@tangle-network/agent-eval 0.140.0 → 0.140.1

package/CHANGELOG.md

@@ -4,6 +4,14 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [0.140.1] - 2026-07-31 - a supervisor journal is read, or reported unreadable
+
+### Fixed
+
+- `parseSupervisorTree` returned zero spawns AND zero invalid rows for the journal `agent-runtime` writes — a positive claim that the run was empty. `agent-runtime/src/durable/spawn-journal.ts:232` writes `{kind:'event', root, event}`, so reading that envelope is a correctness fix rather than leniency.
+- Every non-empty line now lands in exactly one bucket, enforced by an invariant the tests assert on every case: `journalRows === spawns + closes + metered + sum(ignored) + journalInvalidRows`. A `never`-typed switch default makes a future unhandled event kind a compile error instead of a dropped row.
+- `journalInvalidRows` widens to "could not be interpreted at all", so the three existing integrity consumers fail closed on the new case without a second field to remember. `journalMalformedJsonRows` keeps the old narrower meaning, `journalIgnoredRowsByKind` records recognized-but-unmodelled kinds by name, and `journalDialect` (`none | flat | runtime-envelope | mixed`) records that the shape differed.
+
 ## [0.140.0] - 2026-07-31 - the recursive engine runs on real providers
 
 ### Changed

package/dist/analyst/index.d.ts

@@ -693,7 +693,7 @@ declare const ANALYST_BENCHMARK_HELP = "agent-eval analyst-benchmark\n\nRun the
 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 = "1e4c763f2b791c57b37c0e326c56546b433a7a99c0bc7d064abdb58b9cbf0d81";
+declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "701ea59184f3d227cf590ea3f16d24d65dbed2ad5644f8294801ceb66a30d4ef";
 /** The published benchmark evidence was produced at this package version, by
  * the retired one-shot direct runner, before trace analysts moved to the
  * recursive DSPy RLM engine. Both evidence digests below are historical facts

package/dist/analyst/index.js

@@ -1,12 +1,12 @@
 import { i as CostLedger } from "../cost-ledger-DMFxsLKr.js";
-import { C as createChatClient, _ as CONTROL_INTEGRITY_ANALYST, b as behavioralAnalyst, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, t as buildDefaultAnalystRegistry, v as ControlIntegrityAnalyst, x as deriveEfficiencyFindings, y as emitControlIntegrityFindings } from "../default-registry-DCp-6hc-.js";
+import { C as createChatClient, _ as CONTROL_INTEGRITY_ANALYST, b as behavioralAnalyst, f as DEFAULT_TRACE_ANALYST_KINDS, g as FAILURE_MODE_KIND_SPEC, h as IMPROVEMENT_KIND_SPEC, i as assertExactRegistryRunOpts, m as KNOWLEDGE_GAP_KIND_SPEC, n as AnalystRegistry, p as KNOWLEDGE_POISONING_KIND_SPEC, r as ExactAnalystRunExecutionError, t as buildDefaultAnalystRegistry, v as ControlIntegrityAnalyst, x as deriveEfficiencyFindings, y as emitControlIntegrityFindings } from "../default-registry-BOiBXNFh.js";
 import { A as coerceJson, B as renderFindingSubject, D as RawAnalystFindingSchema, E as RawAnalystEvidenceSchema, F as FINDING_SUBJECT_SYNTAX, G as resolveTraceAnalystLimits, I as FindingSubjectStringSchema, L as KIND_EXPECTED_SUBJECTS, M as stripCodeFences, N as FINDING_SUBJECT_GRAMMAR_PROMPT, O as evidenceRefsFromRawFinding, P as FINDING_SUBJECT_KINDS, R as findingSubjectGrammarPromptFor, T as RAW_FINDING_SCHEMA_PROMPT, W as DEFAULT_TRACE_ANALYST_LIMITS, a as buildTraceToolsForGroup, i as runTraceAnalyst, j as coerceToFindingRows, k as parseRawFinding, n as renderPriorFindings, r as renderUpstreamFindings, t as createTraceAnalyst, w as ANALYST_SEVERITIES, z as parseFindingSubject } from "../kind-factory-DB7nIs35.js";
 import { a as computeFindingId, i as validateUsageSettlementTimeout, n as settleUsageReceiptFromCostLedger, o as makeFinding, s as makeProposalFinding } from "../usage-receipt-CgxMEBZq.js";
 import { n as isProposalFinding, t as assertProposalFindings } from "../proposal-findings-2GIUo1et.js";
 import { a as RunCritic, c as buildSkillUsageReport, d as defaultIsMaterial, f as diffFindings, h as defineTraceAnalyst, i as runSemanticConceptJudge, l as emitSkillUsageFindings, m as defineCustomAnalyst, n as SEMANTIC_CONCEPT_JUDGE_VERSION, o as SKILL_USAGE_ANALYST, s as SkillUsageAnalyst, u as FindingsStore } from "../semantic-concept-judge-DJQtFr95.js";
 import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-Bkak4nzo.js";
 import { a as scoreAnalystFindings, n as runAnalystBenchmark, r as traceStoreEvidenceResolver, t as registryBenchmarkRunner } from "../benchmark-CYtcIF2V.js";
-import { A as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, B as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, C as publicBenchmarkSystemPrompt, D as parseVerificationOutcome, E as loadCodeTraceVerificationArtifacts, F as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, G as summarizeAgentRxCalibration, H as ANALYST_BENCHMARK_OBSERVATIONS_FILE, I as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, J as agentRxBenchmarkCase, K as codeTraceBenchCase, L as analystBenchmarkDependencyLockDigest, M as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, N as ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, O as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, P as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, Q as normalizeBenchmarkLabel, R as analystBenchmarkImplementationDigest, S as publicBenchmarkRlmInstructions, T as appendVerificationArtifactsToOtlp, U as AGENT_RX_UPSTREAM_REVISION, V as ANALYST_BENCHMARK_MANIFEST_FILE, W as renderAgentRxCalibrationMarkdown, X as normalizeAgentRxCategory, Y as agentRxPredictionsToFindings, Z as roundAgentRxStep, _ as expandCodeTraceFailureBlocks, a as renderCodeTraceCalibrationMarkdown, b as MAX_INCORRECT_BLOCK_STEPS, c as createPublicBenchmarkRlmRunner, d as preparePublicAnalystBenchmark, f as publicBenchmarkDistributions, g as emptyPublicBenchmarkRunner, h as adaptPublicBenchmarkFindings, i as readAnalystBenchmarkArtifact, j as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, k as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, l as createPublicBenchmarkDirectRunner, m as selectPublicBenchmarkRows, n as runAnalystBenchmarkCommand, o as summarizeCodeTraceCalibration, p as publicBenchmarkSelectionReport, q as codeTracerPredictionsToFindings, r as renderAnalystBenchmarkMarkdown, s as compareAnalystRunners, t as ANALYST_BENCHMARK_HELP, u as loadPublicBenchmarkRows, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, x as publicBenchmarkProtocolSha256, y as MAX_INCORRECT_BLOCKS, z as ANALYST_BENCHMARK_COST_LEDGER_FILE } from "../benchmark-command-CK0UnXAD.js";
+import { A as ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, B as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, C as publicBenchmarkSystemPrompt, D as parseVerificationOutcome, E as loadCodeTraceVerificationArtifacts, F as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, G as summarizeAgentRxCalibration, H as ANALYST_BENCHMARK_OBSERVATIONS_FILE, I as ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, J as agentRxBenchmarkCase, K as codeTraceBenchCase, L as analystBenchmarkDependencyLockDigest, M as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, N as ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, O as ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, P as ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, Q as normalizeBenchmarkLabel, R as analystBenchmarkImplementationDigest, S as publicBenchmarkRlmInstructions, T as appendVerificationArtifactsToOtlp, U as AGENT_RX_UPSTREAM_REVISION, V as ANALYST_BENCHMARK_MANIFEST_FILE, W as renderAgentRxCalibrationMarkdown, X as normalizeAgentRxCategory, Y as agentRxPredictionsToFindings, Z as roundAgentRxStep, _ as expandCodeTraceFailureBlocks, a as renderCodeTraceCalibrationMarkdown, b as MAX_INCORRECT_BLOCK_STEPS, c as createPublicBenchmarkRlmRunner, d as preparePublicAnalystBenchmark, f as publicBenchmarkDistributions, g as emptyPublicBenchmarkRunner, h as adaptPublicBenchmarkFindings, i as readAnalystBenchmarkArtifact, j as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, k as ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, l as createPublicBenchmarkDirectRunner, m as selectPublicBenchmarkRows, n as runAnalystBenchmarkCommand, o as summarizeCodeTraceCalibration, p as publicBenchmarkSelectionReport, q as codeTracerPredictionsToFindings, r as renderAnalystBenchmarkMarkdown, s as compareAnalystRunners, t as ANALYST_BENCHMARK_HELP, u as loadPublicBenchmarkRows, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, x as publicBenchmarkProtocolSha256, y as MAX_INCORRECT_BLOCKS, z as ANALYST_BENCHMARK_COST_LEDGER_FILE } from "../benchmark-command-Bo_hkhQf.js";
 //#region src/analyst/adapters.ts
 /**
 * Adapter factories — lift each existing agent-eval primitive into the

package/dist/{benchmark-command-CK0UnXAD.js → benchmark-command-Bo_hkhQf.js}

@@ -1172,7 +1172,7 @@ const ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES = Object.freeze([
 	"package.json",
 	"pnpm-lock.yaml"
 ]);
-const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "1e4c763f2b791c57b37c0e326c56546b433a7a99c0bc7d064abdb58b9cbf0d81";
+const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "701ea59184f3d227cf590ea3f16d24d65dbed2ad5644f8294801ceb66a30d4ef";
 /** The published benchmark evidence was produced at this package version, by
 * the retired one-shot direct runner, before trace analysts moved to the
 * recursive DSPy RLM engine. Both evidence digests below are historical facts
@@ -4828,4 +4828,4 @@ function shellQuote(value) {
 //#endregion
 export { ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 as A, ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE as B, publicBenchmarkSystemPrompt as C, parseVerificationOutcome as D, loadCodeTraceVerificationArtifacts as E, ANALYST_BENCHMARK_IMPLEMENTATION_FILES as F, summarizeAgentRxCalibration as G, ANALYST_BENCHMARK_OBSERVATIONS_FILE as H, ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 as I, agentRxBenchmarkCase as J, codeTraceBenchCase as K, analystBenchmarkDependencyLockDigest as L, ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 as M, ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION as N, ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM as O, ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM as P, normalizeBenchmarkLabel as Q, analystBenchmarkImplementationDigest as R, publicBenchmarkRlmInstructions as S, appendVerificationArtifactsToOtlp as T, AGENT_RX_UPSTREAM_REVISION as U, ANALYST_BENCHMARK_MANIFEST_FILE as V, renderAgentRxCalibrationMarkdown as W, normalizeAgentRxCategory as X, agentRxPredictionsToFindings as Y, roundAgentRxStep as Z, expandCodeTraceFailureBlocks as _, renderCodeTraceCalibrationMarkdown as a, MAX_INCORRECT_BLOCK_STEPS as b, createPublicBenchmarkRlmRunner as c, preparePublicAnalystBenchmark as d, publicBenchmarkDistributions as f, emptyPublicBenchmarkRunner as g, adaptPublicBenchmarkFindings as h, readAnalystBenchmarkArtifact as i, ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 as j, ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES as k, createPublicBenchmarkDirectRunner as l, selectPublicBenchmarkRows as m, runAnalystBenchmarkCommand as n, summarizeCodeTraceCalibration as o, publicBenchmarkSelectionReport as p, codeTracerPredictionsToFindings as q, renderAnalystBenchmarkMarkdown as r, compareAnalystRunners as s, ANALYST_BENCHMARK_HELP as t, loadPublicBenchmarkRows as u, CODE_TRACE_BENCH_ANALYST_PROMPT as v, DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES as w, publicBenchmarkProtocolSha256 as x, MAX_INCORRECT_BLOCKS as y, ANALYST_BENCHMARK_COST_LEDGER_FILE as z };
 
-//# sourceMappingURL=benchmark-command-CK0UnXAD.js.map
+//# sourceMappingURL=benchmark-command-Bo_hkhQf.js.map