@tangle-network/agent-eval 0.144.4 → 0.144.6

package/CHANGELOG.md

@@ -4,6 +4,20 @@ All notable changes to `@tangle-network/agent-eval` and its sibling `agent-eval-
 
 ---
 
+## [0.144.6] - 2026-08-09 - interaction binding cohort alignment
+
+### Changed
+
+- Align the exact `@tangle-network/agent-core` and `@tangle-network/agent-interface` dependencies with `0.5.4` and `0.46.1`, so consumers share the canonical provider interaction binding without nested contract copies.
+
+## [0.144.5] - 2026-08-09 - structured profile and prime cohort alignment
+
+### Changed
+
+- Align the exact `@tangle-network/agent-core` and `@tangle-network/agent-interface` dependencies with `0.5.3` and `0.46.0`.
+- Preserve the shared profile contract's structured system-prompt capabilities and `prime` harness through Eval's profile matrix without adding a local profile schema or capability copy.
+- Multishot agent requests now preserve the profile's ordered `systemPrompt` and `appendSystemPrompt` values.
+
 ## [0.144.4] - 2026-08-04 - prompt-cache cohort alignment
 
 ### Changed

package/dist/{agent-profile-cell-Cw0PVwDr.d.ts → agent-profile-cell-BOP-iA9Q.d.ts}

@@ -1,4 +1,4 @@
-import { l as ValidationError } from "./errors-DkfjIDvD.js";
+import { l as ValidationError } from "./errors-CKPfb2aH.js";
 import { AgentProfile } from "@tangle-network/agent-interface";
 //#region src/agent-profile-cell.d.ts
 type AgentProfileCellSchemaVersion = 'agent-profile-cell/v1';
@@ -105,4 +105,4 @@ type AgentInterfaceProfileLike = AgentProfile & {
 declare function buildAgentInterfaceProfileCell(profile: AgentInterfaceProfileLike, input: Omit<AgentProfileCellInput, 'profileId' | 'sourceProfile'>): Promise<AgentProfileCell>;
 //#endregion
 export { verifyAgentProfileCell as C, validateAgentProfileCell as S, buildAgentInterfaceProfileCell as _, AgentProfileCellSchemaVersion as a, requireAgentProfileCell as b, AgentProfileHarness as c, AgentProfileKind as d, AgentProfileSource as f, assertRunAgentProfileCell as g, agentProfileCellKey as h, AgentProfileCellInput as i, AgentProfileJson as l, agentProfileCellHashMaterial as m, AgentInterfaceProfileLike as n, AgentProfileCellValidationError as o, AgentProfileSourceInput as p, AgentProfileCell as r, AgentProfileDimensionValue as s, AGENT_PROFILE_KINDS as t, AgentProfileJsonObject as u, buildAgentProfileCell as v, toAgentProfileJson as x, groupRunsByAgentProfileCell as y };
-//# sourceMappingURL=agent-profile-cell-Cw0PVwDr.d.ts.map
+//# sourceMappingURL=agent-profile-cell-BOP-iA9Q.d.ts.map

package/dist/{agent-profile-cell-Cw0PVwDr.d.ts.map → agent-profile-cell-BOP-iA9Q.d.ts.map}

@@ -1 +1 @@
-{"version":3,"file":"agent-profile-cell-Cw0PVwDr.d.ts","names":[],"sources":["../src/agent-profile-cell.ts"],"mappings":";;;KAIY;KAEA;GAA4B,cAAc;;KAE1C,sDAKR,qBACA;KAEQ;UAEK;;EAEf;;EAEA;;UAGe;EACf;;EAEA;;EAEA,UAAU;;UAGK;EACf;EACA;EACA;;UAGe;EACf;EACA,eAAe;EACf,UAAU;EACV;EACA;EACA,aAAa,eAAe;;UAGb;EACf,eAAe;EACf;EACA;EACA,eAAe;EACf,UAAU;EACV;EACA;EACA,aAAa,eAAe;;cAGjB,wCAAwC;WAC1C;EACT,YAAY,iBAAiB;;iBAST,sBACpB,OAAO,wBACN,QAAQ;iBAMK,6BACd,MAAM,mBACL,KAAK;;;;iBASc,uBAAuB,MAAM,mBAAmB;iBAQtD,yBAAyB,iBAAiB;iBAqB1C,wBAAwB;EACtC;EACA,eAAe;IACb;iBAUY,oBAAoB;EAClC;EACA,eAAe;;iBAKK,0BAA0B;EAC9C;EACA;EACA;EACA,eAAe;IACb,QAAQ;iBAuBI,4BACd;EAAY;EAAe,eAAe;GAC1C,kBAAkB,MAAM,YAAY;;;;;cA0NzB;;;;WAIX;;KAGU,2BAA2B,kCAAkC;;;;;iBAMzD,mBAAmB,iBAAiB;;KAoBxC,4BAA4B;EAAiB;EAAc;;;;;;;;;;;iBAWjD,+BACpB,SAAS,2BACT,OAAO,KAAK,wDACX,QAAQ"}
+{"version":3,"file":"agent-profile-cell-BOP-iA9Q.d.ts","names":[],"sources":["../src/agent-profile-cell.ts"],"mappings":";;;KAIY;KAEA;GAA4B,cAAc;;KAE1C,sDAKR,qBACA;KAEQ;UAEK;;EAEf;;EAEA;;UAGe;EACf;;EAEA;;EAEA,UAAU;;UAGK;EACf;EACA;EACA;;UAGe;EACf;EACA,eAAe;EACf,UAAU;EACV;EACA;EACA,aAAa,eAAe;;UAGb;EACf,eAAe;EACf;EACA;EACA,eAAe;EACf,UAAU;EACV;EACA;EACA,aAAa,eAAe;;cAGjB,wCAAwC;WAC1C;EACT,YAAY,iBAAiB;;iBAST,sBACpB,OAAO,wBACN,QAAQ;iBAMK,6BACd,MAAM,mBACL,KAAK;;;;iBASc,uBAAuB,MAAM,mBAAmB;iBAQtD,yBAAyB,iBAAiB;iBAqB1C,wBAAwB;EACtC;EACA,eAAe;IACb;iBAUY,oBAAoB;EAClC;EACA,eAAe;;iBAKK,0BAA0B;EAC9C;EACA;EACA;EACA,eAAe;IACb,QAAQ;iBAuBI,4BACd;EAAY;EAAe,eAAe;GAC1C,kBAAkB,MAAM,YAAY;;;;;cA0NzB;;;;WAIX;;KAGU,2BAA2B,kCAAkC;;;;;iBAMzD,mBAAmB,iBAAiB;;KAoBxC,4BAA4B;EAAiB;EAAc;;;;;;;;;;;iBAWjD,+BACpB,SAAS,2BACT,OAAO,KAAK,wDACX,QAAQ"}

package/dist/analyst/index.d.ts

@@ -1,14 +1,14 @@
 import { a as MultiLayerVerifier, l as VerifyOptions, o as Severity } from "../multi-layer-verifier-BHY1gWAc.js";
-import { A as parseFindingSubject, C as FINDING_SUBJECT_KINDS, D as FindingSubjectStringSchema, E as FindingSubjectKind, F as DefineExactCustomAnalystOptions, G as SemanticConceptJudgeOptions, I as defineCustomAnalyst, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, O as KIND_EXPECTED_SUBJECTS, P as DefineCustomAnalystOptions, S as FINDING_SUBJECT_GRAMMAR_PROMPT, T as FindingSubject, W as SemanticConceptJudgeInput, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, a as SkillUsageScanConfig, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, i as SkillUsageReport, j as renderFindingSubject, k as findingSubjectGrammarPromptFor, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, o as buildSkillUsageReport, p as CONTROL_INTEGRITY_ANALYST, r as SkillUsageRecord, s as emitSkillUsageFindings, t as SKILL_USAGE_ANALYST, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, w as FINDING_SUBJECT_SYNTAX, x as diffFindings, y as PersistedFinding } from "../skill-usage-DtpLou9L.js";
-import { b as CustomTokenPricing, c as CostLedgerHandle } from "../cost-ledger-FuQvHxPm.js";
-import { A as ChatClient, B as SandboxSdkTransportOpts, F as CreateChatClientOpts, I as CustomTransportOpts, L as DirectProviderTransportOpts, M as ChatResponse, N as ChatTransport, P as CliBridgeTransportOpts, R as MockTransportOpts, V as createChatClient, j as ChatRequest, k as ChatCallOpts, m as JudgeInput, p as JudgeFn, z as RouterTransportOpts } from "../types-BjMFz88h.js";
-import { _ as makeFinding, a as AnalystInputKind, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, x as TraceAnalysisStore } from "../types-y8jrxXWd.js";
-import { a as createTraceAnalyst, c as runTraceAnalyst, d as deriveEfficiencyFindings, i as TraceAnalystDefinition, l as BehavioralAnalystOptions, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions, u as behavioralAnalyst } from "../default-registry-iXfu2trt.js";
-import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "../exact-types-BygCBR4L.js";
-import { A as ExactAnalystRunExecutionError, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, T as AnalystHooks, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy } from "../completion-verifier-EJERfFwF.js";
-import { C as scoreAnalystFindings, S as traceStoreEvidenceResolver, _ as AnalystIssueExpectation, a as AnalystBenchmarkLabelState, b as registryBenchmarkRunner, c as AnalystBenchmarkProvenance, d as AnalystBenchmarkSummary, f as AnalystEvidenceExpectation, g as AnalystFindingScore, h as AnalystEvidenceResolver, i as AnalystBenchmarkError, l as AnalystBenchmarkResult, m as AnalystEvidenceResolutionError, n as AnalystBenchmarkDatasetRef, o as AnalystBenchmarkObservation, p as AnalystEvidenceResolution, r as AnalystBenchmarkDescriptor, s as AnalystBenchmarkOutput, t as AnalystBenchmarkCase, u as AnalystBenchmarkRunner, v as AnalystLatencyDistribution, x as runAnalystBenchmark, y as RunAnalystBenchmarkOptions } from "../benchmark-CP6kWfj8.js";
-import { f as ExternalOptimizerModelExecutionObservation, h as ExternalOptimizerRunnerCommand, l as ExternalOptimizerModelCall } from "../external-optimizer-contracts-CdmX2K2S.js";
-import { a as TraceAnalysisEngineRequest, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, f as RawAnalystEvidenceSchema, g as parseRawFinding, h as evidenceRefsFromRawFinding, i as TraceAnalysisEngine, l as ANALYST_SEVERITIES, m as RawAnalystFindingSchema, n as buildTraceToolsForGroup, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, t as TraceToolGroupName, u as RAW_FINDING_SCHEMA_PROMPT } from "../tool-groups-CMmsgTzj.js";
+import { A as parseFindingSubject, C as FINDING_SUBJECT_KINDS, D as FindingSubjectStringSchema, E as FindingSubjectKind, F as DefineExactCustomAnalystOptions, G as SemanticConceptJudgeOptions, I as defineCustomAnalyst, L as defineTraceAnalyst, M as DspyRlmTraceEngineOptions, N as createDspyRlmTraceEngine, O as KIND_EXPECTED_SUBJECTS, P as DefineCustomAnalystOptions, S as FINDING_SUBJECT_GRAMMAR_PROMPT, T as FindingSubject, W as SemanticConceptJudgeInput, Y as RunCritic, Z as RunTrace, _ as FindingsDiff, a as SkillUsageScanConfig, b as defaultIsMaterial, c as DEFAULT_TRACE_ANALYST_KINDS, d as IMPROVEMENT_KIND_SPEC, f as FAILURE_MODE_KIND_SPEC, g as DiffPolicy, h as emitControlIntegrityFindings, i as SkillUsageReport, j as renderFindingSubject, k as findingSubjectGrammarPromptFor, l as KNOWLEDGE_POISONING_KIND_SPEC, m as ControlIntegrityAnalyst, n as SkillUsageAnalyst, o as buildSkillUsageReport, p as CONTROL_INTEGRITY_ANALYST, r as SkillUsageRecord, s as emitSkillUsageFindings, t as SKILL_USAGE_ANALYST, u as KNOWLEDGE_GAP_KIND_SPEC, v as FindingsStore, w as FINDING_SUBJECT_SYNTAX, x as diffFindings, y as PersistedFinding } from "../skill-usage-GlOphAhX.js";
+import { b as CustomTokenPricing, c as CostLedgerHandle } from "../cost-ledger-Bv_e8XHY.js";
+import { A as ChatClient, B as SandboxSdkTransportOpts, F as CreateChatClientOpts, I as CustomTransportOpts, L as DirectProviderTransportOpts, M as ChatResponse, N as ChatTransport, P as CliBridgeTransportOpts, R as MockTransportOpts, V as createChatClient, j as ChatRequest, k as ChatCallOpts, m as JudgeInput, p as JudgeFn, z as RouterTransportOpts } from "../types-XMVEdrE_.js";
+import { _ as makeFinding, a as AnalystInputKind, c as AnalystRunInputs, d as AnalystSeverity, f as AnalystUsageReceipt, g as computeFindingId, h as ProposalFindingOrigin, i as AnalystFinding, l as AnalystRunResult, m as ProposalFinding, n as AnalystContext, o as AnalystRequirements, p as EvidenceRef, r as AnalystCost, s as AnalystRunEvent, t as Analyst, u as AnalystRunSummary, v as makeProposalFinding, x as TraceAnalysisStore } from "../types-BhP9q0Fq.js";
+import { a as createTraceAnalyst, c as runTraceAnalyst, d as deriveEfficiencyFindings, i as TraceAnalystDefinition, l as BehavioralAnalystOptions, n as buildDefaultAnalystRegistry, o as renderPriorFindings, r as CreateTraceAnalystOptions, s as renderUpstreamFindings, t as DefaultAnalystRegistryOptions, u as behavioralAnalyst } from "../default-registry-J9m-_tya.js";
+import { a as ExactAnalystRunPolicySnapshot, c as ExactAnalystSnapshot, d as ExactExecutionComponentSnapshot, i as ExactAnalystRunEvent, l as ExactCapableAnalyst, n as ExactAnalystExecutionPlanSnapshot, o as ExactAnalystRunResult, r as ExactAnalystRunCompletion, s as ExactAnalystRunSummary, t as ExactAnalystBudgetSnapshot, u as ExactExecutionComponentIdentity } from "../exact-types-CBYF5MGd.js";
+import { A as ExactAnalystRunExecutionError, D as AnalystRegistryOptions, E as AnalystRegistry, M as RegistryRunOpts, N as assertExactRegistryRunOpts, O as BudgetPolicy, T as AnalystHooks, j as ExactRegistryRunOpts, k as ExactAnalystBudgetPolicy } from "../completion-verifier-CBiee74w.js";
+import { C as scoreAnalystFindings, S as traceStoreEvidenceResolver, _ as AnalystIssueExpectation, a as AnalystBenchmarkLabelState, b as registryBenchmarkRunner, c as AnalystBenchmarkProvenance, d as AnalystBenchmarkSummary, f as AnalystEvidenceExpectation, g as AnalystFindingScore, h as AnalystEvidenceResolver, i as AnalystBenchmarkError, l as AnalystBenchmarkResult, m as AnalystEvidenceResolutionError, n as AnalystBenchmarkDatasetRef, o as AnalystBenchmarkObservation, p as AnalystEvidenceResolution, r as AnalystBenchmarkDescriptor, s as AnalystBenchmarkOutput, t as AnalystBenchmarkCase, u as AnalystBenchmarkRunner, v as AnalystLatencyDistribution, x as runAnalystBenchmark, y as RunAnalystBenchmarkOptions } from "../benchmark-J9Qe6j2_.js";
+import { f as ExternalOptimizerModelExecutionObservation, h as ExternalOptimizerRunnerCommand, l as ExternalOptimizerModelCall } from "../external-optimizer-contracts-Q9c0L3eY.js";
+import { a as TraceAnalysisEngineRequest, c as resolveTraceAnalystLimits, d as RawAnalystEvidence, f as RawAnalystEvidenceSchema, g as parseRawFinding, h as evidenceRefsFromRawFinding, i as TraceAnalysisEngine, l as ANALYST_SEVERITIES, m as RawAnalystFindingSchema, n as buildTraceToolsForGroup, o as TraceAnalysisEngineResult, p as RawAnalystFinding, r as DEFAULT_TRACE_ANALYST_LIMITS, s as TraceAnalystLimits, t as TraceToolGroupName, u as RAW_FINDING_SCHEMA_PROMPT } from "../tool-groups-CK0JCkqO.js";
 //#region src/analyst/adapters.d.ts
 declare function liftSeverity(s: Severity): AnalystSeverity;
 interface VerifierAdapterOpts<Env> {
@@ -337,6 +337,13 @@ interface PublicAnalystBenchmarkModelConfig {
     responseCacheDir: string;
   };
 }
+/**
+ * Model settings the benchmark command records in the run identity. The
+ * owner-call pair is present exactly when a model-owner module executes the
+ * provider calls (`dspy-rlm` and `direct`); the `prime` analyst owns its own
+ * cli-bridge transport and never receives an owner call path.
+ */
+type PublicAnalystBenchmarkModelSettings = Omit<PublicAnalystBenchmarkModelConfig, 'call' | 'recordExecution'> & Partial<Pick<PublicAnalystBenchmarkModelConfig, 'call' | 'recordExecution'>>;
 interface PreparedPublicAnalystBenchmark {
   cases: AnalystBenchmarkCase<AnalystRunInputs>[];
   sourceRowCount: number;
@@ -506,6 +513,76 @@ declare function createPublicBenchmarkDirectRunner(dataset: PublicAnalystBenchma
 /** Public benchmark candidate that runs the actual recursive trace analyst. */
 declare function createPublicBenchmarkRlmRunner(dataset: PublicAnalystBenchmarkDataset, config: PublicAnalystBenchmarkModelConfig): AnalystBenchmarkRunner<AnalystRunInputs>;
 //#endregion
+//#region src/analyst/prime-bridge-transport.d.ts
+/**
+ * The bridge seam for the prime analyst protocol: one POST to an
+ * OpenAI-compatible `/v1/chat/completions` endpoint, returning the raw status
+ * and body text.
+ *
+ * Deliberately NOT `callLlm` from `../llm-client`, which serves a different
+ * contract:
+ *  - it retries transient failures, and a prime turn holds a single model seat
+ *    for minutes — a silent second attempt doubles the seat time and the spend;
+ *  - it normalizes usage, while the protocol's receipt depends on the bridge's
+ *    non-standard `model_requests` and `estimated` fields;
+ *  - it composes sampling and response-format options the bridge's CLI backends
+ *    reject;
+ *  - it collapses HTTP status, unparseable body, and empty content into two
+ *    error classes, while the protocol classifies them as three distinct
+ *    terminal reasons.
+ */
+interface PrimeBridgeTransportRequest {
+  url: string;
+  body: {
+    model: string;
+    messages: Array<{
+      role: 'user';
+      content: string;
+    }>;
+  };
+  /**
+   * The call's only deadline. The protocol aborts it on the per-call timeout
+   * and on the caller's cancellation, so a transport that imposes a second
+   * deadline of its own competes with this one.
+   */
+  signal: AbortSignal;
+}
+interface PrimeBridgeTransportResult {
+  status: number;
+  text: string;
+}
+/** One POST to the bridge. Injectable so tests run against a fake bridge. */
+type PrimeBridgeTransport = (request: PrimeBridgeTransportRequest) => Promise<PrimeBridgeTransportResult>;
+/**
+ * Default transport on node:http/node:https rather than fetch: undici's fixed
+ * response-header timeout kills prime calls that legitimately run past five
+ * minutes, so the request's AbortSignal is the only deadline.
+ */
+declare function nodeHttpPrimeBridgeTransport(): PrimeBridgeTransport;
+//#endregion
+//#region src/analyst/benchmark-runner-prime.d.ts
+interface PrimeBenchmarkRunnerOptions {
+  /** OpenAI-compatible cli-bridge base URL, e.g. `http://localhost:4181`. */
+  baseUrl: string;
+  /** Bridge model id in `<backend>/<provider>/<model>` form, e.g. `prime/zai/glm-5.2`. */
+  model: string;
+  /** Deadline for one bridge call. Prime analyses routinely exceed 5 minutes. */
+  timeoutMs: number;
+  /** Whether a structurally malformed reply gets one bounded repair turn. */
+  repair: boolean;
+  /** Exact token rates. Default: the agent-eval catalog rates for `model`. */
+  pricing?: CustomTokenPricing;
+  /** Bridge transport. Default: node:http POST (see nodeHttpPrimeBridgeTransport). */
+  transport?: PrimeBridgeTransport;
+}
+/**
+ * Digest of everything this runner can send to the bridge, recorded per
+ * observation so a prime result names the exact contract that produced it.
+ */
+declare function primeAnalystProtocolSha256(): string;
+/** CodeTraceBench-only: the prompt and output contract speak its block grammar. */
+declare function createPrimeBenchmarkRunner(options: PrimeBenchmarkRunnerOptions): AnalystBenchmarkRunner<AnalystRunInputs>;
+//#endregion
 //#region src/analyst/benchmark-comparison.d.ts
 type AnalystComparisonMetric = 'completion' | 'issueRecall' | 'findingPrecision' | 'f1' | 'criticalStepAccuracy' | 'citationCoverage' | 'citationExcerptCoverage' | 'citationLabelAgreement' | 'citationResolution' | 'trustedNegativeAccuracy' | 'latencyMs' | 'calls' | 'inputTokens' | 'outputTokens' | 'reasoningTokens' | 'cachedTokens' | 'cacheWriteTokens' | 'costUsd';
 interface AnalystMetricComparison {
@@ -731,7 +808,8 @@ interface AnalystBenchmarkLocalRunReceipt {
     traceDir: string;
     artifactDir?: string;
     outputDir: string;
-    modelOwnerModule: string;
+    /** Absent when the analyst owns its own transport (`prime`). */
+    modelOwnerModule?: string;
   };
   command: string;
   environment: {
@@ -765,7 +843,7 @@ declare function readAnalystBenchmarkArtifact(path: string): Promise<AnalystBenc
 //#endregion
 //#region src/analyst/benchmark-command.d.ts
 interface AnalystBenchmarkCommandDependencies {
-  createAnalystRunner?: (dataset: PublicAnalystBenchmarkDataset, config: PublicAnalystBenchmarkModelConfig) => AnalystBenchmarkRunner<AnalystRunInputs>;
+  createAnalystRunner?: (dataset: PublicAnalystBenchmarkDataset, config: PublicAnalystBenchmarkModelSettings) => AnalystBenchmarkRunner<AnalystRunInputs>;
   loadModelExecutionOwner?: (moduleRef: string, context: {
     model: string;
     environment: Readonly<NodeJS.ProcessEnv>;
@@ -775,8 +853,15 @@ interface AnalystBenchmarkCommandDependencies {
  * Which analyst produces the scored arm.
  *
  * `dspy-rlm` is the recursive engine. `direct` is the one-shot comparison arm.
+ * `prime` is the RLM coding agent reached through an OpenAI-compatible
+ * cli-bridge (CodeTraceBench only; see docs/prime-analyst.md).
  */
-type AnalystBenchmarkRunnerKind = 'dspy-rlm' | 'direct';
+type AnalystBenchmarkRunnerKind = 'dspy-rlm' | 'direct' | 'prime';
+/** Bridge execution settings, present exactly when the analyst is `prime`. */
+interface PrimeAnalystBridgeConfig {
+  bridgeUrl: string;
+  repair: boolean;
+}
 interface AnalystBenchmarkCommandConfig {
   dataset: PublicAnalystBenchmarkDataset;
   analyst: AnalystBenchmarkRunnerKind;
@@ -786,7 +871,7 @@ interface AnalystBenchmarkCommandConfig {
   outDir: string;
   revision: string;
   split: string;
-  model: PublicAnalystBenchmarkModelConfig;
+  model: PublicAnalystBenchmarkModelSettings;
   limit: number;
   seed: number;
   concurrency: number;
@@ -795,18 +880,21 @@ interface AnalystBenchmarkCommandConfig {
   rlmSamples: number;
   maxCostUsd: number;
   maxArtifactBytes: number;
-  modelOwnerModule: string;
+  /** Absent exactly when the analyst is `prime`: the cli-bridge owns execution. */
+  modelOwnerModule?: string;
+  /** Present exactly when the analyst is `prime`. */
+  prime?: PrimeAnalystBridgeConfig;
   command: string;
   resume: boolean;
 }
 declare function runAnalystBenchmarkCommand(argv: readonly string[], env?: NodeJS.ProcessEnv, dependencies?: AnalystBenchmarkCommandDependencies): Promise<number>;
-declare const ANALYST_BENCHMARK_HELP = "agent-eval analyst-benchmark\n\nRun the recursive DSPy trace analyst against public AgentRx or CodeTraceBench labels.\n\nRequired:\n  --dataset agentrx|codetracebench\n  --analyst dspy-rlm|direct        Scored analyst. Default: dspy-rlm.\n                                   'direct' is the one-shot comparison arm.\n  --labels <dataset.json|dataset.jsonl>\n  --trace-dir <one-trace-per-file OTLP JSONL directory>\n  --artifact-dir <extracted artifact root>  Required for CodeTraceBench\n  --out <new output directory>\n  --revision <full 40- or 64-character hex digest>\n  --split <dataset split>\n  --model-owner-module <module>   Module exporting createModelExecutionOwner;\n                                   the owner keeps provider credentials and policy\n  --model <provider model id>\n  --limit <positive case count>\n\nControls:\n  --resume                         Continue an interrupted run in --out\n  --seed <integer>                 Case-selection and comparison seed. Default: 0\n  --concurrency <positive integer> Parallel benchmark jobs. Default: 1\n  --repetitions <positive integer> Runs per case and runner. Default: 1\n  --rlm-samples <positive integer> Recursive-engine runs per case; above 1 the\n                                   step-level majority consensus is scored\n                                   (CodeTraceBench + dspy-rlm only). Default: 1\n  --instructions-file <path>       Replace the recursive analyst instructions\n                                   with this file's text (dspy-rlm only). The\n                                   recorded protocol digest binds the stock\n                                   protocol to the override text, and\n                                   result.json records instructionsOverrideSha256.\n  --max-output-tokens <positive>   Model output limit per call. Default: 16384\n  --max-reasoning-tokens <integer> Reasoning-token limit per call. Default: 65536\n  --max-model-requests <positive>  Caller-owned model calls per analysis.\n                                   Default: max iterations + model calls + 1\n  --max-model-request-bytes <positive>   Default: 16777216\n  --max-model-response-bytes <positive>  Default: 4194304\n  --model-request-timeout-ms <positive>  Default: --timeout-ms\n  --max-iterations <positive>      Recursive iterations per analysis. Default: 14\n  --max-llm-calls <positive>       DSPy model calls per analysis. Default: 8\n  --max-tool-calls <positive>      Trace-tool calls per analysis. Default: 80\n  --max-analysis-output-chars <positive> Default: 8000\n  --trace-tool-request-bytes <positive>  Default: 1000000\n  --trace-tool-response-bytes <positive> Default: 4000000\n  --trace-tool-timeout-ms <positive>     Default: 60000\n  --max-process-input-bytes <positive>   Default: 67108864\n  --max-process-result-bytes <positive>  Default: 4194304\n  --max-process-output-chars <positive>  Default: 64000\n  --python <executable>             Python with agent-eval-rpc[dspy]. Default: python\n  --timeout-ms <positive>          Model analyst deadline per case. Default: 300000\n  --max-cost-usd <positive>        Run-wide spend limit. Default: 5\n  --max-artifact-bytes <positive>  Final evidence bytes per case. Default: 8388608\n\nWrites result.json with every observation, metric, usage field, error, comparison,\ninput digest, artifact digest, case distribution, selected case id, and explicit\nunknown cost. Limited deterministic-hash subsets are marked non-representative.\nCompleted observations are fsynced to observations.jsonl. Shareable output is in\nresult.json and report.md. Machine-local paths, execution-owner module, and command\nare isolated in run.local.json. Provider credentials never enter this command.";
+declare const ANALYST_BENCHMARK_HELP = "agent-eval analyst-benchmark\n\nRun the recursive DSPy trace analyst against public AgentRx or CodeTraceBench labels.\n\nRequired:\n  --dataset agentrx|codetracebench\n  --analyst dspy-rlm|direct|prime  Scored analyst. Default: dspy-rlm.\n                                   'direct' is the one-shot comparison arm.\n                                   'prime' is the RLM coding agent behind an\n                                   OpenAI-compatible cli-bridge (codetracebench\n                                   only; see docs/prime-analyst.md)\n  --labels <dataset.json|dataset.jsonl>\n  --trace-dir <one-trace-per-file OTLP JSONL directory>\n  --artifact-dir <extracted artifact root>  Required for CodeTraceBench\n  --out <new output directory>\n  --revision <full 40- or 64-character hex digest>\n  --split <dataset split>\n  --model-owner-module <module>   dspy-rlm|direct only. Module exporting\n                                   createModelExecutionOwner; the owner keeps\n                                   provider credentials and policy\n  --model <provider model id>      For prime, the bridge model id in\n                                   <backend>/<provider>/<model> form, e.g.\n                                   prime/zai/glm-5.2\n  --limit <positive case count>\n\nControls:\n  --resume                         Continue an interrupted run in --out\n  --bridge-url <url>               prime only. OpenAI-compatible cli-bridge\n                                   base URL. Default: http://localhost:4181\n  --no-repair                      prime only. Disable the bounded repair turn\n                                   for a structurally malformed reply\n  --seed <integer>                 Case-selection and comparison seed. Default: 0\n  --concurrency <positive integer> Parallel benchmark jobs. Default: 1\n  --repetitions <positive integer> Runs per case and runner. Default: 1\n  --rlm-samples <positive integer> Recursive-engine runs per case; above 1 the\n                                   step-level majority consensus is scored\n                                   (CodeTraceBench + dspy-rlm only). Default: 1\n  --instructions-file <path>       Replace the recursive analyst instructions\n                                   with this file's text (dspy-rlm only). The\n                                   recorded protocol digest binds the stock\n                                   protocol to the override text, and\n                                   result.json records instructionsOverrideSha256.\n  --max-output-tokens <positive>   Model output limit per call. Default: 16384\n  --max-reasoning-tokens <integer> Reasoning-token limit per call. Default: 65536\n  --max-model-requests <positive>  Caller-owned model calls per analysis.\n                                   Default: max iterations + model calls + 1\n  --max-model-request-bytes <positive>   Default: 16777216\n  --max-model-response-bytes <positive>  Default: 4194304\n  --model-request-timeout-ms <positive>  Default: --timeout-ms\n  --max-iterations <positive>      Recursive iterations per analysis. Default: 14\n  --max-llm-calls <positive>       DSPy model calls per analysis. Default: 8\n  --max-tool-calls <positive>      Trace-tool calls per analysis. Default: 80\n  --max-analysis-output-chars <positive> Default: 8000\n  --trace-tool-request-bytes <positive>  Default: 1000000\n  --trace-tool-response-bytes <positive> Default: 4000000\n  --trace-tool-timeout-ms <positive>     Default: 60000\n  --max-process-input-bytes <positive>   Default: 67108864\n  --max-process-result-bytes <positive>  Default: 4194304\n  --max-process-output-chars <positive>  Default: 64000\n  --python <executable>             Python with agent-eval-rpc[dspy]. Default: python\n  --timeout-ms <positive>          Model analyst deadline per case. Default: 300000\n  --max-cost-usd <positive>        Run-wide spend limit. Default: 5\n  --max-artifact-bytes <positive>  Final evidence bytes per case. Default: 8388608\n\nWrites result.json with every observation, metric, usage field, error, comparison,\ninput digest, artifact digest, case distribution, selected case id, and explicit\nunknown cost. Limited deterministic-hash subsets are marked non-representative.\nCompleted observations are fsynced to observations.jsonl. Shareable output is in\nresult.json and report.md. Machine-local paths, execution-owner module, and command\nare isolated in run.local.json. Provider credentials never enter this command.";
 //#endregion
 //#region src/analyst/benchmark-implementation.d.ts
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM = "sha256-canonical-source-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM = "sha256-canonical-file-manifest";
 declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "6d84e54943016e204ddfca4ac7a3cb17508913f3a43201d09c13a82f1f07a318";
+declare const ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256 = "c0525dfe7f6931aeefe0e4c98d366cf6558c176e56477c5271d35274f383cf61";
 /** The published benchmark evidence was produced at this package version, by
  * the retired one-shot direct runner, before trace analysts moved to the
  * recursive DSPy RLM engine. Both evidence digests below are historical facts
@@ -818,7 +906,7 @@ declare const ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION = "0.137.0";
 declare const ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256 = "1e03f2daed356d60316aabefb407ec1e437ac94d408d61eea4ae096e9c6fbb5b";
 declare const ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256 = "4dba263b6256a30d56c7fdb2d992d3a953c0035d731f359b704db806f68f75ac";
 declare const ANALYST_BENCHMARK_IMPLEMENTATION_FILES: readonly string[];
-declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "67cd83aaaa158d44a9fa32e172cbe68fa97f01ae673ccb117469688b628bb9b1";
+declare const ANALYST_BENCHMARK_IMPLEMENTATION_SHA256 = "47592eaceab13f9cf40530e02a4af3d9ff5bcd6a7eb7d92b110508cc6cd323e5";
 declare function analystBenchmarkImplementationDigest(): string;
 declare function analystBenchmarkDependencyLockDigest(): string;
 //#endregion
@@ -872,6 +960,265 @@ declare function coerceJson(text: string): unknown;
  */
 declare function coerceToFindingRows(raw: unknown): unknown[];
 //#endregion
+//#region src/analyst/prime-protocol.d.ts
+/**
+ * The prime analyst protocol: a one-shot RLM reached through an
+ * OpenAI-compatible bridge, given the whole trajectory inline because it has no
+ * REPL and no trace tools, answering in one fenced JSON block of short strings.
+ *
+ * This module owns every part of that protocol that does not depend on what the
+ * rows MEAN: prompt and contract composition, the bounded repair turn, reply
+ * extraction, row decoding order, the render-measure-fall-back-fail-loud
+ * projection ladder, usage normalization, and the protocol identity digest.
+ *
+ * The cut: the core speaks RAW ROWS and never names a finding type. It knows a
+ * prime reply carries an `answer` string plus an array of rows under a named
+ * field, and that each row is decoded by a caller-supplied function. Consumers
+ * — the CodeTraceBench benchmark runner here, a span-grounded trace analyzer
+ * elsewhere — map decoded rows into their own finding shape on the way out. So
+ * nothing in this file may import a block type, a finding type, or a trace
+ * store, and no consumer needs another copy of the protocol.
+ */
+type PrimeRowDecoded<TRow> = {
+  ok: true;
+  row: TRow;
+} | {
+  ok: false;
+  reason: string;
+};
+interface PrimeReplyContract<TRow> {
+  /** Name of the reply's row array, e.g. `blocks` or `findings`. */
+  rowsField: string;
+  /** Row-grammar lines spliced into the first-turn prompt. */
+  contractLines: readonly string[];
+  /** The same grammar restated for the repair turn, which never carries the trajectory. */
+  repairContractLines: readonly string[];
+  /**
+   * Validate and convert one raw row in a single pass. One function rather than
+   * a separate validator and constructor, so the two can never disagree about
+   * what a valid row is.
+   */
+  decodeRow(row: unknown, index: number): PrimeRowDecoded<TRow>;
+  /**
+   * Cap on ACCEPTED rows. The cap is applied after decoding, so malformed rows
+   * never consume an accepted slot; the surplus is reported as `overflow`.
+   * Omit when the consumer's own expansion owns the cap and records the drop.
+   */
+  maxRows?: number;
+}
+interface PrimePromptSpec {
+  question: string;
+  /** Task definition spliced under a `TASK DEFINITION:` heading. Omit when the question is the whole task. */
+  taskDefinition?: string;
+  contractLines: readonly string[];
+  /** Line introducing the trajectory, e.g. `TRAJECTORY (trace_id X; 12 assistant step spans; ...):`. */
+  trajectoryHeader: string;
+  renderedTrajectory: string;
+  /** Material appended after the trajectory, e.g. the final-verification spans. */
+  trailer?: string;
+}
+declare function buildPrimePrompt(spec: PrimePromptSpec): string;
+interface PrimeRepairPromptSpec {
+  defect: string;
+  previousReply: string;
+  repairContractLines: readonly string[];
+}
+/** Carries the malformed reply and the contract — never the trajectory. */
+declare function buildPrimeRepairPrompt(spec: PrimeRepairPromptSpec): string;
+/**
+ * Recover the reply's JSON object.
+ *
+ * Distinct from `extractJsonPayload` in ../llm-client, which serves a response
+ * that DECLARES a JSON root and therefore must not scan onward. A prime reply
+ * is prose plus a fenced block, and when the model emits several fences the
+ * last one is its answer — so fences are scanned in reverse, and only then is a
+ * brace-to-brace slice tried.
+ */
+declare function extractPrimeJsonObject(text: string): Record<string, unknown> | null;
+/** Why the reply cannot be read as a prime answer, or null when it can. */
+declare function primeReplyDefect(parsed: Record<string, unknown> | null, rowsField: string): string | null;
+/**
+ * Token accounting exactly as the bridge reported it. Lossless on purpose: a
+ * side the bridge did not report stays null rather than becoming a zero, and a
+ * side it DID report survives even when its partner is missing.
+ */
+interface PrimeRawUsage {
+  /**
+   * Model-usage records the bridge observed. NOT a count of bridge round trips:
+   * one prime turn routes several internal model calls, so a single turn
+   * typically reports 3-5.
+   */
+  calls: number | null;
+  /** Prompt tokens the bridge reported; null when it reported none. */
+  inputTokens: number | null;
+  /** Completion tokens the bridge reported; null when it reported none. */
+  outputTokens: number | null;
+  /**
+   * True when the bridge DERIVED the counts (from character lengths) because
+   * the backend reported no usage of its own. Pricing derived counts produces a
+   * guess about a guess, which a reader must be able to tell apart from a
+   * rate-estimated cost over exact tokens.
+   */
+  bridgeEstimated: boolean;
+}
+declare function emptyPrimeRawUsage(): PrimeRawUsage;
+/** Read the bridge's OpenAI-shaped `usage` object. */
+declare function normalizePrimeUsage(raw: unknown): PrimeRawUsage;
+/**
+ * Sum two turns. Each side poisons independently: two turns that both report
+ * input and neither report output yield a real input total beside a null
+ * output, because discarding a measured count is as wrong as inventing one.
+ */
+declare function mergePrimeRawUsage(a: PrimeRawUsage, b: PrimeRawUsage): PrimeRawUsage;
+/**
+ * Bind raw prime usage to agent-eval's typed receipt.
+ *
+ * `RunTokenUsage.input` and `.output` are non-nullable, so a one-sided count
+ * cannot round-trip through `tokens` without writing a zero nobody measured.
+ * The complete-accounting field therefore stays null, the reported side is
+ * carried verbatim in `partialTokens`, and its price becomes the receipt's
+ * `knownCostUsd` lower bound.
+ *
+ * Only agent-eval calls this; consumers with no pricing table read
+ * `PrimeRawUsage` directly.
+ */
+declare function analystUsageReceiptFromPrimeUsage(usage: PrimeRawUsage, pricing: CustomTokenPricing): AnalystUsageReceipt;
+/**
+ * Why an exchange ended without a readable reply. `deadline` and `aborted` are
+ * separate members from `transport` on purpose: a model that ran past its
+ * deadline, a caller who cancelled, and a bridge that refused the connection
+ * demand three different responses, and a record that conflates them cannot be
+ * acted on.
+ */
+type PrimeFailure = {
+  kind: 'transport' | 'unparseable-json' | 'no-content' | 'deadline' | 'malformed-reply';
+  message: string;
+} | {
+  kind: 'http-status';
+  message: string;
+  status: number;
+  bodySnippet: string;
+} | {
+  kind: 'aborted';
+  message: string;
+  cause: unknown;
+};
+interface PrimeRepairState {
+  attempted: boolean;
+  /** Null when no repair turn ran, or when the repair turn never returned a reply. */
+  succeeded: boolean | null;
+}
+/** One completed bridge turn. A turn that failed produces no record. */
+interface PrimeTurnRecord {
+  turn: 'first' | 'repair';
+  usage: PrimeRawUsage;
+  /** The reply's `usage` object verbatim; null when the bridge sent none. */
+  rawUsage: unknown;
+}
+interface PrimeRejectedRow {
+  index: number;
+  reason: string;
+}
+type PrimeExchangeOutcome<TRow> = {
+  ok: true;
+  /** The reply's `answer` field; null when absent or not a string. */
+  answer: string | null;
+  rows: TRow[];
+  rejected: PrimeRejectedRow[];
+  /** Rows the model reported, before decoding or the count cap. */
+  reportedRows: number;
+  /** Valid rows dropped by `contract.maxRows`. */
+  overflow: number;
+  usage: PrimeRawUsage;
+  turns: PrimeTurnRecord[];
+  repair: PrimeRepairState;
+  reply: string;
+} | {
+  ok: false;
+  failure: PrimeFailure;
+  usage: PrimeRawUsage;
+  turns: PrimeTurnRecord[];
+  repair: PrimeRepairState;
+  /** The last reply received, when one arrived — the diagnostic artifact for a malformed case. */
+  reply?: string;
+};
+interface PrimeExchangeOptions<TRow> {
+  contract: PrimeReplyContract<TRow>;
+  prompt: string;
+  transport: PrimeBridgeTransport;
+  /** Full endpoint, e.g. `http://localhost:4181/v1/chat/completions`. */
+  url: string;
+  model: string;
+  /** Deadline for ONE turn. Prime analyses routinely exceed five minutes. */
+  timeoutMs: number;
+  /** Whether a structurally malformed reply gets one bounded repair turn. */
+  repair: boolean;
+  signal?: AbortSignal;
+}
+/**
+ * Run the protocol: one call, one bounded repair turn on a structurally
+ * malformed reply, then decode. Zero valid rows from a well-formed reply is an
+ * honest null, not a failure.
+ */
+declare function runPrimeExchange<TRow>(options: PrimeExchangeOptions<TRow>): Promise<PrimeExchangeOutcome<TRow>>;
+/**
+ * Where the inlined trajectory comes from. The protocol needs only two moves —
+ * fetch it all, or fetch a reduced version — so no store, artifact path, or
+ * span type enters the shared signature.
+ */
+interface PrimeProjectionSource<TItem> {
+  /** The full projection, or null when the source declares itself oversized. */
+  full(): Promise<readonly TItem[] | null>;
+  /** A reduced projection of the same material. */
+  capped(): Promise<readonly TItem[]>;
+  /** How `capped` reduces, named for the failure message, e.g. `per-attribute cap 1200`. */
+  cappedDescription: string;
+}
+interface PrimeProjectionDelivery {
+  mode: 'inline-json';
+  fetch: 'full' | 'capped';
+  renderedChars: number;
+}
+type PrimeProjectionOutcome<TItem> = {
+  ok: true;
+  items: readonly TItem[];
+  rendered: string;
+  delivery: PrimeProjectionDelivery;
+} | {
+  ok: false;
+  reason: string;
+  renderedChars: number;
+};
+/**
+ * Render, measure, fall back to the capped projection, re-measure, fail loud.
+ *
+ * Inline is the only delivery prime has, so an oversized trajectory is a
+ * refusal rather than a silent truncation: dropping spans would understate the
+ * trajectory and the analyst would answer a question about a different run.
+ */
+declare function projectPrimeTrajectory<TItem>(source: PrimeProjectionSource<TItem>, limits: {
+  maxInlineChars: number;
+}): Promise<PrimeProjectionOutcome<TItem>>;
+interface PrimeProtocolIdentity {
+  question: string;
+  taskDefinition?: string;
+  contractLines: readonly string[];
+  repairContractLines: readonly string[];
+  /** Numeric limits the contract states, e.g. row and width caps. */
+  limits: Readonly<Record<string, number>>;
+}
+/**
+ * Digest of everything a consumer can send to the bridge under the prime
+ * protocol, recorded per observation so a prime result names the exact contract
+ * that produced it.
+ *
+ * Computed over the ACTUALLY composed contract, so two consumers that both
+ * stamp `analyst_id: 'prime'` while asking materially different questions get
+ * different digests by construction. That is what makes 'prime' a reproducible
+ * claim rather than a label.
+ */
+declare function primeProtocolSha256(identity: PrimeProtocolIdentity): string;
+//#endregion
 //#region src/analyst/proposal-findings.d.ts
 /** True when a finding names a source candidate generation may learn from. */
 declare function isProposalFinding(finding: unknown): finding is ProposalFinding;
@@ -882,5 +1229,5 @@ declare function isProposalFinding(finding: unknown): finding is ProposalFinding
  */
 declare function assertProposalFindings(findings: unknown, context?: string): ReadonlyArray<ProposalFinding>;
 //#endregion
-export { AGENT_RX_UPSTREAM_REVISION, ANALYST_BENCHMARK_COST_LEDGER_FILE, ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, ANALYST_BENCHMARK_HELP, ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, ANALYST_BENCHMARK_IMPLEMENTATION_FILES, ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, ANALYST_BENCHMARK_MANIFEST_FILE, ANALYST_BENCHMARK_OBSERVATIONS_FILE, ANALYST_SEVERITIES, type AgentRxBenchmarkCaseOptions, type AgentRxCalibrationRunnerSummary, type AgentRxCalibrationSummary, type AgentRxFailure, type AgentRxPrediction, type AgentRxPredictionReport, type AgentRxRow, type Analyst, type AnalystBenchmarkArtifact, type AnalystBenchmarkCase, type AnalystBenchmarkCommandConfig, type AnalystBenchmarkCommandDependencies, type AnalystBenchmarkDatasetRef, type AnalystBenchmarkDescriptor, type AnalystBenchmarkError, type AnalystBenchmarkLabelState, type AnalystBenchmarkLocalRunReceipt, type AnalystBenchmarkObservation, type AnalystBenchmarkOutput, type AnalystBenchmarkProgressRow, type AnalystBenchmarkProvenance, type AnalystBenchmarkResult, type AnalystBenchmarkRunIdentity, type AnalystBenchmarkRunManifest, type AnalystBenchmarkRunner, type AnalystBenchmarkSummary, type AnalystComparisonMetric, type AnalystContext, type AnalystCost, type AnalystEvidenceExpectation, type AnalystEvidenceResolution, type AnalystEvidenceResolutionError, type AnalystEvidenceResolver, type AnalystFinding, type AnalystFindingScore, type AnalystHooks, type AnalystInputKind, type AnalystInstructionsOverride, type AnalystIssueExpectation, type AnalystLatencyDistribution, type AnalystMetricComparison, AnalystRegistry, type AnalystRegistryOptions, type AnalystRequirements, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystRunnerComparison, type AnalystSeverity, type AnalystUsageReceipt, type BehavioralAnalystOptions, type BudgetPolicy, CODE_TRACE_BENCH_ANALYST_PROMPT, CONTROL_INTEGRITY_ANALYST, type ChatCallOpts, type ChatClient, type ChatRequest, type ChatResponse, type ChatTransport, type CliBridgeTransportOpts, type CodeTraceBenchCaseOptions, type CodeTraceBenchLabelOptions, type CodeTraceBenchLabelSet, type CodeTraceBenchRow, type CodeTraceBlockDiagnostics, type CodeTraceCalibrationRunnerSummary, type CodeTraceCalibrationSummary, type CodeTraceFailureBlock, type CodeTraceStageAnnotation, type CodeTracerLabelGroup, type CodeTracerPredictionAdapterOptions, type CodeTracerPredictions, type CodeTracerStepLabel, ControlIntegrityAnalyst, type CreateChatClientOpts, type CreateTraceAnalystOptions, type CustomTransportOpts, DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_LIMITS, type DefaultAnalystRegistryOptions, type DefineCustomAnalystOptions, type DefineExactCustomAnalystOptions, type DiffPolicy, type DirectProviderTransportOpts, type DspyRlmTraceEngineOptions, type EvidenceRef, type ExactAnalystBudgetPolicy, type ExactAnalystBudgetSnapshot, type ExactAnalystExecutionPlanSnapshot, type ExactAnalystRunCompletion, type ExactAnalystRunEvent, ExactAnalystRunExecutionError, type ExactAnalystRunPolicySnapshot, type ExactAnalystRunResult, type ExactAnalystRunSummary, type ExactAnalystSnapshot, type ExactCapableAnalyst, type ExactExecutionComponentIdentity, type ExactExecutionComponentSnapshot, type ExactRegistryRunOpts, FAILURE_MODE_KIND_SPEC, FINDING_SUBJECT_GRAMMAR_PROMPT, FINDING_SUBJECT_KINDS, FINDING_SUBJECT_SYNTAX, type FindingSubject, type FindingSubjectKind, FindingSubjectStringSchema, type FindingsDiff, FindingsStore, IMPROVEMENT_KIND_SPEC, type JudgeAdapterOpts, KIND_EXPECTED_SUBJECTS, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type LoadedVerificationArtifacts, MAX_INCORRECT_BLOCKS, MAX_INCORRECT_BLOCK_STEPS, type MockTransportOpts, type PersistedFinding, type PreparedPublicAnalystBenchmark, type ProposalFinding, type ProposalFindingOrigin, type PublicAnalystBenchmarkDataset, type PublicAnalystBenchmarkModelConfig, type PublicAnalystBenchmarkModelOwner, type PublicBenchmarkDistributions, type PublicBenchmarkSelectionReport, type PublicBenchmarkValueDistribution, RAW_FINDING_SCHEMA_PROMPT, type RawAnalystEvidence, RawAnalystEvidenceSchema, type RawAnalystFinding, RawAnalystFindingSchema, type RegistryRunOpts, type RouterTransportOpts, type RunAnalystBenchmarkOptions, type RunCriticAdapterOpts, SKILL_USAGE_ANALYST, type SandboxSdkTransportOpts, type SemanticConceptJudgeAdapterOpts, SkillUsageAnalyst, type SkillUsageRecord, type SkillUsageReport, type SkillUsageScanConfig, type StepLabelAdapterOptions, type TraceAnalysisEngine, type TraceAnalysisEngineRequest, type TraceAnalysisEngineResult, type TraceAnalystDefinition, type TraceAnalystLimits, type TraceToolGroupName, type UpstreamPredictionAdapterOptions, type VerificationArtifactFile, type VerificationArtifactManifest, type VerificationArtifactRole, type VerificationAvailabilitySummary, type VerificationOutcome, type VerificationOutcomeSource, type VerificationOutcomeStatus, type VerificationResultFile, type VerifierAdapterOpts, adaptPublicBenchmarkFindings, agentRxBenchmarkCase, agentRxPredictionsToFindings, analystBenchmarkDependencyLockDigest, analystBenchmarkImplementationDigest, analystInstructionsOverrideFromText, appendVerificationArtifactsToOtlp, assertExactRegistryRunOpts, assertProposalFindings, behavioralAnalyst, buildDefaultAnalystRegistry, buildSkillUsageReport, buildTraceToolsForGroup, codeTraceBenchCase, codeTracerPredictionsToFindings, coerceJson, coerceToFindingRows, compareAnalystRunners, computeFindingId, createChatClient, createDspyRlmTraceEngine, createJudgeAdapter, createPublicBenchmarkDirectRunner, createPublicBenchmarkRlmRunner, createRunCriticAdapter, createSemanticConceptJudgeAdapter, createTraceAnalyst, createVerifierAdapter, defaultIsMaterial, defineCustomAnalyst, defineTraceAnalyst, deriveEfficiencyFindings, diffFindings, effectiveAnalystProtocolSha256, emitControlIntegrityFindings, emitSkillUsageFindings, emptyPublicBenchmarkRunner, evidenceRefsFromRawFinding, expandCodeTraceFailureBlocks, findingSubjectGrammarPromptFor, isProposalFinding, liftSeverity, loadCodeTraceVerificationArtifacts, loadPublicBenchmarkRows, makeFinding, makeProposalFinding, normalizeAgentRxCategory, normalizeBenchmarkLabel, parseFindingSubject, parseRawFinding, parseVerificationOutcome, preparePublicAnalystBenchmark, publicBenchmarkDistributions, publicBenchmarkProtocolSha256, publicBenchmarkRlmInstructions, publicBenchmarkSelectionReport, publicBenchmarkSystemPrompt, readAnalystBenchmarkArtifact, readAnalystInstructionsOverride, registryBenchmarkRunner, renderAgentRxCalibrationMarkdown, renderAnalystBenchmarkMarkdown, renderCodeTraceCalibrationMarkdown, renderFindingSubject, renderPriorFindings, renderUpstreamFindings, resolveTraceAnalystLimits, roundAgentRxStep, runAnalystBenchmark, runAnalystBenchmarkCommand, runTraceAnalyst, scoreAnalystFindings, selectPublicBenchmarkRows, stripCodeFences, summarizeAgentRxCalibration, summarizeAnalystBenchmarkRunner, summarizeCodeTraceCalibration, traceStoreEvidenceResolver };
+export { AGENT_RX_UPSTREAM_REVISION, ANALYST_BENCHMARK_COST_LEDGER_FILE, ANALYST_BENCHMARK_DEPENDENCY_LOCK_DIGEST_ALGORITHM, ANALYST_BENCHMARK_DEPENDENCY_LOCK_FILES, ANALYST_BENCHMARK_DEPENDENCY_LOCK_SHA256, ANALYST_BENCHMARK_EVIDENCE_DEPENDENCY_LOCK_SHA256, ANALYST_BENCHMARK_EVIDENCE_IMPLEMENTATION_SHA256, ANALYST_BENCHMARK_EVIDENCE_PACKAGE_VERSION, ANALYST_BENCHMARK_HELP, ANALYST_BENCHMARK_IMPLEMENTATION_DIGEST_ALGORITHM, ANALYST_BENCHMARK_IMPLEMENTATION_FILES, ANALYST_BENCHMARK_IMPLEMENTATION_SHA256, ANALYST_BENCHMARK_LOCAL_RECEIPT_FILE, ANALYST_BENCHMARK_MANIFEST_FILE, ANALYST_BENCHMARK_OBSERVATIONS_FILE, ANALYST_SEVERITIES, type AgentRxBenchmarkCaseOptions, type AgentRxCalibrationRunnerSummary, type AgentRxCalibrationSummary, type AgentRxFailure, type AgentRxPrediction, type AgentRxPredictionReport, type AgentRxRow, type Analyst, type AnalystBenchmarkArtifact, type AnalystBenchmarkCase, type AnalystBenchmarkCommandConfig, type AnalystBenchmarkCommandDependencies, type AnalystBenchmarkDatasetRef, type AnalystBenchmarkDescriptor, type AnalystBenchmarkError, type AnalystBenchmarkLabelState, type AnalystBenchmarkLocalRunReceipt, type AnalystBenchmarkObservation, type AnalystBenchmarkOutput, type AnalystBenchmarkProgressRow, type AnalystBenchmarkProvenance, type AnalystBenchmarkResult, type AnalystBenchmarkRunIdentity, type AnalystBenchmarkRunManifest, type AnalystBenchmarkRunner, type AnalystBenchmarkSummary, type AnalystComparisonMetric, type AnalystContext, type AnalystCost, type AnalystEvidenceExpectation, type AnalystEvidenceResolution, type AnalystEvidenceResolutionError, type AnalystEvidenceResolver, type AnalystFinding, type AnalystFindingScore, type AnalystHooks, type AnalystInputKind, type AnalystInstructionsOverride, type AnalystIssueExpectation, type AnalystLatencyDistribution, type AnalystMetricComparison, AnalystRegistry, type AnalystRegistryOptions, type AnalystRequirements, type AnalystRunEvent, type AnalystRunInputs, type AnalystRunResult, type AnalystRunSummary, type AnalystRunnerComparison, type AnalystSeverity, type AnalystUsageReceipt, type BehavioralAnalystOptions, type BudgetPolicy, CODE_TRACE_BENCH_ANALYST_PROMPT, CONTROL_INTEGRITY_ANALYST, type ChatCallOpts, type ChatClient, type ChatRequest, type ChatResponse, type ChatTransport, type CliBridgeTransportOpts, type CodeTraceBenchCaseOptions, type CodeTraceBenchLabelOptions, type CodeTraceBenchLabelSet, type CodeTraceBenchRow, type CodeTraceBlockDiagnostics, type CodeTraceCalibrationRunnerSummary, type CodeTraceCalibrationSummary, type CodeTraceFailureBlock, type CodeTraceStageAnnotation, type CodeTracerLabelGroup, type CodeTracerPredictionAdapterOptions, type CodeTracerPredictions, type CodeTracerStepLabel, ControlIntegrityAnalyst, type CreateChatClientOpts, type CreateTraceAnalystOptions, type CustomTransportOpts, DEFAULT_MAX_VERIFICATION_ARTIFACT_BYTES, DEFAULT_TRACE_ANALYST_KINDS, DEFAULT_TRACE_ANALYST_LIMITS, type DefaultAnalystRegistryOptions, type DefineCustomAnalystOptions, type DefineExactCustomAnalystOptions, type DiffPolicy, type DirectProviderTransportOpts, type DspyRlmTraceEngineOptions, type EvidenceRef, type ExactAnalystBudgetPolicy, type ExactAnalystBudgetSnapshot, type ExactAnalystExecutionPlanSnapshot, type ExactAnalystRunCompletion, type ExactAnalystRunEvent, ExactAnalystRunExecutionError, type ExactAnalystRunPolicySnapshot, type ExactAnalystRunResult, type ExactAnalystRunSummary, type ExactAnalystSnapshot, type ExactCapableAnalyst, type ExactExecutionComponentIdentity, type ExactExecutionComponentSnapshot, type ExactRegistryRunOpts, FAILURE_MODE_KIND_SPEC, FINDING_SUBJECT_GRAMMAR_PROMPT, FINDING_SUBJECT_KINDS, FINDING_SUBJECT_SYNTAX, type FindingSubject, type FindingSubjectKind, FindingSubjectStringSchema, type FindingsDiff, FindingsStore, IMPROVEMENT_KIND_SPEC, type JudgeAdapterOpts, KIND_EXPECTED_SUBJECTS, KNOWLEDGE_GAP_KIND_SPEC, KNOWLEDGE_POISONING_KIND_SPEC, type LoadedVerificationArtifacts, MAX_INCORRECT_BLOCKS, MAX_INCORRECT_BLOCK_STEPS, type MockTransportOpts, type PersistedFinding, type PreparedPublicAnalystBenchmark, type PrimeBenchmarkRunnerOptions, type PrimeBridgeTransport, type PrimeBridgeTransportRequest, type PrimeBridgeTransportResult, type PrimeExchangeOptions, type PrimeExchangeOutcome, type PrimeFailure, type PrimeProjectionDelivery, type PrimeProjectionOutcome, type PrimeProjectionSource, type PrimePromptSpec, type PrimeProtocolIdentity, type PrimeRawUsage, type PrimeRejectedRow, type PrimeRepairPromptSpec, type PrimeRepairState, type PrimeReplyContract, type PrimeRowDecoded, type PrimeTurnRecord, type ProposalFinding, type ProposalFindingOrigin, type PublicAnalystBenchmarkDataset, type PublicAnalystBenchmarkModelConfig, type PublicAnalystBenchmarkModelOwner, type PublicAnalystBenchmarkModelSettings, type PublicBenchmarkDistributions, type PublicBenchmarkSelectionReport, type PublicBenchmarkValueDistribution, RAW_FINDING_SCHEMA_PROMPT, type RawAnalystEvidence, RawAnalystEvidenceSchema, type RawAnalystFinding, RawAnalystFindingSchema, type RegistryRunOpts, type RouterTransportOpts, type RunAnalystBenchmarkOptions, type RunCriticAdapterOpts, SKILL_USAGE_ANALYST, type SandboxSdkTransportOpts, type SemanticConceptJudgeAdapterOpts, SkillUsageAnalyst, type SkillUsageRecord, type SkillUsageReport, type SkillUsageScanConfig, type StepLabelAdapterOptions, type TraceAnalysisEngine, type TraceAnalysisEngineRequest, type TraceAnalysisEngineResult, type TraceAnalystDefinition, type TraceAnalystLimits, type TraceToolGroupName, type UpstreamPredictionAdapterOptions, type VerificationArtifactFile, type VerificationArtifactManifest, type VerificationArtifactRole, type VerificationAvailabilitySummary, type VerificationOutcome, type VerificationOutcomeSource, type VerificationOutcomeStatus, type VerificationResultFile, type VerifierAdapterOpts, adaptPublicBenchmarkFindings, agentRxBenchmarkCase, agentRxPredictionsToFindings, analystBenchmarkDependencyLockDigest, analystBenchmarkImplementationDigest, analystInstructionsOverrideFromText, analystUsageReceiptFromPrimeUsage, appendVerificationArtifactsToOtlp, assertExactRegistryRunOpts, assertProposalFindings, behavioralAnalyst, buildDefaultAnalystRegistry, buildPrimePrompt, buildPrimeRepairPrompt, buildSkillUsageReport, buildTraceToolsForGroup, codeTraceBenchCase, codeTracerPredictionsToFindings, coerceJson, coerceToFindingRows, compareAnalystRunners, computeFindingId, createChatClient, createDspyRlmTraceEngine, createJudgeAdapter, createPrimeBenchmarkRunner, createPublicBenchmarkDirectRunner, createPublicBenchmarkRlmRunner, createRunCriticAdapter, createSemanticConceptJudgeAdapter, createTraceAnalyst, createVerifierAdapter, defaultIsMaterial, defineCustomAnalyst, defineTraceAnalyst, deriveEfficiencyFindings, diffFindings, effectiveAnalystProtocolSha256, emitControlIntegrityFindings, emitSkillUsageFindings, emptyPrimeRawUsage, emptyPublicBenchmarkRunner, evidenceRefsFromRawFinding, expandCodeTraceFailureBlocks, extractPrimeJsonObject, findingSubjectGrammarPromptFor, isProposalFinding, liftSeverity, loadCodeTraceVerificationArtifacts, loadPublicBenchmarkRows, makeFinding, makeProposalFinding, mergePrimeRawUsage, nodeHttpPrimeBridgeTransport, normalizeAgentRxCategory, normalizeBenchmarkLabel, normalizePrimeUsage, parseFindingSubject, parseRawFinding, parseVerificationOutcome, preparePublicAnalystBenchmark, primeAnalystProtocolSha256, primeProtocolSha256, primeReplyDefect, projectPrimeTrajectory, publicBenchmarkDistributions, publicBenchmarkProtocolSha256, publicBenchmarkRlmInstructions, publicBenchmarkSelectionReport, publicBenchmarkSystemPrompt, readAnalystBenchmarkArtifact, readAnalystInstructionsOverride, registryBenchmarkRunner, renderAgentRxCalibrationMarkdown, renderAnalystBenchmarkMarkdown, renderCodeTraceCalibrationMarkdown, renderFindingSubject, renderPriorFindings, renderUpstreamFindings, resolveTraceAnalystLimits, roundAgentRxStep, runAnalystBenchmark, runAnalystBenchmarkCommand, runPrimeExchange, runTraceAnalyst, scoreAnalystFindings, selectPublicBenchmarkRows, stripCodeFences, summarizeAgentRxCalibration, summarizeAnalystBenchmarkRunner, summarizeCodeTraceCalibration, traceStoreEvidenceResolver };
 //# sourceMappingURL=index.d.ts.map

package/dist/analyst/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/analyst/adapters.ts","../../src/analyst/benchmark-agentrx-calibration.ts","../../src/analyst/benchmark-dataset-types.ts","../../src/analyst/benchmark-dataset-agentrx.ts","../../src/analyst/benchmark-dataset-codetrace.ts","../../src/analyst/benchmark-dataset-utils.ts","../../src/analyst/benchmark-verification-outcome.ts","../../src/analyst/benchmark-verification-artifacts.ts","../../src/analyst/benchmark-public-types.ts","../../src/analyst/benchmark-public-adapters.ts","../../src/analyst/benchmark-public-data.ts","../../src/analyst/benchmark-public-prompt.ts","../../src/analyst/benchmark-public-model.ts","../../src/analyst/benchmark-public-rlm.ts","../../src/analyst/benchmark-comparison.ts","../../src/analyst/benchmark-public-calibration.ts","../../src/analyst/benchmark-command-artifact.ts","../../src/analyst/benchmark-command-result.ts","../../src/analyst/benchmark-command.ts","../../src/analyst/benchmark-implementation.ts","../../src/analyst/benchmark-instructions-override.ts","../../src/analyst/benchmark-report.ts","../../src/analyst/benchmark-summary.ts","../../src/analyst/parse-tolerant.ts","../../src/analyst/proposal-findings.ts"],"mappings":";;;;;;;;;;;;iBA4CgB,aAAa,GAAG,WAAgB;UAe/B,oBAAoB;EACnC;EACA;EACA,UAAU,mBAAmB;;;;;EAK7B,UAAU,KAAK,cAAc;;iBAGf,sBAAsB,KAAK,MAAM,oBAAoB,OAAO,QAAQ;UAwEnE;EACf;EACA;EACA,SAAS;;EAET;;iBAGc,uBAAuB,OAAM,uBAA4B,QAAQ;UAiEhE;EACf;EACA;EACA,OAAO;;EAEP,MAAM;;EAEN,OAAO;;EAEP;;iBAGc,mBAAmB,MAAM,mBAAmB,QAAQ;UAiDnD;EACf;EACA;;EAEA,UAAU,KAAK;;EAEf;;iBAGc,kCACd,OAAM,kCACL,QAAQ;;;cC5RE;UAEI;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,SAAS;;iBAGK,4BACd,QAAQ,wBACR,2BACC;iBAkBa,iCAAiC,SAAS;;;KCtD9C;UAEK;EACf,YAAY;EACZ;EACA;EACA;EACA;EACA;;UAGe;EACf,eAAe;EACf,mBAAmB;EACnB;IAAe,YAAY;IAAY;;EACvC,wBAAwB;EACxB;EACA;EACA;;UAGe;EACf,UAAU;EACV;EACA;EACA;EACA;;UAGe;EACf,UAAU;EACV,mBAAmB;EACnB;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,iBAAiB;;KAGP,0CAEC,sCACA,iCACA;UAEI;EACf;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA,oCAAoC;;UAGrB;EACf,eAAe;EACf,WAAW,sBAAsB;;KAGvB;UAEK;;;;;EAKf,WAAW;;UAGI,kCACP,yBACN;UAEa,oCAAoC;EACnD;;EAEA;;UAGe,yCAAyC;EACxD;EACA;EACA;EACA;;UAGe,2CACP,kCACN;;;iBCtGY,qBAAqB,QACnC,KAAK,YACL,OAAO,QACP,UAAS,8BACR,qBAAqB;;iBAoGR,6BACd,mBAAmB,YACnB,iBACA,UAAS,mCACR;iBA6Ea,yBAAyB;;iBAoNzB,iBAAiB;;;iBC7YjB,mBAAmB,QACjC,KAAK,mBACL,OAAO,QACP,UAAS,4BACR,qBAAqB;;iBAwFR,gCACd,2BACA,aAAa,uBACb,UAAS,qCACR;;;iBClHa,wBAAwB;;;KCA5B;UAEK;EACf;EACA;EACA,QAAQ;;UAGO;EACf,QAAQ;EACR;EAKA;IAAe;IAAe;;EAC9B,SAAS;EACT;EACA;EACA;EACA;;UAGe;EACf;EACA;;iBA8Ec,yBACd,gBAAgB,2BACf;;;cChGU;KAED;UAEK;EACf,MAAM;EACN;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;EACT;EACA;EACA;EACA;EACA;EACA,OAAO;EACP,cAAc;EACd,UAAU,OAAO;;UAGF;EACf,UAAU;EACV,SAAS;EACT,OAAO,MAAM;IAA6B;;;iBAatB,mCAAmC;EACvD;EACA,KAAK;EACL;IACE,QAAQ;iBAwHI,kCACd,kBACA,iBACA,WAAW,6BACX;;;KC5KU;;;;;;;;UASK;;WAEN;;WAEA;;;UAIM;EACf,MAAM;EACN;EACA,kBAAkB,aAAa;;EAE/B,UAAU;;UAGK;;EAEf,MAAM;;EAEN;;EAEA,kBAAkB,aAAa;EAC/B;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA,UAAU;;EAEV;;EAEA,uBAAuB;EACvB;IACE,SAAS;IACT;IACA;IACA;IACA;IACA;IACA;IACA;IACA;;;;;;IAMA;;EAEF,aAAa;EACb;IACE;IACA;;;UAIa;EACf,OAAO,qBAAqB;EAC5B;EACA;EACA;EACA,YAAY;IACV;IACA;IACA;;EAEF,uBAAuB;EACvB,WAAW;;UAGI;EACf;EACA;EACA,QAAQ;;UAGO;EACf,OAAO;EACP,OAAO;EACP,OAAO;EACP,YAAY;EACZ,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;EACA;EACA,QAAQ;EACR,UAAU;;;;;;;;;;UCpGK;EACf;EACA;;EAEA;EACA;EACA,UAAU;EACV;EACA;EACA;EACA;EACA,WAAW;;;;;;;;;UAUI;EACf;EACA;;EAEA,kCAAkC;;EAElC;;EAEA;;EAEA;;EAEA;;EAEA;;;;;;;;UASe;EACf;EACA,OAAO;;iBAGO,8BAA8B,uBAAuB;iBAiB/C,6BAA6B;EACjD,SAAS;EACT;EACA,mBAAmB;EACnB;EACA,OAAO;EACP,SAAS;IACP;EACF,UAAU;EACV,aAAa;;EAEb,aAAa;;;;;;;;;iBA4MO,6BAA6B;EACjD;EACA,iBAAiB;EACjB,OAAO;EACP;EACA;EACA,SAAS;IACP;EACF,UAAU;EACV,aAAa;EACb,YAAY;;;;iBCtQQ,wBACpB,eACC,QAAQ,MAAM;iBA2BD,0BACd,SAAS,+BACT,eAAe,2BACf;EAAW;EAAe;IACzB,MAAM;iBAwBO,6BACd,SAAS,+BACT,eAAe,4BACd;iBAyCa,+BACd,SAAS,+BACT,iBAAiB,2BACjB,mBAAmB,2BACnB,eACC;iBAcmB,8BAA8B;EAClD,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;IACE,QAAQ;;;;;;;cCxKC;;;;cAKA;cAMA;;iBAiGG,4BAA4B,SAAS;;iBAerC,+BAA+B,SAAS;;;;iBAaxC,8BAA8B,SAAS;;;;iBCjFvC,kCACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;;iBC/BV,+BACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;KCnCd;UAoBK;EACf,QAAQ;EACR;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBASK,sBACd,QAAQ,wBACR;EACE;EACA;EACA;EACA;EACA;IAED;;;UCnEc;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBAGK,8BACd,QAAQ,yBACP;iBAca,mCAAmC,SAAS;;;UCpC3C;EACf;EACA;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D,uBAAuB;IACvB,0BAA0B;IAC1B;MACE;MACA;MACA;MACA,QAAQ;;IAEV;MACE;MACA;;MAEA;MACA;;MAEA;MACA;MACA;MACA;MACA;MACA;MACA;MACA,UAAU;MACV;QACE;QACA;QACA;QACA;QACA;QACA;QACA;QACA;;MAEF;QACE;QACA;QACA;;MAEF;MACA;MACA;;MAEA;MACA;MACA;;;EAGJ,QAAQ;EACR,aAAa;EACb,uBAAuB;EACvB,qBAAqB;;UAGN;EACf;EACA;EACA;EACA;IACE;IACA;IACA;;;UAIa;EACf;IACE,SAAS;IACT;IACA;IACA;MACE;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA,SAAS;MACT;QACE;QACA;QACA;QACA;QACA;QACA;QACA;QACA;;MAEF;QACE;QACA;QACA;;;IAGJ;IACA;IACA;IACA;;IAEA;IACA;IACA;IACA;;IAEA;IACA;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D;IACA;;;UAIa;EACf;EACA;EACA;EACA;EACA,UAAU;;UAGK;EACf;EACA;EACA;EACA;IACE;IACA;IACA;IACA;IACA;;EAEF;EACA;IACE;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA;IACA;IACA;;;UAIa;EACf;EACA;EACA;EACA,aAAa;EACb;;cAGW;cACA;cACA;cACA;;;iBC3KS,6BACpB,eACC,QAAQ;;;UCqEM;EACf,uBACE,SAAS,+BACT,QAAQ,sCACL,uBAAuB;EAC5B,2BACE,mBACA;IACE;IACA,aAAa,SAAS,OAAO;QAE5B,QAAQ;;;;;;;KAQH;UAEK;EACf,SAAS;EACT,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;EACA,OAAO;EACP;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;;iBAKoB,2BACpB,yBACA,MAAK,OAAO,YACZ,eAAc,sCACb;cA0TU;;;cCtcA;cAEA;cAEA;cAOA;;;;;;;;cAUA;cAEA;cAGA;cAGA;cAyFA;iBAGG;iBAIA;;;;iBCpHA,oCAAoC,eAAe;;iBAQnD,gCAAgC,eAAe;;;;;;;;;;;iBAyB/C,+BACd,SAAS,+BACT,WAAW,KAAK;;;iBCzCF,+BACd,QAAQ,wBACR,uBAAsB;;;iBCER,gCACd,kBACA,uBAAuB,gCACtB;;;;;;;;;;;;;;;iBCGa,gBAAgB;;;;;iBAgBhB,WAAW;;;;;;;iBAeX,oBAAoB;;;;iBCXpB,kBAAkB,mBAAmB,WAAW;;;;;;iBAShD,uBACd,mBACA,mBACC,cAAc"}
+{"version":3,"file":"index.d.ts","names":[],"sources":["../../src/analyst/adapters.ts","../../src/analyst/benchmark-agentrx-calibration.ts","../../src/analyst/benchmark-dataset-types.ts","../../src/analyst/benchmark-dataset-agentrx.ts","../../src/analyst/benchmark-dataset-codetrace.ts","../../src/analyst/benchmark-dataset-utils.ts","../../src/analyst/benchmark-verification-outcome.ts","../../src/analyst/benchmark-verification-artifacts.ts","../../src/analyst/benchmark-public-types.ts","../../src/analyst/benchmark-public-adapters.ts","../../src/analyst/benchmark-public-data.ts","../../src/analyst/benchmark-public-prompt.ts","../../src/analyst/benchmark-public-model.ts","../../src/analyst/benchmark-public-rlm.ts","../../src/analyst/prime-bridge-transport.ts","../../src/analyst/benchmark-runner-prime.ts","../../src/analyst/benchmark-comparison.ts","../../src/analyst/benchmark-public-calibration.ts","../../src/analyst/benchmark-command-artifact.ts","../../src/analyst/benchmark-command-result.ts","../../src/analyst/benchmark-command.ts","../../src/analyst/benchmark-implementation.ts","../../src/analyst/benchmark-instructions-override.ts","../../src/analyst/benchmark-report.ts","../../src/analyst/benchmark-summary.ts","../../src/analyst/parse-tolerant.ts","../../src/analyst/prime-protocol.ts","../../src/analyst/proposal-findings.ts"],"mappings":";;;;;;;;;;;;iBA4CgB,aAAa,GAAG,WAAgB;UAe/B,oBAAoB;EACnC;EACA;EACA,UAAU,mBAAmB;;;;;EAK7B,UAAU,KAAK,cAAc;;iBAGf,sBAAsB,KAAK,MAAM,oBAAoB,OAAO,QAAQ;UAwEnE;EACf;EACA;EACA,SAAS;;EAET;;iBAGc,uBAAuB,OAAM,uBAA4B,QAAQ;UAiEhE;EACf;EACA;EACA,OAAO;;EAEP,MAAM;;EAEN,OAAO;;EAEP;;iBAGc,mBAAmB,MAAM,mBAAmB,QAAQ;UAiDnD;EACf;EACA;;EAEA,UAAU,KAAK;;EAEf;;iBAGc,kCACd,OAAM,kCACL,QAAQ;;;cC5RE;UAEI;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,SAAS;;iBAGK,4BACd,QAAQ,wBACR,2BACC;iBAkBa,iCAAiC,SAAS;;;KCtD9C;UAEK;EACf,YAAY;EACZ;EACA;EACA;EACA;EACA;;UAGe;EACf,eAAe;EACf,mBAAmB;EACnB;IAAe,YAAY;IAAY;;EACvC,wBAAwB;EACxB;EACA;EACA;;UAGe;EACf,UAAU;EACV;EACA;EACA;EACA;;UAGe;EACf,UAAU;EACV,mBAAmB;EACnB;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA;EACA,iBAAiB;;KAGP,0CAEC,sCACA,iCACA;UAEI;EACf;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA,oCAAoC;;UAGrB;EACf,eAAe;EACf,WAAW,sBAAsB;;KAGvB;UAEK;;;;;EAKf,WAAW;;UAGI,kCACP,yBACN;UAEa,oCAAoC;EACnD;;EAEA;;UAGe,yCAAyC;EACxD;EACA;EACA;EACA;;UAGe,2CACP,kCACN;;;iBCtGY,qBAAqB,QACnC,KAAK,YACL,OAAO,QACP,UAAS,8BACR,qBAAqB;;iBAoGR,6BACd,mBAAmB,YACnB,iBACA,UAAS,mCACR;iBA6Ea,yBAAyB;;iBAoNzB,iBAAiB;;;iBC7YjB,mBAAmB,QACjC,KAAK,mBACL,OAAO,QACP,UAAS,4BACR,qBAAqB;;iBAwFR,gCACd,2BACA,aAAa,uBACb,UAAS,qCACR;;;iBClHa,wBAAwB;;;KCA5B;UAEK;EACf;EACA;EACA,QAAQ;;UAGO;EACf,QAAQ;EACR;EAKA;IAAe;IAAe;;EAC9B,SAAS;EACT;EACA;EACA;EACA;;UAGe;EACf;EACA;;iBA8Ec,yBACd,gBAAgB,2BACf;;;cChGU;KAED;UAEK;EACf,MAAM;EACN;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;EACT;EACA;EACA;EACA;EACA;EACA,OAAO;EACP,cAAc;EACd,UAAU,OAAO;;UAGF;EACf,UAAU;EACV,SAAS;EACT,OAAO,MAAM;IAA6B;;;iBAatB,mCAAmC;EACvD;EACA,KAAK;EACL;IACE,QAAQ;iBAwHI,kCACd,kBACA,iBACA,WAAW,6BACX;;;KC5KU;;;;;;;;UASK;;WAEN;;WAEA;;;UAIM;EACf,MAAM;EACN;EACA,kBAAkB,aAAa;;EAE/B,UAAU;;UAGK;;EAEf,MAAM;;EAEN;;EAEA,kBAAkB,aAAa;EAC/B;EACA;EACA;;EAEA;;EAEA;;EAEA;;EAEA;;EAEA,UAAU;;EAEV;;EAEA,uBAAuB;EACvB;IACE,SAAS;IACT;IACA;IACA;IACA;IACA;IACA;IACA;IACA;;;;;;IAMA;;EAEF,aAAa;EACb;IACE;IACA;;;;;;;;;KAUQ,sCAAsC,KAChD,iEAGA,QAAQ,KAAK;UAEE;EACf,OAAO,qBAAqB;EAC5B;EACA;EACA;EACA,YAAY;IACV;IACA;IACA;;EAEF,uBAAuB;EACvB,WAAW;;UAGI;EACf;EACA;EACA,QAAQ;;UAGO;EACf,OAAO;EACP,OAAO;EACP,OAAO;EACP,YAAY;EACZ,QAAQ;;UAGO;EACf;EACA;EACA;EACA;EACA;EACA;EACA,QAAQ;EACR,UAAU;;;;;;;;;;UChHK;EACf;EACA;;EAEA;EACA;EACA,UAAU;EACV;EACA;EACA;EACA;EACA,WAAW;;;;;;;;;UAUI;EACf;EACA;;EAEA,kCAAkC;;EAElC;;EAEA;;EAEA;;EAEA;;EAEA;;;;;;;;UASe;EACf;EACA,OAAO;;iBAGO,8BAA8B,uBAAuB;iBAiB/C,6BAA6B;EACjD,SAAS;EACT;EACA,mBAAmB;EACnB;EACA,OAAO;EACP,SAAS;IACP;EACF,UAAU;EACV,aAAa;;EAEb,aAAa;;;;;;;;;iBA4MO,6BAA6B;EACjD;EACA,iBAAiB;EACjB,OAAO;EACP;EACA;EACA,SAAS;IACP;EACF,UAAU;EACV,aAAa;EACb,YAAY;;;;iBCtQQ,wBACpB,eACC,QAAQ,MAAM;iBA2BD,0BACd,SAAS,+BACT,eAAe,2BACf;EAAW;EAAe;IACzB,MAAM;iBAwBO,6BACd,SAAS,+BACT,eAAe,4BACd;iBAyCa,+BACd,SAAS,+BACT,iBAAiB,2BACjB,mBAAmB,2BACnB,eACC;iBAcmB,8BAA8B;EAClD,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;IACE,QAAQ;;;;;;;cCxKC;;;;cAKA;cAMA;;iBAiGG,4BAA4B,SAAS;;iBAerC,+BAA+B,SAAS;;;;iBAaxC,8BAA8B,SAAS;;;;iBCjFvC,kCACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;;iBC/BV,+BACd,SAAS,+BACT,QAAQ,oCACP,uBAAuB;;;;;;;;;;;;;;;;;;;;UClBT;EACf;EACA;IACE;IACA,UAAU;MAAQ;MAAc;;;;;;;;EAOlC,QAAQ;;UAGO;EACf;EACA;;;KAIU,wBACV,SAAS,gCACN,QAAQ;;;;;;iBAOG,gCAAgC;;;UC2B/B;;EAEf;;EAEA;;EAEA;;EAEA;;EAEA,UAAU;;EAEV,YAAY;;;;;;iBA8EE;;iBAKA,2BACd,SAAS,8BACR,uBAAuB;;;KC1Kd;UAoBK;EACf,QAAQ;EACR;;EAEA;;EAEA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBASK,sBACd,QAAQ,wBACR;EACE;EACA;EACA;EACA;EACA;IAED;;;UCnEc;EACf;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;EAEA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;EACA;;UAGe;EACf;EACA;EACA,SAAS;;iBAGK,8BACd,QAAQ,yBACP;iBAca,mCAAmC,SAAS;;;UCpC3C;EACf;EACA;EACA;IACE,SAAS;IACT;IACA;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D,uBAAuB;IACvB,0BAA0B;IAC1B;MACE;MACA;MACA;MACA,QAAQ;;IAEV;MACE;MACA;;MAEA;MACA;;MAEA;MACA;MACA;MACA;MACA;MACA;MACA;MACA,UAAU;MACV;QACE;QACA;QACA;QACA;QACA;QACA;QACA;QACA;;MAEF;QACE;QACA;QACA;;MAEF;MACA;MACA;;MAEA;MACA;MACA;;;EAGJ,QAAQ;EACR,aAAa;EACb,uBAAuB;EACvB,qBAAqB;;UAGN;EACf;EACA;EACA;EACA;IACE;IACA;IACA;;;UAIa;EACf;IACE,SAAS;IACT;IACA;IACA;MACE;MACA;MACA;MACA;MACA;MACA;MACA;MACA;MACA,SAAS;MACT;QACE;QACA;QACA;QACA;QACA;QACA;QACA;QACA;;MAEF;QACE;QACA;QACA;;;IAGJ;IACA;IACA;IACA;;IAEA;IACA;IACA;IACA;;IAEA;IACA;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA,YAAY;MAAQ;MAAiB;MAAsB;;IAC3D;IACA;;;UAIa;EACf;EACA;EACA;EACA;EACA,UAAU;;UAGK;EACf;EACA;EACA;EACA;IACE;IACA;IACA;IACA;;IAEA;;EAEF;EACA;IACE;IACA;IACA;;EAEF;IACE;IACA;IACA;IACA;IACA;IACA;;;UAIa;EACf;EACA;EACA;EACA,aAAa;EACb;;cAGW;cACA;cACA;cACA;;;iBC5KS,6BACpB,eACC,QAAQ;;;UCuEM;EACf,uBACE,SAAS,+BACT,QAAQ,wCACL,uBAAuB;EAC5B,2BACE,mBACA;IACE;IACA,aAAa,SAAS,OAAO;QAE5B,QAAQ;;;;;;;;;KAUH;;UAGK;EACf;EACA;;UAGe;EACf,SAAS;EACT,SAAS;EACT;EACA;EACA;EACA;EACA;EACA;EACA,OAAO;EACP;EACA;EACA;EACA;;EAEA;EACA;EACA;;EAEA;;EAEA,QAAQ;EACR;EACA;;iBAKoB,2BACpB,yBACA,MAAK,OAAO,YACZ,eAAc,sCACb;cAsUU;;;cC/dA;cAEA;cAEA;cAOA;;;;;;;;cAUA;cAEA;cAGA;cAGA;cA8FA;iBAGG;iBAIA;;;;iBCzHA,oCAAoC,eAAe;;iBAQnD,gCAAgC,eAAe;;;;;;;;;;;iBAyB/C,+BACd,SAAS,+BACT,WAAW,KAAK;;;iBCzCF,+BACd,QAAQ,wBACR,uBAAsB;;;iBCER,gCACd,kBACA,uBAAuB,gCACtB;;;;;;;;;;;;;;;iBCGa,gBAAgB;;;;;iBAgBhB,WAAW;;;;;;;iBAeX,oBAAoB;;;;;;;;;;;;;;;;;;;;;KClBxB,gBAAgB;EAAU;EAAU,KAAK;;EAAW;EAAW;;UAE1D,mBAAmB;;EAElC;;EAEA;;EAEA;;;;;;EAMA,UAAU,cAAc,gBAAgB,gBAAgB;;;;;;EAMxD;;UAKe;EACf;;EAEA;EACA;;EAEA;EACA;;EAEA;;iBAGc,iBAAiB,MAAM;UAatB;EACf;EACA;EACA;;;iBAIc,uBAAuB,MAAM;;;;;;;;;;iBAuB7B,uBAAuB,eAAe;;iBAkBtC,iBACd,QAAQ,gCACR;;;;;;UAyBe;;;;;;EAMf;;EAEA;;EAEA;;;;;;;EAOA;;iBAGc,sBAAsB;;iBAKtB,oBAAoB,eAAe;;;;;;iBAgBnC,mBAAmB,GAAG,eAAe,GAAG,gBAAgB;;;;;;;;;;;;;iBA6BxD,kCACd,OAAO,eACP,SAAS,qBACR;;;;;;;;KAqCS;EAEN;EACA;;EAEA;EAAqB;EAAiB;EAAgB;;EACtD;EAAiB;EAAiB;;UAEvB;EACf;;EAEA;;;UAIe;EACf;EACA,OAAO;;EAEP;;UAGe;EACf;EACA;;KAGU,qBAAqB;EAE3B;;EAEA;EACA,MAAM;EACN,UAAU;;EAEV;;EAEA;EACA,OAAO;EACP,OAAO;EACP,QAAQ;EACR;;EAGA;EACA,SAAS;EACT,OAAO;EACP,OAAO;EACP,QAAQ;;EAER;;UAGW,qBAAqB;EACpC,UAAU,mBAAmB;EAC7B;EACA,WAAW;;EAEX;EACA;;EAEA;;EAEA;EACA,SAAS;;;;;;;iBAQW,iBAAiB,MACrC,SAAS,qBAAqB,QAC7B,QAAQ,qBAAqB;;;;;;UAwMf,sBAAsB;;EAErC,QAAQ,iBAAiB;;EAEzB,UAAU,iBAAiB;;EAE3B;;UAGe;EACf;EACA;EACA;;KAGU,uBAAuB;EAE7B;EACA,gBAAgB;EAChB;EACA,UAAU;;EAEV;EAAW;EAAgB;;;;;;;;;iBASX,uBAAuB,OAC3C,QAAQ,sBAAsB,QAC9B;EAAU;IACT,QAAQ,uBAAuB;UA8BjB;EACf;EACA;EACA;EACA;;EAEA,QAAQ,SAAS;;;;;;;;;;;;iBAaH,oBAAoB,UAAU;;;;iBCzkB9B,kBAAkB,mBAAmB,WAAW;;;;;;iBAShD,uBACd,mBACA,mBACC,cAAc"}