@tangle-network/agent-eval 0.126.0 → 0.126.2

package/CHANGELOG.md

@@ -6,6 +6,32 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ## [Unreleased]
 
+## [0.126.2] - 2026-07-24 - fail-closed candidate ranking
+
+### Added
+
+- `runOptimization()` and `selfImprove()` accept a fixed-length `selectionRankKey` so domain-specific reliability metrics can choose candidates during every generation.
+
+### Fixed
+
+- Candidate rank keys must be non-empty, fixed-length, and finite.
+- A candidate must strictly beat the incumbent on the configured rank key before it can become the next parent or final winner.
+- Method-reported spend above `costCeiling` is rejected before final scoring.
+
+## [0.126.1] - 2026-07-24 - optimizer lifecycle integrity
+
+### Changed
+
+- `compareOptimizationMethods()` uses one caller-visible spend limit across optimizer models, train and selection evaluation, and final scoring.
+- Candidate surfaces are detached before every optimizer and scoring callback.
+- SkillOpt receives complete non-secret model settings automatically while provider credentials remain in the local Node proxy.
+
+### Fixed
+
+- Aborting a comparison now stops official optimizer subprocess groups, callback work, and model proxy requests.
+- Resumed final scoring retains the original cost identity and reports cumulative spend instead of resetting cached calls to zero.
+- Model-backed GEPA fails before search when the full upstream dependency set is missing instead of silently returning the baseline after swallowed reflection errors.
+
 ## [0.126.0] - 2026-07-24 - official optimizer engines
 
 ### Added

package/dist/analyst/index.js

@@ -9,7 +9,7 @@ import {
   diffFindings,
   emitSkillUsageFindings,
   runSemanticConceptJudge
-} from "../chunk-4B7ZZHPX.js";
+} from "../chunk-KE2VWPZX.js";
 import {
   ANALYST_SEVERITIES,
   AnalystRegistry,
@@ -50,9 +50,9 @@ import {
   stripCodeFences,
   structureFindings,
   validateUsageSettlementTimeout
-} from "../chunk-CM4OILD2.js";
+} from "../chunk-LUNF2SEL.js";
 import "../chunk-HHWE3POT.js";
-import "../chunk-UI4YMIN2.js";
+import "../chunk-WGXIEX7P.js";
 import "../chunk-J5SQWP6Y.js";
 import {
   CostLedger

package/dist/benchmarks/index.js

@@ -17,10 +17,10 @@ import {
   runBenchmarkAdapter,
   summarizeBenchmarkCampaign
 } from "../chunk-W4L6C2XT.js";
-import "../chunk-KO2PZOGP.js";
-import "../chunk-NTOV7RU5.js";
+import "../chunk-Y5CLI4PY.js";
+import "../chunk-7AN2E7BU.js";
 import "../chunk-UCLVDLCH.js";
-import "../chunk-UI4YMIN2.js";
+import "../chunk-WGXIEX7P.js";
 import "../chunk-ARU2PZFM.js";
 import "../chunk-J5SQWP6Y.js";
 import "../chunk-PJQFMIOX.js";

package/dist/campaign/index.d.ts

@@ -2479,8 +2479,8 @@ declare function planCampaignRun<TScenario extends Scenario, TArtifact>(opts: Pl
  * data and compared with paired confidence intervals.
  */
 
-/** Per-method campaign settings. Each method receives its own spend account. */
-type OptimizationMethodRunOptions<TScenario extends Scenario, TArtifact> = Omit<RunCampaignOptions<TScenario, TArtifact>, 'costLedger' | 'dispatch' | 'judges' | 'runDir' | 'scenarios' | 'seed'>;
+/** Shared campaign settings applied to every optimization method. */
+type OptimizationMethodRunOptions<TScenario extends Scenario, TArtifact> = Omit<RunCampaignOptions<TScenario, TArtifact>, 'costCeiling' | 'costLedger' | 'dispatch' | 'judges' | 'runDir' | 'scenarios' | 'seed'>;
 /** Cost reported by a method or by final test scoring. */
 interface ComparisonCost {
     totalCostUsd: number;
@@ -2553,7 +2553,7 @@ interface OptimizationMethodInput<TScenario extends Scenario, TArtifact> {
     readonly seed: number;
     /** Shared defaults for every method. A method may override them explicitly. */
     readonly runOptions: Readonly<OptimizationMethodRunOptions<TScenario, TArtifact>>;
-    /** Durable spend account shared by the method's model and evaluation calls. */
+    /** Durable spend account shared by every method and final scoring. */
     readonly costLedger: CostLedgerHandle;
 }
 interface OptimizationMethodResult {
@@ -2662,8 +2662,7 @@ interface CompareOptimizationMethodsOptions<TScenario extends Scenario, TArtifac
     /** Simultaneous confidence across method-vs-baseline and method-vs-method contrasts.
      *  Each bootstrap interval is Bonferroni-adjusted. Default 0.95. */
     confidence?: number;
-    /** Shared spend limit across baseline and winner scoring on the final test partition.
-     *  Each method owns its optimization budget through `optimizationRunOptions.costCeiling`. */
+    /** Shared spend limit across every method's optimizer and evaluation calls plus final scoring. */
     costCeiling?: number;
 }
 /**
@@ -4458,6 +4457,24 @@ interface RunOptimizationBaseOptions<TScenario extends Scenario, TArtifact> exte
         costLedger?: CostLedgerHandle;
         costPhase?: string;
     }) => Promise<unknown[]>;
+    /**
+     * Optional override for how the WINNER is selected among coverage-complete
+     * candidates (and how the incumbent bar is set). Returns a lexicographic rank
+     * key — each element higher-is-better; candidates are ranked by descending key
+     * (`compareRankKeys`) and the top must STRICTLY beat the incumbent's key to
+     * promote. Defaults to `[campaignMeanComposite(campaign)]`, i.e. the historical
+     * scalar-mean ranking (single-element key ⇒ identical behavior).
+     *
+     * A binary-with-replicates consumer (e.g. swe-arena, whose ship-gate counts an
+     * instance resolved only when EVERY replicate resolved) passes a fail-closed
+     * key built from the SAME reduction its gate uses, so winner-selection and the
+     * ship-gate rank on the identical metric and can never invert — the selector
+     * cannot promote a flaky per-cell-mean candidate the gate would reject over a
+     * fail-closed candidate the gate would accept. Only the winner CHOICE changes;
+     * the descriptive `composite` (mean) on every record and the Pareto objective
+     * vectors are untouched, so proposer diversity and reporting are unaffected.
+     */
+    selectionRankKey?: (campaign: CampaignResult<TArtifact, TScenario>) => number[];
 }
 type RunOptimizationOptions<TScenario extends Scenario, TArtifact> = RunOptimizationBaseOptions<TScenario, TArtifact>;
 interface RunOptimizationResult<TArtifact, TScenario extends Scenario> {
@@ -5488,6 +5505,10 @@ declare function selectDiscriminative(signals: ScenarioSignal[], k: number, opts
  *  descriptive aggregate with NaN. Cells with no valid scores are skipped.
  *  Empty ⇒ 0. */
 declare function campaignMeanComposite<TArtifact, TScenario extends Scenario>(campaign: CampaignResult<TArtifact, TScenario>): number;
+/** Compare fixed-length lexicographic rank keys where each element is higher-is-better.
+ *  Returns a positive number when `a` ranks above `b`, negative when below, and
+ *  zero when equal. */
+declare function compareRankKeys(a: readonly number[], b: readonly number[]): number;
 interface CampaignBreakdown {
     /** Mean score per judge dimension across all cells. */
     dimensions: Record<string, number>;
@@ -6055,4 +6076,4 @@ declare function verifyCodeSurface(surface: CodeSurface, worktreeDir?: string):
  *  identity against the checkout at `worktreeRef`. */
 declare function resolveWorktreePath(surface: CodeSurface, worktreeDir?: string): string;
 
-export { type AnalystArtifact, type AnalystScenario, type AnalyzeCrossSurfaceInteractionsInput, type AxisEvidence, type AxisVerdict, type BuildAnalystSurfaceDispatchOptions, type BuildEvidenceVectorOptions, type BuildLoopProvenanceArgs, type CampaignAggregates, type CampaignArtifactWriter, type CampaignBreakdown, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignRunPlan, type CampaignRunPlanCell, type CampaignScenarioIdentity, type CampaignStorage, type CampaignTokenUsage, type CampaignTraceWriter, type CodeSurface, type CodeSurfaceVerification, type CompareOptimizationMethodsOptions, type ComparisonCost, type ComponentSurface, type CostLedgerHandle, type CrossSurfaceAdditionDecision, type CrossSurfaceAdditionRejectionReason, type CrossSurfaceAttemptCompleteness, type CrossSurfaceBestSingleSelection, type CrossSurfaceBootstrapPolicy, type CrossSurfaceCandidate, type CrossSurfaceCandidateComparison, type CrossSurfaceCandidateEvidence, type CrossSurfaceCandidateOutcome, type CrossSurfaceCandidateSummary, type CrossSurfaceComponent, type CrossSurfaceComponentEvidence, type CrossSurfaceCompositionStep, type CrossSurfaceDistribution, type CrossSurfaceEligibility, type CrossSurfaceEvidenceBreakdown, type CrossSurfaceIneligibilityReason, type CrossSurfaceInteractionAwareSelection, type CrossSurfaceInteractionEffect, type CrossSurfaceInteractionPath, type CrossSurfaceInteractionReport, type CrossSurfaceInteractionTask, type CrossSurfaceNaiveStackSelection, type CrossSurfacePairCompatibility, type CrossSurfacePairEvidence, type CrossSurfacePairIncompatibilityReason, type CrossSurfacePairwiseEntry, type CrossSurfaceRankedSingle, type CrossSurfaceRelativeCost, type CrossSurfaceSelectionPolicy, type CrossSurfaceSelections, type CrossSurfaceTaskRow, type DefaultProductionGateOptions, type DimensionRegression, type DiscriminationScore, type DispatchContext, type DispatchFn, type EmitLoopProvenanceArgs, type EmitLoopProvenanceResult, type EvalFixture, type EvalFixtureFile, type EvalFixtureLoadOptions, type EvalFixtureRunPlan, type EvalFixtureScenario, type EvalFixtureValidationMode, type EvidenceVector, type ExternalOptimizationExample, type ExternalTextEvaluationResponse, type ExternalTextOptimizationMethodConfig, type ExternalTextOptimizerContext, type ExternalTextOptimizerResult, type FailureModeRecallJudgeOptions, FileSearchLedger, FsLabeledScenarioStore, type FsLabeledScenarioStoreOptions, type Gate, type GateContext, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaAdaptiveEngineRun, type GepaEngineOptions, type GepaEngineRun, type GepaOptimizationMethodConfig, type GepaOptimizationRecipe, type GepaRunnerCommand, type GitWorktreeAdapterOptions, type HeldOutGateOptions, type HeldoutSignificance, type HeldoutSignificanceOptions, type JudgeAggregate, type JudgeConfig, type JudgeDimension, type JudgeScore, type LabelTrust, type LabeledScenarioRecord, type LabeledScenarioSampleArgs, type LabeledScenarioSource, type LabeledScenarioStore, LabeledScenarioStoreError, type LabeledScenarioWrite, type LlmJudgeDimension, type LlmJudgeOptions, type LoadEvalFixtureScenariosOptions, type LoopProvenanceArgsFromResult, type LoopProvenanceBackend, type LoopProvenanceCandidate, type LoopProvenanceEvidence, type LoopProvenanceOptimizationMethod, type LoopProvenanceRecord, type MutableSurface, type NeutralizationGateOptions, type ObjectiveSource, type OpenAICompatibleOptimizerModel, type OpenAutoPrOptions, type OpenAutoPrResult, type OpenSearchLedgerOptions, type OptimizationMethod, type OptimizationMethodComparison, type OptimizationMethodInput, type OptimizationMethodPairwise, type OptimizationMethodProvenance, type OptimizationMethodResult, type OptimizationMethodRunOptions, type OptimizationMethodScore, type OptimizationPackageSource, type OptimizationProposer, type OptimizationTokenUsage, type OptimizerConfig, type OptimizerModelBudget, type PairedHoldout, type ParetoParent, type ParetoSignificanceGateOptions, type PendingCostCallView, type PlanCampaignRunOptions, type PlanEvalFixtureRunOptions, type PlaybackContext, type PlaybackDriver, type PlaybackStep, type PowerPreflight, type PowerPreflightOptions, type PremeasuredOptimizationBaseline, type ProfileDispatchFn, ProfileMatrixError, type ProfileSummary, type PromotionObjective, type PromotionPolicy, type ProposalTrackContext, type ProposeContext, type ProposedCandidate, type RedactionStatus, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceScenario, type RolloutArgumentDiff, type RolloutArgumentDiffOptions, type RolloutCall, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunOptimizationOptions, type RunOptimizationResult, type RunProfileMatrixOptions, type RunProfileMatrixResult, SEARCH_LEDGER_SCHEMA, type Scenario, type ScenarioAggregate, type ScenarioRollup, type ScenarioSignal, type ScoreboardRenderOptions, type ScoreboardRow, type ScoreboardSummary, type ScoredRollout, type ScoredSurfaceOutcome, type SearchAccountingAudit, type SearchArtifactRef, type SearchAttemptAccounting, type SearchCandidateDecidedEvent, type SearchCandidateLineage, type SearchCandidateRegisteredEvent, type SearchCandidateSlot, type SearchCandidateSlotClosedEvent, type SearchCandidateSurface, type SearchCompletedEvent, type SearchCostAccounting, type SearchFailureReason, type SearchLedger, type SearchLedgerAppendResult, SearchLedgerConflictError, type SearchLedgerEntry, SearchLedgerError, type SearchLedgerEvent, type SearchLedgerHash, SearchLedgerIntegrityError, type SearchLedgerReplay, type SearchModelIdentity, type SearchOperationKind, type SearchOperationRecordedEvent, type SearchPlan, type SearchPlannedEvent, type SearchPlannedOperation, type SearchPlannedTask, type SearchSourceRef, type SearchSurfaceEffect, type SearchSurfaceEvidence, type SearchSurfaceKind, type SearchTaskAttemptedEvent, type SearchTaskOutcome, type SearchTokenAccounting, type SequentialDecideFn, type SequentialDecideOptions, type SequentialDecision, type SequentialObservation, type SequentialPairedGate, type SequentialPairedGateOptions, type SessionScript, type SingleRunLock, type SingleRunLockOptions, type SkillOptOptimizationMethodConfig, type SkillOptRunnerCommand, type SkillOptTrainerConfig, type SurfaceProposer, type TraceSpan, type TransientFailureOptions, type UngroundedLiteralReport, type UserStory, type UserStoryVerdict, type Worktree, type WorktreeAdapter, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, buildAnalystSurfaceDispatch, buildEvidenceVector, buildLoopProvenanceRecord, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, compareOptimizationMethods, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createReferenceEquivalenceJudge, createRunCostLedger, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, failureModeRecallJudge, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, renderScoreboardMarkdown, renderSurfaceDiff, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };
+export { type AnalystArtifact, type AnalystScenario, type AnalyzeCrossSurfaceInteractionsInput, type AxisEvidence, type AxisVerdict, type BuildAnalystSurfaceDispatchOptions, type BuildEvidenceVectorOptions, type BuildLoopProvenanceArgs, type CampaignAggregates, type CampaignArtifactWriter, type CampaignBreakdown, type CampaignCellResult, type CampaignCostMeter, type CampaignResult, type CampaignRunPlan, type CampaignRunPlanCell, type CampaignScenarioIdentity, type CampaignStorage, type CampaignTokenUsage, type CampaignTraceWriter, type CodeSurface, type CodeSurfaceVerification, type CompareOptimizationMethodsOptions, type ComparisonCost, type ComponentSurface, type CostLedgerHandle, type CrossSurfaceAdditionDecision, type CrossSurfaceAdditionRejectionReason, type CrossSurfaceAttemptCompleteness, type CrossSurfaceBestSingleSelection, type CrossSurfaceBootstrapPolicy, type CrossSurfaceCandidate, type CrossSurfaceCandidateComparison, type CrossSurfaceCandidateEvidence, type CrossSurfaceCandidateOutcome, type CrossSurfaceCandidateSummary, type CrossSurfaceComponent, type CrossSurfaceComponentEvidence, type CrossSurfaceCompositionStep, type CrossSurfaceDistribution, type CrossSurfaceEligibility, type CrossSurfaceEvidenceBreakdown, type CrossSurfaceIneligibilityReason, type CrossSurfaceInteractionAwareSelection, type CrossSurfaceInteractionEffect, type CrossSurfaceInteractionPath, type CrossSurfaceInteractionReport, type CrossSurfaceInteractionTask, type CrossSurfaceNaiveStackSelection, type CrossSurfacePairCompatibility, type CrossSurfacePairEvidence, type CrossSurfacePairIncompatibilityReason, type CrossSurfacePairwiseEntry, type CrossSurfaceRankedSingle, type CrossSurfaceRelativeCost, type CrossSurfaceSelectionPolicy, type CrossSurfaceSelections, type CrossSurfaceTaskRow, type DefaultProductionGateOptions, type DimensionRegression, type DiscriminationScore, type DispatchContext, type DispatchFn, type EmitLoopProvenanceArgs, type EmitLoopProvenanceResult, type EvalFixture, type EvalFixtureFile, type EvalFixtureLoadOptions, type EvalFixtureRunPlan, type EvalFixtureScenario, type EvalFixtureValidationMode, type EvidenceVector, type ExternalOptimizationExample, type ExternalTextEvaluationResponse, type ExternalTextOptimizationMethodConfig, type ExternalTextOptimizerContext, type ExternalTextOptimizerResult, type FailureModeRecallJudgeOptions, FileSearchLedger, FsLabeledScenarioStore, type FsLabeledScenarioStoreOptions, type Gate, type GateContext, type GateDecision, type GateResult, type GenerationCandidate, type GenerationRecord, type GepaAdaptiveEngineRun, type GepaEngineOptions, type GepaEngineRun, type GepaOptimizationMethodConfig, type GepaOptimizationRecipe, type GepaRunnerCommand, type GitWorktreeAdapterOptions, type HeldOutGateOptions, type HeldoutSignificance, type HeldoutSignificanceOptions, type JudgeAggregate, type JudgeConfig, type JudgeDimension, type JudgeScore, type LabelTrust, type LabeledScenarioRecord, type LabeledScenarioSampleArgs, type LabeledScenarioSource, type LabeledScenarioStore, LabeledScenarioStoreError, type LabeledScenarioWrite, type LlmJudgeDimension, type LlmJudgeOptions, type LoadEvalFixtureScenariosOptions, type LoopProvenanceArgsFromResult, type LoopProvenanceBackend, type LoopProvenanceCandidate, type LoopProvenanceEvidence, type LoopProvenanceOptimizationMethod, type LoopProvenanceRecord, type MutableSurface, type NeutralizationGateOptions, type ObjectiveSource, type OpenAICompatibleOptimizerModel, type OpenAutoPrOptions, type OpenAutoPrResult, type OpenSearchLedgerOptions, type OptimizationMethod, type OptimizationMethodComparison, type OptimizationMethodInput, type OptimizationMethodPairwise, type OptimizationMethodProvenance, type OptimizationMethodResult, type OptimizationMethodRunOptions, type OptimizationMethodScore, type OptimizationPackageSource, type OptimizationProposer, type OptimizationTokenUsage, type OptimizerConfig, type OptimizerModelBudget, type PairedHoldout, type ParetoParent, type ParetoSignificanceGateOptions, type PendingCostCallView, type PlanCampaignRunOptions, type PlanEvalFixtureRunOptions, type PlaybackContext, type PlaybackDriver, type PlaybackStep, type PowerPreflight, type PowerPreflightOptions, type PremeasuredOptimizationBaseline, type ProfileDispatchFn, ProfileMatrixError, type ProfileSummary, type PromotionObjective, type PromotionPolicy, type ProposalTrackContext, type ProposeContext, type ProposedCandidate, type RedactionStatus, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceScenario, type RolloutArgumentDiff, type RolloutArgumentDiffOptions, type RolloutCall, type RunCampaignOptions, type RunEvalOptions, type RunImprovementLoopOptions, type RunImprovementLoopResult, type RunOptimizationOptions, type RunOptimizationResult, type RunProfileMatrixOptions, type RunProfileMatrixResult, SEARCH_LEDGER_SCHEMA, type Scenario, type ScenarioAggregate, type ScenarioRollup, type ScenarioSignal, type ScoreboardRenderOptions, type ScoreboardRow, type ScoreboardSummary, type ScoredRollout, type ScoredSurfaceOutcome, type SearchAccountingAudit, type SearchArtifactRef, type SearchAttemptAccounting, type SearchCandidateDecidedEvent, type SearchCandidateLineage, type SearchCandidateRegisteredEvent, type SearchCandidateSlot, type SearchCandidateSlotClosedEvent, type SearchCandidateSurface, type SearchCompletedEvent, type SearchCostAccounting, type SearchFailureReason, type SearchLedger, type SearchLedgerAppendResult, SearchLedgerConflictError, type SearchLedgerEntry, SearchLedgerError, type SearchLedgerEvent, type SearchLedgerHash, SearchLedgerIntegrityError, type SearchLedgerReplay, type SearchModelIdentity, type SearchOperationKind, type SearchOperationRecordedEvent, type SearchPlan, type SearchPlannedEvent, type SearchPlannedOperation, type SearchPlannedTask, type SearchSourceRef, type SearchSurfaceEffect, type SearchSurfaceEvidence, type SearchSurfaceKind, type SearchTaskAttemptedEvent, type SearchTaskOutcome, type SearchTokenAccounting, type SequentialDecideFn, type SequentialDecideOptions, type SequentialDecision, type SequentialObservation, type SequentialPairedGate, type SequentialPairedGateOptions, type SessionScript, type SingleRunLock, type SingleRunLockOptions, type SkillOptOptimizationMethodConfig, type SkillOptRunnerCommand, type SkillOptTrainerConfig, type SurfaceProposer, type TraceSpan, type TransientFailureOptions, type UngroundedLiteralReport, type UserStory, type UserStoryVerdict, type Worktree, type WorktreeAdapter, WorktreeAdapterError, acquireSingleRunLock, analyzeCrossSurfaceInteractions, assertCampaignDesign, assertCampaignSplitIdentity, assertCodeSurfaceIdentity, assertComponentSurface, buildAnalystSurfaceDispatch, buildEvidenceVector, buildLoopProvenanceRecord, campaignBreakdown, campaignMeanComposite, campaignMeasurementDigest, campaignScenarioIdentity, campaignSplitDigest, campaignSplitDigestFromIdentities, canonicalDigest, classifyUngroundedLiterals, codeSurfaceIdentityMaterial, compareOptimizationMethods, compareRankKeys, componentSurfaceIdentityMaterial, composeGate, costFromLedgerSummary, createReferenceEquivalenceJudge, createRunCostLedger, defaultProductionGate, detectScale, dimensionRegressions, discoverEvalFixtures, emitLoopProvenance, externalTextOptimizationMethod, failureModeRecallJudge, fsCampaignStorage, gepaOptimizationMethod, gitWorktreeAdapter, heldOutGate, heldoutSignificance, inMemoryCampaignStorage, isProposedCandidate, isTransientTransportFailure, labelTrustRank, llmJudge, loadEvalFixture, loadEvalFixtureScenarios, loopProvenanceArgsFromResult, loopProvenanceSpans, makePlaybackDispatch, neutralizationGate, neutralizeText, openAutoPr, openSearchLedger, optimizationTokenUsageFromSummary, pairHoldout, paretoPolicy, paretoSignificanceGate, planCampaignRun, planEvalFixtureRun, powerPreflight, provenanceRecordPath, provenanceSpansPath, renderScoreboardMarkdown, renderSurfaceDiff, resolveRunDir, resolveWorktreePath, rolloutArgumentDiff, runCampaign, runEval, runImprovementLoop, runOptimization, runProfileMatrix, scoreDiscrimination, scoreUserStory, scoreboardSummary, selectDiscriminative, sequentialDecide, sequentialPairedGate, skillOptOptimizationMethod, surfaceContentHash, surfaceHash, tangleTracesRoot, userStoryScoreboard, validateSearchLedgerEvent, verifyCodeSurface, verifyLoopProvenanceRecord };

package/dist/campaign/index.js

@@ -32,7 +32,7 @@ import {
   userStoryScoreboard,
   validateSearchLedgerEvent,
   verifyCodeSurface
-} from "../chunk-KO2PZOGP.js";
+} from "../chunk-Y5CLI4PY.js";
 import {
   acquireSingleRunLock,
   assertCodeSurfaceIdentity,
@@ -45,6 +45,7 @@ import {
   canonicalDigest,
   codeSurfaceIdentityMaterial,
   compareOptimizationMethods,
+  compareRankKeys,
   componentSurfaceIdentityMaterial,
   composeGate,
   costFromLedgerSummary,
@@ -78,7 +79,7 @@ import {
   surfaceContentHash,
   surfaceHash,
   verifyLoopProvenanceRecord
-} from "../chunk-NTOV7RU5.js";
+} from "../chunk-7AN2E7BU.js";
 import {
   SearchLedgerConflictError,
   SearchLedgerError,
@@ -96,7 +97,7 @@ import {
   runCampaign,
   tangleTracesRoot
 } from "../chunk-UCLVDLCH.js";
-import "../chunk-UI4YMIN2.js";
+import "../chunk-WGXIEX7P.js";
 import "../chunk-ARU2PZFM.js";
 import "../chunk-J5SQWP6Y.js";
 import "../chunk-PJQFMIOX.js";
@@ -141,6 +142,7 @@ export {
   classifyUngroundedLiterals,
   codeSurfaceIdentityMaterial,
   compareOptimizationMethods,
+  compareRankKeys,
   componentSurfaceIdentityMaterial,
   composeGate,
   costFromLedgerSummary,