@tangle-network/agent-eval 0.116.0 → 0.117.1

package/dist/contract/index.js

@@ -12,11 +12,14 @@ import {
 } from "../chunk-ZZUXHH3R.js";
 import {
   analyzeRuns
-} from "../chunk-J6P6PK2R.js";
+} from "../chunk-FQNLDL4D.js";
 import {
+  REFERENCE_EQUIVALENCE_INPUT_LIMITS,
+  REFERENCE_EQUIVALENCE_JUDGE_VERSION,
   buildEvidenceVector,
   campaignMeanComposite,
   composeGate,
+  createReferenceEquivalenceJudge,
   defaultProductionGate,
   emitLoopProvenance,
   evolutionaryProposer,
@@ -27,19 +30,23 @@ import {
   powerPreflight,
   runEval,
   runImprovementLoop,
+  runReferenceEquivalenceJudge,
   surfaceContentHash,
   surfaceHash
-} from "../chunk-3274WNK7.js";
-import "../chunk-VI2UW6B6.js";
+} from "../chunk-HQPHZGL6.js";
 import {
+  createRunCostLedger,
   fsCampaignStorage,
   inMemoryCampaignStorage,
+  resolveRunDir,
   runCampaign
-} from "../chunk-FAOEFFRT.js";
+} from "../chunk-IDZTTFRR.js";
 import {
-  buildDefaultAnalystRegistry
-} from "../chunk-7GKEAIAD.js";
-import "../chunk-CIUOICJT.js";
+  buildDefaultAnalystRegistry,
+  createChatClient
+} from "../chunk-VF3XSYTI.js";
+import "../chunk-HHWE3POT.js";
+import "../chunk-MGEHEHSN.js";
 import {
   FileSystemOutcomeStore,
   InMemoryOutcomeStore
@@ -47,25 +54,38 @@ import {
 import "../chunk-ARU2PZFM.js";
 import "../chunk-DPZAEKA6.js";
 import "../chunk-PJQFMIOX.js";
-import "../chunk-MHNQWM4I.js";
+import "../chunk-LQUTGLOZ.js";
 import "../chunk-GGE4NNQT.js";
 import {
   LLM_INPUT_TOKEN_ATTR_KEYS,
   LLM_MODEL_ATTR_KEYS,
   LLM_OUTPUT_TOKEN_ATTR_KEYS
-} from "../chunk-LNQEP766.js";
+} from "../chunk-S2F4J57L.js";
 import {
   parseRunRecordSafe
 } from "../chunk-5UF54T55.js";
 import "../chunk-XJYR7XFV.js";
 import "../chunk-VSMTAMNK.js";
-import "../chunk-GY4SYVPJ.js";
+import "../chunk-NJC7U437.js";
+import "../chunk-VCTY3W6J.js";
+import "../chunk-VI2UW6B6.js";
 import "../chunk-PC4UYEBM.js";
 import "../chunk-ONWEPEDO.js";
 import "../chunk-PZ5AY32C.js";
 
 // src/contract/self-improve.ts
 import { createHash } from "crypto";
+var SelfImproveRunError = class extends Error {
+  cost;
+  receipts;
+  constructor(cause, ledger) {
+    const original = cause instanceof Error ? cause : new Error(String(cause));
+    super(original.message, { cause: original });
+    this.name = "SelfImproveRunError";
+    this.cost = ledger.summary();
+    this.receipts = ledger.list();
+  }
+};
 function splitTrainHoldout(scenarios, fraction) {
   function hash(s) {
     let h = 2166136261 >>> 0;
@@ -96,12 +116,26 @@ function meanComposite(byScenario) {
 }
 async function selfImprove(opts) {
   const startedAt = Date.now();
+  const requestedRunDir = opts.runDir ?? `mem://selfImprove-${startedAt}`;
+  const runDir = resolveRunDir(requestedRunDir);
+  const storage = opts.storage ?? (runDir.startsWith("mem://") ? inMemoryCampaignStorage() : fsCampaignStorage());
+  const costLedger = createRunCostLedger({
+    storage,
+    runDir,
+    costCeilingUsd: opts.budget?.dollars
+  });
+  try {
+    return await runSelfImprove(opts, costLedger, startedAt, runDir, storage);
+  } catch (error) {
+    throw new SelfImproveRunError(error, costLedger);
+  }
+}
+async function runSelfImprove(opts, costLedger, startedAt, runDir, storage) {
   const budget = opts.budget ?? {};
   const generations = budget.generations ?? 3;
   const populationSize = budget.populationSize ?? 2;
   const maxConcurrency = budget.maxConcurrency ?? 2;
   const holdoutFraction = budget.holdoutFraction ?? 0.25;
-  const costCeiling = budget.dollars;
   const expectUsage = opts.expectUsage ?? "assert";
   const explicitHoldout = budget.holdoutScenarios;
   const { train, holdout } = explicitHoldout ? {
@@ -131,9 +165,6 @@ async function selfImprove(opts) {
     holdoutScenarios: holdout,
     deltaThreshold: 0.05
   });
-  const runDir = opts.runDir ?? `mem://selfImprove-${startedAt}`;
-  const isMemRunDir = runDir.startsWith("mem://");
-  const storage = opts.storage ?? (isMemRunDir ? inMemoryCampaignStorage() : fsCampaignStorage());
   if (opts.onProgress) {
     opts.onProgress({ kind: "baseline.started", scenarios: opts.scenarios.length });
   }
@@ -157,7 +188,7 @@ async function selfImprove(opts) {
     runDir,
     maxConcurrency,
     cellPlacement: opts.cellPlacement,
-    costCeiling,
+    costLedger,
     expectUsage,
     labeledStore: opts.labeledStore,
     captureSource: opts.captureSource,
@@ -201,10 +232,8 @@ async function selfImprove(opts) {
       lift: winnerStats.compositeMean - baseline.compositeMean
     });
   }
-  const totalCost = result.baselineCampaign.aggregates.totalCostUsd + result.generations.reduce(
-    (sum, gen) => sum + gen.surfaces.reduce((s, sf) => s + sf.campaign.aggregates.totalCostUsd, 0),
-    0
-  );
+  const cost = result.cost;
+  const totalCost = cost.totalCostUsd;
   const insight = await analyzeRuns({
     runs: [
       ...cellsToRunRecords(result.baselineCampaign.cells, "baseline", runDir, opts.baselineSurface),
@@ -256,6 +285,8 @@ async function selfImprove(opts) {
     generationsExplored: result.generations.length,
     durationMs,
     totalCostUsd: totalCost,
+    cost,
+    receipts: costLedger.list(),
     insight,
     ...power ? { power } : {},
     raw: result
@@ -988,10 +1019,15 @@ function collectNumericAttrs(spans) {
 export {
   FileSystemOutcomeStore,
   InMemoryOutcomeStore,
+  REFERENCE_EQUIVALENCE_INPUT_LIMITS,
+  REFERENCE_EQUIVALENCE_JUDGE_VERSION,
+  SelfImproveRunError,
   analyzeRuns,
   buildDefaultAnalystRegistry,
   buildEvidenceVector,
   composeGate,
+  createChatClient,
+  createReferenceEquivalenceJudge,
   defaultProductionGate,
   defineAgentEval,
   diffGenerations,
@@ -1020,6 +1056,7 @@ export {
   runCampaign,
   runEval,
   runImprovementLoop,
+  runReferenceEquivalenceJudge,
   selfImprove
 };
 //# sourceMappingURL=index.js.map