js";
 import { INPUT_VALUE, LLM_MODEL_ATTR_KEYS, OUTPUT_VALUE, SPAN_KIND_ATTR_KEYS, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS } from "./trace-attributes.js";
-import { i as validateUsageSettlementTimeout, n as settleUsageReceiptFromCostLedger, o as makeFinding } from "./usage-receipt-CgxMEBZq.js";
+import { i as validateUsageSettlementTimeout, n as settleUsageReceiptFromCostLedger, o as makeFinding } from "./usage-receipt-EVI8B8Xu.js";
 import { i as hashCanonical, r as canonicalString } from "./canonical-D011XM8r.js";
 import { z } from "zod";
 import { RE2JS } from "re2js";
@@ -2140,4 +2140,4 @@ function truncateForContext(value, max) {
 //#endregion
 export { spanEpochMillis as $, coerceJson as A, renderFindingSubject as B, TraceNotFoundError as C, RawAnalystFindingSchema as D, RawAnalystEvidenceSchema as E, FINDING_SUBJECT_SYNTAX as F, resolveTraceAnalystLimits as G, snapshotExactExecutionPlan as H, FindingSubjectStringSchema as I, extractOtlpAttributes as J, asString as K, KIND_EXPECTED_SUBJECTS as L, stripCodeFences as M, FINDING_SUBJECT_GRAMMAR_PROMPT as N, evidenceRefsFromRawFinding as O, FINDING_SUBJECT_KINDS as P, readOtlpStatus as Q, findingSubjectGrammarPromptFor as R, TraceFileTooLargeError as S, RAW_FINDING_SCHEMA_PROMPT as T, deepFreezeCanonicalJson as U, snapshotExactExecutionComponentIdentity as V, DEFAULT_TRACE_ANALYST_LIMITS as W, inferOtlpKind as X, firstStringAttr as Y, projectOtlpFlatLine as Z, TraceAnalysisLimitError as _, buildTraceToolsForGroup as a, TraceFileMalformedError as b, traceAnalystFunctionGroup as c, truncateForBudget as d, stringField as et, validateInteger as f, SpanNotFoundError as g, TRACE_ANALYSIS_LIMITS as h, runTraceAnalyst as i, traceSpanKindToOpenInferenceKind as it, coerceToFindingRows as j, parseRawFinding as k, createBoundedTraceAnalysisStore as l, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX as m, renderPriorFindings as n, classifyOtlpSpanRole as nt, TRACE_ANALYST_TOOL_NAMESPACE as o, DEFAULT_TRACE_ANALYST_BUDGETS as p, compareSpanTime as q, renderUpstreamFindings as r, isOtlpModelCall as rt, buildTraceAnalysisToolDescriptors as s, createTraceAnalyst as t, applyToolSpanOtlpAttributes as tt, compileSearchRegex as u, TraceAnalysisStoreContractError as v, ANALYST_SEVERITIES as w, TraceFileMissingError as x, TraceAnalysisValidationError as y, parseFindingSubject as z };
 
-//# sourceMappingURL=kind-factory-Bvwe3pup.js.map
+//# sourceMappingURL=kind-factory-BHIgPmzS.js.map