agent-inspect 6.31.16 → 6.31.18

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## 6.31.18
+
+### Patch Changes
+
+- c65817e: Fix recovery result-availability ordering and mixed attemptId counting; emit bounded OTLP operation/attempt identity without broad attribute dump; add real MCP SDK packed e2e and Fix-Owes recipe corpus (variable paths, timeout ledger, access disclosure, browser observed outcomes, NestJS tenant, causal failure, headsign, Keniel B8).
+
+## 6.31.17
+
+### Patch Changes
+
+- dc9aaf2: Harden sensitive-key complete-marker handling so arbitrary `[REDACTED:…]` / `[HASH:…]` payloads are re-scrubbed (library-emitted markers only), and ship the already-merged Windows `--policy` drive-path fix.
+
 ## 6.31.16
 
 ### Patch Changes

package/README.md

@@ -213,7 +213,7 @@ The root package is enough for custom capture, the CLI, checks, and Evidence wor
 
 ## Status and documentation
 
-**Current published baseline:** **6.31.16** · persisted schema `1.0` · Node.js `>=20` · MIT.
+**Current published baseline:** **6.31.18** · persisted schema `1.0` · Node.js `>=20` · MIT.
 
 Legacy v0.1 and v0.2 traces remain readable. Check the npm badge and [changelog](CHANGELOG.md) for the current published version.
 

package/SECURITY.md

@@ -62,6 +62,7 @@ AgentInspect aims to be safe-by-default for **log-derived attributes** and **exp
 
 - Log ingestion applies redaction to parsed attributes using configured rules (with conservative defaults).
 - Exporters default to redacted output and bounded attribute previews.
+- Sensitive-key complete-marker bypass trusts only library-emitted placeholders (`[REDACTED]`, `[HASH:<8 hex>]`, `[HASH:unknown]`). Arbitrary `[REDACTED:…]` / `[HASH:…]` strings are re-scrubbed. Report residual marker-grammar gaps via a private advisory when possible.
 
 Important limitations:
 

package/docs/COMPARE.md

@@ -100,6 +100,16 @@ AgentInspect avoids SDK/collector setup for local debugging:
 - Use OpenTelemetry when you need organization-wide production telemetry pipelines.
 - Local OTLP JSON export can help bridge later, but AgentInspect is not an OpenTelemetry SDK replacement.
 
+## AgentInspect vs OrcaReplay
+
+OrcaReplay is a local CLI that records an agent run at the model-provider boundary and serves the recorded turns back later without contacting the provider.
+
+AgentInspect is local-first CLI debugging:
+
+- Use AgentInspect to inspect and gate a run's trajectory; the two capture at different layers. AgentInspect builds a local execution tree from in-process instrumentation and from existing log/trace readers; OrcaReplay proxies the provider connection and records the bytes on the wire. Simultaneous AgentInspect + OrcaReplay capture on one run is possible in principle but **untested** here until a retained joint experiment exists.
+- If you reach for OrcaReplay, two limits are worth knowing: a matching replay is not a determinism result, and replay blocks model-provider egress only — it is not a sandbox, so recorded tool calls still execute and a tool that opens its own connection still reaches the network.
+- AgentInspect does not replay a recorded run; the README lists "a replay engine" among the things it is not.
+
 ## Quick decision table
 
 | Need | AgentInspect fit |
@@ -112,6 +122,7 @@ AgentInspect avoids SDK/collector setup for local debugging:
 | Share-safe local redaction copy | Good fit |
 | Existing structured logs to trees | Good fit |
 | VS Code trace review (in-repo extension) | Good fit; Marketplace listing separate |
+| Offline replay of a recorded run | Not the goal |
 | Production dashboards | Not the goal |
 | Hosted eval datasets | Not the goal |
 | Prompt management | Not the goal |

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-inspect",
-  "version": "6.31.16",
+  "version": "6.31.18",
   "license": "MIT",
   "type": "module",
   "description": "Local evidence debugger and trajectory-test toolkit for TypeScript AI agents — execution trees, TraceContract checks, Evidence v2, and read-only MCP",
@@ -236,7 +236,7 @@
     "test:all": "pnpm run typecheck && pnpm run linked-versions:check && pnpm run build && pnpm run test && pnpm run size",
     "prepublish:checks": "pnpm run typecheck && pnpm run build && pnpm run test && pnpm run test:coverage && pnpm run fixtures:check && pnpm run recipes:check && pnpm run test:integration-offline && pnpm run size && pnpm run linked-versions:check && pnpm run repo:health && pnpm run pack:smoke",
     "pack:dry-run": "pnpm run build && npm pack --dry-run",
-    "pack:smoke": "pnpm run package-licenses:check && pnpm run build && node scripts/package-smoke.mjs && node scripts/packed-openai-agents-e2e.mjs && node scripts/packed-ai-sdk-e2e.mjs && node scripts/packed-mcp-e2e.mjs && node scripts/packed-quickstart-e2e.mjs && node scripts/packed-semantic-loop-e2e.mjs && node scripts/packed-swarm-loop-e2e.mjs && node scripts/evidence-ci-golden-paths.mjs",
+    "pack:smoke": "pnpm run package-licenses:check && pnpm run build && node scripts/package-smoke.mjs && node scripts/packed-openai-agents-e2e.mjs && node scripts/packed-ai-sdk-e2e.mjs && node scripts/packed-mcp-e2e.mjs && node scripts/packed-mcp-sdk-e2e.mjs && node scripts/packed-quickstart-e2e.mjs && node scripts/packed-semantic-loop-e2e.mjs && node scripts/packed-swarm-loop-e2e.mjs && node scripts/evidence-ci-golden-paths.mjs",
     "linked-versions:check": "node scripts/check-linked-versions.mjs",
     "docs:commands": "node scripts/validate-doc-commands.mjs",
     "docs:links": "node scripts/validate-doc-links.mjs",

package/packages/cli/dist/{chunk-ZAYVAHIH.mjs → chunk-AHARYDFO.mjs}

@@ -930,7 +930,7 @@ var Redactor = class {
 };
 
 // packages/core/src/safety/credential-value-patterns.ts
-var KEY_VALUE_SECRET_PATTERN = /\b(?:api[_-]?key|internal[_-]?token|access[_-]?token|auth[_-]?token|password|secret|token)=(?!\[(?:REDACTED(?::[^\]]*)?|HASH:[0-9a-f]{8})\](?![^\s"'\\&#?]))([^\s"'\\]{8,})/i;
+var KEY_VALUE_SECRET_PATTERN = /\b(?:api[_-]?key|internal[_-]?token|access[_-]?token|auth[_-]?token|password|secret|token)=(?!\[(?:REDACTED|HASH:(?:[0-9a-f]{8}|unknown))\](?![^\s"'\\&#?]))([^\s"'\\]{8,})/i;
 var HIGH_CONFIDENCE_CREDENTIAL_PATTERNS = [
   /^(?:basic|bearer|digest|apikey)\s+[a-z0-9._~+/=-]+$/i,
   /\bbearer\s+[a-z0-9._~+/=-]{12,}\b/i,
@@ -5737,24 +5737,26 @@ function attemptIdOf(event) {
   return typeof value === "string" && value.trim() !== "" ? value : void 0;
 }
 function countOperationAttempts(members) {
-  const attemptIds = new Set(
-    members.map((event) => attemptIdOf(event)).filter((value) => value !== void 0)
-  );
-  if (attemptIds.size > 0) {
-    return attemptIds.size;
+  const identified = /* @__PURE__ */ new Set();
+  let unidentified = 0;
+  for (const event of members) {
+    if (event.kind !== "TOOL" && event.kind !== "LLM") continue;
+    const id = attemptIdOf(event);
+    if (id !== void 0) {
+      identified.add(id);
+      continue;
+    }
+    unidentified += 1;
+  }
+  if (identified.size > 0) {
+    return identified.size + unidentified;
   }
   const numbers = members.map((event) => attemptNumberOf(event)).filter((value) => value !== void 0).sort((a, b) => a - b);
   if (numbers.length > 0) {
     const unique = [...new Set(numbers)];
-    const min = unique[0];
-    const max = unique[unique.length - 1];
-    const contiguous = unique.length === max - min + 1 && unique.every((value, index) => value === min + index);
-    if (contiguous) {
-      return unique.length;
-    }
     return unique.length;
   }
-  return members.filter((event) => event.kind === "TOOL" || event.kind === "LLM").length;
+  return unidentified;
 }
 function classifyRetryAttempt(event, priorOrdered, allMembers) {
   const workflow = workflowFor(event);
@@ -6029,8 +6031,30 @@ function eventEvidence2(event) {
 function eventTime2(event) {
   return event.startedAt ?? event.timestamp ?? "";
 }
+function parseComparableMs(value) {
+  if (!value) return void 0;
+  const parsed = Date.parse(value);
+  return Number.isFinite(parsed) ? parsed : void 0;
+}
+function consumerStartMs(event) {
+  return parseComparableMs(event.startedAt) ?? parseComparableMs(event.timestamp);
+}
+function resultAvailableMs(event) {
+  const ended = parseComparableMs(event.endedAt);
+  if (ended !== void 0) return ended;
+  const started = parseComparableMs(event.startedAt) ?? parseComparableMs(event.timestamp);
+  if (started !== void 0 && event.durationMs !== void 0 && Number.isFinite(event.durationMs)) {
+    return started + event.durationMs;
+  }
+  return parseComparableMs(event.timestamp);
+}
 function sortByTime2(events) {
-  return [...events].sort((a, b) => eventTime2(a).localeCompare(eventTime2(b)));
+  return [...events].sort((a, b) => {
+    const aMs = consumerStartMs(a) ?? 0;
+    const bMs = consumerStartMs(b) ?? 0;
+    if (aMs !== bMs) return aMs - bMs;
+    return eventTime2(a).localeCompare(eventTime2(b));
+  });
 }
 function errorCodeOf(event) {
   if (typeof event.error?.code === "string" && event.error.code.trim() !== "") {
@@ -6098,13 +6122,20 @@ function sameArguments(left, right) {
   }
   return { ok: true };
 }
+var UNEVALUABLE_WRITE_ERROR_CODES = /* @__PURE__ */ new Set([
+  "TIMEOUT",
+  "ETIMEDOUT",
+  "UNKNOWN_COMPLETION",
+  "DEADLINE_EXCEEDED"
+]);
 function isUnevaluableWriteCompletion(event) {
   if (event.status === "running" || event.status === "unknown") return true;
   const attrs = event.attributes ?? {};
   if (attrs.timeout === true) return true;
+  if (attrs.unknownCompletion === true) return true;
   if (attrs.completionState === "unknown" || attrs.completionState === "timeout") return true;
   const code = errorCodeOf(event)?.toUpperCase();
-  if (code === "TIMEOUT" || code === "UNKNOWN_COMPLETION" || code === "DEADLINE_EXCEEDED") {
+  if (code && UNEVALUABLE_WRITE_ERROR_CODES.has(code)) {
     return true;
   }
   return false;
@@ -6152,14 +6183,17 @@ function collectExplicitReferences(event) {
   return refs;
 }
 function llmReferencesTool(llm, toolEvent, requireExplicit) {
-  if (!requireExplicit) {
-    return eventTime2(llm) >= eventTime2(toolEvent);
-  }
-  const refs = collectExplicitReferences(llm);
-  if (refs.has(toolEvent.eventId)) return true;
-  const toolCallId = workflowFor2(toolEvent).toolCallId;
-  if (toolCallId && refs.has(toolCallId)) return true;
-  return false;
+  if (llm.runId !== toolEvent.runId) return false;
+  if (requireExplicit) {
+    const refs = collectExplicitReferences(llm);
+    const toolCallId = workflowFor2(toolEvent).toolCallId;
+    const hasRef = refs.has(toolEvent.eventId) || toolCallId !== void 0 && refs.has(toolCallId);
+    if (!hasRef) return false;
+  }
+  const resultMs = resultAvailableMs(toolEvent);
+  const consumerMs = consumerStartMs(llm);
+  if (resultMs === void 0 || consumerMs === void 0) return false;
+  return resultMs <= consumerMs;
 }
 function attemptIdOf2(event) {
   const value = workflowFor2(event).attemptId;
@@ -7525,7 +7559,7 @@ var DEFAULT_SECRET_PATTERNS = [
   // Keep in sync with packages/redact/src/key-value-secret.ts (KEY_VALUE_SECRET_PATTERN_SOURCE).
   {
     id: "key-value-secret",
-    pattern: /\b(?:api[_-]?key|internal[_-]?token|access[_-]?token|auth[_-]?token|password|secret|token)=(?!\[(?:REDACTED(?::[^\]]*)?|HASH:[0-9a-f]{8})\](?![^\s"'\\&#?]))([^\s"'\\]{8,})/i
+    pattern: /\b(?:api[_-]?key|internal[_-]?token|access[_-]?token|auth[_-]?token|password|secret|token)=(?!\[(?:REDACTED|HASH:(?:[0-9a-f]{8}|unknown))\](?![^\s"'\\&#?]))([^\s"'\\]{8,})/i
   }
 ];
 function compareStrings(a, b) {
@@ -7966,8 +8000,8 @@ function isFullyRedactedValue(value, markers = ["[REDACTED]", "[REDACTED:"]) {
   if (typeof value !== "string") return false;
   const trimmed = value.trim();
   if (trimmed.length === 0) return false;
-  if (/^\[HASH:[A-Za-z0-9_-]+\]$/.test(trimmed)) return true;
-  if (/^\[REDACTED:[^\]]*\]$/.test(trimmed)) return true;
+  if (trimmed === "[HASH:unknown]") return true;
+  if (/^\[HASH:[0-9a-f]{8}\]$/.test(trimmed)) return true;
   for (const marker of markers) {
     if (marker.endsWith(":")) continue;
     if (trimmed === marker) return true;
@@ -15547,5 +15581,5 @@ function renderGateReport(result, options = {}) {
 }
 
 export { ATTRIBUTION_CONFIDENCES, COHORT_METRIC_IDS, DEFAULT_REDACT_KEYS, DEFAULT_SUITE_ARTIFACTS_DIR, EVIDENCE_FORMAT_VERSION, EVIDENCE_HTML_FILENAME, EVIDENCE_MANIFEST_FILENAME, Redactor, TraceDirectory, TraceReadError, TreeBuilder, aggregateBundleSafeStatus, aggregateSessionCheckResults, analyzeCohort, applyProfileMetadataCaps, assertBundlePathContained, assertEvidenceRelativePath, buildActivitySummary, buildBundleMetadata, buildBundleSummaryMarkdown, buildEvidenceCausalFailureViewHtml, buildEvidenceCiPackage, buildEvidenceCircuitViewHtml, buildEvidenceContractsViewHtml, buildEvidenceDiffViewHtml, buildEvidenceHtmlShell, buildEvidenceManifest, buildEvidenceOutcomesViewHtml, buildEvidenceProvenanceViewHtml, buildEvidenceSafetyViewHtml, buildEvidenceTimelineViewHtml, buildEvidenceToolsLlmViewHtml, buildEvidenceTreeViewHtml, buildLocalExplanation, buildPlaceholderArtifact, buildRunSummary, buildRunTimeline, buildRunWhatSummary, buildSessionIndex, buildTraceStats, buildZipArchive, bundleFailsOnSafety, bundleRunAssetRelativePath, collectTraceSchemaVersions, compactAttributes, createBaselineRegressionRule, createLlmUsageRule, createMaxStepDurationRule, createObservedOutcomeRule, createRequireCompletedRule, createRunDepthRule, createRunDurationRule, createRunStatusRule, createSafetyOversizedAttributeRule, createSafetyRawContentRule, createSafetyRedactionRule, createSafetySecretPatternRule, createStallDetectionRule, createStructureCycleRule, createStructureOrphanRule, createStructureParallelWidthRule, createStructureRelationshipRule, createToolUsageRule, defaultBundleOutputPath, defaultSuiteConfigTemplate, defineTraceContract, diffRuns, diffTraceEvents, enrichSessionRunRecord, escapeHtml, escapeMarkdown, evaluateTraceContract, extractMetadata, extractOutcomesFromTraceEvents, filterMetasBySessionScope, filterTraces, flattenTree, formatDuration2 as formatDuration, formatStepLabel, formatTimestamp, gateHasThresholds, getIndent, getTraceFilePath, inferEvidenceFileRole, isAgentInspectTrace, isAttributionConfidence, isCredentialSensitiveKey, isPersistedInspectEvent, loadSessionRunRecords, loadSuiteConfig, loadTraceMetadataList, manualTraceEventsToComparableRun, nanoid, normalizeBundleOutputPath, openTrace, parseCohortMetricList, parseDuration, parseDurationFilter, parseGateList, parseTraceJsonl, persistedInspectEventsToRunTrees, persistedInspectEventsToTraceEvents, projectLogicalEvents, renderActivitySummaryHuman, renderCohortReport, renderErrorLine, renderGateReport, renderObservedOutcomesHtml, renderObservedOutcomesMarkdown, renderRunDiff, renderRunWhat, renderStepLine, renderSuiteReport, renderTimeline, renderTraceStats, resolveBundleRunIds, resolveRedactionProfile, resolveSuiteTemplate, resolveTraceDir, runGate, runSuite, runTraceChecks, safeString, sanitizeBundleRunId, searchTraces, serializeEvidenceManifest, sha256Hex, stableJson, stringContainsHighConfidenceCredential, summarizeObservedOutcomes, summarizeSemanticParity, traceEventToPersistedInspectEvent, truncateName, truncateStringForProfile, validateEvent, validateSuiteConfig, verifyEvidenceDirectory, zeroKinds };
-//# sourceMappingURL=chunk-ZAYVAHIH.mjs.map
-//# sourceMappingURL=chunk-ZAYVAHIH.mjs.map
+//# sourceMappingURL=chunk-AHARYDFO.mjs.map
+//# sourceMappingURL=chunk-AHARYDFO.mjs.map