@tangle-network/agent-eval 0.223.0 → 1.0.0

package/CHANGELOG.md

@@ -4,6 +4,106 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [1.0.0] — 2026-10-26
+
+**1.0.0 ships the 0.223.0 API unchanged.** From here a breaking change is a major release, and a caret range (`^1.0.0`) takes every later compatible release, so a consumer no longer needs a release of its own for each minor of this package. The dependencies move onto the 1.0 releases of agent-core and harness-sessions. The commits below are everything since the v0.220.2 tag.
+
+- test(adopt-stack): explain a refusal by any peer whose latest is an excluded major; post-pin cases use a manifest with no 0.x peer
+- test(adopt-stack): the live-registry plans hold across a package's first 1.0 release
+- deps: repin the analyst benchmark's dependency-lock digest
+- deps: lock agent-core 1.0.0, harness-sessions 1.0.1, agent-interface 5.8.1 and agent-trace-contract 2.2.0
+- deps: floor agent-interface at 5.8.1 and agent-trace-contract at 2.2.0
+- deps: agent-core ^1.0.0 and harness-sessions ^1.0.1
+- chore(release): 0.223.0 (#1140)
+- feat(storyboard): one tool classifier, classifyTool, shared by scenes and extractVisual
+- refactor: use statistics mean helpers instead of private copies (#1137)
+- fix(adopt-stack): restart a branch that outlived another's squash merge (#1136)
+- fix(analyst): repin the benchmark dependency lock after the zod range (#1135)
+- chore(release): 0.222.1 (#1134)
+- fix(deps): depend on zod ^4.6.6 so consumers resolve one copy (#1133)
+- chore(test): retire incomplete Runtime adoption snapshot
+- fix(release): build public exports before integration tests
+- chore(release): 0.222.0
+- feat(campaign): share concurrency for measured scenario adapters
+- fix(supervisor-run): staged play-check reward is stages passed, read from the SCORE line only (#1126)
+- refactor(statistics): delete duplicate numeric summary helpers
+- fix(rollout): import survives a damaged deliverables archive; named ids are not keys (#1125)
+- feat(supervisor-run): rollouts importer for Discovery runs, harness sessions and gate receipts (#1124)
+- feat(statistics): extend series summaries with nearest-rank evidence
+- test(rollout): isolate bundle permission probe from runner identity
+- fix(rollout): license class is provenance and an opt-in filter, not a default exclusion (#1121)
+- fix(rollout): allow nonroot images to read the base bundle
+- feat(rollout): license class on rollout lines and whole-row privacy drop in rollout-release (#1119)
+- test(rollout): inherit configured identity for Git fixtures
+- test(rollout): capture empty tool definitions at the ledger boundary
+- feat(rollout): write verified episodes as Harbor tasks
+- feat(rollout): project captured policy steps into Tinker data
+- adopt-stack: delete the adoption branch when its squash merge lands (#1115)
+- chore(deps)!: require Python 3.11 for patched locks and refresh dependencies
+- feat(rollout): read retained Codex sessions through the shared reader
+- chore(release): 0.221.8
+- test(adopt): retain failed browser child outcomes
+- docs: leave changelog preparation to the release workflow
+- chore(release): 0.221.7
+- feat(multishot): conversation legs report the USD they spent
+- chore(release): 0.221.6
+- docs(traces): bound the incomplete usage coverage claim
+- feat: own judge-panel calibration, multishot conversation legs and served-panel identity
+- test(traces): typecheck OTLP usage integration proof
+- fix(traces): retain incomplete OTLP token usage
+- chore(release): 0.221.5
+- ci: retain private adopter failure evidence (#1106)
+- test(process): isolate process-tree fixture environment
+- fix(adopt): invoke each resolved Playwright CLI (#1096)
+- test(systemone): close disposable provider connections
+- chore(release): 0.221.4
+- test(adoption): inherit the maintained integration deadline
+- fix(adoption): scope guard allowance to disposable Git fixtures
+- fix(analyst): carry root-cause target into failure-mode rubric
+- ci: check locally and in publication paths
+- test(adopt): give real-git fixtures their own commit identity (#1101)
+- chore(release): 0.221.3 (#1099)
+- 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
+
 ## [0.223.0] — 2026-10-26
 
 - feat(storyboard): one tool classifier, classifyTool, shared by scenes and extractVisual

package/dist/analyst/index.js

@@ -11,7 +11,7 @@ import { t as createDspyRlmTraceEngine } from "../dspy-rlm-engine-jDiWkul6.js";
 import { n as diffFindings, r as createChatTraceEngine, t as FindingsStore } from "../findings-store-B75Dspaq.js";
 import { a as systemOneAnalyst, p as validDecisionIdentity } from "../decision-point-lhMgODOR.js";
 import { t as createChatClient } from "../chat-client-CO3uB33i.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-CH86twZj.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-DWkc5mzq.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-CrhpXRz9.js";
 import { t as nodeHttpPrimeBridgeTransport } from "../prime-bridge-transport-Csd5SQg7.js";
 import { t as clusterBootstrapMean } from "../cluster-bootstrap-SbaeM44s.js";

package/dist/{benchmark-command-CH86twZj.js → benchmark-command-DWkc5mzq.js}

@@ -1863,7 +1863,7 @@ Object.freeze([
 	"package.json",
 	"pnpm-lock.yaml"
 ]);
-const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "42c7216e1c748e20ec3c64a79f410a587064caf2cac74c49ed457223dc21d921";
+const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "52fe1dfdf8e688008c0f800f731bc3c26c416d211a0438d00d9b6e75d45790e1";
 const ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 = "1e03f2daed356d60316aabefb407ec1e437ac94d408d61eea4ae096e9c6fbb5b";
 const ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 = "4dba263b6256a30d56c7fdb2d992d3a953c0035d731f359b704db806f68f75ac";
 const ANALYST_BENCHMARK_IMPLEMENTATION_FILES = Object.freeze([
@@ -6854,4 +6854,4 @@ function shellQuote(value) {
 }
 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-CH86twZj.js.map
+//# sourceMappingURL=benchmark-command-DWkc5mzq.js.map