@tangle-network/agent-eval 0.191.0 → 0.193.0

package/CHANGELOG.md

@@ -6,6 +6,66 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ## Unreleased
 
+## [0.193.0] — 2026-09-25
+
+### Added
+
+- A search with a test split ends in its claim ([search ledger](./docs/search-ledger.md#the-claim)).
+  When expansion stops, `runSearch` fixes the claim's design before any test cell exists (`planSearchClaim`, stored as the `claim-plan` blob of the `claim` operation): at most 3 finalists that scored every selection unit and beat the root there, ranked by selection mean; the estimator; and a power check that simulates the claim's own `decidePairedPromotion` call at the claim's `minimumEffect` with the search's pooled selection variance (`pairedPromotionPower`).
+  The finalists are decided `finalist`, the root's and their test cells run together ahead of everything else, and `decideSearchClaim` tests each finalist against the root at Bonferroni confidence `1 - 0.05 / k`.
+  A test split that cannot resolve the minimum effect closes as `test-cannot-resolve` with no test spend.
+  In 200 simulated 50-node searches in which no node differs from the root (12 selection and 24 test units, minimum effect 0.1), the claim shipped 5 times: 2.5 %, Wilson 95 % interval [1.1 %, 5.7 %]; per finalist, 5 of 584 tests promoted. With one child planted 0.15 better, it shipped a truly better node in 200 of 200.
+- The divergence rule: a node whose train mean rose over its parent's while its selection interval against the parent lies wholly on the worse side is decided `invalid` with rule `divergence` (`searchDivergence`), and never becomes a parent or a finalist.
+- `searchClaimReserveUsd({ testTasks, reps, cellUsd })` sizes `budget.reservedClaimUsd`; `SearchRunResult.claim` returns the claim.
+- `renderSearchSummary(state, { split })` (`/campaign`) renders a search as compact text: the leading nodes against the root, recently discarded nodes with the measurement that discarded them, and recent proposals; `agent-eval search show <ledger>` verifies a ledger and prints it. `ProposeContext` gains `parents`, `train` (a `searchProposerView`, which reads only the train split) and `summary`, and `runOptimization`'s proposal step fills all three.
+- `scripts/search-sim.ts claims` runs many simulated searches on in-memory ledgers and tallies their claims; `--null`, `--plant-gain`, `--plant-divergence`, `--test` and `--min-effect` shape them.
+
+### Changed
+
+- **Breaking:** `runSearch` refuses to start a search with a test split that has no selection split, no claim `minimumEffect`, or an unpinned judge, and, under a cap, one whose `reservedClaimUsd` cannot hold the root and 3 finalists on every test task or whose cap cannot also cover the root and one screening round.
+- **Breaking:** the `search-closed` claim records its `rule`, family-wise `confidence` and `reason`; its power check records the target power, finalist count, pooled variance and estimator; each finalist records its deciding `test` (pairs, confidence, method, delta, interval).
+- **Breaking:** `SearchAllocator` declares `reps`, the repeats the claim's test cells run at.
+- The projector refuses a finalist that is the root, a fourth finalist, a finalist decided after a claim cell, any operation, node, edge or non-claim cell after the first finalist, an edge from a node decided `invalid`, a claim that omits a finalist or tests one at the wrong Bonferroni confidence, and a `ship` claim without a pinned judge or with a test unit the root or the shipped finalist did not score.
+- `operation-started` binds its artifacts, which `SearchOperation.artifacts` lists before the recorded ones.
+- Migration: add `reps` to a custom allocator; give a search with a test split a selection split, `objective.claim.minimumEffect`, a pinned `objective.judge` and, when capped, `budget.reservedClaimUsd` of at least `searchClaimReserveUsd(...)`.
+
+## [0.192.0] — 2026-09-25
+
+### Added
+
+- `runSearch` (`/campaign`) is the one search kernel: an event-driven loop with no generation barrier that separates where to expand (`SearchPolicy`), where to spend rollouts (`SearchAllocator`) and what to claim ([search ledger](./docs/search-ledger.md#run-a-search-the-kernel)).
+  A lane that frees up takes the next allocated cell (claim, then rung and root, then screen, then train); the policy proposes when no cell waits and the cap admits one proposal and its expected screens.
+  Its ports are an executor (`lanes`, `place`, `run`, `adopt`), a proposer, and an artifact codec.
+  Every reservation passes the ledger's admission rule before it is asked for: a hard lane holds its enforced per-cell maximum, an estimate lane 1.5 times the p99 of its settled cells once 20 settled (else its prior), and spend above a hold is recorded as overspend.
+  The ledger is the only checkpoint: rerun on an open ledger, the kernel records an interrupted proposal as failed at an unknown cost, finishes a recorded proposal from its stored output, and offers every unsettled cell to `executor.adopt` before it runs it.
+  Killed with SIGKILL 8 times at random ledger positions, a simulated search (7 nodes, 56 cells, 10 % environment faults, an $8 cap, estimate lanes) resumed to the same nodes, edges, cells, scores, decisions and close as an uninterrupted run; 10 attempts that finished before a kill were adopted, none finished twice, and committed spend ($2.55) stayed within the cap plus recorded overspend ($0.18).
+  A real `runOptimization` (runCampaign, CostLedger, cell cache, file ledger; 7 nodes, 28 cells) killed 6 times, once inside a proposal, returned the same winner and generations as an uninterrupted run and dispatched the same 28 cells to completion.
+  With 1.5 s cells on 4 slots, all 4 slots were busy for 93 % of the time any cell ran.
+- `incumbent({ patience })` and `crowdedFrontierParent({ seed })` are `SearchPolicy`s; `uniform({ reps })` is the fixed-plan `SearchAllocator`.
+- `SearchStateView.budget` gives the admission rule's room (`headroomUsd`, `claimReserveUsd`), and `SearchOperation` carries its recorded outcome and artifacts.
+- `SearchRecorder.readBlob(ref)` reads a stored blob back and checks its digest and length.
+- `openSearchLedger({ store })` keeps a ledger in an in-process text store (an in-memory `CampaignStorage`) for runs without a filesystem, with the file ledger's chain, parse and state rules.
+- `scripts/search-sim.ts` runs the real kernel over a seeded synthetic objective, proposer and executor; `kill-resume` SIGKILLs it at random ledger positions and compares the resumed search with an uninterrupted one.
+
+### Changed
+
+- **Breaking:** `runOptimization` is a search on the kernel with `incumbent()` (or `policy`) and `uniform({ reps })`.
+  Its search ledger is always written (`<runDir>/search/ledger.jsonl`, held in `storage` unless that is the filesystem, or where `searchLedger` puts it), and `searchHistory` is always returned.
+  Running it again on the same run directory continues an interrupted search and reruns nothing that settled; after a closed search it starts the next one beside it.
+  A generation is one proposal of `populationSize` candidates; each candidate runs one cell at a time on lanes of `candidateConcurrency * maxConcurrency` slots.
+  `analyzeGeneration` runs just before the proposal it feeds, inside that proposal's operation, so its spend is booked there.
+  A candidate identical to a surface the search holds is a second edge into that node and is not measured again; it used to throw.
+  The kept surface is the policy's leader: the candidate that scored every scenario and beats the leader on the scenarios they share.
+- **Breaking:** `selfImprove({ selectParent })` is `selfImprove({ policy })`, and `ProposeContext` gains `operator`.
+- `CampaignStorage.kind` names a filesystem or memory storage.
+- Migration: replace `selectParent: crowdedFrontierParent({ seed })` with `policy: crowdedFrontierParent({ seed })`; drop `selectionRankKey` (a domain metric belongs in the judge's composite, which the leader rule ranks); read a candidate's lineage from the search ledger's edges and its contrast from `estimateNode(state, nodeId, { against, split: 'train' })` instead of `parentSurfaceHash`, `parentComposite` and `observedDeltaFromParent`.
+
+### Removed
+
+- The generation loop body of `runOptimization`, its `compareRankKeys` promotion and the `selectionRankKey` option, `compareRankKeys`, `assertFiniteRankKey`, the `ParentSelector` and `ParentSelectionContext` types, and `OptimizationSearch`.
+- `GenerationCandidate.ci95` (always null), `parentSurfaceHash`, `parentComposite` and the unpaired `observedDeltaFromParent`, with their `LoopProvenanceCandidate` copies and span attributes.
+- The `runOptimization`, proposal-findings, parent-selection, loop-provenance-integrity and `selfImprove` unit tests. The proof is the search simulator's kill-resume run and a real `runOptimization` (runCampaign, CostLedger, cell cache, file ledger) killed and resumed against an uninterrupted run.
+
 ## [0.191.0] — 2026-09-25
 
 ### Changed
@@ -85,6 +145,7 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 - `knownSecrets` are cut out where they stand instead of replacing the whole string, so `redactText('cleanup failed: <key> was still set', { knownSecrets: [key] })` keeps `cleanup failed:` and `was still set`.
   A whole-string base64 payload that holds a known secret at any byte offset is still replaced whole, and value shapes such as bearer tokens still replace the whole string.
 
+
 ### Removed
 
 - The candidate-slot ledger: `SearchPlan`, candidate slots, `search-planned`, `search-plan-extended`, `candidate-registered`, `candidate-slot-closed`, `task-attempted`, `candidate-decided`, `search-completed`, `lineageNodeId`, `SearchLedgerReplay`, `SearchLedgerAudit`, `search-ledger-projector.ts`, and `recordCandidatePopulationSearch`.

package/dist/adapters/http.d.ts

@@ -1,6 +1,6 @@
 import { m as CostReceipt } from "../cost-ledger-DbQdN3nO.js";
-import { R as Scenario, d as DispatchContext, f as DispatchFn } from "../types-BmQRs661.js";
-import "../index-DH-MWH5f.js";
+import { R as Scenario, d as DispatchContext, f as DispatchFn } from "../types-BWwRXofp.js";
+import "../index-DWQNDqPd.js";
 //#region src/adapters/http.d.ts
 interface HttpDispatchOptions<TScenario extends Scenario, _TArtifact> {
   /** Static endpoint URL. Mutually exclusive with `resolveUrl`. */

package/dist/analyst/index.d.ts

@@ -4,10 +4,10 @@ import { A as createChatClient, C as ChatResponse, D as MockTransportOpts, E as
 import { A as runAnalystBenchmark, C as AnalystEvidenceResolutionError, D as AnalystLatencyDistribution, E as AnalystIssueExpectation, M as scoreAnalystFindings, O as RunAnalystBenchmarkOptions, S as AnalystEvidenceResolution, T as AnalystFindingScore, V as ExternalOptimizerModelCall, W as ExternalOptimizerModelExecutionObservation, _ as AnalystBenchmarkProvenance, b as AnalystBenchmarkSummary, d as AnalystBenchmarkDatasetRef, f as AnalystBenchmarkDescriptor, g as AnalystBenchmarkOutput, h as AnalystBenchmarkObservation, j as traceStoreEvidenceResolver, k as registryBenchmarkRunner, m as AnalystBenchmarkLabelState, p as AnalystBenchmarkError, q as ExternalOptimizerRunnerCommand, t as AgentProfile, u as AnalystBenchmarkCase, v as AnalystBenchmarkResult, w as AnalystEvidenceResolver, x as AnalystEvidenceExpectation, y as AnalystBenchmarkRunner } from "../agent-profile-DDfIjdyS.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-KGxBWH4b.js";
 import { a as ExactAnalystBudgetPolicy, c as RegistryRunOpts, i as BudgetPolicy, n as AnalystRegistry, o as ExactAnalystRunExecutionError, r as AnalystRegistryOptions, s as ExactRegistryRunOpts, t as AnalystHooks } from "../registry-Bu3lNole.js";
-import { a as createTraceAnalyst, c as BehavioralAnalystOptions, i as TraceAnalystDefinition, l as behavioralAnalyst, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as runTraceAnalyst, t as DefaultAnalystRegistryOptions, u as deriveEfficiencyFindings } from "../default-registry-DK_9aVJf.js";
-import { a as TraceAnalystLimits, c as RawAnalystEvidence, d as evidenceRefsFromRawFinding, f as parseRawFinding, i as TraceAnalysisEngineResult, l as RawAnalystFinding, n as TraceAnalysisEngine, o as resolveTraceAnalystLimits, r as TraceAnalysisEngineRequest, s as RAW_FINDING_SCHEMA_PROMPT, t as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, u as RawAnalystFindingSchema } from "../engine-vNiM02Ub.js";
-import { n as buildTraceToolsForGroup, t as TraceToolGroupName } from "../tool-groups-D-mkeN0N.js";
-import { A as SemanticConceptJudgeInput, C as DspyRlmTraceEngineOptions, E as createChatTraceEngine, S as renderFindingSubject, T as ChatTraceEngineOptions, _ as FindingSubject, a as FAILURE_MODE_KIND_SPEC, b as findingSubjectGrammarPromptFor, c as emitControlIntegrityFindings, d as FindingsStore, f as PersistedFinding, g as FINDING_SUBJECT_SYNTAX, h as FINDING_SUBJECT_KINDS, i as IMPROVEMENT_KIND_SPEC, j as SemanticConceptJudgeOptions, l as DiffPolicy, m as diffFindings, n as KNOWLEDGE_POISONING_KIND_SPEC, o as CONTROL_INTEGRITY_ANALYST, p as defaultIsMaterial, r as KNOWLEDGE_GAP_KIND_SPEC, s as ControlIntegrityAnalyst, t as DEFAULT_TRACE_ANALYST_KINDS, u as FindingsDiff, v as FindingSubjectKind, w as createDspyRlmTraceEngine, x as parseFindingSubject, y as KIND_EXPECTED_SUBJECTS } from "../index-D-TN7ZRM.js";
+import { a as createTraceAnalyst, c as BehavioralAnalystOptions, i as TraceAnalystDefinition, l as behavioralAnalyst, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as runTraceAnalyst, t as DefaultAnalystRegistryOptions, u as deriveEfficiencyFindings } from "../default-registry-C3HTE08O.js";
+import { a as TraceAnalystLimits, c as RawAnalystEvidence, d as evidenceRefsFromRawFinding, f as parseRawFinding, i as TraceAnalysisEngineResult, l as RawAnalystFinding, n as TraceAnalysisEngine, o as resolveTraceAnalystLimits, r as TraceAnalysisEngineRequest, s as RAW_FINDING_SCHEMA_PROMPT, t as DEFAULT_TRACE_ANALYST_OUTPUT_TOKENS, u as RawAnalystFindingSchema } from "../engine-DjRV55F9.js";
+import { n as buildTraceToolsForGroup, t as TraceToolGroupName } from "../tool-groups-BC9O7Epn.js";
+import { A as SemanticConceptJudgeInput, C as DspyRlmTraceEngineOptions, E as createChatTraceEngine, S as renderFindingSubject, T as ChatTraceEngineOptions, _ as FindingSubject, a as FAILURE_MODE_KIND_SPEC, b as findingSubjectGrammarPromptFor, c as emitControlIntegrityFindings, d as FindingsStore, f as PersistedFinding, g as FINDING_SUBJECT_SYNTAX, h as FINDING_SUBJECT_KINDS, i as IMPROVEMENT_KIND_SPEC, j as SemanticConceptJudgeOptions, l as DiffPolicy, m as diffFindings, n as KNOWLEDGE_POISONING_KIND_SPEC, o as CONTROL_INTEGRITY_ANALYST, p as defaultIsMaterial, r as KNOWLEDGE_GAP_KIND_SPEC, s as ControlIntegrityAnalyst, t as DEFAULT_TRACE_ANALYST_KINDS, u as FindingsDiff, v as FindingSubjectKind, w as createDspyRlmTraceEngine, x as parseFindingSubject, y as KIND_EXPECTED_SUBJECTS } from "../index-0jEtpSM7.js";
 import { A as ReplyRowDecoded, C as primeProtocolSha256, D as DecodedReply, E as runPrimeExchange, M as decodeReplyRows, O as ReplyContract, S as normalizePrimeUsage, T as projectPrimeTrajectory, _ as buildPrimePrompt, a as PrimeProjectionOutcome, b as extractPrimeJsonObject, c as PrimeProtocolIdentity, d as PrimeRepairPromptSpec, f as PrimeRepairState, g as analystUsageReceiptFromPrimeUsage, h as PrimeTurnRecord, i as PrimeProjectionDelivery, j as ReplyRowRejection, k as ReplyEnvelope, l as PrimeRawUsage, m as PrimeRowDecoded, n as PrimeExchangeOutcome, o as PrimeProjectionSource, p as PrimeReplyContract, r as PrimeFailure, s as PrimePromptSpec, t as PrimeExchangeOptions, u as PrimeRejectedRow, v as buildPrimeRepairPrompt, w as primeReplyDefect, x as mergePrimeRawUsage, y as emptyPrimeRawUsage } from "../prime-protocol-T0i6xcAs.js";
 import { i as nodeHttpPrimeBridgeTransport, n as PrimeBridgeTransportRequest, r as PrimeBridgeTransportResult, t as PrimeBridgeTransport } from "../prime-bridge-transport-6feEglLf.js";
 import { z } from "zod";
@@ -1181,11 +1181,11 @@ declare function runAnalystBenchmarkCommand(argv: readonly string[], env?: NodeJ
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM = "sha256-canonical-source-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM = "sha256-canonical-file-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "f02c0d075ada4e1046658d9591f00e431aefdde78779971cc568ccc23b2333ac";
+declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "0ff3eda790af1c3811ac4e084fe57c9d64254966fd0ebf53374f40e504bc28be";
 declare const ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 = "1e03f2daed356d60316aabefb407ec1e437ac94d408d61eea4ae096e9c6fbb5b";
 declare const ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 = "4dba263b6256a30d56c7fdb2d992d3a953c0035d731f359b704db806f68f75ac";
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "14264d6b6f09cc0314c0e930083ec947123d54b64b021bb107bfe5d8859efe20";
+declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "00002cbbc8a749281287418f0abc03e4fc6a78ef3c1b4f21e33f2452629283da";
 declare function analystBenchmarkImplementationDigest(): string;
 declare function analystBenchmarkDependencyLockDigest(): string;
 //#endregion

package/dist/analyst/index.js

@@ -4,11 +4,11 @@ import { B as FINDING_SUBJECT_SYNTAX, F as parseRawFinding, G as DEFAULT_TRACE_A
 import { i as validateUsageSettlementTimeout, n as settleUsageReceiptFromCostLedger } from "../usage-receipt-Cx3xNeyB.js";
 import { a as KNOWLEDGE_POISONING_KIND_SPEC, c as FAILURE_MODE_KIND_SPEC, d as emitControlIntegrityFindings, f as behavioralAnalyst, i as DEFAULT_TRACE_ANALYST_KINDS, l as CONTROL_INTEGRITY_ANALYST, n as AnalystRegistry, o as KNOWLEDGE_GAP_KIND_SPEC, p as deriveEfficiencyFindings, r as ExactAnalystRunExecutionError, s as IMPROVEMENT_KIND_SPEC, t as buildDefaultAnalystRegistry, u as ControlIntegrityAnalyst } from "../default-registry-xMLX160E.js";
 import { i as makeProposalFinding, n as computeFindingId, r as makeFinding } from "../types-CW5b90JH.js";
-import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-MfpwoEaj.js";
-import { a as diffFindings, i as defaultIsMaterial, n as runSemanticConceptJudge, o as createChatTraceEngine, r as FindingsStore, t as SEMANTIC_CONCEPT_JUDGE_VERSION } from "../semantic-concept-judge-Bt7bXAAo.js";
+import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-BX34vPVY.js";
+import { a as diffFindings, i as defaultIsMaterial, n as runSemanticConceptJudge, o as createChatTraceEngine, r as FindingsStore, t as SEMANTIC_CONCEPT_JUDGE_VERSION } from "../semantic-concept-judge-Bg-560H0.js";
 import { t as createChatClient } from "../chat-client-B_ItsSBw.js";
 import { a as scoreAnalystFindings, i as summarizeAnalystBenchmarkRunner, n as runAnalystBenchmark, r as traceStoreEvidenceResolver, t as registryBenchmarkRunner } from "../benchmark-BeQJyrRC.js";
-import { $ as ANALYST_BENCHMARK_MANIFEST_FILE, A as readAnalystInstructionsOverride, B as parseVerificationOutcome, C as emptyPublicBenchmarkRunner, D as summarizeCodeTraceCalibration, E as renderCodeTraceCalibrationMarkdown, F as publicBenchmarkRlmInstructions, G as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, H as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, I as publicBenchmarkSystemPrompt, J as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, K as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, L as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, M as MAX_INCORRECT_BLOCKS, N as MAX_INCORRECT_BLOCK_STEPS, O as analystInstructionsOverrideFromText, P as publicBenchmarkProtocolSha256, Q as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, R as appendVerificationArtifactsToOtlp, S as adaptPublicBenchmarkFindings, T as readAnalystBenchmarkArtifact, U as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, V as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, W as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, X as analystBenchmarkImplementationDigest, Y as analystBenchmarkDependencyLockDigest, Z as ANALYST_BENCHMARK_COST_LEDGER_FILE, _ as publicDirectAnalystDefinition, a as primeCodeTraceAnalystDefinition, at as codeTraceBenchCase, b as analystDefinitionAsymmetries, c as preparePublicAnalystBenchmark, ct as agentRxPredictionsToFindings, d as selectPublicBenchmarkRows, dt as normalizeBenchmarkLabel, et as ANALYST_BENCHMARK_OBSERVATIONS_FILE, f as createPublicBenchmarkRlmRunner, g as createPublicBenchmarkDirectRunner, h as runReplVariableAnalystDefinition, i as primeAnalystProtocolSha256, it as summarizeAgentRxCalibration, j as CODE_TRACE_BENCH_ANALYST_PROMPT, k as effectiveAnalystProtocolSha256, l as publicBenchmarkDistributions, lt as normalizeAgentRxCategory, m as rlmEngineLimits, n as renderAnalystBenchmarkMarkdown, nt as AGENT_RX_UPSTREAM_REVISION, o as runInlineAnalystDefinition, ot as codeTracerPredictionsToFindings, p as publicRlmAnalystDefinition, q as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, r as createPrimeBenchmarkRunner, rt as renderAgentRxCalibrationMarkdown, s as loadPublicBenchmarkRows, st as agentRxBenchmarkCase, t as runAnalystBenchmarkCommand, tt as compareAnalystRunners, u as publicBenchmarkSelectionReport, ut as roundAgentRxStep, v as runChunkedAnalystDefinition, w as expandCodeTraceFailureBlocks, x as analystDefinitionProtocolSha256, y as AnalystExpressivenessError, z as loadCodeTraceVerificationArtifacts } from "../benchmark-command-3tZJnD82.js";
+import { $ as ANALYST_BENCHMARK_MANIFEST_FILE, A as readAnalystInstructionsOverride, B as parseVerificationOutcome, C as emptyPublicBenchmarkRunner, D as summarizeCodeTraceCalibration, E as renderCodeTraceCalibrationMarkdown, F as publicBenchmarkRlmInstructions, G as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, H as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, I as publicBenchmarkSystemPrompt, J as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, K as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, L as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, M as MAX_INCORRECT_BLOCKS, N as MAX_INCORRECT_BLOCK_STEPS, O as analystInstructionsOverrideFromText, P as publicBenchmarkProtocolSha256, Q as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, R as appendVerificationArtifactsToOtlp, S as adaptPublicBenchmarkFindings, T as readAnalystBenchmarkArtifact, U as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, V as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, W as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, X as analystBenchmarkImplementationDigest, Y as analystBenchmarkDependencyLockDigest, Z as ANALYST_BENCHMARK_COST_LEDGER_FILE, _ as publicDirectAnalystDefinition, a as primeCodeTraceAnalystDefinition, at as codeTraceBenchCase, b as analystDefinitionAsymmetries, c as preparePublicAnalystBenchmark, ct as agentRxPredictionsToFindings, d as selectPublicBenchmarkRows, dt as normalizeBenchmarkLabel, et as ANALYST_BENCHMARK_OBSERVATIONS_FILE, f as createPublicBenchmarkRlmRunner, g as createPublicBenchmarkDirectRunner, h as runReplVariableAnalystDefinition, i as primeAnalystProtocolSha256, it as summarizeAgentRxCalibration, j as CODE_TRACE_BENCH_ANALYST_PROMPT, k as effectiveAnalystProtocolSha256, l as publicBenchmarkDistributions, lt as normalizeAgentRxCategory, m as rlmEngineLimits, n as renderAnalystBenchmarkMarkdown, nt as AGENT_RX_UPSTREAM_REVISION, o as runInlineAnalystDefinition, ot as codeTracerPredictionsToFindings, p as publicRlmAnalystDefinition, q as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, r as createPrimeBenchmarkRunner, rt as renderAgentRxCalibrationMarkdown, s as loadPublicBenchmarkRows, st as agentRxBenchmarkCase, t as runAnalystBenchmarkCommand, tt as compareAnalystRunners, u as publicBenchmarkSelectionReport, ut as roundAgentRxStep, v as runChunkedAnalystDefinition, w as expandCodeTraceFailureBlocks, x as analystDefinitionProtocolSha256, y as AnalystExpressivenessError, z as loadCodeTraceVerificationArtifacts } from "../benchmark-command-DVhUPU7H.js";
 import { a as extractPrimeJsonObject, c as primeProtocolSha256, d as runPrimeExchange, f as decodeReplyRows, i as emptyPrimeRawUsage, l as primeReplyDefect, n as buildPrimePrompt, o as mergePrimeRawUsage, r as buildPrimeRepairPrompt, s as normalizePrimeUsage, t as analystUsageReceiptFromPrimeUsage, u as projectPrimeTrajectory } from "../prime-protocol-DcAK9pC8.js";
 import { t as nodeHttpPrimeBridgeTransport } from "../prime-bridge-transport-Z3cpKGoV.js";
 import { existsSync, readFileSync, readdirSync, statSync } from "node:fs";

package/dist/{analyze-runs-SCSWYpy0.js → analyze-runs-B6gv7vAT.js}

@@ -5,7 +5,7 @@ import { i as requiredPairedSampleSize, r as pairedMde } from "./power-and-mde-B
 import { r as pairRunRecords } from "./paired-arms-JSDcs2zx.js";
 import { o as pairedTTest, r as pairedBootstrap } from "./paired-tests-C8iCsioC.js";
 import { r as welchsTTest } from "./baseline-BC-eBZ7U.js";
-import { n as aggregatePairedHoldout } from "./statistical-heldout-GcLIPxk4.js";
+import { n as aggregatePairedHoldout } from "./statistical-heldout-DANVmyug.js";
 import "./query-DlV0pdx4.js";
 import { r as observedSplitScore } from "./reward-nw2xZGZG.js";
 import { u as validateRunRecord } from "./run-record-BRlM99rX.js";
@@ -982,4 +982,4 @@ function buildRecommendations(ctx) {
 //#endregion
 export { summarizeExecution as n, checkCanaries as r, analyzeRuns as t };
 
-//# sourceMappingURL=analyze-runs-SCSWYpy0.js.map
+//# sourceMappingURL=analyze-runs-B6gv7vAT.js.map