test: add workbench p1 reliability inventory

package.json

@@ -9,6 +9,7 @@
     "check:plan": "node scripts/check-runner.mjs --profile check --list",
     "check:cloud-api": "node scripts/check-runner.mjs --profile check --group cloud-api",
     "check:cloud-web": "node scripts/check-runner.mjs --profile check --group cloud-web",
+    "workbench:p1:inventory": "node scripts/workbench-long-reliability-p1-inventory.mjs",
     "dev-runtime-base:build": "node scripts/dev-runtime-base-image.mjs",
     "cloud-api:smoke": "node scripts/run-bun.mjs scripts/cloud-api-runtime-smoke.mjs",
     "l2:smoke": "node scripts/l2-runtime-contract-smoke.mjs",

scripts/workbench-long-reliability-p1-inventory.mjs

@@ -0,0 +1,151 @@
+#!/usr/bin/env node
+// SPEC: PJ2026-0104010803 Workbench唯一投影 draft-2026-06-20-p0-durable-facts-model.
+// Responsibility: P1 source inventory for HWLAB#1651/#1658. It intentionally exits non-zero while read-side projection assembly remains.
+
+import fs from "node:fs";
+import path from "node:path";
+
+const args = new Set(process.argv.slice(2));
+const allowRed = args.has("--allow-red");
+const json = args.has("--json");
+const root = process.cwd();
+const maxMatchesPerCheck = 8;
+
+const checks = [
+  {
+    id: "workbench-get-result-for-trace",
+    phase: "P4",
+    paths: ["internal/cloud/server-workbench-http.ts", "internal/cloud/workbench-read-model.ts"],
+    pattern: /resultForTrace\s*[:(]/gu,
+    expectation: "GET/read-model path must not assemble turn facts from mutable Code Agent result cache."
+  },
+  {
+    id: "workbench-get-trace-snapshot",
+    phase: "P4",
+    paths: ["internal/cloud/server-workbench-http.ts", "internal/cloud/workbench-read-model.ts"],
+    pattern: /traceSnapshot\s*[:(]/gu,
+    expectation: "GET/read-model path must read durable Workbench facts, not hydrate trace snapshots on demand."
+  },
+  {
+    id: "compact-session-trace-result",
+    phase: "P4",
+    paths: ["internal/cloud/server-workbench-http.ts"],
+    pattern: /compactSessionTraceResult|compactSessionTraceSnapshot/gu,
+    expectation: "Session list/detail must not reconstruct trace result or snapshot from compact session payloads."
+  },
+  {
+    id: "facts-store-runtime-read-through",
+    phase: "P2",
+    paths: ["internal/cloud/workbench-facts-store.ts"],
+    pattern: /runtimeStore\.queryAgentTraceEvents|durableTraceSnapshot\s*\(|shouldPreferDurableTrace|mergeTraceProjectionDiagnostic/gu,
+    expectation: "WorkbenchFactsStore should expose already persisted facts; runtime trace read-through belongs to projector/finalizer."
+  },
+  {
+    id: "get-handler-projection-assembly",
+    phase: "P4",
+    paths: ["internal/cloud/server-workbench-http.ts"],
+    pattern: /createWorkbenchTurnProjection\s*\(|sessionProjectionOptions\s*\(|sessionListProjectionOptions\s*\(/gu,
+    expectation: "GET handlers should serialize a read model DTO, not run projection assembly in the request path."
+  },
+  {
+    id: "frontend-trace-final-response-fallback",
+    phase: "P5",
+    paths: ["web/hwlab-cloud-web/src/composables/useTraceSubscription.ts", "web/hwlab-cloud-web/src/stores/workbench.ts"],
+    pattern: /finalResponse\s*:\s*[^\n]*\?\?|finalResponseText\(result\.finalResponse\)|traceSnapshotWithTurnStatus|mergeTerminalResultTrace/gu,
+    expectation: "Frontend must render sealed final facts from the server DTO, not infer final response from prior trace/result objects."
+  },
+  {
+    id: "frontend-status-fallback",
+    phase: "P5",
+    paths: ["web/hwlab-cloud-web/src/composables/useTraceSubscription.ts", "web/hwlab-cloud-web/src/stores/workbench.ts"],
+    pattern: /status\s*:\s*[^\n]*\?\?|\.status\s*\?\?\s*[^\n]*/gu,
+    expectation: "Frontend status authority must come from the Workbench DTO/reducer, not status fallback chains."
+  },
+  {
+    id: "probe-session-repair",
+    phase: "P7",
+    paths: ["scripts/web-live-dom-probe.mjs"],
+    pattern: /sessionRepair/gu,
+    expectation: "The final golden probe should not rely on helper-side session repair to mask default-route/read-model gaps."
+  }
+];
+
+const findings = [];
+const missingFiles = [];
+
+for (const check of checks) {
+  const matches = [];
+  for (const relativePath of check.paths) {
+    const filePath = path.resolve(root, relativePath);
+    if (!fs.existsSync(filePath)) {
+      missingFiles.push(relativePath);
+      continue;
+    }
+    const source = fs.readFileSync(filePath, "utf8");
+    const lineStarts = lineStartOffsets(source);
+    for (const match of source.matchAll(check.pattern)) {
+      const line = lineNumberForOffset(lineStarts, match.index ?? 0);
+      const snippet = source.split("\n")[line - 1]?.trim() ?? "";
+      matches.push({ path: relativePath, line, text: match[0], snippet });
+      if (matches.length >= maxMatchesPerCheck) break;
+    }
+  }
+  if (matches.length > 0) findings.push({ ...check, pattern: String(check.pattern), matches });
+}
+
+const report = {
+  ok: findings.length === 0 && missingFiles.length === 0,
+  status: findings.length > 0 ? "red" : missingFiles.length > 0 ? "blocked" : "green",
+  issue: "pikasTech/HWLAB#1658",
+  parentIssue: "pikasTech/HWLAB#1651",
+  checkedAt: new Date().toISOString(),
+  checkedFiles: [...new Set(checks.flatMap((check) => check.paths))],
+  missingFiles,
+  redFindingCount: findings.length,
+  findings,
+  nextPhases: [...new Set(findings.map((finding) => finding.phase))].sort(),
+  valuesPrinted: false
+};
+
+if (json) {
+  console.log(JSON.stringify(report, null, 2));
+} else {
+  printTextReport(report);
+}
+
+if (!report.ok && !allowRed) process.exit(1);
+
+function printTextReport(report) {
+  console.log(`Workbench P1 inventory: ${report.status}`);
+  console.log(`redFindingCount=${report.redFindingCount} missingFiles=${report.missingFiles.length}`);
+  if (report.missingFiles.length > 0) console.log(`missingFiles=${report.missingFiles.join(",")}`);
+  for (const finding of report.findings) {
+    console.log(`\n[${finding.phase}] ${finding.id}`);
+    console.log(`expectation: ${finding.expectation}`);
+    for (const match of finding.matches) {
+      console.log(`- ${match.path}:${match.line}: ${match.snippet}`);
+    }
+  }
+  if (report.findings.length > 0) {
+    console.log("\nThis is the expected P1 red inventory. Use --allow-red when collecting evidence without failing the shell.");
+  }
+}
+
+function lineStartOffsets(source) {
+  const starts = [0];
+  for (let index = 0; index < source.length; index += 1) {
+    if (source.charCodeAt(index) === 10) starts.push(index + 1);
+  }
+  return starts;
+}
+
+function lineNumberForOffset(starts, offset) {
+  let low = 0;
+  let high = starts.length - 1;
+  while (low <= high) {
+    const mid = Math.floor((low + high) / 2);
+    if (starts[mid] <= offset) low = mid + 1;
+    else high = mid - 1;
+  }
+  return Math.max(1, high + 1);
+}

web/hwlab-cloud-web/tests/workbench-e2e/fixtures/real-captures/d601-v03-workbench-p1-long-reliability-redacted.json

@@ -0,0 +1,103 @@
+{
+  "metadata": {
+    "captureId": "d601-v03-workbench-p1-long-reliability-redacted-001",
+    "capturedFrom": {
+      "node": "D601",
+      "lane": "HWLAB v0.3",
+      "origin": "https://hwlab.pikapython.com",
+      "source": "UniDesk controlled web-probe plus existing redacted Workbench fake-server captures",
+      "webProbe": {
+        "command": "bun scripts/cli.ts hwlab nodes web-probe script --node D601 --lane v03",
+        "scriptSha256": "sha256:5b0537b26266937b84eeebaef23985e1aa7507019ccd30d11dec78228e0d76ed",
+        "reportPath": "/home/ubuntu/workspace/hwlab-v03/.state/web-probe-script/run.6US9qW/web-probe-script-report.json",
+        "reportSha256": "sha256:00b6543859599fa416328aee509f2c2a9158a286132e30a2e916045ed501fcd9",
+        "valuesPrinted": false
+      }
+    },
+    "capturedAt": "2026-06-20T00:24:00.000Z",
+    "schemaVersion": "workbench-p1-long-reliability-inventory.v1",
+    "redactionVersion": "metadata-only-redaction.v1",
+    "derivedFrom": [
+      "web/hwlab-cloud-web/tests/workbench-e2e/fixtures/real-captures/d601-v03-redacted.json",
+      "web/hwlab-cloud-web/scripts/workbench-e2e-server.ts#sealed-final-response-diagnostics",
+      "web/hwlab-cloud-web/scripts/workbench-e2e-server.ts#trace-hydration-timeout-diagnostic",
+      "web/hwlab-cloud-web/scripts/workbench-e2e-server.ts#projection-degraded-diagnostics",
+      "web/hwlab-cloud-web/scripts/workbench-e2e-server.ts#projector-resume-from-checkpoint"
+    ],
+    "valuesPrinted": false,
+    "notes": [
+      "This fixture intentionally stores endpoint shape, scenario IDs, and source references only.",
+      "Session IDs in live evidence are reduced to prefix/shape in issue comments; full values remain in local redacted probe artifacts.",
+      "No cookie, Authorization header, API key, provider token, DSN, raw prompt, or user secret is stored."
+    ]
+  },
+  "liveReadModelShape": {
+    "finalUrlKind": "workbench-session-route",
+    "selectedSessionShape": {
+      "sessionIdPrefix": "ses_79e7",
+      "status": "idle",
+      "hasLastTraceId": false
+    },
+    "endpoints": [
+      {
+        "path": "/auth/session",
+        "status": 200,
+        "topLevelKeys": ["authenticated", "authMethod", "identityAuthority", "sessionKind"]
+      },
+      {
+        "path": "/v1/workbench/sessions?limit=5",
+        "status": 200,
+        "topLevelKeys": ["ok", "status", "contractVersion", "sessions"]
+      },
+      {
+        "path": "/v1/workbench/sessions/<selected-session>",
+        "status": 200,
+        "topLevelKeys": ["ok", "status", "contractVersion", "session"]
+      },
+      {
+        "path": "/v1/workbench/sessions/<selected-session>/messages?limit=100",
+        "status": 200,
+        "topLevelKeys": ["ok", "status", "contractVersion", "sessionId"]
+      }
+    ],
+    "valuesPrinted": false
+  },
+  "redScenarios": [
+    {
+      "scenarioId": "sealed-final-response-diagnostics",
+      "purpose": "completed sealed final response remains visible while trace hydration diagnostics are shown separately",
+      "covers": ["completed sealed after polling", "trace hydration error does not overwrite main message"]
+    },
+    {
+      "scenarioId": "trace-hydration-timeout-diagnostic",
+      "purpose": "GET lag or trace hydration timeout surfaces diagnostics instead of terminal/status inference",
+      "covers": ["projection lag", "diagnostic visibility", "no terminal inference"]
+    },
+    {
+      "scenarioId": "projection-degraded-diagnostics",
+      "purpose": "AgentRun result lag stays a projection diagnostic and keeps lifecycle running",
+      "covers": ["projection blocker", "running lifecycle authority"]
+    },
+    {
+      "scenarioId": "projector-resume-from-checkpoint",
+      "purpose": "future P2/P3 projector catch-up can turn P1 red inventory green without GET repair",
+      "covers": ["durable checkpoint", "pagination consistency", "no legacy request ledger"]
+    }
+  ],
+  "negativeSourceScan": {
+    "script": "scripts/workbench-long-reliability-p1-inventory.mjs",
+    "entrypoint": "npm run workbench:p1:inventory -- --json",
+    "expectedCurrentStatus": "red",
+    "trackedTerms": [
+      "resultForTrace",
+      "traceSnapshot",
+      "compactSessionTraceResult",
+      "compactSessionTraceSnapshot",
+      "status ??",
+      "finalResponse ?? previous",
+      "sessionRepair",
+      "GET handler projection assembly"
+    ],
+    "valuesPrinted": false
+  }
+}