@tangle-network/agent-eval 0.221.7 → 0.222.0

package/dist/contract/index.js

@@ -1,32 +1,32 @@
-import { a as hashCanonical } from "../canonical-Dq2hOQ8L.js";
-import "../run-record-DhEJgoPz.js";
-import { r as summarizeExecution, t as analyzeRuns } from "../analyze-runs-CHmsykNJ.js";
-import { i as pairedBootstrap } from "../mean-BbA_JCzW.js";
-import { t as decidePairedPromotion } from "../paired-promotion-decision-D0SHb-ZL.js";
-import { a as heldoutSignificance } from "../statistical-heldout-Bksv92Th.js";
-import { B as defaultProductionGate, Q as runCampaign, Z as runEval, i as transientDispatchFailure, o as clamp01, p as runImprovementLoop, t as llmJudge } from "../llm-judge-Cybq29bL.js";
-import { i as isModelPriced, n as estimateCost } from "../metrics-BKaPBtIG.js";
-import { t as CostLedger } from "../cost-ledger-BGZB6yY2.js";
-import { z as campaignSplitDigest } from "../campaign-evidence-DunzcJGv.js";
-import { o as mapConcurrentRange } from "../journal-C7om8Gum.js";
-import { M as fsCampaignStorage, N as inMemoryCampaignStorage } from "../external-optimizer-subprocess-B_ukPtRD.js";
-import { a as makeProposalFinding } from "../types-CnWmj8NW.js";
-import { n as isRecord } from "../json-BhdsJaOQ.js";
-import { n as SelfImproveRunError, r as selfImprove, t as defineAgentEval } from "../define-agent-eval-CCbT7Kfr.js";
-import { t as buildDefaultAnalystRegistry } from "../default-registry-BBTDgkDM.js";
-import { t as extractUsage } from "../extract-usage-BZLU01kR.js";
-import { t as createChatClient } from "../chat-client-BNATB3rJ.js";
-import { n as InMemoryOutcomeStore, t as FileSystemOutcomeStore } from "../outcome-store-DUcMj-3D.js";
-import { c as composeGate, l as externalTextOptimizationMethod, n as compareOptimizationMethods, r as gepaOptimizationMethod, s as heldOutGate, t as skillOptOptimizationMethod } from "../skillopt-optimization-method-sB_-yi1G.js";
-import { n as paretoPolicy, r as paretoSignificanceGate, t as buildEvidenceVector } from "../promotion-policy-6A8osGr_.js";
-import { t as fromOtelSpans } from "../otel-spans-CdN909k1.js";
+import { a as hashCanonical } from "../canonical-CicJlINA.js";
+import "../run-record-DEjx43cI.js";
+import { r as summarizeExecution, t as analyzeRuns } from "../analyze-runs-CTheASYV.js";
+import { i as pairedBootstrap } from "../mean-CdDYG-us.js";
+import { t as decidePairedPromotion } from "../paired-promotion-decision-FGvHA4p_.js";
+import { a as heldoutSignificance } from "../statistical-heldout-TP6ndL0B.js";
+import { f as mapConcurrentRange } from "../journal-C1uLJi0w.js";
+import { X as runEval, Z as runCampaign, a as clamp01, f as runImprovementLoop, r as transientDispatchFailure, t as llmJudge, z as defaultProductionGate } from "../llm-judge-cgbhTKQp.js";
+import { i as isModelPriced, n as estimateCost } from "../metrics-DNCAUtuB.js";
+import { t as CostLedger } from "../cost-ledger-CS8R3IZj.js";
+import { A as campaignSplitDigest } from "../campaign-evidence-DN8rWuBj.js";
+import { M as fsCampaignStorage, N as inMemoryCampaignStorage } from "../external-optimizer-subprocess-U616hV_l.js";
+import { a as makeProposalFinding } from "../types-BK0blWhe.js";
+import { n as isRecord } from "../json-DtvZNAU9.js";
+import { n as SelfImproveRunError, r as selfImprove, t as defineAgentEval } from "../define-agent-eval-BLoP4Lf9.js";
+import { t as buildDefaultAnalystRegistry } from "../default-registry-C1tPEjZ4.js";
+import { t as extractUsage } from "../extract-usage-ISFqhfDz.js";
+import { t as createChatClient } from "../chat-client-CO3uB33i.js";
+import { n as InMemoryOutcomeStore, t as FileSystemOutcomeStore } from "../outcome-store-TarrcfHG.js";
+import { c as composeGate, l as externalTextOptimizationMethod, n as compareOptimizationMethods, r as gepaOptimizationMethod, s as heldOutGate, t as skillOptOptimizationMethod } from "../skillopt-optimization-method-qB1M554K.js";
+import { n as paretoPolicy, r as paretoSignificanceGate, t as buildEvidenceVector } from "../promotion-policy-CBYeD4zi.js";
+import { t as fromOtelSpans } from "../otel-spans-BsGzTWIK.js";
 import { createHash } from "node:crypto";
 import { agentCandidateBenchmarkSuiteSchema, agentCandidateBenchmarkTaskSchema, agentCandidateBundleSchema, agentCandidateEvaluationPolicySchema, agentCandidateExperimentSchema, agentImprovementMeasuredComparisonSchema, agentProfileImprovementExperimentSchema, agentProfileImprovementMeasuredComparisonSchema, agentProfileImprovementRunCellSchema, agentProfileImprovementRunReceiptSchema, agentProfileImprovementSuiteInputsSchema, agentProfileImprovementSuiteSchema, agentProfileImprovementTaskSchema, candidateExecutionEvidenceSchema, canonicalCandidateDigest, canonicalCandidateJson, numbersApproximatelyEqual, omitTopLevelDigest } from "@tangle-network/agent-interface";
 import { createReadStream } from "node:fs";
 import "node:path";
 import "node:fs/promises";
 import { readSessionInput, summarizeSessionInput } from "@tangle-network/harness-sessions";
-//#region src/contract/fixed-spend.ts
+// src/contract/fixed-spend.ts
 function addFixedSpend(left, right) {
 	return {
 		inputTokens: left.inputTokens + right.inputTokens,
@@ -59,8 +59,7 @@ function pairedFixedSpendReceipt(measurements, model, usageOf) {
 		...usage.costProvenance === "observed" ? { actualCostUsd: costUsd } : { estimatedCostUsd: costUsd }
 	};
 }
-//#endregion
-//#region src/contract/candidate-experiment-evidence.ts
+// src/contract/candidate-experiment-evidence.ts
 /** Verify signed candidate execution receipts and read their measured usage. */
 const candidateExecutionEvidenceAdapter = {
 	score: (evidence) => evidence.receipt.benchmarkResult.material.score,
@@ -187,8 +186,7 @@ function combinedUsage(evidence) {
 function candidateMeasurementCostReceipt(measurements, model) {
 	return pairedFixedSpendReceipt(measurements, model, combinedUsage);
 }
-//#endregion
-//#region src/contract/measured-comparison-record.ts
+// src/contract/measured-comparison-record.ts
 /** Keep published timing and cost accounting identical across experiment domains. */
 function comparisonEvaluation(options, evaluation) {
 	return {
@@ -221,8 +219,7 @@ function comparisonRecomputationInput(comparison) {
 		...comparison.metadata ? { metadata: comparison.metadata } : {}
 	};
 }
-//#endregion
-//#region src/contract/concurrent-map.ts
+// src/contract/concurrent-map.ts
 /** Map both arms while bounding the actual executions, not the number of pairs. */
 async function mapPairedConcurrent(options) {
 	if (!Number.isSafeInteger(options.count) || options.count < 0) throw new Error(`${options.label} pair count must be a non-negative integer`);
@@ -243,8 +240,7 @@ async function mapPairedConcurrent(options) {
 		candidate: executions[index * 2 + 1]
 	}));
 }
-//#endregion
-//#region src/contract/paid-paired-measurement.ts
+// src/contract/paid-paired-measurement.ts
 /**
 * Reserve the complete signed suite before its first cell starts.
 *
@@ -299,8 +295,7 @@ function resolveCostLedger(options) {
 function assertNonNegativeFinite(value, label) {
 	if (!Number.isFinite(value) || value < 0) throw new Error(`${label} must be a non-negative finite number`);
 }
-//#endregion
-//#region src/contract/paired-measurement-input.ts
+// src/contract/paired-measurement-input.ts
 function projectPairedMeasurement(measurement, index, adapter) {
 	if (typeof measurement.cellId !== "string" || !measurement.cellId.trim()) throw new Error(`paired measurement ${index} requires a cell id`);
 	return {
@@ -358,8 +353,7 @@ function assertKnownCost(cost, label) {
 	if (!Number.isFinite(cost.usd) || cost.usd < 0) throw new Error(`${label} must be a non-negative number`);
 	if (cost.provenance !== "observed" && cost.provenance !== "estimated") throw new Error(`${label} provenance must be observed or estimated`);
 }
-//#endregion
-//#region src/contract/paired-measurement-evaluation.ts
+// src/contract/paired-measurement-evaluation.ts
 /** Shared significance, regression, precision, and budget decisions on paired receipts. */
 /**
 * Calculate the shared paired decision from any complete receipt shape.
@@ -734,8 +728,7 @@ function mean(values) {
 	if (values.length === 0) throw new Error("candidate experiment requires measured values");
 	return values.reduce((sum, value) => sum + value, 0) / values.length;
 }
-//#endregion
-//#region src/contract/candidate-experiment.ts
+// src/contract/candidate-experiment.ts
 /** Seal held-out experiments, execute their signed cells, and publish verified comparisons. */
 /** Content-address one task before any measured execution can see it. */
 function sealCandidateBenchmarkTask(material) {
@@ -970,8 +963,7 @@ function cellIds(experiment) {
 		return `${tasks[taskIndex]?.scenario.id ?? taskIndex}:${repetition}`;
 	});
 }
-//#endregion
-//#region src/contract/profile-measured-comparison.ts
+// src/contract/profile-measured-comparison.ts
 /** Content-address one held-out profile task before either state can execute it. */
 function sealAgentProfileImprovementTask(material) {
 	return agentProfileImprovementTaskSchema.parse({
@@ -1208,8 +1200,7 @@ const profileReceiptAdapter = {
 	completed: (receipt) => receipt.outcome.status === "succeeded",
 	passed: (receipt) => receipt.grading.passed
 };
-//#endregion
-//#region src/contract/intake/agent-trace.ts
+// src/contract/intake/agent-trace.ts
 function rangeLines(r) {
 	return Math.max(0, r.end_line - r.start_line + 1);
 }
@@ -1288,8 +1279,7 @@ function partitionRunsByAuthoringModel(runs, index) {
 		unattributed
 	};
 }
-//#endregion
-//#region src/contract/intake/code-agent-fields.ts
+// src/contract/intake/code-agent-fields.ts
 function record(value) {
 	return isRecord(value) ? value : null;
 }
@@ -1308,8 +1298,7 @@ function timestampMs(value) {
 		return Number.isFinite(parsed) ? parsed : void 0;
 	}
 }
-//#endregion
-//#region src/contract/intake/code-agent-identity.ts
+// src/contract/intake/code-agent-identity.ts
 function sessionIdFromEntries(source, entries) {
 	for (const entry of entries) {
 		if (source === "pi" && entry.type === "session") {
@@ -1376,8 +1365,7 @@ function hashString(value) {
 function stableSeed(value) {
 	return createHash("sha256").update(value).digest().readUInt32BE(0);
 }
-//#endregion
-//#region src/contract/intake/code-agent-observation.ts
+// src/contract/intake/code-agent-observation.ts
 /**
 * Where a session came from. `pi` is a native Pi session (`~/.pi/agent/sessions/*.jsonl`), read
 * through @tangle-network/harness-sessions; `pi-graph` is the graph IR a Pi graph run exports
@@ -2055,8 +2043,7 @@ function looksLikeError(value) {
 function nonEmpty(value) {
 	return value !== void 0 && value.length > 0 ? value : null;
 }
-//#endregion
-//#region src/contract/intake/code-agent-session-metrics.ts
+// src/contract/intake/code-agent-session-metrics.ts
 function emptyMetrics(entries) {
 	return {
 		entries,
@@ -2194,8 +2181,7 @@ function resolveSessionCost(explicit, metrics, model) {
 		}
 	};
 }
-//#endregion
-//#region src/contract/intake/code-agent-codex-metrics.ts
+// src/contract/intake/code-agent-codex-metrics.ts
 function codexMetrics(entries) {
 	const metrics = emptyMetrics(entries.length);
 	const startedToolIds = /* @__PURE__ */ new Set();
@@ -2314,8 +2300,7 @@ function codexProcessScore(metrics) {
 	const terminalTurns = metrics.turnsCompleted + metrics.turnsAborted;
 	return penalizeErrors(terminalTurns > 0 ? metrics.turnsCompleted / terminalTurns : metrics.assistantMessages > 0 ? .5 : 0, metrics);
 }
-//#endregion
-//#region src/contract/intake/code-agent-harness-metrics.ts
+// src/contract/intake/code-agent-harness-metrics.ts
 function claudeCodeMetrics(entries) {
 	const metrics = emptyMetrics(entries.length);
 	let startedAt;
@@ -2540,8 +2525,7 @@ function terminalProcessScore(metrics) {
 function maxOptional(current, next) {
 	return current === void 0 ? next : Math.max(current, next);
 }
-//#endregion
-//#region src/contract/intake/code-agent-jsonl.ts
+// src/contract/intake/code-agent-jsonl.ts
 /** The single per-line rule. Both the string path and the file path call this,
 *  so malformed-line handling and entry validation cannot drift apart. */
 function readCodeAgentJsonlLine(line, lineNumber) {
@@ -2621,8 +2605,7 @@ async function parseCodeAgentJsonlFile(path) {
 		malformedLines
 	};
 }
-//#endregion
-//#region src/contract/intake/code-agent-session.ts
+// src/contract/intake/code-agent-session.ts
 function fromCodexSession(options) {
 	return fromCodeAgentSession("codex", options);
 }
@@ -2868,8 +2851,7 @@ function firstUserText(entries) {
 function finiteScore(value) {
 	return value !== void 0 && Number.isFinite(value) ? clamp01(value) : void 0;
 }
-//#endregion
-//#region src/contract/intake/feedback-table.ts
+// src/contract/intake/feedback-table.ts
 function fromFeedbackTable(opts) {
 	const { ratings, meta = [], scale, emitRaterScores = true } = opts;
 	const metaByRun = new Map(meta.map((m) => [m.runId, m]));
@@ -2947,7 +2929,6 @@ function fromFeedbackTable(opts) {
 		raterScores
 	};
 }
-//#endregion
 export { FileSystemOutcomeStore, InMemoryOutcomeStore, SelfImproveRunError, analyzeRuns, buildDefaultAnalystRegistry, buildEvidenceVector, campaignSplitDigest, compareOptimizationMethods, composeGate, createChatClient, defaultProductionGate, defineAgentEval, evaluatePairedMeasurements, externalTextOptimizationMethod, fromClaudeCodeSession, fromCodexSession, fromFeedbackTable, fromKimiCodeSession, fromOpenCodeSession, fromOtelSpans, fromPiSession, fsCampaignStorage, gepaOptimizationMethod, heldOutGate, inMemoryCampaignStorage, llmJudge, makeProposalFinding, measuredComparisonFromAgentProfileImprovementExperiment, measuredComparisonFromCandidateExperiment, observeCodeAgentSession, paretoPolicy, paretoSignificanceGate, parseAgentTrace, parseCodeAgentJsonl, parseCodeAgentJsonlFile, partitionRunsByAuthoringModel, runAgentProfileImprovementExperiment, runCampaign, runCandidateExperiment, runEval, runImprovementLoop, sealAgentProfileImprovementExperiment, sealAgentProfileImprovementSuite, sealAgentProfileImprovementTask, sealCandidateBenchmarkSuite, sealCandidateBenchmarkTask, sealCandidateExperiment, selfImprove, skillOptOptimizationMethod, streamCodeAgentJsonlFile, summarizeExecution, transientDispatchFailure, verifyAgentProfileImprovementExperimentComparison, verifyCandidateExperiment, verifyCandidateExperimentComparison };
 
 //# sourceMappingURL=index.js.map