@tangle-network/agent-eval 0.221.1 → 0.221.3

package/CHANGELOG.md

@@ -4,6 +4,85 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [0.221.3] — 2026-10-24
+
+- fix(analyst): repin the benchmark implementation digest after #1097 (#1098)
+- fix(trace-analyst): build OTLP stores from bytes, not one joined string
+- fix(adopt): prepare caller-locked Chromium before gates
+- test(adopt): retain admitted identity in real Git fixtures
+- fix(adopt): retain the sparse caller entrypoint
+- fix(adopt): settle pushed proposal reads against remote refs
+- chore(release): 0.221.2
+- Return plain analyst findings from decision replay
+- chore(release): 0.221.1 (#1091)
+- fix(deps): pin zod 4.6.6 (GHSA-mrhx-c7xc-m3w5) (#1088)
+- Give the disposable adoption fixture its own Git author
+- Fix real adoption drift fixture under the declared package manager
+- Capture supporting trace events in the original native decision
+- test: inherit the integration default for behavioral questions
+- test: set the integration default from the observed loaded tail
+- chore(release): 0.221.0
+- test: own the measured integration deadline in Vitest config
+- test: await observed integration completion across tool paths
+- feat: preserve finding composition and assigned arm populations
+- test: await decision-point HTTP and ledger completion
+- test: await durable readiness in filesystem integration cases
+- test(process): wait for argv child readiness before abort
+- feat(analyst): add bounded calibrated trace question observations
+- fix(analyst): reject surrounding whitespace in caller gold ids
+- test(analyst): reproduce whitespace source-id bypass
+- test(optimizer): retire fake CLI metering harness
+- test(analyst): isolate recorded gold admission scenarios
+- fix(systemone): validate decision boundaries before routing
+- ci: pin the eval publication dispatcher body
+- fix(analyst): reserve excluded source ids across caller gold
+- fix(analyst): validate root provenance before gold admission
+- fix(analyst): preserve actor boundaries in AgentRx calibration gold
+- Preserve manual drafts observed while recording failed adoption (#1076)
+- Recover updater-owned failure drafts while preserving manual holds (#1075)
+- feat(systemone): persist calibrated decision points and outcomes
+- ci: move eval verification and Python publishing to owned runners
+- chore(deps): bump opentelemetry-api in /clients/python (#1068)
+- fix(adopt): quote optional workflow flags
+- fix(adopt): preserve shallow history and caller-owned dependency references
+- fix(adopt): reconcile effective workspace pins and preserve proposal history
+
+## [0.221.2] — 2026-10-24
+
+- Return plain analyst findings from decision replay
+- chore(release): 0.221.1 (#1091)
+- fix(deps): pin zod 4.6.6 (GHSA-mrhx-c7xc-m3w5) (#1088)
+- Give the disposable adoption fixture its own Git author
+- Fix real adoption drift fixture under the declared package manager
+- Capture supporting trace events in the original native decision
+- test: inherit the integration default for behavioral questions
+- test: set the integration default from the observed loaded tail
+- chore(release): 0.221.0
+- test: own the measured integration deadline in Vitest config
+- test: await observed integration completion across tool paths
+- feat: preserve finding composition and assigned arm populations
+- test: await decision-point HTTP and ledger completion
+- test: await durable readiness in filesystem integration cases
+- test(process): wait for argv child readiness before abort
+- feat(analyst): add bounded calibrated trace question observations
+- fix(analyst): reject surrounding whitespace in caller gold ids
+- test(analyst): reproduce whitespace source-id bypass
+- test(optimizer): retire fake CLI metering harness
+- test(analyst): isolate recorded gold admission scenarios
+- fix(systemone): validate decision boundaries before routing
+- ci: pin the eval publication dispatcher body
+- fix(analyst): reserve excluded source ids across caller gold
+- fix(analyst): validate root provenance before gold admission
+- fix(analyst): preserve actor boundaries in AgentRx calibration gold
+- Preserve manual drafts observed while recording failed adoption (#1076)
+- Recover updater-owned failure drafts while preserving manual holds (#1075)
+- feat(systemone): persist calibrated decision points and outcomes
+- ci: move eval verification and Python publishing to owned runners
+- chore(deps): bump opentelemetry-api in /clients/python (#1068)
+- fix(adopt): quote optional workflow flags
+- fix(adopt): preserve shallow history and caller-owned dependency references
+- fix(adopt): reconcile effective workspace pins and preserve proposal history
+
 ## [0.221.1] — 2026-10-23
 
 - fix(deps): pin zod 4.6.6 (GHSA-mrhx-c7xc-m3w5) (#1088)

package/dist/analyst/index.js

@@ -6,12 +6,12 @@ import { A as spanningWindow, C as UNKNOWN_LINEAGE, D as mostSevere, E as mergeE
 import { a as makeProposalFinding, i as makeFinding, r as computeFindingId } from "../types-CnWmj8NW.js";
 import { t as assertProposalFindings } from "../proposal-findings-D0_zkQ2l.js";
 import { A as RawAnalystFindingSchema, B as parseFindingSubject, F as stripCodeFences, I as FINDING_SUBJECT_KINDS, L as FINDING_SUBJECT_SYNTAX, M as parseRawFinding, N as parseTraceSpanEvidenceUri, R as KIND_EXPECTED_SUBJECTS, V as renderFindingSubject, j as evidenceRefsFromRawFinding, k as RAW_FINDING_SCHEMA_PROMPT, n as runTraceAnalyst, r as buildTraceToolsForGroup, t as createTraceAnalyst, z as findingSubjectGrammarPromptFor } from "../kind-factory-Dl0PmNQL.js";
-import { _ as behavioralTraceQuestionDigest, a as KNOWLEDGE_POISONING_KIND_SPEC, c as FAILURE_MODE_KIND_SPEC, d as emitControlIntegrityFindings, f as behavioralAnalyst, g as BEHAVIORAL_TRACE_QUESTIONS_VERSION, h as BEHAVIORAL_TRACE_QUESTIONS, i as DEFAULT_TRACE_ANALYST_KINDS, l as CONTROL_INTEGRITY_ANALYST, m as behavioralTraceSupportSchema, n as AnalystRegistry, o as KNOWLEDGE_GAP_KIND_SPEC, p as validBehavioralCalibration, r as ExactAnalystRunExecutionError, s as IMPROVEMENT_KIND_SPEC, t as buildDefaultAnalystRegistry, u as ControlIntegrityAnalyst } from "../default-registry-BbFCkqvB.js";
+import { _ as behavioralTraceQuestionDigest, a as KNOWLEDGE_POISONING_KIND_SPEC, c as FAILURE_MODE_KIND_SPEC, d as emitControlIntegrityFindings, f as behavioralAnalyst, g as BEHAVIORAL_TRACE_QUESTIONS_VERSION, h as BEHAVIORAL_TRACE_QUESTIONS, i as DEFAULT_TRACE_ANALYST_KINDS, l as CONTROL_INTEGRITY_ANALYST, m as behavioralTraceSupportSchema, n as AnalystRegistry, o as KNOWLEDGE_GAP_KIND_SPEC, p as validBehavioralCalibration, r as ExactAnalystRunExecutionError, s as IMPROVEMENT_KIND_SPEC, t as buildDefaultAnalystRegistry, u as ControlIntegrityAnalyst } from "../default-registry-BZ5FyYl9.js";
 import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-BTy7u0jQ.js";
 import { n as diffFindings, r as createChatTraceEngine, t as FindingsStore } from "../findings-store-DRiFdoid.js";
-import { a as systemOneAnalyst, p as validDecisionIdentity } from "../decision-point-Co3u3W8k.js";
+import { a as systemOneAnalyst, p as validDecisionIdentity } from "../decision-point-0nJubuoM.js";
 import { t as createChatClient } from "../chat-client-CMQ8_h4J.js";
-import { A as agentRxBenchmarkCase, C as ANALYST_BENCHMARK_COST_LEDGER_FILE, D as AGENT_RX_UPSTREAM_REVISION, E as ANALYST_BENCHMARK_OBSERVATIONS_FILE, F as runAnalystBenchmark, I as traceStoreEvidenceResolver, M as normalizeAgentRxCategory, N as externalId, O as codeTraceBenchCase, P as positiveStep, S as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, T as ANALYST_BENCHMARK_MANIFEST_FILE, _ as readAnalystBenchmarkArtifact, a as runInlineAnalystDefinition, b as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, c as publicRlmAnalystDefinition, d as publicDirectAnalystDefinition, f as runChunkedAnalystDefinition, g as expandCodeTraceFailureBlocks, h as analystDefinitionProtocolSha256, i as primeCodeTraceAnalystDefinition, j as agentRxPredictionsToFindings, k as codeTracerPredictionsToFindings, l as runReplVariableAnalystDefinition, m as analystDefinitionAsymmetries, n as createPrimeBenchmarkRunner, o as loadPublicBenchmarkRows, p as AnalystExpressivenessError, r as primeAnalystProtocolSha256, s as createPublicBenchmarkRlmRunner, t as runAnalystBenchmarkCommand, u as createPublicBenchmarkDirectRunner, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, x as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, y as publicBenchmarkProtocolSha256 } from "../benchmark-command-DiGqnGUt.js";
+import { A as agentRxBenchmarkCase, C as ANALYST_BENCHMARK_COST_LEDGER_FILE, D as AGENT_RX_UPSTREAM_REVISION, E as ANALYST_BENCHMARK_OBSERVATIONS_FILE, F as runAnalystBenchmark, I as traceStoreEvidenceResolver, M as normalizeAgentRxCategory, N as externalId, O as codeTraceBenchCase, P as positiveStep, S as ANALYST_BENCHMARK_IMPLEMENTATION_FILES, T as ANALYST_BENCHMARK_MANIFEST_FILE, _ as readAnalystBenchmarkArtifact, a as runInlineAnalystDefinition, b as ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, c as publicRlmAnalystDefinition, d as publicDirectAnalystDefinition, f as runChunkedAnalystDefinition, g as expandCodeTraceFailureBlocks, h as analystDefinitionProtocolSha256, i as primeCodeTraceAnalystDefinition, j as agentRxPredictionsToFindings, k as codeTracerPredictionsToFindings, l as runReplVariableAnalystDefinition, m as analystDefinitionAsymmetries, n as createPrimeBenchmarkRunner, o as loadPublicBenchmarkRows, p as AnalystExpressivenessError, r as primeAnalystProtocolSha256, s as createPublicBenchmarkRlmRunner, t as runAnalystBenchmarkCommand, u as createPublicBenchmarkDirectRunner, v as CODE_TRACE_BENCH_ANALYST_PROMPT, w as ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, x as ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, y as publicBenchmarkProtocolSha256 } from "../benchmark-command-BDq0VgoP.js";
 import { a as normalizePrimeUsage, c as runPrimeExchange, i as mergePrimeRawUsage, n as buildPrimePrompt, o as primeProtocolSha256, r as buildPrimeRepairPrompt, s as projectPrimeTrajectory, t as analystUsageReceiptFromPrimeUsage } from "../prime-protocol-BFgGrEMP.js";
 import { t as nodeHttpPrimeBridgeTransport } from "../prime-bridge-transport-Z3cpKGoV.js";
 import { t as clusterBootstrapMean } from "../cluster-bootstrap-DEwwhl0C.js";

package/dist/{benchmark-command-DiGqnGUt.js → benchmark-command-BDq0VgoP.js}

@@ -13,7 +13,7 @@ import { a as LlmCallError, n as isRecord, o as LlmResponseError } from "./json-
 import { A as RawAnalystFindingSchema, U as assertValidAnalystUsageReceipt, W as usageReceiptFromCostLedger, j as evidenceRefsFromRawFinding, k as RAW_FINDING_SCHEMA_PROMPT, m as TRACE_ANALYSIS_LIMITS, n as runTraceAnalyst } from "./kind-factory-Dl0PmNQL.js";
 import { n as callLlmJson } from "./llm-client-DXuOTseU.js";
 import { t as createDspyRlmTraceEngine } from "./dspy-rlm-engine-BTy7u0jQ.js";
-import { i as otlpTextToTraceAnalysisStore, n as DEFAULT_MAX_TRACE_FILE_BYTES, t as BufferTraceStore } from "./store-otlp-BVZfR168.js";
+import { i as otlpTextToTraceAnalysisStore, n as DEFAULT_MAX_TRACE_FILE_BYTES, t as BufferTraceStore } from "./store-otlp-CFf27E8_.js";
 import "./external-optimizer-process-E5yY2SYS.js";
 import { t as assertEqualDeclarativeTerms } from "./equal-terms-DbPGcVtI.js";
 import { c as runPrimeExchange, l as decodeReplyRows, n as buildPrimePrompt, o as primeProtocolSha256, s as projectPrimeTrajectory, t as analystUsageReceiptFromPrimeUsage } from "./prime-protocol-BFgGrEMP.js";
@@ -2031,7 +2031,7 @@ const ANALYST_BENCHMARK_IMPLEMENTATION_FILES = Object.freeze([
 	"src/trace/redaction/detectors.ts",
 	"src/verdict-cache.ts"
 ]);
-const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "eea1626b12169ab6305bf535eb19e74312d358fc31ad27f06a416e4832917437";
+const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "b53310dc335856788568fd97afa30f48448098d03fa7ef4504e0555bdfa3481c";
 //#endregion
 //#region src/analyst/benchmark-evidence-validation.ts
 const MIN_ACTION_EXCERPT_CHARACTERS = 12;
@@ -6892,4 +6892,4 @@ function shellQuote(value) {
 //#endregion
 export { agentRxBenchmarkCase as A, ANALYST_BENCHMARK_COST_LEDGER_FILE as C, AGENT_RX_UPSTREAM_REVISION as D, ANALYST_BENCHMARK_OBSERVATIONS_FILE as E, runAnalystBenchmark as F, traceStoreEvidenceResolver as I, normalizeAgentRxCategory as M, externalId as N, codeTraceBenchCase as O, positiveStep as P, ANALYST_BENCHMARK_IMPLEMENTATION_FILES as S, ANALYST_BENCHMARK_MANIFEST_FILE as T, readAnalystBenchmarkArtifact as _, runInlineAnalystDefinition as a, ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 as b, publicRlmAnalystDefinition as c, publicDirectAnalystDefinition as d, runChunkedAnalystDefinition as f, expandCodeTraceFailureBlocks as g, analystDefinitionProtocolSha256 as h, primeCodeTraceAnalystDefinition as i, agentRxPredictionsToFindings as j, codeTracerPredictionsToFindings as k, runReplVariableAnalystDefinition as l, analystDefinitionAsymmetries as m, createPrimeBenchmarkRunner as n, loadPublicBenchmarkRows as o, AnalystExpressivenessError as p, primeAnalystProtocolSha256 as r, createPublicBenchmarkRlmRunner as s, runAnalystBenchmarkCommand as t, createPublicBenchmarkDirectRunner as u, CODE_TRACE_BENCH_ANALYST_PROMPT as v, ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE as w, ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 as x, publicBenchmarkProtocolSha256 as y };
 
-//# sourceMappingURL=benchmark-command-DiGqnGUt.js.map
+//# sourceMappingURL=benchmark-command-BDq0VgoP.js.map