agent-inspect 6.31.20 → 6.31.21
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +6 -0
- package/README.md +1 -1
- package/package.json +1 -1
- package/packages/cli/dist/{chunk-EVSIYDB2.mjs → chunk-N5JN26MW.mjs} +30 -10
- package/packages/cli/dist/chunk-N5JN26MW.mjs.map +1 -0
- package/packages/cli/dist/index.cjs +29 -9
- package/packages/cli/dist/index.cjs.map +1 -1
- package/packages/cli/dist/index.mjs +4 -4
- package/packages/cli/dist/index.mjs.map +1 -1
- package/packages/cli/dist/{src-NWDDK72G.mjs → src-VMGRSTG7.mjs} +3 -3
- package/packages/cli/dist/{src-NWDDK72G.mjs.map → src-VMGRSTG7.mjs.map} +1 -1
- package/packages/core/dist/advanced.mjs +2 -2
- package/packages/core/dist/checks.cjs +28 -8
- package/packages/core/dist/checks.cjs.map +1 -1
- package/packages/core/dist/checks.mjs +1 -1
- package/packages/core/dist/{chunk-GFQPPNSF.mjs → chunk-4CDX5GDS.mjs} +30 -10
- package/packages/core/dist/chunk-4CDX5GDS.mjs.map +1 -0
- package/packages/cli/dist/chunk-EVSIYDB2.mjs.map +0 -1
- package/packages/core/dist/chunk-GFQPPNSF.mjs.map +0 -1
|
@@ -1,4 +1,4 @@
|
|
|
1
|
-
export { buildTraceFacts, createBaselineRegressionRule, createDecisionRule, createGuardrailRule, createLlmUsageRule, createMaxStepDurationRule, createObservedOutcomeRule, createRequireCompletedRule, createRetrievalRule, createRunDepthRule, createRunDurationRule, createRunEventCountRule, createRunStatusRule, createSafetyOversizedAttributeRule, createSafetyRawContentRule, createSafetyRedactionRule, createSafetySecretPatternRule, createStallDetectionRule, createStepOrderingRule, createStructureCycleRule, createStructureIncompleteRule, createStructureOrphanRule, createStructureParallelWidthRule, createStructureRelationshipRule, createToolFailureRule, createToolOrderingRule, createToolUsageRule, defineTraceContract, deriveFailureFacts, deriveRelationshipFacts, evaluateToolArgumentValue, evaluateTraceContract, evaluateTraceContractRead, explainTraceContract, extractToolArgumentPayload, lintTraceContract, resolveJsonPointer, resolveTraceContractScope, runTraceChecks, summarizeSemanticParity, workflowMetadataForRun } from './chunk-
|
|
1
|
+
export { buildTraceFacts, createBaselineRegressionRule, createDecisionRule, createGuardrailRule, createLlmUsageRule, createMaxStepDurationRule, createObservedOutcomeRule, createRequireCompletedRule, createRetrievalRule, createRunDepthRule, createRunDurationRule, createRunEventCountRule, createRunStatusRule, createSafetyOversizedAttributeRule, createSafetyRawContentRule, createSafetyRedactionRule, createSafetySecretPatternRule, createStallDetectionRule, createStepOrderingRule, createStructureCycleRule, createStructureIncompleteRule, createStructureOrphanRule, createStructureParallelWidthRule, createStructureRelationshipRule, createToolFailureRule, createToolOrderingRule, createToolUsageRule, defineTraceContract, deriveFailureFacts, deriveRelationshipFacts, evaluateToolArgumentValue, evaluateTraceContract, evaluateTraceContractRead, explainTraceContract, extractToolArgumentPayload, lintTraceContract, resolveJsonPointer, resolveTraceContractScope, runTraceChecks, summarizeSemanticParity, workflowMetadataForRun } from './chunk-4CDX5GDS.mjs';
|
|
2
2
|
import './chunk-77UZIMMD.mjs';
|
|
3
3
|
import './chunk-EOXUEQB6.mjs';
|
|
4
4
|
import './chunk-VLVCLAPB.mjs';
|
|
@@ -1318,13 +1318,33 @@ function classifyRetryAttempt(event, priorOrdered, allMembers) {
|
|
|
1318
1318
|
}
|
|
1319
1319
|
return { kind: "unknown", reason: "ambiguous attempt identity" };
|
|
1320
1320
|
}
|
|
1321
|
-
function
|
|
1322
|
-
const
|
|
1321
|
+
function redeliveryFingerprint(event) {
|
|
1322
|
+
const workflow = workflowFor(event);
|
|
1323
|
+
const stableEventId = (event.eventId ?? "").replace(/:\d+$/, "");
|
|
1324
|
+
return JSON.stringify({
|
|
1325
|
+
stableEventId,
|
|
1326
|
+
runId: event.runId,
|
|
1327
|
+
kind: event.kind,
|
|
1328
|
+
name: event.name,
|
|
1329
|
+
status: event.status,
|
|
1330
|
+
operationId: workflow.operationId ?? null,
|
|
1331
|
+
attemptId: workflow.attemptId ?? null,
|
|
1332
|
+
attemptNumber: workflow.attemptNumber ?? null,
|
|
1333
|
+
retryOf: workflow.retryOf ?? null
|
|
1334
|
+
});
|
|
1335
|
+
}
|
|
1336
|
+
function hasConflictingAttemptIdReuse(members) {
|
|
1337
|
+
const byAttempt = /* @__PURE__ */ new Map();
|
|
1323
1338
|
for (const event of members) {
|
|
1339
|
+
if (event.kind !== "TOOL" && event.kind !== "LLM") continue;
|
|
1324
1340
|
const id = attemptIdOf(event);
|
|
1325
1341
|
if (!id) continue;
|
|
1326
|
-
|
|
1327
|
-
|
|
1342
|
+
const fingerprints = byAttempt.get(id) ?? /* @__PURE__ */ new Set();
|
|
1343
|
+
fingerprints.add(redeliveryFingerprint(event));
|
|
1344
|
+
byAttempt.set(id, fingerprints);
|
|
1345
|
+
}
|
|
1346
|
+
for (const fingerprints of byAttempt.values()) {
|
|
1347
|
+
if (fingerprints.size > 1) return true;
|
|
1328
1348
|
}
|
|
1329
1349
|
return false;
|
|
1330
1350
|
}
|
|
@@ -1341,11 +1361,11 @@ function evaluateRetrySafetyRules(events, rules, _runEvidence) {
|
|
|
1341
1361
|
}
|
|
1342
1362
|
if (rules.maxAttempts !== void 0) {
|
|
1343
1363
|
for (const [operationId, members] of byOperation) {
|
|
1344
|
-
if (
|
|
1364
|
+
if (hasConflictingAttemptIdReuse(members)) {
|
|
1345
1365
|
findings.push(
|
|
1346
1366
|
fail2(
|
|
1347
1367
|
"contract.retry.duplicate-attempt-id",
|
|
1348
|
-
`Operation ${operationId} has
|
|
1368
|
+
`Operation ${operationId} has conflicting attemptId reuse (distinct executions).`,
|
|
1349
1369
|
members.slice(0, 4).map(eventEvidence),
|
|
1350
1370
|
"unique attemptId",
|
|
1351
1371
|
"duplicate"
|
|
@@ -1730,7 +1750,7 @@ function attemptIdOf2(event) {
|
|
|
1730
1750
|
const value = workflowFor2(event).attemptId;
|
|
1731
1751
|
return typeof value === "string" && value.trim() !== "" ? value.trim() : void 0;
|
|
1732
1752
|
}
|
|
1733
|
-
function
|
|
1753
|
+
function hasDuplicateAttemptIds(members) {
|
|
1734
1754
|
const seen = /* @__PURE__ */ new Set();
|
|
1735
1755
|
for (const event of members) {
|
|
1736
1756
|
const id = attemptIdOf2(event);
|
|
@@ -1826,7 +1846,7 @@ function evaluateRecoveryOperations(events, operations) {
|
|
|
1826
1846
|
}
|
|
1827
1847
|
for (const [operationKey, members] of groups) {
|
|
1828
1848
|
const ordered = sortByTime2(members);
|
|
1829
|
-
if (
|
|
1849
|
+
if (hasDuplicateAttemptIds(members)) {
|
|
1830
1850
|
findings.push(
|
|
1831
1851
|
fail3(
|
|
1832
1852
|
"contract.retry.operations.duplicate-attempt-id",
|
|
@@ -5318,5 +5338,5 @@ function runTraceChecks(input, options = {}) {
|
|
|
5318
5338
|
}
|
|
5319
5339
|
|
|
5320
5340
|
export { buildTraceFacts, createBaselineRegressionRule, createDecisionRule, createGuardrailRule, createLlmUsageRule, createMaxStepDurationRule, createObservedOutcomeRule, createRequireCompletedRule, createRetrievalRule, createRunDepthRule, createRunDurationRule, createRunEventCountRule, createRunStatusRule, createSafetyOversizedAttributeRule, createSafetyRawContentRule, createSafetyRedactionRule, createSafetySecretPatternRule, createStallDetectionRule, createStepOrderingRule, createStructureCycleRule, createStructureIncompleteRule, createStructureOrphanRule, createStructureParallelWidthRule, createStructureRelationshipRule, createToolFailureRule, createToolOrderingRule, createToolUsageRule, defineTraceContract, deriveFailureFacts, deriveRelationshipFacts, evaluateToolArgumentValue, evaluateTraceContract, evaluateTraceContractRead, explainTraceContract, extractSessionWorkflowMetadata, extractToolArgumentPayload, lintTraceContract, resolveJsonPointer, resolveTraceContractScope, runTraceChecks, sessionKeyForRun, summarizeSemanticParity, workflowMetadataForRun };
|
|
5321
|
-
//# sourceMappingURL=chunk-
|
|
5322
|
-
//# sourceMappingURL=chunk-
|
|
5341
|
+
//# sourceMappingURL=chunk-4CDX5GDS.mjs.map
|
|
5342
|
+
//# sourceMappingURL=chunk-4CDX5GDS.mjs.map
|