@tangle-network/agent-eval 0.135.3 → 0.136.0

package/CHANGELOG.md

@@ -4,6 +4,33 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [0.136.0] - 2026-07-29 - preserve recursive evidence and complete profile changes
+
+### Fixed
+
+- The npm package pins `@tangle-network/agent-core` 0.4.28 and `@tangle-network/agent-interface` 0.39.0 as one compatible cohort.
+  Profile-improvement experiments validate every existing `AgentProfileDiff` axis with the same schema that constructs those diffs.
+- Recursive supervisor-run reports now join worker artifacts by optional stable `workerId`, falling back to `label` only for older stores.
+  Retry and reaction counts are computed within each parent, so identical labels and settlements in separate branches no longer collide.
+  Per-worker and steer rows expose the joined `workerId`; custom report consumers should read it instead of treating the display label as identity.
+- Spawned invocations retain an explicit `supervisor` or `worker` role, and structured verdicts retain their numeric score in rollout rewards and per-worker report rows.
+- Accepted-patch counts are unavailable when the source did not retain worker deliverables, even when a worker event claimed patch bytes.
+- Manager and worker token totals have independent unavailable reasons, so an uncaptured channel is not reported as zero and a captured zero remains zero.
+
+### Changed — BREAKING
+
+- Custom `SupervisorRunSources` readers must add `managerTokens` and `workerTokens` to `SourceLimits`.
+  Set each field to `null` only when that role's aggregate token channel is complete; otherwise set the reason it is unavailable.
+  Readers with no source limitations can continue to use `NO_SOURCE_LIMITS`.
+
+## [0.135.4] - 2026-07-29 - keep rich source evidence in one schema cohort
+
+### Fixed
+
+- The npm package pins `@tangle-network/agent-core` 0.4.26 and `@tangle-network/agent-interface` 0.37.0 as one compatible cohort.
+  Candidate and profile-improvement contracts accept and retain licensed, attributed, noticed, and transformed public-source evidence without installing an older nested Interface schema.
+- Packed-package verification installs the release archive into a fresh npm consumer and refuses duplicate or incorrect Core and Interface versions.
+
 ## [0.135.3] - 2026-07-29 - preserve real tool evidence for trace learning
 
 ### Added

package/dist/{index-C61Wi7yg.d.ts → index-CyC1BTmn.d.ts}

@@ -10,8 +10,16 @@ declare function unavailable(reason: string): Unavailable;
 declare function isUnavailable(v: unknown): v is Unavailable;
 /** Render a measured scalar for the markdown/headline: `0` and `unavailable` stay distinct. */
 declare function showMeasured(v: Measured<number | string | boolean | null>): string;
+/** The two invocation roles a recursive supervision tree can contain. */
+type SupervisorRunNodeRole = 'supervisor' | 'worker';
 /** One worker's logs, as read. `null` = the artifact did not exist. */
 interface WorkerLogSource {
+  /**
+   * Stable journal node id. Readers should set this whenever their source has
+   * one; `label` remains the compatibility join for older stores.
+   */
+  readonly workerId?: string;
+  /** Human-readable task label. It is not required to be unique. */
   readonly label: string;
   /** Worker event stream — started / progress / finished / message events (JSONL). */
   readonly events: string | null;
@@ -41,6 +49,10 @@ interface WorkerLogSource {
  * `null` on a field means the source DOES carry that fact.
  */
 interface SourceLimits {
+  /** Reason manager input/output token totals are unavailable (null = recorded). */
+  readonly managerTokens: string | null;
+  /** Reason worker input/output token totals are unavailable (null = recorded). */
+  readonly workerTokens: string | null;
   /** Reason inference spend has no price in this store (null = the store prices it). */
   readonly spendUsd: string | null;
   /** Reason workers carry no pass/fail verdict (null = verdicts are recorded). */
@@ -67,7 +79,11 @@ interface SupervisorRunSources {
   readonly arm: string | null;
   /** Identity of the supervision-tree store this was read from; null = none found. */
   readonly supRunDir: string | null;
-  /** Supervision journal — spawned / settled / cancelled / metered events (JSONL). */
+  /**
+   * Supervision journal — spawned / settled / cancelled / metered events (JSONL).
+   * Recursive readers put `role: 'supervisor' | 'worker'` on spawned rows;
+   * settled `verdict` may be a legacy string or `{ valid, score, ... }`.
+   */
   readonly journal: string | null;
   /** Per-brain-call tap (JSONL): finish_reason, completion tokens, requested max tokens. */
   readonly brainLog: string | null;
@@ -134,6 +150,8 @@ interface SupervisorRunReader {
 declare const SUPERVISOR_RUN_SCHEMA = "tangle.supervisor-run@1";
 declare const SUPERVISOR_RUN_ROLLUP_SCHEMA = "tangle.supervisor-run-rollup@1";
 interface SteerBreakdown {
+  /** Stable journal node id when the reader retained one. */
+  readonly workerId: string | null;
   readonly worker: string;
   /** Steer requests durably queued to this worker's inbox. */
   readonly queued: number;
@@ -158,9 +176,9 @@ interface OrchestrationMetrics {
   readonly waves: Measured<number>;
   readonly waveSizes: Measured<readonly number[]>;
   readonly maxConcurrency: Measured<number>;
-  /** Worker spawns issued after the first settlement — the retry/respawn tail. */
+  /** Direct-child spawns issued after that parent's first direct-child settlement. */
   readonly respawns: Measured<number>;
-  /** Labels spawned more than once (a literal retry of the same subtask). */
+  /** Labels spawned more than once by the same parent. */
   readonly repeatedLabels: Measured<readonly string[]>;
   /** Longest parent chain below the root, in worker hops. */
   readonly delegationDepth: Measured<number>;
@@ -181,9 +199,9 @@ interface DecisionMetrics {
   readonly rejected: Measured<number>;
   /** Worker verified green but delivered no patch bytes — output with nothing to accept. */
   readonly emptyPass: Measured<number>;
-  /** Settlements the brain observed before issuing its next spawn (evidence→respawn). */
+  /** Direct-child settlements a parent observed before issuing its next direct-child spawn. */
   readonly observeThenRespawn: Measured<number>;
-  /** Respawns with no settled evidence in front of them. */
+  /** Parent-local respawns with no direct-child settlement in front of them. */
   readonly respawnWithoutEvidence: Measured<number>;
   /** Steer + question traffic on the live down/up legs — the only "review while running" signal. */
   readonly reviewActions: Measured<number>;
@@ -203,7 +221,11 @@ interface RoleSpend {
   readonly source: string;
 }
 interface PerWorkerRow {
+  /** Stable journal node id when the reader retained one. */
+  readonly workerId: string | null;
   readonly worker: string;
+  /** Explicit journal role after the worker source was joined to its spawn. */
+  readonly role: SupervisorRunNodeRole | null;
   readonly wallMs: number | null;
   /** `null` = this store does not attribute tokens per worker (NOT "zero tokens"). */
   readonly tokensIn: number | null;
@@ -211,6 +233,8 @@ interface PerWorkerRow {
   readonly usd: number | null;
   readonly patchBytes: number | null;
   readonly passed: boolean | null;
+  /** Numeric verdict score exactly as recorded; null means no score was recorded. */
+  readonly score: number | null;
 }
 interface WallDistribution {
   readonly n: number;
@@ -310,8 +334,8 @@ interface SupervisorRunRollup {
 /**
  * A supervision tree expressed in the canonical rollout row type: one
  * `RolloutLine` per invocation, joined by `parent_rollout_id`. The root row
- * carries `role: 'supervisor'`; every spawned worker carries `role: 'worker'`
- * with the root as its parent.
+ * carries `role: 'supervisor'`; nested supervisors retain that role and leaf
+ * invocations carry `role: 'worker'`.
  */
 interface SupervisorRunTree {
   readonly rootId: string | null;
@@ -337,13 +361,21 @@ interface SpawnRow {
   id: string;
   parent: string | null;
   label: string;
+  role: SupervisorRunNodeRole;
   at: number | null;
 }
 interface CloseRow {
   id: string;
   kind: 'settled' | 'cancelled';
   status: string | null;
+  /** String verdict from legacy journals, or valid/invalid for a structured verdict. */
   verdict: string | null;
+  /** Structured verdict validity, when recorded. */
+  valid: boolean | null;
+  /** Structured verdict score, preserved without boolean coercion. */
+  score: number | null;
+  /** The verdict exactly as the journal carried it. */
+  rawVerdict: unknown | null;
   at: number | null;
   spend: SpendLike;
   /** False when the close event carried no spend object — not "spent nothing". */
@@ -355,6 +387,8 @@ interface WorkerLogFacts {
   finished: boolean;
   finishedAt: number | null;
   passed: boolean | null;
+  /** Numeric score exactly as the finished event recorded it. */
+  score: number | null;
   /** `patchBytes` as reported by the finished event (not the patch file's size). */
   finishedPatchBytes: number | null;
   evidenceBytes: number;
@@ -543,5 +577,5 @@ interface SupervisorRolloutOptions {
  */
 declare function supervisorRunRolloutLines(src: SupervisorRunSources, opts?: SupervisorRolloutOptions): SupervisorRunTree;
 //#endregion
-export { isUnavailable as $, rollupSupervisorRuns as A, RollupCellRow as B, CloseRow as C, analyzeSupervisorRunSources as D, WorkerLogFacts as E, OrchestrationMetrics as F, SupervisorRunReader as G, SUPERVISOR_RUN_SCHEMA as H, OutcomeMetrics as I, SupervisorRunSources as J, SupervisorRunReport as K, PatchStats as L, EconomicsMetrics as M, Measured as N, parsePatch as O, NO_SOURCE_LIMITS as P, WorkerLogSource as Q, PerWorkerRow as R, readClaudeCodeSupervisorRun as S, SupervisorTreeFacts as T, SourceLimits as U, SUPERVISOR_RUN_ROLLUP_SCHEMA as V, SteerBreakdown as W, Unavailable as X, SupervisorRunTree as Y, WallDistribution as Z, ClaudeCodeReaderOptions as _, renderSupervisorRunMarkdown as a, DEFAULT_STEER_TOOLS as b, analyzeSupervisorRun as c, loopsSupervisorRunReader as d, showMeasured as et, readLoopsSupervisorRun as f, writeSupervisorRunReportSafe as g, writeSupervisorRunReport as h, renderSupervisorRunHeadline as i, DecisionMetrics as j, parseSupervisorTree as k, findSupervisorRunDirIn as l, supervisorReportStem as m, supervisorRunRolloutLines as n, LoopsReaderOptions as o, reportSupervisorRound as p, SupervisorRunRollup as q, renderSupervisorRollupMarkdown as r, WriteSupervisorRunOptions as s, SupervisorRolloutOptions as t, unavailable as tt, findSupervisorRunDirs as u, DEFAULT_CANCEL_TOOLS as v, SpawnRow as w, claudeCodeSupervisorRunReader as x, DEFAULT_SPAWN_TOOLS as y, RoleSpend as z };
-//# sourceMappingURL=index-C61Wi7yg.d.ts.map
+export { WorkerLogSource as $, rollupSupervisorRuns as A, RollupCellRow as B, CloseRow as C, analyzeSupervisorRunSources as D, WorkerLogFacts as E, OrchestrationMetrics as F, SupervisorRunNodeRole as G, SUPERVISOR_RUN_SCHEMA as H, OutcomeMetrics as I, SupervisorRunRollup as J, SupervisorRunReader as K, PatchStats as L, EconomicsMetrics as M, Measured as N, parsePatch as O, NO_SOURCE_LIMITS as P, WallDistribution as Q, PerWorkerRow as R, readClaudeCodeSupervisorRun as S, SupervisorTreeFacts as T, SourceLimits as U, SUPERVISOR_RUN_ROLLUP_SCHEMA as V, SteerBreakdown as W, SupervisorRunTree as X, SupervisorRunSources as Y, Unavailable as Z, ClaudeCodeReaderOptions as _, renderSupervisorRunMarkdown as a, DEFAULT_STEER_TOOLS as b, analyzeSupervisorRun as c, loopsSupervisorRunReader as d, isUnavailable as et, readLoopsSupervisorRun as f, writeSupervisorRunReportSafe as g, writeSupervisorRunReport as h, renderSupervisorRunHeadline as i, DecisionMetrics as j, parseSupervisorTree as k, findSupervisorRunDirIn as l, supervisorReportStem as m, supervisorRunRolloutLines as n, unavailable as nt, LoopsReaderOptions as o, reportSupervisorRound as p, SupervisorRunReport as q, renderSupervisorRollupMarkdown as r, WriteSupervisorRunOptions as s, SupervisorRolloutOptions as t, showMeasured as tt, findSupervisorRunDirs as u, DEFAULT_CANCEL_TOOLS as v, SpawnRow as w, claudeCodeSupervisorRunReader as x, DEFAULT_SPAWN_TOOLS as y, RoleSpend as z };
+//# sourceMappingURL=index-CyC1BTmn.d.ts.map

package/dist/index-CyC1BTmn.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index-CyC1BTmn.d.ts","names":[],"sources":["../src/supervisor-run/types.ts","../src/supervisor-run/analyze.ts","../src/supervisor-run/claude-code-reader.ts","../src/supervisor-run/loops-reader.ts","../src/supervisor-run/render.ts","../src/supervisor-run/rollout-nodes.ts"],"mappings":";;;UAiCiB;WACN;;;KAIC,SAAS,KAAK,IAAI;iBAEd,YAAY,iBAAiB;iBAI7B,cAAc,aAAa,KAAK;;iBAKhC,aAAa,GAAG;;KAWpB;;UAGK;;;;;WAKN;;WAEA;;WAEA;;WAEA;;WAEA;;WAEA;;WAEA;;WAEA;WACA;WACA;WACA;;;;;;;;;;;;;UAcM;;WAEN;;WAEA;;WAEA;;WAEA;;WAEA;;;cAIE,kBAAkB;;;;;;;;;;UAiBd;;WAEN;WACA;;WAEA;;WAEA;;;;;;WAMA;;WAEA;;WAEA;;WAEA;;WAEA,kBAAkB;;WAElB;;WAEA;;;;;;;WAOA;;WAEA;;WAEA;;WAEA;;;;;WAKA;IACP;IACA;IACA;IACA;;IAEA;IACA;;WAEO;;WAEA,QAAQ;;;;;;WAMR;;;;;WAKA;;;;;;UAOM;;WAEN;EACT,QAAQ,QAAQ;;cAOL;cACA;UAEI;;WAEN;WACA;;WAEA;;WAEA;;UAGM;WACN,gBAAgB;WAChB,gBAAgB;WAChB,kBAAkB;;WAElB,QAAQ;WACR,iBAAiB;WACjB,gBAAgB,kBAAkB;;WAElC,kBAAkB;;;;;;WAMlB,OAAO;WACP,WAAW;WACX,gBAAgB;;WAEhB,UAAU;;WAEV,gBAAgB;;WAEhB,iBAAiB;WACjB,oBAAoB;WACpB,kBAAkB;;WAElB,QAAQ;WACR,SAAS;;WAET,mBAAmB;;UAGb;WACN,iBAAiB,SAAS;WAC1B,iBAAiB,SAAS;;WAE1B,UAAU;;WAEV,UAAU;;WAEV,WAAW;;WAEX,oBAAoB;;WAEpB,wBAAwB;;WAExB,eAAe;WACf,qBAAqB;;UAGf;WACN,UAAU;WACV,WAAW;;;;;;WAMX,WAAW;WACX,YAAY;WACZ,KAAK;WACL;;UAGM;;WAEN;WACA;;WAEA,MAAM;WACN;;WAEA;WACA;WACA;WACA;WACA;;WAEA;;UAGM;WACN;WACA;WACA;WACA;WACA;WACA;;UAGM;;WAEN,OAAO;;;;;;;;WAQP,kBAAkB;;WAElB,SAAS;WACT,UAAU;;;;;;WAMV;WACA,yBAAyB;WACzB,0BAA0B,SAAS;WACnC,WAAW,kBAAkB;;UAGvB;WACN;WACA;WACA;WACA;;UAGM;WACN,WAAW;WACX,YAAY;WACZ,WAAW;WACX,eAAe;WACf,YAAY;WACZ,aAAa;WACb,YAAY;WACZ,YAAY;WACZ,UAAU;WACV,OAAO,SAAS;;WAEhB;;UAGM;WACN,eAAe;;WAEf;WACA;WACA;WACA,cAAc;WACd;WACA,eAAe;WACf,UAAU;WACV,WAAW;WACX,SAAS;;WAET;;WAEA;;UAGM;WACN;WACA;WACA,QAAQ;WACR,OAAO;WACP,aAAa;WACb,SAAS;WACT,UAAU;WACV,KAAK;;UAGC;WACN,eAAe;WACf;WACA,aAAa;WACb,iBAAiB;WACjB;WACA,WAAW;WACX,mBAAmB;WACnB,iBAAiB;WACjB,aAAa;WACb,qBAAqB;WACrB,eAAe;WACf,UAAU;WACV,eAAe;WACf,kBAAkB;;;;;;;;UASZ;WACN;WACA,gBAAgB;;WAEhB;;;;UCpXD;EACR;EACA;EACA;EACA;;EAEA;;UAGQ;EACR,QAAQ;EACR;;UAoEe;EACf;EACA;EACA;EACA,MAAM;EACN;;UAGe;EACf;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;EACA;EACA,OAAO;;EAEP;;UAGe;EACf;;EAEA;EACA;EACA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;;;;;;UAOe;WACN;WACA,iBAAiB;WACjB,iBAAiB;WACjB,uBAAuB;WACvB,uBAAuB;WACvB;IACP;IACA;IACA;IACA;;IAEA;IACA;IACA;;WAEO,YAAY,oBAAoB;WAChC;WACA;;iBA6BK,oBAAoB,KAAK,uBAAuB;;;;;iBAoJhD,4BACd,KAAK,sBACL,qBACC;;iBAqmBa,WAAW,eAAe;;;;;;iBA+C1B,qBAAqB,kBAAkB,wBAAwB;;;;cCj8BlE;;cAEA;;cAEA;UASI;;WAEN;;;;;WAKA;WACA;WACA;WACA;WACA;WACA;WACA;;;;;;iBAmOW,4BACpB,MAAM,0BACL,QAAQ;;iBAiUK,8BAA8B,MAAM,0BAA0B;;;;iBC5lBxD,uBAAuB,aAAa;UAOzC;;WAEN;;WAEA;;WAEA;;WAEA;;;;;;;iBAQW,uBACpB,gBACA,OAAM,qBACL,QAAQ;;iBAmKK,yBACd,gBACA,OAAM,qBACL;;;;;;iBAkCmB,qBACpB,gBAAgB,sBAAsB,sBACtC,OAAM,qBACL,QAAQ;UAYM,kCAAkC;;WAExC;;WAEA;;;;;;WAMA;;;;;;;;iBASW,yBACpB,gBACA,OAAM,4BACL,QAAQ;;;;;;;iBAuBK,qBAAqB;;;;;iBAiBf,6BACpB,gBACA,OAAM,4BACL,QAAQ;;;;;;iBAgBW,sBACpB,gBACA,OAAM;EAA8B;IACnC,QAAQ;;iBA0BW,sBAAsB,eAAe;;;;;;;iBCrX3C,4BAA4B,GAAG;iBAgC/B,4BAA4B,GAAG;iBAwK/B,+BACd,QAAQ,qBACR;;;UC3Le;;WAEN;;WAEA,QAAQ;;WAER;;WAEA;;WAEA;;WAEA;;WAEA;;WAEA;;WAEA;WACA;WACA;WACA;WACA;;WAEA;;;;;;;iBAwBK,0BACd,KAAK,sBACL,OAAM,2BACL"}

package/dist/index.d.ts

@@ -35,7 +35,7 @@ import { A as ActionExecutionPolicy, C as ReviewMemoryStore, D as inMemoryReview
 import { _ as ReleaseConfidenceStatus, a as JudgeReplayGateArgs, b as assertReleaseConfidence, c as judgeReplayGate, d as ReleaseConfidenceAxis, f as ReleaseConfidenceAxisName, g as ReleaseConfidenceScorecard, h as ReleaseConfidenceMetrics, i as BootstrapResult, l as ActionableSideInfo, m as ReleaseConfidenceIssue, n as renderReleaseReport, o as Verdict, p as ReleaseConfidenceInput, r as BootstrapOptions, s as bootstrapCi, t as RenderReleaseReportOptions, u as AsiSeverity, v as ReleaseConfidenceThresholds, x as evaluateReleaseConfidence, y as ReleaseTraceEvidence } from "./release-report-DpBxGGI1.js";
 import { A as isTrainableSplit, D as assertRolloutLine, E as assertMintedLines, T as assertMinted, b as RolloutSplit, h as RolloutLine, i as ChatToolCall, j as validateRolloutLine, k as isRolloutLine, n as ChatMessage, o as MintedRolloutLine, p as RolloutCapture, s as MintedRolloutOutcome, u as ROLLOUT_SCHEMA, w as ToolDef, x as RolloutStep, y as RolloutRole } from "./schema-Cef2cFmb.js";
 import { $ as ScoreOrigin, $t as toSftRows, Ct as HarborSubagentTrajectoryRef, Dt as relabelImportedSplit, Et as fromHarborTrajectory, Gt as SftRow, Ht as RewardRow, J as MintRolloutOptions, Ot as toHarborTrajectories, Q as unmintableReasons, St as HarborStepSource, Tt as HarborTrajectory, Wt as SftExportOptions, X as RolloutScrubber, Xt as toRewardRows, Y as MintRolloutResult, Yt as toJsonl, Z as mintRolloutRows, _t as HarborImageSource, at as trainingReward, bt as HarborObservationResult, dt as ATIF_SCHEMA_VERSION, et as ScorePreference, ft as FromHarborOptions, gt as HarborFinalMetrics, ht as HarborContentPart, it as scoreOrigin, kt as toHarborTrajectory, mt as HarborAgent, nt as observedScore, ot as trainingScore, pt as HARBOR_IMPORT_GAP, rt as observedSplitScore, tt as isRealnessGated, vt as HarborMetrics, wt as HarborToolCall, xt as HarborStep, yt as HarborObservation } from "./index-AbhwHp0V.js";
-import { $ as isUnavailable, A as rollupSupervisorRuns, D as analyzeSupervisorRunSources, G as SupervisorRunReader, H as SUPERVISOR_RUN_SCHEMA, J as SupervisorRunSources, K as SupervisorRunReport, N as Measured, S as readClaudeCodeSupervisorRun, U as SourceLimits, X as Unavailable, Y as SupervisorRunTree, a as renderSupervisorRunMarkdown, c as analyzeSupervisorRun, et as showMeasured, h as writeSupervisorRunReport, i as renderSupervisorRunHeadline, n as supervisorRunRolloutLines, q as SupervisorRunRollup, x as claudeCodeSupervisorRunReader } from "./index-C61Wi7yg.js";
+import { A as rollupSupervisorRuns, D as analyzeSupervisorRunSources, G as SupervisorRunNodeRole, H as SUPERVISOR_RUN_SCHEMA, J as SupervisorRunRollup, K as SupervisorRunReader, N as Measured, S as readClaudeCodeSupervisorRun, U as SourceLimits, X as SupervisorRunTree, Y as SupervisorRunSources, Z as Unavailable, a as renderSupervisorRunMarkdown, c as analyzeSupervisorRun, et as isUnavailable, h as writeSupervisorRunReport, i as renderSupervisorRunHeadline, n as supervisorRunRolloutLines, q as SupervisorRunReport, tt as showMeasured, x as claudeCodeSupervisorRunReader } from "./index-CyC1BTmn.js";
 import { a as WelchTestResult, c as iqr, d as TrajectoryStep, f as buildTrajectory, g as computeToolUseMetrics, h as ToolUseOptions, i as MetricVerdict, l as welchsTTest, m as ToolUseMetrics, n as BaselineReport, o as WelchTestStatus, p as ToolStats, r as MetricSamples, s as compareToBaseline, t as BaselineOptions, u as Trajectory } from "./baseline-D_fT6277.js";
 import { n as SeriesConvergenceResult, r as analyzeSeries, t as SeriesConvergenceOptions } from "./series-convergence-ofsqPWhs.js";
 import { _ as EvalCampaignResult, a as FailureMode, c as SteeringChange, d as CampaignRunContext, f as CampaignRunOutcome, g as EvalCampaignOptions, h as CampaignVariant, i as ExperimentResult, l as CampaignFactoryParams, m as CampaignScenario, n as CallbackResearcherOptions, o as NoopResearcher, p as CampaignRunner, r as ExperimentPlan, s as Researcher, t as CallbackResearcher, u as CampaignIntegrityPolicy, v as FailedRun, y as runEvalCampaign } from "./researcher-Doo95b50.js";
@@ -5967,5 +5967,5 @@ type CachedJudge<TArtifact, TScenario extends Scenario$1 = Scenario$1> = JudgeCo
  */
 declare function cachedJudge<TArtifact, TScenario extends Scenario$1 = Scenario$1>(judge: JudgeConfig$1<TArtifact, TScenario>, store: VerdictCacheStore, options: CachedJudgeOptions): CachedJudge<TArtifact, TScenario>;
 //#endregion
-export { AGENT_PROFILE_KINDS, ATIF_SCHEMA_VERSION, ATTESTATION_ALGORITHM, type ActionExecutionPolicy, type ActionPolicyDecision, type ActionableSideInfo, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, type AgentEvalErrorCode, type AgentInterfaceProfileLike, type AgentProfile, type AgentProfileCell, type AgentProfileCellInput, type AgentProfileCellSchemaVersion, AgentProfileCellValidationError, type AgentProfileDimensionValue, type AgentProfileHarness, type AgentProfileJson, type AgentProfileJsonObject, type AgentProfileKind, type AgentProfileRuntimeReceipt, type AgentProfileSource, type AgentProfileSourceInput, type AlignmentOp, type Analyst, type AnalystContext, type AnalystCost, type AnalystFinding, type AnalystHooks, type AnalystInputKind, AnalystRegistry, type AnalystRegistryOptions, type AnalystRequirements, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystSeverity, type AnalystUsageReceipt, type AnalyzeTracesInput, type AnalyzeTracesOptions, type AnalyzeTracesResult, type AnalyzeTracesTurnSnapshot, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, Artifact, type ArtifactCheck, type Artifact$1 as ArtifactCheckArtifact, type ArtifactEventLike, type ArtifactResult, type ArtifactValidator, type AsiSeverity, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, BENCHMARK_SPLIT_SEED, BOOTSTRAP_GATE_MIN_N, type BackendDescriptor, BackendIntegrityError, type BackendIntegrityReport, type BaselineOptions, type BaselineReport, type BehaviorAssertion, type BehavioralMetrics, type BehavioralTokenSequence, type BenchmarkAdapter, type BenchmarkDatasetItem, type BenchmarkEvaluation, type BenchmarkFamily, type BenchmarkReport, type BenchmarkResponder, BenchmarkRunner, type BenchmarkRunnerConfig, type BenchmarkScenario, type BenchmarkSource, type BenchmarkTaskKind, type BisectOptions, type BisectResult, type BisectStep, type BlendWeights, type BootstrapOptions, type BootstrapResult, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, type BudgetPolicy, BudgetSpec, CODING_HARNESSES, type CachedJudge, type CachedJudgeOptions, type CalibrationResult, type CallExpectation, CallbackResearcher, type CallbackResearcherOptions, type CampaignFactoryParams, type CampaignIntegrityPolicy, type CampaignRunContext, type CampaignRunOutcome, type CampaignRunner, type CampaignScenario, type CampaignVariant, type CanaryAlert, type CanaryEvaluation, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateComparison, type CandidateScenario, type CandidateScore, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, CaptureFetchContext, CaptureFetchOptions, CaptureIntegrityError, type CausalAttributionReport, type CellVerdict, type ChannelRollup, type ChatCallOpts, type ChatClient, type ChatMessage, type ChatRequest, type ChatResponse, type ChatToolCall, type ChatTransport, type CheckResult, type CliBridgeTransportOpts, type CliffsMagnitude, type ClusterBootstrapInterval, type ClusterSignFlipAlternative, type ClusterSignFlipResult, type ClusteredBinaryCluster, type ClusteredMatchedPair, type ClusteredPairedBinaryOptions, type ClusteredPairedBinaryResult, type ClusteredPairedBinaryStatistics, type CollectedArtifacts, type CommandRunner, type ComparePairedArmsOptions, type CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptComplexity, type ConceptFinding, type ConceptSpec, type ConceptWeightStrategy, ConfigError, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContinuousAgreement, type ContinuousAgreementOptions, type ContinuousCalibrationResult, type ContractCheckResult, type ContractJudgeOptions, type ContractMetric, type ContractReport, type ContractRule, type ContractRuleKind, type ContractSpan, type ContractVerdict, type ContractViolation, type ControlActionFailureMode, type ControlActionOutcome, type ControlBudget, type ControlContext, type ControlDecision, type ControlEvalResult, type ControlRunResult, type ControlRunToRunRecordOptions, type ControlRuntimeConfig, type ControlRuntimeError, type ControlSeverity, type ControlStep, type ControlStopPolicies, ConvergenceTracker, type CorpusAgreementOptions, type CorpusAgreementPerDimension, type CorpusAgreementReport, type CorpusScoreRecord, type CorrectnessChecker, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, type CostChannel, type CostEntry, CostLedger, type CostLedgerFilter, type CostLedgerHandle, type CostLedgerOptions, type CostLedgerPersistence, CostLedgerPersistenceError, type CostLedgerSummary, type CostProvenance, type CostReceipt, CostReceiptCaptureError, type CostReceiptInput, type CostReport, CostReservationExceededError, type CostResult, type CostSummary, CostTracker, type CostUsage, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateAnalystAiConfig, type CreateChatClientOpts, type CreateDefaultReviewerOptions, type CreateExperimentInput, type CreateSandboxPoolOpts, type CreateTraceAnalystKindOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, type CustomTokenPricing, type CustomTransportOpts, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_AGENT_SLOS, DEFAULT_COMPLEXITY_WEIGHTS, DEFAULT_RULES as DEFAULT_FAILURE_RULES, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PERMUTATIONS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_REDACTION_RULES, DEFAULT_RED_TEAM_CORPUS, DEFAULT_RUN_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DataAcquisitionPlan, Dataset, type DatasetDifficulty, type DatasetManifest, type DatasetOverview, type DatasetProvenance, type DatasetScenario, type DatasetSplit, type DecideNextUserTurnOpts, type DefaultAnalystRegistryOptions, type DefaultVerdict, type DeltaStatistic, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DescriptionLengthCandidate, type DescriptionLengthConfig, type DescriptionLengthDecision, type DescriptionLengthEvidence, DescriptionLengthGate, type DescriptionLengthRejectionCode, type DetectorEvent, type DetectorSeverity, type DetectorSignal, type DiffPolicy, type DiffScorecardOptions, type DirEntry, type DirectProviderTransportOpts, type Direction, type DiscoverPersonasOptions, type DiscoveredPersona, DockerSandboxDriver, type DriverResult, type DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, type EProcess, type EProcessOptions, type EProcessState, type EProcessStep, ERROR_COUNT_PATTERNS, type EnsembleAggregate, type EnsembleJudgeOptions, type ErrorCluster, type ErrorCountPattern, type ErrorStreakOptions, type EvalCampaignOptions, type EvalCampaignResult, type EvalResult, type EvalToolDef, EvalTraceStore, EventFilter, EventKind, type EvidenceRef, type EvolutionRound, type ExactRiskDifferenceResult, type ExecutorConfig, type Expectation, type Experiment, type ExperimentPlan, type ExperimentProvenance, type ExperimentRep, type ExperimentResult, type ExperimentStats, type ExperimentStore, ExperimentTracker, type ExperimentTrackerOptions, type ExperimentVerdict, ExportableSpan, type ExportedRewardModel, type ExtractOptions, type ExtractResult, ExtractUsageFromSseOptions, ExtractedUsage, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, type FactorContribution, type FactorialCell, type FailedRun, FailureClass, type FailureClassification, type FailureContext, type FailureMode, type FailureRule, type FeedbackArtifactType, type FeedbackAttempt, type FeedbackLabel, type FeedbackLabelKind, type FeedbackLabelSource, type FeedbackOptimizerRow, type FeedbackOutcome, type FeedbackPattern, type FeedbackReplayAdapter, type FeedbackReplayResult, type FeedbackSeverity, type FeedbackSplitPolicy, type FeedbackTask, type FeedbackTrajectory, type FeedbackTrajectoryFilter, type FeedbackTrajectoryStore, type FieldDestination, type FileChange, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemRawProviderSinkOptions, FileSystemTraceStore, FileSystemTraceStoreOptions, type Finding, type FindingSubject, type FindingSubjectKind, type FindingsDiff, FindingsStore, type FlattenOtlpOptions, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type FromHarborOptions, type GainDistributionBin, type GainDistributionFigureSpec, type GainDistributionOptions, type GateDecision, type GateEvidence, GenericSpan, type GhCliClientOptions, type GoldenItem, type GoldenSeverity, type GoldenSpec, HARBOR_IMPORT_GAP, HARNESS_NATIVE_MODEL, type HarborAgent, type HarborContentPart, type HarborFinalMetrics, type HarborImageSource, type HarborMetrics, type HarborObservation, type HarborObservationResult, type HarborStep, type HarborStepSource, type HarborSubagentTrajectoryRef, type HarborToolCall, type HarborTrajectory, type HarnessAdapter, type HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessType, type HarnessVariant, type HarnessVariantReport, type HeadroomClass, type HeadroomInput, HeldOutGate, type HeldOutGateConfig, type HeldOutGateRejectionCode, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, HoldoutAuditor, HoldoutLockedError, type HttpGithubClientOptions, type HypothesisManifest, type HypothesisResult, IMPROVEMENT_KIND_SPEC, INPUT_VALUE, INTENT_MATCH_JUDGE_VERSION, type ImageData, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryRawProviderSinkOptions, InMemoryTraceStore, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, type InterimReleaseConfidence, type InterimReleaseConfidenceInput, type JudgeConfig, JudgeError, type JudgeFamily, type JudgeFleetOptions, type JudgeFn, type JudgeInput, JudgeParseError, type JudgeReplayGateArgs, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, type JudgeRubric, JudgeRunner, type JudgeScore, type JudgeScoreInput, type JudgeScoresRecord, JudgeSpan, type JudgeVerdict, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, KnowledgeAcquisitionMode, KnowledgeBundle, KnowledgeFallbackPolicy, KnowledgeFreshness, KnowledgeImportance, KnowledgeReadinessReport, KnowledgeRecommendedAction, KnowledgeRequirement, KnowledgeRequirementCategory, KnowledgeResponsibleSurface, KnowledgeSensitivity, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_CACHE_WRITE_TOKENS, LLM_CACHE_WRITE_TOKEN_ATTR_KEYS, LLM_CONTEXT_TOKENS, LLM_COST_ATTR_KEYS, LLM_COST_USD, LLM_INPUT_TOKENS, LLM_INPUT_TOKEN_ATTR_KEYS, LLM_MODEL_ATTR_KEYS, LLM_MODEL_NAME, LLM_OUTPUT_TOKENS, LLM_OUTPUT_TOKEN_ATTR_KEYS, LLM_REASONING_TOKENS, LLM_REASONING_TOKEN_ATTR_KEYS, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, type Layer, type LayerResult, type LayerStatus, type LeaderboardOptions, type LeaderboardRow, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmCallError, type LlmCallMetadata, type LlmCallRequest, type LlmCallResult, LlmClient, type LlmClientOptions, type LlmCorrectnessCheckerOpts, type LlmJsonCall, type LlmJudgeDimension, type LlmJudgeOptions, type LlmMessage, LlmResponseError, type LlmReviewerConfig, LlmRouteAssertionError, type LlmRouteRequirements, LlmSpan, LlmSpanOtlpInput, type LlmUsage, LockedJsonlAppender, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, MODEL_PRICING, type MakeEvalToolsConfig, type MannWhitneyResult, type MatchResult, type MatchedPair, type MatchedRunRecordPair, type MatcherResult, type MaximumCharge, type McNemarResult, type Measured, type MeasurementPolicy, type MergeOptions, Message, type MetricSamples, type MetricVerdict, MetricsCollector, type MintRolloutOptions, type MintRolloutResult, type MintedRolloutLine, type MintedRolloutOutcome, type MockTransportOpts, type ModelCostRollup, type ModelPreflight, type ModelSeats, ModelsUnreachableError, type MuffledFinder, type MuffledFinding, MultiLayerVerifier, type MultiToolchainLayerConfig, type Mutator, Mutex, type NoLeakOptions, type NoProgressOptions, NoopRawProviderSink, NoopResearcher, NotFoundError, OPENINFERENCE_SPAN_KIND, OTEL_AGENT_EVAL_SCOPE, OUTPUT_VALUE, type Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, OtlpExport, OtlpFileTraceStore, type OtlpFileTraceStoreOptions, type OtlpFlatLine, OtlpResourceSpans, OtlpSpan, OtlpSpanRole, OtlpSpanRoleInput, type OtlpToRunRecordsOptions, type OtlpTraceRunRecord, type PaidCallResult, type PairArmsOptions, type PairArmsResult, type PairRunRecordsResult, type PairedArmRow, type PairedArmsComparison, type PairedBootstrapOptions, type PairedBootstrapResult, type PairedCorrectness, type PairedDecisionMethod, type PairedDecisionShape, type PairedDecisionStatistic, type PairedDeltaTestOptions, type PairedDeltaTestResult, type PairedEvalueOptions, type PairedEvalueSequence, type PairedEvalueStep, type PairedMcNemarEvidence, type PairedMetricDelta, type PairedPromotionDecision, type PairedPromotionDecisionOptions, type PairedSignTestResult, type PairedTTestResult, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, type ParetoFigureSpec, type ParetoPoint, type ParetoResult, type PartitionHeldOutOptions, type PendingCostCall, type PendingCostCallView, type PersistedFinding, type PersonaConfig, type PersonaRigor, type Playbook, type PlaybookEntry, type PoolSlot, type PositionalBiasResult, type PrReviewAuditCase, type PrReviewBenchmarkSummary, type PrReviewComment, type PrReviewMatchedFinding, type PrReviewOutcome, type PrReviewReferenceFinding, type PrReviewScore, type PrReviewScoreWeights, type PrReviewSeverity, type PrReviewSource, type PreferenceMemoryEntry, type PreflightModelsOptions, type PreflightOutcome, type ProducedProposal, type ProducedState, type ProductBenchmarkArm, type ProductBenchmarkArtifactPaths, type ProductBenchmarkBudgets, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkProfileRef, type ProductBenchmarkRecord, type ProductBenchmarkRepoRef, type ProductBenchmarkRunInput, type ProductBenchmarkScenario, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkSubstrateVersions, type ProductBenchmarkValidationReport, ProductClient, type ProductClientConfig, type ProfileAxisSpec, type ProjectRuntimeTrajectoryEvidenceOptions, type ProjectedOtlpSpan, type PromptHandle, PromptRegistry, type ProportionInterval, type ProposalEventLike, type ProposalFinding, type ProposalFindingOrigin, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, type ProposeFn, type ProposeInput, type ProposeOutput, type ProposeReviewConfig, type ProposeReviewControlAction, type ProposeReviewControlConfig, type ProposeReviewControlResult, type ProposeReviewControlState, type ProposeReviewReport, type ProposeReviewShot, type ProposedSideEffect, type ProvenanceReader, ProviderRedactor, type QueryTracesPage, REDACTION_VERSION, REFERENCE_EQUIVALENCE_INPUT_LIMITS, REFERENCE_EQUIVALENCE_JUDGE_VERSION, RESEARCH_REPORT_HARD_PAIR_FLOOR, ROLLOUT_SCHEMA, RUN_COST_ATTR_KEYS, type RankTestMethod, type RankTestMethodRequest, type RankTestOptions, type RawAnalystEvidence, type RawAnalystFinding, RawProviderDirection, RawProviderEvent, RawProviderSink, RawProviderSinkFilter, type RecordRunsOptions, type RedTeamCase, type RedTeamCategory, type RedTeamFinding, type RedTeamPayload, type RedTeamReport, RedactionReport, RedactionRule, type ReferenceEquivalenceJudgeInput, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceJudgeResult, type ReferenceEquivalenceScenario, type ReferenceMatchResult, type ReferenceReplayAdapter, type ReferenceReplayAdapterFn, type ReferenceReplayAdapterLike, type ReferenceReplayAggregate, type ReferenceReplayCandidate, type ReferenceReplayCase, type ReferenceReplayCaseRun, type ReferenceReplayExecutionScenario, type ReferenceReplayItem, type ReferenceReplayMatch, type ReferenceReplayMatchStrategy, type ReferenceReplayMatcher, type ReferenceReplayPromotionDecision, type ReferenceReplayPromotionPolicy, type ReferenceReplayRun, type ReferenceReplayRunContext, type ReferenceReplayRunOptions, type ReferenceReplayRunStore, type ReferenceReplayScenario, type ReferenceReplayScenarioScore, type ReferenceReplayScore, type ReferenceReplayScoreOptions, type ReferenceReplaySplit, type ReferenceReplaySplitComparison, type ReferenceReplaySteeringRowsOptions, type ReflectionContext, type ReflectionProposal, type RegistryRunOpts, type ReleaseConfidenceAxis, type ReleaseConfidenceAxisName, type ReleaseConfidenceInput, type ReleaseConfidenceIssue, type ReleaseConfidenceMetrics, type ReleaseConfidenceScorecard, type ReleaseConfidenceStatus, type ReleaseConfidenceThresholds, type ReleaseTraceEvidence, type RenderReleaseReportOptions, type RepeatedActionOptions, ReplayCache, type ReplayCacheEntry, ReplayCacheMissError, type ReplayCacheStats, ReplayError, type ReplayFetchOptions, type RepoRef, type RequirementCheck, type ResearchReport, type ResearchReportCandidate, type ResearchReportDecision, type ResearchReportMethodology, type ResearchReportOptions, type ResearchReportRecommendation, type Researcher, RetrievalSpan, type Review, type ReviewFn, type ReviewInput, type ReviewMemoryEntry, type ReviewMemoryStore, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RewardRow, type RiskDifferenceResult, type RobustnessResult, type RolloutCapture, type RolloutLine, type RolloutRole, type RolloutScrubber, type RolloutSplit, type RolloutStep, type RouteMap, type RoutedField, type RouterTransportOpts, type RubricDimension, Run, type RunCommandInput, type RunCommandResult, RunCompleteHook, RunCompleteHookContext, type RunCostProvenance, RunCritic, type RunCriticOptions, type RunEvidenceMetadata, RunFilter, RunIntegrityError, RunIntegrityExpectations, RunIntegrityIssue, RunIntegrityIssueCode, RunIntegrityReport, type RunJudgeMetadata, RunLayer, RunOutcome, type RunPaidCallInput, type RunRecord, type RunRecordBackend, type RunRecordFilter, RunRecordValidationError, type RunScore, type RunScoreWeights, type RunSplitTag, RunStatus, type RunTaskFailure, type RunTerminalOutcome, type RunTokenUsage, type RunTrace, type RuntimeEventLike, type RuntimeResolution, type RuntimeTrajectoryEvidenceProjection, type RuntimeTrajectoryEvidenceSummary, type RuntimeTrajectoryHookEvent, type RuntimeTrajectoryRecord, type RuntimeTrajectoryRunRecord, SEMANTIC_CONCEPT_JUDGE_VERSION, SKILL_USAGE_ANALYST, SPAN_KIND_ATTR_KEYS, SUPERVISOR_RUN_SCHEMA, type SandboxDriver, SandboxHarness, type SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type SandboxResult, type SandboxSdkTransportOpts, SandboxSpan, type SatisfiedBy, type ScanOptions, type Scenario, type ScenarioCost, type ScenarioFile, ScenarioRegistry, type ScenarioResult, ScoreKnowledgeReadinessOptions, type ScoreOrigin, type ScorePreference, type ScoreRiskDifferenceResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SearchSpanResult, type SearchTraceResult, type SeatName, type SeatPresetName, SeatUnsetError, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SelfPreferenceResult, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SequentialDecision, type SerializedRegex, type SeriesConvergenceOptions, type SeriesConvergenceResult, type Severity, type SftExportOptions, type SftRow, type SignTestAlternative, type SignedManifest, type SignedManifestAlgo, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, SkillUsageAnalyst, type SliceOptions, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, type SourceLimits, Span, SpanBase, SpanFilter, SpanHandle, SpanKind, type SpanMatchRecord, SpanNotFoundError, type SpanPredicate, SpanStatus, type SplitCoverage, SseUsageMode, type SteeringBundle, type SteeringChange, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type StopDecision, type StreamingDetector, type SuboptimalCode, type SuboptimalSignal, SubprocessSandboxDriver, type SubprocessSandboxDriverOptions, type SummaryTable, type SummaryTableOptions, type SummaryTableRow, type SupervisorRunReader, type SupervisorRunReport, type SupervisorRunRollup, type SupervisorRunSources, type SupervisorRunTree, type SynthesisReason, type SynthesisTarget, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS, TRACE_ANALYST_ACTOR_DESCRIPTION, TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, TRACE_SCHEMA_VERSION, type TaskGold, type TaskHeadroom, type TestGradedRunOptions, type TestGradedRunResult, type TestGradedScenario, type TestOutputParser, type TestResult, type TextMatcher, type ThresholdContract, TokenCounter, type TokenSpec, type ToolCallEventLike, type ToolDef, type ToolMatcher, ToolSpan, ToolSpanOtlpInput, type ToolSpansToTraceAnalysisStoreOptions, type ToolStats, ToolTraceMissingError, type ToolUseMetrics, type ToolUseOptions, type TraceAggregate, type TraceAnalysisStore, type TraceAnalystByteBudgets, type TraceAnalystFilters, type TraceAnalystGolden, type TraceAnalystHookOptions, type TraceAnalystKindSpec, type TraceAnalystSpan, type TraceAnalystSpanKind, type TraceAnalystSpanStatus, type TraceAnalystTraceSummary, type TraceContract, TraceContractBuilder, TraceEmitter, TraceEmitterOptions, TraceEvent, TraceFileMissingError, type TraceInsightContext, type TraceInsightFinding, type TraceInsightPanelRole, type TraceInsightPromptInput, type TraceInsightQualityGate, type TraceInsightQuestion, type TraceInsightReadiness, type TraceInsightSuite, type TraceInsightTask, TraceNotFoundError, TraceStore, TraceStoreSource, TraceStoreToOtlpOptions, type TracedAnalystOptions, type TracedJudgeOptions, TracesToOtlpResult, type Trajectory, type TrajectoryStep, type TreatmentClass, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, type Turn, type TurnMetrics, type TurnResult, UI_FINDING_SEVERITIES, UI_LENSES, UNIVERSAL_FINDERS, type UiFinding, type UiFindingScreenshot, type UiFindingSeverity, type UiLens, type Unavailable, UserQuestion, type ValidationContext, ValidationError, type ValidationIssue, type ValidationResult, type VerbosityBiasResult, type Verdict, type VerdictCacheStats, type VerdictCacheStore, type Verification, VerificationError, type VerificationReport, type VerifyContext, type VerifyFn, type VerifyOptions, type ViewSpansResult, type ViewTraceOversized, type ViewTraceResult, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, WILCOXON_EXACT_MAX_N, type WeightedCompositeInput, type WeightedCompositeResult, type WelchTestResult, type WelchTestStatus, type WilcoxonSignedRankResult, type WorkerDriverContext, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, acquisitionPlansForKnowledgeGaps, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, agentProfileModelId, agentVisibleFields, aggregateJudgeVerdicts, aggregateLlm, aggregatePrReviewScore, aggregateRunScore, allCriticalPassed, analyzeAntiSlop, analyzeSeries, analyzeSupervisorRun, analyzeSupervisorRunSources, analyzeTraces, appendScorecard, applyLlmSpanOtlpAttributes, applyToolSpanOtlpAttributes, argHash, asNumber, asString, assertCapabilityHeadroom, assertCrossFamily, assertLlmRoute, assertMinted, assertMintedLines, assertModelsServed, assertNoHiddenLeak, assertProductBenchmarkRun, assertRealAgentReceipts, assertRealBackend, assertReleaseConfidence, assertRolloutLine, assertRunAgentProfileCell, assertRunCaptured, assertSingleBackend, assignFeedbackSplit, assignHeldOutTag, attachCostToReport, attest, attributeCounterfactuals, backoffMs, deterministicSplit as benchmarkDeterministicSplit, index_d_exports as benchmarks, benjaminiHochberg, bisect, blendHeldout, blockingKnowledgeEval, bonferroni, bootstrapCi, buildAgentInterfaceProfileCell, buildAgentProfileCell, buildDefaultAnalystRegistry, buildDriverSystemPrompt, buildProductBenchmarkManifest, buildReflectionPrompt, buildReviewerPrompt, buildTraceAnalystTools, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, buildWorkerDriverSystemPrompt, byteLengthRange, cachedJudge, calibrateJudge, calibrateJudgeContinuous, callLlm, callLlmJson, canaryLeakView, canonicalJson, canonicalize, capabilityHeadroom, captureFetchToRawSink, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, checkTraceContracts, clamp01, classifyFailure, classifyOtlpSpanRole, classifyTreatment, claudeCodeSupervisorRunReader, cliffsDelta, clusteredPairedBinary, cohensD, collectionPreserved, commentsForSource, commitBisect, comparePairedArms, compareReferenceReplay, compareToBaseline, compilerJudge, completionVerdict, composeParsers, composeValidators, computeExperimentStats, computeFindingId, computeToolUseMetrics, computeTraceMetrics, confidenceInterval, containsAll, contentHash, contextInputTokens, continuousAgreement, contractJudge, controlFailureClassFromVerification, controlRunToFeedbackTrajectory, controlRunToRunRecord, convertTraceStoresToOtlp, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, costReport, createAnalystAi, createAntiSlopJudge, createChatClient, createDefaultReviewer, createFeedbackTrajectory, createIntentMatchJudge, createLlmCorrectnessChecker, createLlmReviewer, createOtelExporter, createOtelTracingStore, createReferenceEquivalenceJudge, createReplayFetch, createSandboxPool, createSemanticConceptJudge, createTokenRecallChecker, createTraceAnalystKind, crossTraceDiff, crowdingDistance, dataDescriptionBits, decideNextUserTurn, decidePairedPromotion, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultBlendWeights, defaultIsMaterial, defaultProviderRedactor, defaultReferenceReplayMatcher, defaultTraceInsightPanel, deployGateLayer, describeTraceInsightScope, diffFindings, diffScorecard, discoverPersonas, distillPlaybook, domainEvidencePattern, dominates, eProcess, ensembleJudge, errorStreakDetector, estimateCost, estimateTokens, evaluateActionPolicy, evaluateContract, evaluateHypothesis, evaluateInterimReleaseConfidence, evaluateOracles, evaluateReleaseConfidence, evaluateTraceContract, executeScenario, expandProfileAxes, expectAgent, exportProductBenchmark, exportProductBenchmarkRuns, exportRewardModel, exportRunAsOtlp, extractAssetUrls, extractErrorCount, extractOtlpAttributes, extractProducedState, extractUsage, extractUsageFromResponse, extractUsageFromSse, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToDatasetScenario, feedbackTrajectoryToOptimizerRow, fileContains, fileExists, fileExperimentStore, fileVerdictCache, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findProductBenchmarkArtifacts, findSkipCountsAsPass, firstNumberAttr, firstStringAttr, flattenOtlpExportToNdjson, flowLayer, fnv1a32, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, fromHarborTrajectory, gainHistogram, gateTreatmentApplied, gateTreatmentFromMetrics, gateTreatmentFromSpans, gateTreatmentFromToolSpans, ghCliClient, gitProvenanceReader, precision as goldenPrecision, gradeOnHidden, gradeSemanticStatus, groupBy, groupRunsByAgentProfileCell, harnessAxisOf, hasCapturedToolArgs, hashContent, hashJson, hashScenarios, hashToUnit, hiddenGrade, holm, htmlContainsElement, httpGithubClient, improvementVerdict, inMemoryExperimentStore, inMemoryReferenceReplayStore, inMemoryReviewStore, inMemoryRunRecordBackend, inMemoryVerdictCache, inferDomainKeywords, inferOtlpKind, interRaterReliability, interpretCliffs, iqr, isBinaryOutcomeVector, isHiddenDestination, isJudgeSpan, isLlmSpan, isModelPriced, isOtelConfigured, isOtlpModelCall, isRealnessGated, isRetrievalSpan, isRolloutLine, isRunRecord, isSandboxSpan, isToolSpan, isTrainableSplit, isTransientLlmError, isUnavailable, iterateRawCalls, jestTestParser, jsonHasKeys, jsonShape, jsonlReferenceReplayStore, jsonlReviewStore, jsonlRunRecordBackend, judgeFamily, judgeReplayGate, judgeSpans, keyPreserved, knowledgeReadinessTracePayload, leaderboard, linterJudge, llmJudge, llmSpanFromProvider, llmSpans, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, makeEvalTools, makeFinding, makeProposalFinding, mannWhitneyU, mapConcurrent, matchGoldens, matchSpan, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, mergeLayerResults, mergeSteeringBundle, minimumPairsForPairedDeltaTest, mintRolloutRows, modelDescriptionBits, modelHasSnapshot, modelPriceKey, mulberry32, multiToolchainLayer, noProgressDetector, normalCdf, normalizeScores, notBlocked, objectiveEval, observeAll, observedScore, observedSplitScore, otelRunCompleteHook, otlpRowsToRunRecords, otlpRowsToTraceRunRecords, otlpToRunRecords, otlpToTraceRunRecords, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedCohensDz, pairedDecisionShape, pairedDeltaTest, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paraphraseRobustness, paraphraseRobustnessScenarios, paretoChart, paretoFrontier, paretoFrontierWithCrowding, parseCorrectnessResponse, parseFeedbackTrajectoriesJsonl, parseReflectionResponse, parseRunRecordSafe, parseRuntimeTrajectoryHookEvent, partialCredit, partitionHeldOut, passAtK, passOrthogonality, pearsonR, pixelDeltaRatio, planTraceInsightQuestions, politenessPrefixMutator, positionalBias, preflightModels, printDriverSummary, probeLlm, productBenchmarkIntegrityFailures, productBenchmarkMutableSurfaces, productBenchmarkRepoIdentity, productBenchmarkSplits, index_d_exports$1 as profile, projectOtlpFlatLine, projectRuntimeTrajectoryEvidence, promptBisect, proposeSynthesisTargets, providerFromBaseUrl, pytestTestParser, ranks, readClaudeCodeSupervisorRun, readOtlpStatus, readProductBenchmarkManifest, readProductBenchmarkRecords, recordRuns, recordRunsToScorecard, redTeamDataset, redTeamReport, redactString, redactValue, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatch, regexMatches, relabelImportedSplit, renderMarkdownReport, renderPlaybookMarkdown, renderPreferenceMemoryMarkdown, renderPriorFindings, renderReleaseReport, renderSteeringText, renderSupervisorRunHeadline, renderSupervisorRunMarkdown, renderUpstreamFindings, repeatedActionDetector, replayFeedbackTrajectories, replayFeedbackTrajectory, replayScorerOverCorpus, replayTraceThroughJudge, requireAgentProfileCell, requiredPairedSampleSize, requiredSampleSize, researchReport, resolveModelPricing, resolveSeat, rollupSupervisorRuns, roundTripRunRecord, routeFields, rowCount, rowWhere, runAgentControlLoop, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runE2EWorkflow, runEvalCampaign, runExpectations, runFailureClass, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runProposeReview, runProposeReviewAsControlLoop, runRecordToProductBenchmarkRecord, runReferenceEquivalenceJudge, runReferenceReplay, runScore, runSelfPlay, runSemanticConceptJudge, runTaskScore, runTestGradedScenario, runsForScenario, scalarScore, scanForMuffledGates, scoreContinuity, scoreFromEvals, scoreKnowledgeReadiness, scoreOrigin, scorePrReviewComments, scorePrReviewSource, scoreRedTeamOutput, scoreReferenceReplay, scoreTraceInsightReadiness, seatPresets, securityJudge, selectHarnessVariant, selfPreference, sentenceReorderMutator, serializeFeedbackTrajectoriesJsonl, showMeasured, signManifest, spearmanR, statusAdvanced, stopOnNoProgress, stopOnRepeatedAction, stringField, stripFencedJson, studentTCdf, subjectiveEval, summarizeAgentReceiptIntegrity, summarizeBackendIntegrity, summarizeHarnessResults, summarizePrReviewBenchmark, summarizePreferenceMemory, summaryTable, supervisorRunRolloutLines, testJudge, textInSnapshot, throwIfRunIncomplete, toAgentProfileJson, toHarborTrajectories, toHarborTrajectory, toJsonl, toLangfuseEnvelope, toOpenAiTool, toPrometheusText, toRewardRows, toSftRows, tokenizeDomainWords, toolNamesForRun, toolSpans, toolSpansToTraceAnalysisStore, traceAnalystFunctionGroup, traceAnalystOnRunComplete, traceContract, traceJudge, traceJudgeEnsemble, traceSpanKindToOpenInferenceKind, tracedAnalyzeTraces, trainingReward, trainingScore, typoMutator, unmintableReasons, urlContains, userQuestionsForKnowledgeGaps, validateAgentProfileCell, validateProductBenchmarkManifest, validateProductBenchmarkRecord, validateProductBenchmarkRun, validateRolloutLine, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, verifyManifest, visualDiff, viteDeployRunner, vitestTestParser, weightedComposite, weightedMean, weightedRecall, welchsTTest, whitespaceCollapseMutator, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit, withHeldoutBlend, withJudgeRetry, withOtelPipeline, wranglerDeployRunner, writeSupervisorRunReport };
+export { AGENT_PROFILE_KINDS, ATIF_SCHEMA_VERSION, ATTESTATION_ALGORITHM, type ActionExecutionPolicy, type ActionPolicyDecision, type ActionableSideInfo, type ActiveLearningOptions, type AdapterRun, AgentDriver, type AgentDriverConfig, AgentEvalError, type AgentEvalErrorCode, type AgentInterfaceProfileLike, type AgentProfile, type AgentProfileCell, type AgentProfileCellInput, type AgentProfileCellSchemaVersion, AgentProfileCellValidationError, type AgentProfileDimensionValue, type AgentProfileHarness, type AgentProfileJson, type AgentProfileJsonObject, type AgentProfileKind, type AgentProfileRuntimeReceipt, type AgentProfileSource, type AgentProfileSourceInput, type AlignmentOp, type Analyst, type AnalystContext, type AnalystCost, type AnalystFinding, type AnalystHooks, type AnalystInputKind, AnalystRegistry, type AnalystRegistryOptions, type AnalystRequirements, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystSeverity, type AnalystUsageReceipt, type AnalyzeTracesInput, type AnalyzeTracesOptions, type AnalyzeTracesResult, type AnalyzeTracesTurnSnapshot, type AntiSlopConfig, type AntiSlopIssue, type AntiSlopReport, Artifact, type ArtifactCheck, type Artifact$1 as ArtifactCheckArtifact, type ArtifactEventLike, type ArtifactResult, type ArtifactValidator, type AsiSeverity, type AssertCapabilityHeadroomOptions, type AssertCrossFamilyOptions, type AssertSingleBackendOptions, type AttestationProvenance, type AttestationVerification, type AttestedReport, type AutoPrClient, AxGepaSteeringOptimizer, type AxSteeringOptimizerConfig, BENCHMARK_SPLIT_SEED, BOOTSTRAP_GATE_MIN_N, type BackendDescriptor, BackendIntegrityError, type BackendIntegrityReport, type BaselineOptions, type BaselineReport, type BehaviorAssertion, type BehavioralMetrics, type BehavioralTokenSequence, type BenchmarkAdapter, type BenchmarkDatasetItem, type BenchmarkEvaluation, type BenchmarkFamily, type BenchmarkReport, type BenchmarkResponder, BenchmarkRunner, type BenchmarkRunnerConfig, type BenchmarkScenario, type BenchmarkSource, type BenchmarkTaskKind, type BisectOptions, type BisectResult, type BisectStep, type BlendWeights, type BootstrapOptions, type BootstrapResult, BudgetBreachError, BudgetGuard, BudgetLedgerEntry, type BudgetPolicy, BudgetSpec, CODING_HARNESSES, type CachedJudge, type CachedJudgeOptions, type CalibrationResult, type CallExpectation, CallbackResearcher, type CallbackResearcherOptions, type CampaignFactoryParams, type CampaignIntegrityPolicy, type CampaignRunContext, type CampaignRunOutcome, type CampaignRunner, type CampaignScenario, type CampaignVariant, type CanaryAlert, type CanaryEvaluation, type CanaryKind, type CanaryLeak, type CanaryOptions, type CanaryReport, type CanarySeverity, type CandidateComparison, type CandidateScenario, type CandidateScore, type CapabilityHeadroomOptions, type CapabilityHeadroomResult, CaptureFetchContext, CaptureFetchOptions, CaptureIntegrityError, type CausalAttributionReport, type CellVerdict, type ChannelRollup, type ChatCallOpts, type ChatClient, type ChatMessage, type ChatRequest, type ChatResponse, type ChatToolCall, type ChatTransport, type CheckResult, type CliBridgeTransportOpts, type CliffsMagnitude, type ClusterBootstrapInterval, type ClusterSignFlipAlternative, type ClusterSignFlipResult, type ClusteredBinaryCluster, type ClusteredMatchedPair, type ClusteredPairedBinaryOptions, type ClusteredPairedBinaryResult, type ClusteredPairedBinaryStatistics, type CollectedArtifacts, type CommandRunner, type ComparePairedArmsOptions, type CompletionCriterion, type CompletionRequirement, type CompletionVerdict, type ConceptComplexity, type ConceptFinding, type ConceptSpec, type ConceptWeightStrategy, ConfigError, type ContinuityCheck, type ContinuityCheckResult, type ContinuityReport, type ContinuitySnapshotPair, type ContinuousAgreement, type ContinuousAgreementOptions, type ContinuousCalibrationResult, type ContractCheckResult, type ContractJudgeOptions, type ContractMetric, type ContractReport, type ContractRule, type ContractRuleKind, type ContractSpan, type ContractVerdict, type ContractViolation, type ControlActionFailureMode, type ControlActionOutcome, type ControlBudget, type ControlContext, type ControlDecision, type ControlEvalResult, type ControlRunResult, type ControlRunToRunRecordOptions, type ControlRuntimeConfig, type ControlRuntimeError, type ControlSeverity, type ControlStep, type ControlStopPolicies, ConvergenceTracker, type CorpusAgreementOptions, type CorpusAgreementPerDimension, type CorpusAgreementReport, type CorpusScoreRecord, type CorrectnessChecker, CostAccountingIncompleteError, CostCallConflictError, CostCeilingReachedError, type CostChannel, type CostEntry, CostLedger, type CostLedgerFilter, type CostLedgerHandle, type CostLedgerOptions, type CostLedgerPersistence, CostLedgerPersistenceError, type CostLedgerSummary, type CostProvenance, type CostReceipt, CostReceiptCaptureError, type CostReceiptInput, type CostReport, CostReservationExceededError, type CostResult, type CostSummary, CostTracker, type CostUsage, type CounterfactualContext, type CounterfactualMutation, type CounterfactualResult, type CounterfactualRunner, type CreateAnalystAiConfig, type CreateChatClientOpts, type CreateDefaultReviewerOptions, type CreateExperimentInput, type CreateSandboxPoolOpts, type CreateTraceAnalystKindOpts, CrossFamilyError, type CrossTraceDiff, type CrossTraceDiffOptions, type CustomTokenPricing, type CustomTransportOpts, DECISION_PAIRED_DELTA_STATISTIC, DEFAULT_AGENT_SLOS, DEFAULT_COMPLEXITY_WEIGHTS, DEFAULT_RULES as DEFAULT_FAILURE_RULES, DEFAULT_FINDERS, DEFAULT_HARNESS_OBJECTIVES, DEFAULT_MUTATION_PRIMITIVES, DEFAULT_MUTATORS, DEFAULT_PERMUTATIONS, DEFAULT_PR_REVIEW_SCORE_WEIGHTS, DEFAULT_REDACTION_RULES, DEFAULT_RED_TEAM_CORPUS, DEFAULT_RUN_SCORE_WEIGHTS, DEFAULT_SEVERITY_WEIGHTS, DEFAULT_TRACE_ANALYST_BUDGETS, DEFAULT_TRACE_ANALYST_KINDS, DataAcquisitionPlan, Dataset, type DatasetDifficulty, type DatasetManifest, type DatasetOverview, type DatasetProvenance, type DatasetScenario, type DatasetSplit, type DecideNextUserTurnOpts, type DefaultAnalystRegistryOptions, type DefaultVerdict, type DeltaStatistic, type DeployFamily, type DeployGateLayerInput, type DeployRunResult, type DeployRunner, type DescriptionLengthCandidate, type DescriptionLengthConfig, type DescriptionLengthDecision, type DescriptionLengthEvidence, DescriptionLengthGate, type DescriptionLengthRejectionCode, type DetectorEvent, type DetectorSeverity, type DetectorSignal, type DiffPolicy, type DiffScorecardOptions, type DirEntry, type DirectProviderTransportOpts, type Direction, type DiscoverPersonasOptions, type DiscoveredPersona, DockerSandboxDriver, type DriverResult, type DriverState, DualAgentBench, type DualAgentBenchConfig, type DualAgentReport, type DualAgentRound, type DualAgentScenario, type DualAgentScenarioResult, type EProcess, type EProcessOptions, type EProcessState, type EProcessStep, ERROR_COUNT_PATTERNS, type EnsembleAggregate, type EnsembleJudgeOptions, type ErrorCluster, type ErrorCountPattern, type ErrorStreakOptions, type EvalCampaignOptions, type EvalCampaignResult, type EvalResult, type EvalToolDef, EvalTraceStore, EventFilter, EventKind, type EvidenceRef, type EvolutionRound, type ExactRiskDifferenceResult, type ExecutorConfig, type Expectation, type Experiment, type ExperimentPlan, type ExperimentProvenance, type ExperimentRep, type ExperimentResult, type ExperimentStats, type ExperimentStore, ExperimentTracker, type ExperimentTrackerOptions, type ExperimentVerdict, ExportableSpan, type ExportedRewardModel, type ExtractOptions, type ExtractResult, ExtractUsageFromSseOptions, ExtractedUsage, FAILURE_CLASSES, FAILURE_MODE_KIND_SPEC, type FactorContribution, type FactorialCell, type FailedRun, FailureClass, type FailureClassification, type FailureContext, type FailureMode, type FailureRule, type FeedbackArtifactType, type FeedbackAttempt, type FeedbackLabel, type FeedbackLabelKind, type FeedbackLabelSource, type FeedbackOptimizerRow, type FeedbackOutcome, type FeedbackPattern, type FeedbackReplayAdapter, type FeedbackReplayResult, type FeedbackSeverity, type FeedbackSplitPolicy, type FeedbackTask, type FeedbackTrajectory, type FeedbackTrajectoryFilter, type FeedbackTrajectoryStore, type FieldDestination, type FileChange, FileSystemFeedbackTrajectoryStore, FileSystemRawProviderSink, FileSystemRawProviderSinkOptions, FileSystemTraceStore, FileSystemTraceStoreOptions, type Finding, type FindingSubject, type FindingSubjectKind, type FindingsDiff, FindingsStore, type FlattenOtlpOptions, type FlowAction, type FlowLayerEnv, type FlowLayerFactoryInput, type FlowRunner, type FlowRunnerStepResult, type FlowSpec, type FlowStep, type FromHarborOptions, type GainDistributionBin, type GainDistributionFigureSpec, type GainDistributionOptions, type GateDecision, type GateEvidence, GenericSpan, type GhCliClientOptions, type GoldenItem, type GoldenSeverity, type GoldenSpec, HARBOR_IMPORT_GAP, HARNESS_NATIVE_MODEL, type HarborAgent, type HarborContentPart, type HarborFinalMetrics, type HarborImageSource, type HarborMetrics, type HarborObservation, type HarborObservationResult, type HarborStep, type HarborStepSource, type HarborSubagentTrajectoryRef, type HarborToolCall, type HarborTrajectory, type HarnessAdapter, type HarnessConfig, type HarnessExperimentConfig, type HarnessExperimentResult, type HarnessIntervention, type HarnessRunRequest, type HarnessRunResult, type HarnessScenario, type HarnessSelection, type HarnessType, type HarnessVariant, type HarnessVariantReport, type HeadroomClass, type HeadroomInput, HeldOutGate, type HeldOutGateConfig, type HeldOutGateRejectionCode, type HeldOutPartition, type HiddenCriteriaGrader, type HiddenGradeResult, type HiddenLeak, HoldoutAuditor, HoldoutLockedError, type HttpGithubClientOptions, type HypothesisManifest, type HypothesisResult, IMPROVEMENT_KIND_SPEC, INPUT_VALUE, INTENT_MATCH_JUDGE_VERSION, type ImageData, type ImprovementThresholds, type ImprovementVerdictResult, InMemoryFeedbackTrajectoryStore, InMemoryRawProviderSink, InMemoryRawProviderSinkOptions, InMemoryTraceStore, InMemoryWorkspaceInspector, type InferenceScorer, type InspectorContext, type IntentMatchInput, type IntentMatchOptions, type IntentMatchResult, type InteractionContribution, type InterimReleaseConfidence, type InterimReleaseConfidenceInput, type JudgeConfig, JudgeError, type JudgeFamily, type JudgeFleetOptions, type JudgeFn, type JudgeInput, JudgeParseError, type JudgeReplayGateArgs, type JudgeReplayResult, type JudgeRetryOutcome, type JudgeRetryPolicy, type JudgeRubric, JudgeRunner, type JudgeScore, type JudgeScoreInput, type JudgeScoresRecord, JudgeSpan, type JudgeVerdict, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type KeywordConceptSpec, type KeywordCoverageFinding, type KeywordCoverageOptions, type KeywordCoverageResult, KnowledgeAcquisitionMode, KnowledgeBundle, KnowledgeFallbackPolicy, KnowledgeFreshness, KnowledgeImportance, KnowledgeReadinessReport, KnowledgeRecommendedAction, KnowledgeRequirement, KnowledgeRequirementCategory, KnowledgeResponsibleSurface, KnowledgeSensitivity, LLM_CACHED_TOKENS, LLM_CACHED_TOKEN_ATTR_KEYS, LLM_CACHE_WRITE_TOKENS, LLM_CACHE_WRITE_TOKEN_ATTR_KEYS, LLM_CONTEXT_TOKENS, LLM_COST_ATTR_KEYS, LLM_COST_USD, LLM_INPUT_TOKENS, LLM_INPUT_TOKEN_ATTR_KEYS, LLM_MODEL_ATTR_KEYS, LLM_MODEL_NAME, LLM_OUTPUT_TOKENS, LLM_OUTPUT_TOKEN_ATTR_KEYS, LLM_REASONING_TOKENS, LLM_REASONING_TOKEN_ATTR_KEYS, type LangfuseEnvelope, type LangfuseGeneration, type LangfuseScore, type Layer, type LayerResult, type LayerStatus, type LeaderboardOptions, type LeaderboardRow, type LiveProofArtifact, type LiveProofConfig, type LiveProofContext, type LiveProofResult, LlmCallError, type LlmCallMetadata, type LlmCallRequest, type LlmCallResult, LlmClient, type LlmClientOptions, type LlmCorrectnessCheckerOpts, type LlmJsonCall, type LlmJudgeDimension, type LlmJudgeOptions, type LlmMessage, LlmResponseError, type LlmReviewerConfig, LlmRouteAssertionError, type LlmRouteRequirements, LlmSpan, LlmSpanOtlpInput, type LlmUsage, LockedJsonlAppender, MANN_WHITNEY_EXACT_MAX_STATES, MANN_WHITNEY_EXACT_MAX_WORK, MODEL_PRICING, type MakeEvalToolsConfig, type MannWhitneyResult, type MatchResult, type MatchedPair, type MatchedRunRecordPair, type MatcherResult, type MaximumCharge, type McNemarResult, type Measured, type MeasurementPolicy, type MergeOptions, Message, type MetricSamples, type MetricVerdict, MetricsCollector, type MintRolloutOptions, type MintRolloutResult, type MintedRolloutLine, type MintedRolloutOutcome, type MockTransportOpts, type ModelCostRollup, type ModelPreflight, type ModelSeats, ModelsUnreachableError, type MuffledFinder, type MuffledFinding, MultiLayerVerifier, type MultiToolchainLayerConfig, type Mutator, Mutex, type NoLeakOptions, type NoProgressOptions, NoopRawProviderSink, NoopResearcher, NotFoundError, OPENINFERENCE_SPAN_KIND, OTEL_AGENT_EVAL_SCOPE, OUTPUT_VALUE, type Objective, type Oracle, type OracleObservation, type OracleReport, type OracleResult, type OrthogonalityInput, type OrthogonalityResult, OtelExportConfig, OtelExporter, type OtelPipelineHandle, type OtelPipelineOptions, OtlpExport, OtlpFileTraceStore, type OtlpFileTraceStoreOptions, type OtlpFlatLine, OtlpResourceSpans, OtlpSpan, OtlpSpanRole, OtlpSpanRoleInput, type OtlpToRunRecordsOptions, type OtlpTraceRunRecord, type PaidCallResult, type PairArmsOptions, type PairArmsResult, type PairRunRecordsResult, type PairedArmRow, type PairedArmsComparison, type PairedBootstrapOptions, type PairedBootstrapResult, type PairedCorrectness, type PairedDecisionMethod, type PairedDecisionShape, type PairedDecisionStatistic, type PairedDeltaTestOptions, type PairedDeltaTestResult, type PairedEvalueOptions, type PairedEvalueSequence, type PairedEvalueStep, type PairedMcNemarEvidence, type PairedMetricDelta, type PairedPromotionDecision, type PairedPromotionDecisionOptions, type PairedSignTestResult, type PairedTTestResult, PairwiseSteeringOptimizer, type ParaphraseRobustnessScenarioInput, type ParaphraseRobustnessScenarioResult, type ParetoFigureSpec, type ParetoPoint, type ParetoResult, type PartitionHeldOutOptions, type PendingCostCall, type PendingCostCallView, type PersistedFinding, type PersonaConfig, type PersonaRigor, type Playbook, type PlaybookEntry, type PoolSlot, type PositionalBiasResult, type PrReviewAuditCase, type PrReviewBenchmarkSummary, type PrReviewComment, type PrReviewMatchedFinding, type PrReviewOutcome, type PrReviewReferenceFinding, type PrReviewScore, type PrReviewScoreWeights, type PrReviewSeverity, type PrReviewSource, type PreferenceMemoryEntry, type PreflightModelsOptions, type PreflightOutcome, type ProducedProposal, type ProducedState, type ProductBenchmarkArm, type ProductBenchmarkArtifactPaths, type ProductBenchmarkBudgets, type ProductBenchmarkExportOptions, type ProductBenchmarkExportResult, type ProductBenchmarkManifest, type ProductBenchmarkProfileRef, type ProductBenchmarkRecord, type ProductBenchmarkRepoRef, type ProductBenchmarkRunInput, type ProductBenchmarkScenario, type ProductBenchmarkSingleRunExportOptions, type ProductBenchmarkSplit, type ProductBenchmarkSubstrateVersions, type ProductBenchmarkValidationReport, ProductClient, type ProductClientConfig, type ProfileAxisSpec, type ProjectRuntimeTrajectoryEvidenceOptions, type ProjectedOtlpSpan, type PromptHandle, PromptRegistry, type ProportionInterval, type ProposalEventLike, type ProposalFinding, type ProposalFindingOrigin, type ProposeAutomatedPullRequestInput, type ProposeAutomatedPullRequestResult, type ProposeFn, type ProposeInput, type ProposeOutput, type ProposeReviewConfig, type ProposeReviewControlAction, type ProposeReviewControlConfig, type ProposeReviewControlResult, type ProposeReviewControlState, type ProposeReviewReport, type ProposeReviewShot, type ProposedSideEffect, type ProvenanceReader, ProviderRedactor, type QueryTracesPage, REDACTION_VERSION, REFERENCE_EQUIVALENCE_INPUT_LIMITS, REFERENCE_EQUIVALENCE_JUDGE_VERSION, RESEARCH_REPORT_HARD_PAIR_FLOOR, ROLLOUT_SCHEMA, RUN_COST_ATTR_KEYS, type RankTestMethod, type RankTestMethodRequest, type RankTestOptions, type RawAnalystEvidence, type RawAnalystFinding, RawProviderDirection, RawProviderEvent, RawProviderSink, RawProviderSinkFilter, type RecordRunsOptions, type RedTeamCase, type RedTeamCategory, type RedTeamFinding, type RedTeamPayload, type RedTeamReport, RedactionReport, RedactionRule, type ReferenceEquivalenceJudgeInput, type ReferenceEquivalenceJudgeOptions, type ReferenceEquivalenceJudgeResult, type ReferenceEquivalenceScenario, type ReferenceMatchResult, type ReferenceReplayAdapter, type ReferenceReplayAdapterFn, type ReferenceReplayAdapterLike, type ReferenceReplayAggregate, type ReferenceReplayCandidate, type ReferenceReplayCase, type ReferenceReplayCaseRun, type ReferenceReplayExecutionScenario, type ReferenceReplayItem, type ReferenceReplayMatch, type ReferenceReplayMatchStrategy, type ReferenceReplayMatcher, type ReferenceReplayPromotionDecision, type ReferenceReplayPromotionPolicy, type ReferenceReplayRun, type ReferenceReplayRunContext, type ReferenceReplayRunOptions, type ReferenceReplayRunStore, type ReferenceReplayScenario, type ReferenceReplayScenarioScore, type ReferenceReplayScore, type ReferenceReplayScoreOptions, type ReferenceReplaySplit, type ReferenceReplaySplitComparison, type ReferenceReplaySteeringRowsOptions, type ReflectionContext, type ReflectionProposal, type RegistryRunOpts, type ReleaseConfidenceAxis, type ReleaseConfidenceAxisName, type ReleaseConfidenceInput, type ReleaseConfidenceIssue, type ReleaseConfidenceMetrics, type ReleaseConfidenceScorecard, type ReleaseConfidenceStatus, type ReleaseConfidenceThresholds, type ReleaseTraceEvidence, type RenderReleaseReportOptions, type RepeatedActionOptions, ReplayCache, type ReplayCacheEntry, ReplayCacheMissError, type ReplayCacheStats, ReplayError, type ReplayFetchOptions, type RepoRef, type RequirementCheck, type ResearchReport, type ResearchReportCandidate, type ResearchReportDecision, type ResearchReportMethodology, type ResearchReportOptions, type ResearchReportRecommendation, type Researcher, RetrievalSpan, type Review, type ReviewFn, type ReviewInput, type ReviewMemoryEntry, type ReviewMemoryStore, type ReviewerMemoryEntry, type ReviewerOutput, type ReviewerPromptInput, type ReviewerSoftFailDefaults, type ReviewerVerificationSummary, type RewardRow, type RiskDifferenceResult, type RobustnessResult, type RolloutCapture, type RolloutLine, type RolloutRole, type RolloutScrubber, type RolloutSplit, type RolloutStep, type RouteMap, type RoutedField, type RouterTransportOpts, type RubricDimension, Run, type RunCommandInput, type RunCommandResult, RunCompleteHook, RunCompleteHookContext, type RunCostProvenance, RunCritic, type RunCriticOptions, type RunEvidenceMetadata, RunFilter, RunIntegrityError, RunIntegrityExpectations, RunIntegrityIssue, RunIntegrityIssueCode, RunIntegrityReport, type RunJudgeMetadata, RunLayer, RunOutcome, type RunPaidCallInput, type RunRecord, type RunRecordBackend, type RunRecordFilter, RunRecordValidationError, type RunScore, type RunScoreWeights, type RunSplitTag, RunStatus, type RunTaskFailure, type RunTerminalOutcome, type RunTokenUsage, type RunTrace, type RuntimeEventLike, type RuntimeResolution, type RuntimeTrajectoryEvidenceProjection, type RuntimeTrajectoryEvidenceSummary, type RuntimeTrajectoryHookEvent, type RuntimeTrajectoryRecord, type RuntimeTrajectoryRunRecord, SEMANTIC_CONCEPT_JUDGE_VERSION, SKILL_USAGE_ANALYST, SPAN_KIND_ATTR_KEYS, SUPERVISOR_RUN_SCHEMA, type SandboxDriver, SandboxHarness, type SandboxHarnessResult, type SandboxJudgeKind, type SandboxJudgeResult, type SandboxJudgeSpec, type SandboxPool, type SandboxResult, type SandboxSdkTransportOpts, SandboxSpan, type SatisfiedBy, type ScanOptions, type Scenario, type ScenarioCost, type ScenarioFile, ScenarioRegistry, type ScenarioResult, ScoreKnowledgeReadinessOptions, type ScoreOrigin, type ScorePreference, type ScoreRiskDifferenceResult, type Scorecard, type ScorecardCell, type ScorecardCellDiff, type ScorecardDiff, type ScorecardEntry, type ScorecardLogLine, type ScoredTarget, type SearchSpanResult, type SearchTraceResult, type SeatName, type SeatPresetName, SeatUnsetError, type SelfPlayOptions, type SelfPlayProposer, type SelfPlayScorer, type SelfPreferenceResult, type SemanticConceptJudgeInput, type SemanticConceptJudgeOptions, type SemanticConceptJudgeResult, type SequentialDecision, type SerializedRegex, type SeriesConvergenceOptions, type SeriesConvergenceResult, type Severity, type SftExportOptions, type SftRow, type SignTestAlternative, type SignedManifest, type SignedManifestAlgo, type SingleBackendDivergence, SingleBackendError, type SingleBackendField, type SingleBackendReport, SkillUsageAnalyst, type SliceOptions, type Slo, type SloCheckResult, type SloComparator, type SloReport, type SloSeverity, type SlopCategory, type SlotFactory, type SourceLimits, Span, SpanBase, SpanFilter, SpanHandle, SpanKind, type SpanMatchRecord, SpanNotFoundError, type SpanPredicate, SpanStatus, type SplitCoverage, SseUsageMode, type SteeringBundle, type SteeringChange, type SteeringDelta, type SteeringOptimizationResult, type SteeringOptimizationRow, type SteeringOptimizationSelector, type SteeringOptimizerBackend, type SteeringOptimizerConfig, type SteeringRolePrompt, type StepAttribution, type StopDecision, type StreamingDetector, type SuboptimalCode, type SuboptimalSignal, SubprocessSandboxDriver, type SubprocessSandboxDriverOptions, type SummaryTable, type SummaryTableOptions, type SummaryTableRow, type SupervisorRunNodeRole, type SupervisorRunReader, type SupervisorRunReport, type SupervisorRunRollup, type SupervisorRunSources, type SupervisorRunTree, type SynthesisReason, type SynthesisTarget, TOOL_ARGS_CAPTURED, TOOL_LATENCY_MS, TOOL_NAME, TOOL_NAME_ATTR_KEYS, TRACE_ANALYST_ACTOR_DESCRIPTION, TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, TRACE_ANALYST_TRUNCATION_MARKER_PREFIX, TRACE_SCHEMA_VERSION, type TaskGold, type TaskHeadroom, type TestGradedRunOptions, type TestGradedRunResult, type TestGradedScenario, type TestOutputParser, type TestResult, type TextMatcher, type ThresholdContract, TokenCounter, type TokenSpec, type ToolCallEventLike, type ToolDef, type ToolMatcher, ToolSpan, ToolSpanOtlpInput, type ToolSpansToTraceAnalysisStoreOptions, type ToolStats, ToolTraceMissingError, type ToolUseMetrics, type ToolUseOptions, type TraceAggregate, type TraceAnalysisStore, type TraceAnalystByteBudgets, type TraceAnalystFilters, type TraceAnalystGolden, type TraceAnalystHookOptions, type TraceAnalystKindSpec, type TraceAnalystSpan, type TraceAnalystSpanKind, type TraceAnalystSpanStatus, type TraceAnalystTraceSummary, type TraceContract, TraceContractBuilder, TraceEmitter, TraceEmitterOptions, TraceEvent, TraceFileMissingError, type TraceInsightContext, type TraceInsightFinding, type TraceInsightPanelRole, type TraceInsightPromptInput, type TraceInsightQualityGate, type TraceInsightQuestion, type TraceInsightReadiness, type TraceInsightSuite, type TraceInsightTask, TraceNotFoundError, TraceStore, TraceStoreSource, TraceStoreToOtlpOptions, type TracedAnalystOptions, type TracedJudgeOptions, TracesToOtlpResult, type Trajectory, type TrajectoryStep, type TreatmentClass, type TreatmentGate, type TreatmentGateInput, type TreatmentGateOptions, type TrialTrace, type Turn, type TurnMetrics, type TurnResult, UI_FINDING_SEVERITIES, UI_LENSES, UNIVERSAL_FINDERS, type UiFinding, type UiFindingScreenshot, type UiFindingSeverity, type UiLens, type Unavailable, UserQuestion, type ValidationContext, ValidationError, type ValidationIssue, type ValidationResult, type VerbosityBiasResult, type Verdict, type VerdictCacheStats, type VerdictCacheStore, type Verification, VerificationError, type VerificationReport, type VerifyContext, type VerifyFn, type VerifyOptions, type ViewSpansResult, type ViewTraceOversized, type ViewTraceResult, type VisualDiffOptions, type VisualDiffResult, type ViteDeployRunnerInput, WILCOXON_EXACT_MAX_N, type WeightedCompositeInput, type WeightedCompositeResult, type WelchTestResult, type WelchTestStatus, type WilcoxonSignedRankResult, type WorkerDriverContext, type WorkflowTopology, type WorkspaceAssertion, type WorkspaceAssertionResult, type WorkspaceInspector, type WorkspaceSnapshot, type WranglerDeployRunnerInput, acquisitionPlansForKnowledgeGaps, agentProfileCellHashMaterial, agentProfileCellKey, agentProfileHash, agentProfileId, agentProfileModelId, agentVisibleFields, aggregateJudgeVerdicts, aggregateLlm, aggregatePrReviewScore, aggregateRunScore, allCriticalPassed, analyzeAntiSlop, analyzeSeries, analyzeSupervisorRun, analyzeSupervisorRunSources, analyzeTraces, appendScorecard, applyLlmSpanOtlpAttributes, applyToolSpanOtlpAttributes, argHash, asNumber, asString, assertCapabilityHeadroom, assertCrossFamily, assertLlmRoute, assertMinted, assertMintedLines, assertModelsServed, assertNoHiddenLeak, assertProductBenchmarkRun, assertRealAgentReceipts, assertRealBackend, assertReleaseConfidence, assertRolloutLine, assertRunAgentProfileCell, assertRunCaptured, assertSingleBackend, assignFeedbackSplit, assignHeldOutTag, attachCostToReport, attest, attributeCounterfactuals, backoffMs, deterministicSplit as benchmarkDeterministicSplit, index_d_exports as benchmarks, benjaminiHochberg, bisect, blendHeldout, blockingKnowledgeEval, bonferroni, bootstrapCi, buildAgentInterfaceProfileCell, buildAgentProfileCell, buildDefaultAnalystRegistry, buildDriverSystemPrompt, buildProductBenchmarkManifest, buildReflectionPrompt, buildReviewerPrompt, buildTraceAnalystTools, buildTraceInsightContext, buildTraceInsightPrompt, buildTrajectory, buildWorkerDriverSystemPrompt, byteLengthRange, cachedJudge, calibrateJudge, calibrateJudgeContinuous, callLlm, callLlmJson, canaryLeakView, canonicalJson, canonicalize, capabilityHeadroom, captureFetchToRawSink, causalAttribution, checkBehavioralCanary, checkCanaries, checkSlos, checkTraceContracts, clamp01, classifyFailure, classifyOtlpSpanRole, classifyTreatment, claudeCodeSupervisorRunReader, cliffsDelta, clusteredPairedBinary, cohensD, collectionPreserved, commentsForSource, commitBisect, comparePairedArms, compareReferenceReplay, compareToBaseline, compilerJudge, completionVerdict, composeParsers, composeValidators, computeExperimentStats, computeFindingId, computeToolUseMetrics, computeTraceMetrics, confidenceInterval, containsAll, contentHash, contextInputTokens, continuousAgreement, contractJudge, controlFailureClassFromVerification, controlRunToFeedbackTrajectory, controlRunToRunRecord, convertTraceStoresToOtlp, corpusInterRaterAgreement, corpusInterRaterAgreementFromJudgeScores, costForTokenPricing, costForUsage, costReceiptFromLlm, costReceiptFromLlmError, costReport, createAnalystAi, createAntiSlopJudge, createChatClient, createDefaultReviewer, createFeedbackTrajectory, createIntentMatchJudge, createLlmCorrectnessChecker, createLlmReviewer, createOtelExporter, createOtelTracingStore, createReferenceEquivalenceJudge, createReplayFetch, createSandboxPool, createSemanticConceptJudge, createTokenRecallChecker, createTraceAnalystKind, crossTraceDiff, crowdingDistance, dataDescriptionBits, decideNextUserTurn, decidePairedPromotion, decideReferenceReplayPromotion, decideReferenceReplayRunPromotion, defaultBlendWeights, defaultIsMaterial, defaultProviderRedactor, defaultReferenceReplayMatcher, defaultTraceInsightPanel, deployGateLayer, describeTraceInsightScope, diffFindings, diffScorecard, discoverPersonas, distillPlaybook, domainEvidencePattern, dominates, eProcess, ensembleJudge, errorStreakDetector, estimateCost, estimateTokens, evaluateActionPolicy, evaluateContract, evaluateHypothesis, evaluateInterimReleaseConfidence, evaluateOracles, evaluateReleaseConfidence, evaluateTraceContract, executeScenario, expandProfileAxes, expectAgent, exportProductBenchmark, exportProductBenchmarkRuns, exportRewardModel, exportRunAsOtlp, extractAssetUrls, extractErrorCount, extractOtlpAttributes, extractProducedState, extractUsage, extractUsageFromResponse, extractUsageFromSse, feedbackTrajectoriesToDatasetScenarios, feedbackTrajectoriesToOptimizerRows, feedbackTrajectoryToDatasetScenario, feedbackTrajectoryToOptimizerRow, fileContains, fileExists, fileExperimentStore, fileVerdictCache, findAutoMatchNoExpectation, findConstructorCwdDropped, findFallbackToPass, findLiteralTruePass, findProductBenchmarkArtifacts, findSkipCountsAsPass, firstNumberAttr, firstStringAttr, flattenOtlpExportToNdjson, flowLayer, fnv1a32, formatBenchmarkReport, formatDriverReport, formatFindings, formatScorecardDiff, fromHarborTrajectory, gainHistogram, gateTreatmentApplied, gateTreatmentFromMetrics, gateTreatmentFromSpans, gateTreatmentFromToolSpans, ghCliClient, gitProvenanceReader, precision as goldenPrecision, gradeOnHidden, gradeSemanticStatus, groupBy, groupRunsByAgentProfileCell, harnessAxisOf, hasCapturedToolArgs, hashContent, hashJson, hashScenarios, hashToUnit, hiddenGrade, holm, htmlContainsElement, httpGithubClient, improvementVerdict, inMemoryExperimentStore, inMemoryReferenceReplayStore, inMemoryReviewStore, inMemoryRunRecordBackend, inMemoryVerdictCache, inferDomainKeywords, inferOtlpKind, interRaterReliability, interpretCliffs, iqr, isBinaryOutcomeVector, isHiddenDestination, isJudgeSpan, isLlmSpan, isModelPriced, isOtelConfigured, isOtlpModelCall, isRealnessGated, isRetrievalSpan, isRolloutLine, isRunRecord, isSandboxSpan, isToolSpan, isTrainableSplit, isTransientLlmError, isUnavailable, iterateRawCalls, jestTestParser, jsonHasKeys, jsonShape, jsonlReferenceReplayStore, jsonlReviewStore, jsonlRunRecordBackend, judgeFamily, judgeReplayGate, judgeSpans, keyPreserved, knowledgeReadinessTracePayload, leaderboard, linterJudge, llmJudge, llmSpanFromProvider, llmSpans, loadScorecard, loadScorerFromGrader, localCommandRunner, lowercaseMutator, makeEvalTools, makeFinding, makeProposalFinding, mannWhitneyU, mapConcurrent, matchGoldens, matchSpan, maximumChargeForLlmRequest, mcnemar, mcnemarPower, mcnemarRequiredN, mergeLayerResults, mergeSteeringBundle, minimumPairsForPairedDeltaTest, mintRolloutRows, modelDescriptionBits, modelHasSnapshot, modelPriceKey, mulberry32, multiToolchainLayer, noProgressDetector, normalCdf, normalizeScores, notBlocked, objectiveEval, observeAll, observedScore, observedSplitScore, otelRunCompleteHook, otlpRowsToRunRecords, otlpRowsToTraceRunRecords, otlpToRunRecords, otlpToTraceRunRecords, pairArms, pairRunRecords, pairedBinaryScale, pairedBootstrap, pairedCohensDz, pairedDecisionShape, pairedDeltaTest, pairedDeltaTieFraction, pairedEvalueSequence, pairedMde, pairedRiskDifference, pairedRiskDifferenceExact, pairedRiskDifferenceScore, pairedSignTest, pairedTTest, paraphraseRobustness, paraphraseRobustnessScenarios, paretoChart, paretoFrontier, paretoFrontierWithCrowding, parseCorrectnessResponse, parseFeedbackTrajectoriesJsonl, parseReflectionResponse, parseRunRecordSafe, parseRuntimeTrajectoryHookEvent, partialCredit, partitionHeldOut, passAtK, passOrthogonality, pearsonR, pixelDeltaRatio, planTraceInsightQuestions, politenessPrefixMutator, positionalBias, preflightModels, printDriverSummary, probeLlm, productBenchmarkIntegrityFailures, productBenchmarkMutableSurfaces, productBenchmarkRepoIdentity, productBenchmarkSplits, index_d_exports$1 as profile, projectOtlpFlatLine, projectRuntimeTrajectoryEvidence, promptBisect, proposeSynthesisTargets, providerFromBaseUrl, pytestTestParser, ranks, readClaudeCodeSupervisorRun, readOtlpStatus, readProductBenchmarkManifest, readProductBenchmarkRecords, recordRuns, recordRunsToScorecard, redTeamDataset, redTeamReport, redactString, redactValue, referenceReplayRunsToSteeringRows, referenceReplayScenarioToRunScore, regexMatch, regexMatches, relabelImportedSplit, renderMarkdownReport, renderPlaybookMarkdown, renderPreferenceMemoryMarkdown, renderPriorFindings, renderReleaseReport, renderSteeringText, renderSupervisorRunHeadline, renderSupervisorRunMarkdown, renderUpstreamFindings, repeatedActionDetector, replayFeedbackTrajectories, replayFeedbackTrajectory, replayScorerOverCorpus, replayTraceThroughJudge, requireAgentProfileCell, requiredPairedSampleSize, requiredSampleSize, researchReport, resolveModelPricing, resolveSeat, rollupSupervisorRuns, roundTripRunRecord, routeFields, rowCount, rowWhere, runAgentControlLoop, runAssertions, runBehavioralCanaries, runCanaries, runCounterfactual, runE2EWorkflow, runEvalCampaign, runExpectations, runFailureClass, runHarnessExperiment, runIntentMatchJudge, runJudgeFleet, runKeywordCoverageJudge, runKeywordCoverageJudgeUrl, runLiveProof, runProposeReview, runProposeReviewAsControlLoop, runRecordToProductBenchmarkRecord, runReferenceEquivalenceJudge, runReferenceReplay, runScore, runSelfPlay, runSemanticConceptJudge, runTaskScore, runTestGradedScenario, runsForScenario, scalarScore, scanForMuffledGates, scoreContinuity, scoreFromEvals, scoreKnowledgeReadiness, scoreOrigin, scorePrReviewComments, scorePrReviewSource, scoreRedTeamOutput, scoreReferenceReplay, scoreTraceInsightReadiness, seatPresets, securityJudge, selectHarnessVariant, selfPreference, sentenceReorderMutator, serializeFeedbackTrajectoriesJsonl, showMeasured, signManifest, spearmanR, statusAdvanced, stopOnNoProgress, stopOnRepeatedAction, stringField, stripFencedJson, studentTCdf, subjectiveEval, summarizeAgentReceiptIntegrity, summarizeBackendIntegrity, summarizeHarnessResults, summarizePrReviewBenchmark, summarizePreferenceMemory, summaryTable, supervisorRunRolloutLines, testJudge, textInSnapshot, throwIfRunIncomplete, toAgentProfileJson, toHarborTrajectories, toHarborTrajectory, toJsonl, toLangfuseEnvelope, toOpenAiTool, toPrometheusText, toRewardRows, toSftRows, tokenizeDomainWords, toolNamesForRun, toolSpans, toolSpansToTraceAnalysisStore, traceAnalystFunctionGroup, traceAnalystOnRunComplete, traceContract, traceJudge, traceJudgeEnsemble, traceSpanKindToOpenInferenceKind, tracedAnalyzeTraces, trainingReward, trainingScore, typoMutator, unmintableReasons, urlContains, userQuestionsForKnowledgeGaps, validateAgentProfileCell, validateProductBenchmarkManifest, validateProductBenchmarkRecord, validateProductBenchmarkRun, validateRolloutLine, validateRunRecord, verbosityBias, verifyAgentProfileCell, verifyAttestation, verifyCompletion, verifyManifest, visualDiff, viteDeployRunner, vitestTestParser, weightedComposite, weightedMean, weightedRecall, welchsTTest, whitespaceCollapseMutator, wilcoxonSignedRank, wilson, withAssignedFeedbackSplit, withHeldoutBlend, withJudgeRetry, withOtelPipeline, wranglerDeployRunner, writeSupervisorRunReport };
 //# sourceMappingURL=index.d.ts.map

package/dist/index.js

@@ -25,7 +25,7 @@ import { i as toRewardRows, r as toJsonl, s as toSftRows } from "./exporters-q9i
 import { a as toHarborTrajectories, i as relabelImportedSplit, n as HARBOR_IMPORT_GAP, o as toHarborTrajectory, r as fromHarborTrajectory, t as ATIF_SCHEMA_VERSION } from "./rollout-DLSUIWLu.js";
 import { t as buildTrajectory } from "./trajectory-D_7rLrvE.js";
 import { n as unmintableReasons, t as mintRolloutRows } from "./mint-DyRUc9k6.js";
-import { D as showMeasured, E as isUnavailable, S as rollupSupervisorRuns, T as SUPERVISOR_RUN_SCHEMA, _ as claudeCodeSupervisorRunReader, f as renderSupervisorRunHeadline, l as writeSupervisorRunReport, n as analyzeSupervisorRun, p as renderSupervisorRunMarkdown, t as supervisorRunRolloutLines, v as readClaudeCodeSupervisorRun, y as analyzeSupervisorRunSources } from "./supervisor-run-B7lUGoyZ.js";
+import { D as showMeasured, E as isUnavailable, S as rollupSupervisorRuns, T as SUPERVISOR_RUN_SCHEMA, _ as claudeCodeSupervisorRunReader, f as renderSupervisorRunHeadline, l as writeSupervisorRunReport, n as analyzeSupervisorRun, p as renderSupervisorRunMarkdown, t as supervisorRunRolloutLines, v as readClaudeCodeSupervisorRun, y as analyzeSupervisorRunSources } from "./supervisor-run-Dr5HnTup.js";
 import { n as TRACE_ANALYST_ACTOR_DESCRIPTION, r as TRACE_ANALYST_ACTOR_DESCRIPTION_VERSION, t as analyzeTraces } from "./analyst-j5je5J7c.js";
 import { C as planTraceInsightQuestions, E as traceAnalystOnRunComplete, S as inferDomainKeywords, T as tokenizeDomainWords, _ as buildTraceInsightContext, a as convertTraceStoresToOtlp, b as describeTraceInsightScope, c as otelRunCompleteHook, d as captureFetchToRawSink, f as otlpRowsToRunRecords, g as flattenOtlpExportToNdjson, h as otlpToTraceRunRecords, i as iterateRawCalls, l as OTEL_AGENT_EVAL_SCOPE, m as otlpToRunRecords, n as ReplayCacheMissError, o as createOtelExporter, p as otlpRowsToTraceRunRecords, r as createReplayFetch, s as createOtelTracingStore, t as ReplayCache, u as exportRunAsOtlp, v as buildTraceInsightPrompt, w as scoreTraceInsightReadiness, x as domainEvidencePattern, y as defaultTraceInsightPanel } from "./replay-C6wRg47C.js";
 import { n as extractUsageFromResponse, r as extractUsageFromSse, t as extractUsage } from "./extract-usage-DIQpN-ww.js";

package/dist/openapi.json

@@ -2,7 +2,7 @@
   "openapi": "3.1.0",
   "info": {
     "title": "@tangle-network/agent-eval — wire protocol",
-    "version": "0.135.3",
+    "version": "0.136.0",
     "description": "HTTP and stdio RPC interface to agent-eval. The TypeScript runtime is the source of truth; this spec is the contract that cross-language clients (Python, Rust, Go) generate from.\n\nWire-protocol version: 1.0.0. Bumps on breaking changes to request/response schemas.",
     "contact": {
       "name": "Tangle Network",

package/dist/supervisor-run/index.d.ts

@@ -1,2 +1,2 @@
-import { $ as isUnavailable, A as rollupSupervisorRuns, B as RollupCellRow, C as CloseRow, D as analyzeSupervisorRunSources, E as WorkerLogFacts, F as OrchestrationMetrics, G as SupervisorRunReader, H as SUPERVISOR_RUN_SCHEMA, I as OutcomeMetrics, J as SupervisorRunSources, K as SupervisorRunReport, L as PatchStats, M as EconomicsMetrics, N as Measured, O as parsePatch, P as NO_SOURCE_LIMITS, Q as WorkerLogSource, R as PerWorkerRow, S as readClaudeCodeSupervisorRun, T as SupervisorTreeFacts, U as SourceLimits, V as SUPERVISOR_RUN_ROLLUP_SCHEMA, W as SteerBreakdown, X as Unavailable, Y as SupervisorRunTree, Z as WallDistribution, _ as ClaudeCodeReaderOptions, a as renderSupervisorRunMarkdown, b as DEFAULT_STEER_TOOLS, c as analyzeSupervisorRun, d as loopsSupervisorRunReader, et as showMeasured, f as readLoopsSupervisorRun, g as writeSupervisorRunReportSafe, h as writeSupervisorRunReport, i as renderSupervisorRunHeadline, j as DecisionMetrics, k as parseSupervisorTree, l as findSupervisorRunDirIn, m as supervisorReportStem, n as supervisorRunRolloutLines, o as LoopsReaderOptions, p as reportSupervisorRound, q as SupervisorRunRollup, r as renderSupervisorRollupMarkdown, s as WriteSupervisorRunOptions, t as SupervisorRolloutOptions, tt as unavailable, u as findSupervisorRunDirs, v as DEFAULT_CANCEL_TOOLS, w as SpawnRow, x as claudeCodeSupervisorRunReader, y as DEFAULT_SPAWN_TOOLS, z as RoleSpend } from "../index-C61Wi7yg.js";
-export { type ClaudeCodeReaderOptions, type CloseRow, DEFAULT_CANCEL_TOOLS, DEFAULT_SPAWN_TOOLS, DEFAULT_STEER_TOOLS, type DecisionMetrics, type EconomicsMetrics, type LoopsReaderOptions, type Measured, NO_SOURCE_LIMITS, type OrchestrationMetrics, type OutcomeMetrics, type PatchStats, type PerWorkerRow, type RoleSpend, type RollupCellRow, SUPERVISOR_RUN_ROLLUP_SCHEMA, SUPERVISOR_RUN_SCHEMA, type SourceLimits, type SpawnRow, type SteerBreakdown, type SupervisorRolloutOptions, type SupervisorRunReader, type SupervisorRunReport, type SupervisorRunRollup, type SupervisorRunSources, type SupervisorRunTree, type SupervisorTreeFacts, type Unavailable, type WallDistribution, type WorkerLogFacts, type WorkerLogSource, type WriteSupervisorRunOptions, analyzeSupervisorRun, analyzeSupervisorRunSources, claudeCodeSupervisorRunReader, findSupervisorRunDirIn, findSupervisorRunDirs, isUnavailable, loopsSupervisorRunReader, parsePatch, parseSupervisorTree, readClaudeCodeSupervisorRun, readLoopsSupervisorRun, renderSupervisorRollupMarkdown, renderSupervisorRunHeadline, renderSupervisorRunMarkdown, reportSupervisorRound, rollupSupervisorRuns, showMeasured, supervisorReportStem, supervisorRunRolloutLines, unavailable, writeSupervisorRunReport, writeSupervisorRunReportSafe };
+import { $ as WorkerLogSource, A as rollupSupervisorRuns, B as RollupCellRow, C as CloseRow, D as analyzeSupervisorRunSources, E as WorkerLogFacts, F as OrchestrationMetrics, G as SupervisorRunNodeRole, H as SUPERVISOR_RUN_SCHEMA, I as OutcomeMetrics, J as SupervisorRunRollup, K as SupervisorRunReader, L as PatchStats, M as EconomicsMetrics, N as Measured, O as parsePatch, P as NO_SOURCE_LIMITS, Q as WallDistribution, R as PerWorkerRow, S as readClaudeCodeSupervisorRun, T as SupervisorTreeFacts, U as SourceLimits, V as SUPERVISOR_RUN_ROLLUP_SCHEMA, W as SteerBreakdown, X as SupervisorRunTree, Y as SupervisorRunSources, Z as Unavailable, _ as ClaudeCodeReaderOptions, a as renderSupervisorRunMarkdown, b as DEFAULT_STEER_TOOLS, c as analyzeSupervisorRun, d as loopsSupervisorRunReader, et as isUnavailable, f as readLoopsSupervisorRun, g as writeSupervisorRunReportSafe, h as writeSupervisorRunReport, i as renderSupervisorRunHeadline, j as DecisionMetrics, k as parseSupervisorTree, l as findSupervisorRunDirIn, m as supervisorReportStem, n as supervisorRunRolloutLines, nt as unavailable, o as LoopsReaderOptions, p as reportSupervisorRound, q as SupervisorRunReport, r as renderSupervisorRollupMarkdown, s as WriteSupervisorRunOptions, t as SupervisorRolloutOptions, tt as showMeasured, u as findSupervisorRunDirs, v as DEFAULT_CANCEL_TOOLS, w as SpawnRow, x as claudeCodeSupervisorRunReader, y as DEFAULT_SPAWN_TOOLS, z as RoleSpend } from "../index-CyC1BTmn.js";
+export { type ClaudeCodeReaderOptions, type CloseRow, DEFAULT_CANCEL_TOOLS, DEFAULT_SPAWN_TOOLS, DEFAULT_STEER_TOOLS, type DecisionMetrics, type EconomicsMetrics, type LoopsReaderOptions, type Measured, NO_SOURCE_LIMITS, type OrchestrationMetrics, type OutcomeMetrics, type PatchStats, type PerWorkerRow, type RoleSpend, type RollupCellRow, SUPERVISOR_RUN_ROLLUP_SCHEMA, SUPERVISOR_RUN_SCHEMA, type SourceLimits, type SpawnRow, type SteerBreakdown, type SupervisorRolloutOptions, type SupervisorRunNodeRole, type SupervisorRunReader, type SupervisorRunReport, type SupervisorRunRollup, type SupervisorRunSources, type SupervisorRunTree, type SupervisorTreeFacts, type Unavailable, type WallDistribution, type WorkerLogFacts, type WorkerLogSource, type WriteSupervisorRunOptions, analyzeSupervisorRun, analyzeSupervisorRunSources, claudeCodeSupervisorRunReader, findSupervisorRunDirIn, findSupervisorRunDirs, isUnavailable, loopsSupervisorRunReader, parsePatch, parseSupervisorTree, readClaudeCodeSupervisorRun, readLoopsSupervisorRun, renderSupervisorRollupMarkdown, renderSupervisorRunHeadline, renderSupervisorRunMarkdown, reportSupervisorRound, rollupSupervisorRuns, showMeasured, supervisorReportStem, supervisorRunRolloutLines, unavailable, writeSupervisorRunReport, writeSupervisorRunReportSafe };

package/dist/supervisor-run/index.js

@@ -1,2 +1,2 @@
-import { C as NO_SOURCE_LIMITS, D as showMeasured, E as isUnavailable, O as unavailable, S as rollupSupervisorRuns, T as SUPERVISOR_RUN_SCHEMA, _ as claudeCodeSupervisorRunReader, a as loopsSupervisorRunReader, b as parsePatch, c as supervisorReportStem, d as renderSupervisorRollupMarkdown, f as renderSupervisorRunHeadline, g as DEFAULT_STEER_TOOLS, h as DEFAULT_SPAWN_TOOLS, i as findSupervisorRunDirs, l as writeSupervisorRunReport, m as DEFAULT_CANCEL_TOOLS, n as analyzeSupervisorRun, o as readLoopsSupervisorRun, p as renderSupervisorRunMarkdown, r as findSupervisorRunDirIn, s as reportSupervisorRound, t as supervisorRunRolloutLines, u as writeSupervisorRunReportSafe, v as readClaudeCodeSupervisorRun, w as SUPERVISOR_RUN_ROLLUP_SCHEMA, x as parseSupervisorTree, y as analyzeSupervisorRunSources } from "../supervisor-run-B7lUGoyZ.js";
+import { C as NO_SOURCE_LIMITS, D as showMeasured, E as isUnavailable, O as unavailable, S as rollupSupervisorRuns, T as SUPERVISOR_RUN_SCHEMA, _ as claudeCodeSupervisorRunReader, a as loopsSupervisorRunReader, b as parsePatch, c as supervisorReportStem, d as renderSupervisorRollupMarkdown, f as renderSupervisorRunHeadline, g as DEFAULT_STEER_TOOLS, h as DEFAULT_SPAWN_TOOLS, i as findSupervisorRunDirs, l as writeSupervisorRunReport, m as DEFAULT_CANCEL_TOOLS, n as analyzeSupervisorRun, o as readLoopsSupervisorRun, p as renderSupervisorRunMarkdown, r as findSupervisorRunDirIn, s as reportSupervisorRound, t as supervisorRunRolloutLines, u as writeSupervisorRunReportSafe, v as readClaudeCodeSupervisorRun, w as SUPERVISOR_RUN_ROLLUP_SCHEMA, x as parseSupervisorTree, y as analyzeSupervisorRunSources } from "../supervisor-run-Dr5HnTup.js";
 export { DEFAULT_CANCEL_TOOLS, DEFAULT_SPAWN_TOOLS, DEFAULT_STEER_TOOLS, NO_SOURCE_LIMITS, SUPERVISOR_RUN_ROLLUP_SCHEMA, SUPERVISOR_RUN_SCHEMA, analyzeSupervisorRun, analyzeSupervisorRunSources, claudeCodeSupervisorRunReader, findSupervisorRunDirIn, findSupervisorRunDirs, isUnavailable, loopsSupervisorRunReader, parsePatch, parseSupervisorTree, readClaudeCodeSupervisorRun, readLoopsSupervisorRun, renderSupervisorRollupMarkdown, renderSupervisorRunHeadline, renderSupervisorRunMarkdown, reportSupervisorRound, rollupSupervisorRuns, showMeasured, supervisorReportStem, supervisorRunRolloutLines, unavailable, writeSupervisorRunReport, writeSupervisorRunReportSafe };