fix: surface OTel and sentinel visibility gaps

scripts/src/hwlab-node-web-observe-analyzer-source.ts

@@ -2743,7 +2743,7 @@ function buildFindings(samples, control, network, errors, sampleMetrics, promptN
   const effectiveApiDomLag = apiDomLag || buildApiDomLagReport(samples, network);
   if (commandFailures.length > 0) findings.push({ id: "observer-command-failed", severity: "red", summary: "observer control commands failed; analyze must surface command failure instead of hiding it in command artifacts", count: commandFailures.length, commands: commandFailures.slice(0, 20) });
   findings.push(...buildFrontendFreezeFindings(errors, control));
-  findings.push(...buildBrowserProcessFindings(browserProcess));
+  findings.push(...buildBrowserProcessFindings(browserProcess, runtimeAlerts));
   findings.push(...buildControlledNavigationRootCauseFindings(control, manifest));
   findings.push(...buildSessionInvariantFindings(control, manifest));
   const commandTimes = control
@@ -3220,10 +3220,11 @@ function buildBrowserProcessReport(rows) {
   };
 }
 
-function buildBrowserProcessFindings(report) {
+function buildBrowserProcessFindings(report, runtimeAlerts = null) {
   const summary = report?.summary || {};
   if (!summary || Number(summary.sampleCount ?? 0) <= 0) return [];
   const findings = [];
+  const rootCauseSignals = browserRootCauseSignals(report, runtimeAlerts);
   const maxTotalRssMb = Number(summary.maxTotalRssMb ?? 0);
   const maxProcessRssMb = Number(summary.maxProcessRssMb ?? 0);
   if (maxTotalRssMb >= alertThresholds.browserTotalRssRedMb || maxProcessRssMb >= alertThresholds.browserProcessRssRedMb) {
@@ -3240,6 +3241,7 @@ function buildBrowserProcessFindings(report) {
       rootCause: "frontend_browser_process_memory_pressure",
       rootCauseStatus: "confirmed-from-runner-process-rss",
       rootCauseConfidence: "high",
+      rootCauseSignals,
       fallbackAllowed: false,
       valuesRedacted: true,
     });
@@ -3260,6 +3262,7 @@ function buildBrowserProcessFindings(report) {
       rootCause: "frontend_browser_process_memory_leak_or_unbounded_render_growth",
       rootCauseStatus: "confirmed-from-runner-process-rss-growth",
       rootCauseConfidence: "high",
+      rootCauseSignals,
       fallbackAllowed: false,
       valuesRedacted: true,
     });
@@ -3285,6 +3288,7 @@ function buildBrowserProcessFindings(report) {
       rootCause: "frontend_browser_page_unresponsive_to_playwright",
       rootCauseStatus: "confirmed-from-cdp-runtime-evaluate",
       rootCauseConfidence: "high",
+      rootCauseSignals,
       fallbackAllowed: false,
       valuesRedacted: true,
     });
@@ -3303,6 +3307,7 @@ function buildBrowserProcessFindings(report) {
       rootCause: "frontend_browser_cdp_metrics_unresponsive",
       rootCauseStatus: "confirmed-from-cdp-metrics-timeouts",
       rootCauseConfidence: "high",
+      rootCauseSignals,
       fallbackAllowed: false,
       valuesRedacted: true,
     });
@@ -3310,6 +3315,32 @@ function buildBrowserProcessFindings(report) {
   return findings;
 }
 
+function browserRootCauseSignals(report, runtimeAlerts) {
+  const browserSummary = report?.summary || {};
+  const runtimeSummary = runtimeAlerts?.summary || {};
+  const sampleCount = Number(browserSummary.sampleCount ?? 0);
+  const sessionListReadCount = Number(runtimeSummary.workbenchSessionListReadCount ?? 0);
+  const traceEventsReadCount = Number(runtimeSummary.workbenchTraceEventsReadCount ?? 0);
+  const webPerformanceBeaconFailureCount = Number(runtimeSummary.webPerformanceBeaconFailureCount ?? 0);
+  const eventSourceFailureCount = Number(runtimeSummary.workbenchEventSourceFailureCount ?? 0);
+  const suspectedFrontendRefreshStorm = sessionListReadCount >= Math.max(20, sampleCount * 2);
+  return {
+    suspectedFrontendRefreshStorm,
+    sessionListReadCount,
+    traceEventsReadCount,
+    webPerformanceBeaconFailureCount,
+    eventSourceFailureCount,
+    requestFailedCount: runtimeSummary.significantRequestFailedCount ?? runtimeSummary.requestFailedCount ?? 0,
+    httpErrorCount: runtimeSummary.httpErrorCount ?? 0,
+    topRequestFailedPaths: (runtimeAlerts?.networkSignificantRequestFailedByPath ?? runtimeAlerts?.networkRequestFailedByPath ?? []).slice(0, 5),
+    topHttpErrorPaths: (runtimeAlerts?.networkHttpErrorsByPath ?? []).slice(0, 5),
+    note: suspectedFrontendRefreshStorm
+      ? "suspected frontend refresh storm: session list reads exceed the sample-derived budget during a browser red finding"
+      : "root-cause signals are included so memory/responsiveness/CDP red findings can be correlated without manual grep",
+    valuesRedacted: true,
+  };
+}
+
 function computeBrowserProcessGrowth(samples, windowMs) {
   const budgetMs = Math.max(1000, Number(windowMs || 0));
   const sorted = (samples || []).filter((item) => Number.isFinite(item.tsMs)).sort((a, b) => a.tsMs - b.tsMs);
@@ -4154,6 +4185,18 @@ function buildRuntimeAlerts(samples, control, network, consoleEvents, errors) {
   const requestFailed = naturalNetwork
     .filter((item) => item?.type === "requestfailed")
     .map((item) => networkAlertEvent(item, promptTimes));
+  const workbenchSessionListReadCount = naturalNetwork
+    .filter((item) => urlPath(item?.url) === "/v1/workbench/sessions")
+    .length;
+  const workbenchTraceEventsReadCount = naturalNetwork
+    .filter((item) => /^\/v1\/workbench\/traces\/[^/]+\/events$/u.test(urlPath(item?.url)))
+    .length;
+  const webPerformanceBeaconFailureCount = naturalNetwork
+    .filter((item) => urlPath(item?.url) === "/v1/web-performance" && (item?.type === "requestfailed" || Number(item?.status) >= 400))
+    .length;
+  const workbenchEventSourceFailureCount = naturalNetwork
+    .filter((item) => urlPath(item?.url) === "/v1/workbench/events" && (item?.type === "requestfailed" || Number(item?.status) >= 400))
+    .length;
   const significantRequestFailed = requestFailed.filter(
     (item) => !isBenignLongLivedStreamClosureAlert(item) && !isObserverRefreshClosureAlert(item, observerRefreshTimes),
   );
@@ -4284,6 +4327,10 @@ function buildRuntimeAlerts(samples, control, network, consoleEvents, errors) {
       httpErrorCount: httpErrors.length,
       requestFailedCount: requestFailed.length,
       significantRequestFailedCount: significantRequestFailed.length,
+      workbenchSessionListReadCount,
+      workbenchTraceEventsReadCount,
+      webPerformanceBeaconFailureCount,
+      workbenchEventSourceFailureCount,
       benignLongLivedStreamClosureCount: requestFailed.length - significantRequestFailed.length,
       domDiagnosticSampleCount: domDiagnostics.length,
       domDiagnosticGroupCount: groupDomDiagnostics(domDiagnostics).length,