perf(codex): index rollout transcript ids

.github/workflows/openclaw-release-checks.yml

@@ -44,11 +44,6 @@ on:
         required: false
         default: false
         type: boolean
-      run_maturity_scorecard:
-        description: Render advisory maturity scorecard release docs; default release checks rely on dedicated package, QA, live, and E2E gates
-        required: false
-        default: false
-        type: boolean
       rerun_group:
         description: Release check group to run
         required: false
@@ -111,7 +106,6 @@ jobs:
       mode: ${{ steps.inputs.outputs.mode }}
       release_profile: ${{ steps.inputs.outputs.release_profile }}
       run_release_soak: ${{ steps.inputs.outputs.run_release_soak }}
-      run_maturity_scorecard: ${{ steps.inputs.outputs.run_maturity_scorecard }}
       rerun_group: ${{ steps.inputs.outputs.rerun_group }}
       live_suite_filter: ${{ steps.inputs.outputs.live_suite_filter }}
       cross_os_suite_filter: ${{ steps.inputs.outputs.cross_os_suite_filter }}
@@ -285,7 +279,6 @@ jobs:
           RELEASE_MODE_INPUT: ${{ inputs.mode }}
           RELEASE_PROFILE_INPUT: ${{ inputs.release_profile }}
           RELEASE_RUN_RELEASE_SOAK_INPUT: ${{ inputs.run_release_soak }}
-          RELEASE_RUN_MATURITY_SCORECARD_INPUT: ${{ inputs.run_maturity_scorecard }}
           RELEASE_RERUN_GROUP_INPUT: ${{ inputs.rerun_group }}
           RELEASE_LIVE_SUITE_FILTER_INPUT: ${{ inputs.live_suite_filter }}
           RELEASE_CROSS_OS_SUITE_FILTER_INPUT: ${{ inputs.cross_os_suite_filter }}
@@ -326,12 +319,6 @@ jobs:
           else
             run_release_soak=true
           fi
-          run_maturity_scorecard="$(printf '%s' "$RELEASE_RUN_MATURITY_SCORECARD_INPUT" | tr '[:upper:]' '[:lower:]')"
-          if [[ "$run_maturity_scorecard" != "true" && "$run_maturity_scorecard" != "1" && "$run_maturity_scorecard" != "yes" ]]; then
-            run_maturity_scorecard=false
-          else
-            run_maturity_scorecard=true
-          fi
           release_profile="$RELEASE_PROFILE_INPUT"
           if [[ "$release_profile" == "minimum" ]]; then
             release_profile=beta
@@ -435,7 +422,6 @@ jobs:
             printf 'mode=%s\n' "$RELEASE_MODE_INPUT"
             printf 'release_profile=%s\n' "$release_profile"
             printf 'run_release_soak=%s\n' "$run_release_soak"
-            printf 'run_maturity_scorecard=%s\n' "$run_maturity_scorecard"
             printf 'rerun_group=%s\n' "$RELEASE_RERUN_GROUP_INPUT"
             printf 'live_suite_filter=%s\n' "$RELEASE_LIVE_SUITE_FILTER_INPUT"
             printf 'cross_os_suite_filter=%s\n' "$RELEASE_CROSS_OS_SUITE_FILTER_INPUT"
@@ -458,7 +444,6 @@ jobs:
           RELEASE_MODE: ${{ inputs.mode }}
           RELEASE_PROFILE: ${{ steps.inputs.outputs.release_profile }}
           RUN_RELEASE_SOAK: ${{ steps.inputs.outputs.run_release_soak }}
-          RUN_MATURITY_SCORECARD: ${{ steps.inputs.outputs.run_maturity_scorecard }}
           RELEASE_RERUN_GROUP: ${{ inputs.rerun_group }}
           RELEASE_LIVE_SUITE_FILTER: ${{ inputs.live_suite_filter }}
           RELEASE_CROSS_OS_SUITE_FILTER: ${{ inputs.cross_os_suite_filter }}
@@ -476,7 +461,6 @@ jobs:
             echo "- Cross-OS mode: \`${RELEASE_MODE}\`"
             echo "- Release profile: \`${RELEASE_PROFILE}\`"
             echo "- Release soak lanes: \`${RUN_RELEASE_SOAK}\`"
-            echo "- Maturity scorecard docs: \`${RUN_MATURITY_SCORECARD}\`"
             echo "- Rerun group: \`${RELEASE_RERUN_GROUP}\`"
             if [[ -n "${RELEASE_LIVE_SUITE_FILTER// }" ]]; then
               echo "- Live suite filter: \`${RELEASE_LIVE_SUITE_FILTER}\`"
@@ -786,7 +770,7 @@ jobs:
   maturity_scorecard_release_checks:
     name: Render maturity scorecard release docs
     needs: [resolve_target]
-    if: contains(fromJSON('["all","qa"]'), needs.resolve_target.outputs.rerun_group) && needs.resolve_target.outputs.run_maturity_scorecard == 'true'
+    if: contains(fromJSON('["all","qa"]'), needs.resolve_target.outputs.rerun_group)
     permissions:
       actions: read
       contents: read

extensions/codex/src/app-server/native-subagent-monitor.ts

@@ -91,6 +91,9 @@ const DEFAULT_COMPLETION_DELIVERY_RETRY_DELAYS_MS = [
 ];
 const DEFAULT_TASK_ROW_RECONCILE_INTERVAL_MS = 10_000;
 const RECENT_TERMINAL_TASK_RECONCILE_GRACE_MS = 60_000;
+// Codex's recorder uses this filename contract; non-canonical names keep the
+// legacy substring fallback for older or test-created transcript files.
+const CODEX_ROLLOUT_FILENAME_RE = /^rollout-\d{4}-\d{2}-\d{2}T\d{2}-\d{2}-\d{2}-(.+)\.jsonl$/u;
 
 const defaultRuntime: NativeSubagentMonitorRuntime = {
   createAgentHarnessTaskRuntime,
@@ -1188,8 +1191,9 @@ async function findTranscriptPaths(params: {
 }): Promise<Map<string, string>> {
   const sessionsDir = path.join(params.codexHome, "sessions");
   const found = new Map<string, string>();
+  const remaining = new Set(params.childThreadIds);
   const stack = [sessionsDir];
-  while (stack.length > 0 && found.size < params.childThreadIds.size) {
+  while (stack.length > 0 && remaining.size > 0) {
     const dir = stack.pop()!;
     let entries: Array<{ name: string; isDirectory(): boolean; isFile(): boolean }>;
     try {
@@ -1206,10 +1210,20 @@ async function findTranscriptPaths(params: {
       if (!entry.isFile() || !entry.name.endsWith(".jsonl")) {
         continue;
       }
-      for (const childThreadId of params.childThreadIds) {
-        if (!found.has(childThreadId) && entry.name.includes(childThreadId)) {
+      const rolloutMatch = entry.name.match(CODEX_ROLLOUT_FILENAME_RE);
+      if (rolloutMatch) {
+        const childThreadId = rolloutMatch[1];
+        if (remaining.delete(childThreadId)) {
           found.set(childThreadId, entryPath);
         }
+        continue;
+      }
+      for (const childThreadId of remaining) {
+        if (entry.name.includes(childThreadId)) {
+          found.set(childThreadId, entryPath);
+          remaining.delete(childThreadId);
+          break;
+        }
       }
     }
   }
@@ -1236,10 +1250,13 @@ async function findTranscriptPath(params: {
         stack.push(entryPath);
         continue;
       }
+      const rolloutMatch = entry.name.match(CODEX_ROLLOUT_FILENAME_RE);
       if (
         entry.isFile() &&
         entry.name.endsWith(".jsonl") &&
-        entry.name.includes(params.childThreadId)
+        (rolloutMatch
+          ? rolloutMatch[1] === params.childThreadId
+          : entry.name.includes(params.childThreadId))
       ) {
         return entryPath;
       }

scripts/check-gateway-cpu-scenarios.mjs

@@ -24,13 +24,6 @@ const DEFAULT_QA_SCENARIOS = [
   "memory-failure-fallback",
   "gateway-restart-inflight-run",
 ];
-const SINGLE_VALUE_FLAGS = new Set([
-  "--cpu-core-warn",
-  "--hot-wall-warn-ms",
-  "--output-dir",
-  "--runs",
-  "--warmup",
-]);
 const DEFAULT_CPU_CORE_WARN = 0.9;
 const DEFAULT_HOT_WALL_WARN_MS = 30_000;
 const PRIVATE_QA_REQUIRED_DIST_ENTRIES = [
@@ -56,15 +49,8 @@ function parseArgs(argv) {
     cpuCoreWarn: DEFAULT_CPU_CORE_WARN,
     hotWallWarnMs: DEFAULT_HOT_WALL_WARN_MS,
   };
-  const seenSingleValueFlags = new Set();
   for (let index = 0; index < args.length; index += 1) {
     const arg = args[index];
-    if (SINGLE_VALUE_FLAGS.has(arg)) {
-      if (seenSingleValueFlags.has(arg)) {
-        throw new Error(`${arg} was provided more than once`);
-      }
-      seenSingleValueFlags.add(arg);
-    }
     const readValue = () => {
       const value = args[index + 1];
       if (!value || value.startsWith("-")) {

scripts/check-plugin-gateway-gauntlet.mjs

@@ -35,24 +35,6 @@ const DEFAULT_CPU_CORE_WARN = 0.9;
 const DEFAULT_HOT_WALL_WARN_MS = 30_000;
 const DEFAULT_MAX_RSS_WARN_MB = 1536;
 const DEFAULT_QA_PLUGIN_CHUNK_SIZE = 12;
-const SINGLE_VALUE_FLAGS = new Set([
-  "--build-timeout-ms",
-  "--command-timeout-ms",
-  "--cpu-core-warn",
-  "--hot-wall-warn-ms",
-  "--limit",
-  "--max-rss-warn-mb",
-  "--output-dir",
-  "--qa-cpu-regression-multiplier",
-  "--qa-plugin-chunk-size",
-  "--qa-timeout-ms",
-  "--qa-wall-regression-multiplier",
-  "--repo-root",
-  "--rss-anomaly-multiplier",
-  "--shard-index",
-  "--shard-total",
-  "--wall-anomaly-multiplier",
-]);
 const COMMAND_OUTPUT_MAX_BUFFER_BYTES = 16 * 1024 * 1024;
 const MAX_TIMER_TIMEOUT_MS = 2_147_000_000;
 const ANSI_PATTERN = new RegExp(String.raw`\u001B\[[0-9;]*m`, "gu");
@@ -97,15 +79,8 @@ export function parseArgs(argv) {
   };
   const envIds = normalizeCsv(process.env.OPENCLAW_PLUGIN_GATEWAY_GAUNTLET_IDS);
   options.pluginIds.push(...envIds);
-  const seenSingleValueFlags = new Set();
   parseArgv: for (let index = 0; index < args.length; index += 1) {
     const arg = args[index];
-    if (SINGLE_VALUE_FLAGS.has(arg)) {
-      if (seenSingleValueFlags.has(arg)) {
-        throw new Error(`${arg} was provided more than once`);
-      }
-      seenSingleValueFlags.add(arg);
-    }
     const readValue = () => {
       const value = args[index + 1];
       if (!value || value.startsWith("-")) {

test/scripts/check-gateway-cpu-scenarios.test.ts

@@ -111,19 +111,6 @@ describe("gateway CPU scenario guard", () => {
     }
   });
 
-  it("rejects duplicate single-value controls before running scenarios", () => {
-    expect(() =>
-      testing.parseArgs(["--output-dir", makeTempRoot(), "--output-dir", makeTempRoot()]),
-    ).toThrow("--output-dir was provided more than once");
-
-    const result = runCli("--runs", "1", "--runs", "2");
-
-    expect(result.status).toBe(1);
-    expect(result.stdout).toBe("");
-    expect(result.stderr.trim()).toBe("--runs was provided more than once");
-    expectNoNodeStack(result.stderr);
-  });
-
   it("reports CLI argument errors without a Node stack trace", () => {
     const result = runCli("--wat");
 

test/scripts/ci-workflow-guards.test.ts

@@ -1,6 +1,5 @@
 // Ci Workflow Guards tests cover ci workflow guards script behavior.
-import { execFileSync } from "node:child_process";
-import { existsSync, readdirSync, readFileSync } from "node:fs";
+import { readdirSync, readFileSync } from "node:fs";
 import { describe, expect, it } from "vitest";
 import { parse } from "yaml";
 
@@ -38,17 +37,10 @@ function readCriticalQualityWorkflow() {
   return readFileSync(".github/workflows/codeql-critical-quality.yml", "utf8");
 }
 
-function readTrackedText(relativePath: string): string {
-  if (existsSync(relativePath)) {
-    return readFileSync(relativePath, "utf8");
-  }
-  return execFileSync("git", ["show", `:${relativePath}`], { encoding: "utf8" });
-}
-
-function readAndroidCompileSdk(relativePath: string): number {
-  const match = readTrackedText(relativePath).match(/^\s*compileSdk\s*=\s*(\d+)\s*$/mu);
+function readAndroidCompileSdk(path: string): number {
+  const match = readFileSync(path, "utf8").match(/^\s*compileSdk\s*=\s*(\d+)\s*$/mu);
   if (!match) {
-    throw new Error(`Missing compileSdk in ${relativePath}`);
+    throw new Error(`Missing compileSdk in ${path}`);
   }
   return Number(match[1]);
 }
@@ -735,33 +727,18 @@ describe("ci workflow guards", () => {
     expect(openDocsPrStep.if).toBe("${{ github.event_name == 'workflow_dispatch' }}");
   });
 
-  it("keeps maturity scorecard release docs opt-in from release checks", () => {
+  it("runs maturity scorecard from release checks", () => {
     const releaseWorkflow = readReleaseChecksWorkflow();
     const job = releaseWorkflow.jobs.maturity_scorecard_release_checks;
     const summaryJob = releaseWorkflow.jobs.summary;
     const verifyStep = summaryJob.steps.find(
       (step) => step.name === "Verify release check results",
     );
-    const inputs = releaseWorkflow.on.workflow_dispatch.inputs;
-    const resolveJob = releaseWorkflow.jobs.resolve_target;
-    const summarizeStep = resolveJob.steps.find((step) => step.name === "Summarize validated ref");
 
     expect(releaseWorkflow.jobs).not.toHaveProperty("qa_profile_release_evidence_release_checks");
-    expect(inputs.run_maturity_scorecard).toMatchObject({
-      required: false,
-      default: false,
-      type: "boolean",
-    });
-    expect(resolveJob.outputs.run_maturity_scorecard).toBe(
-      "${{ steps.inputs.outputs.run_maturity_scorecard }}",
-    );
-    expect(summarizeStep.env.RUN_MATURITY_SCORECARD).toBe(
-      "${{ steps.inputs.outputs.run_maturity_scorecard }}",
-    );
-    expect(summarizeStep.run).toContain("- Maturity scorecard docs:");
     expect(job.name).toBe("Render maturity scorecard release docs");
     expect(job.if).toBe(
-      "contains(fromJSON('[\"all\",\"qa\"]'), needs.resolve_target.outputs.rerun_group) && needs.resolve_target.outputs.run_maturity_scorecard == 'true'",
+      'contains(fromJSON(\'["all","qa"]\'), needs.resolve_target.outputs.rerun_group)',
     );
     expect(job.permissions).toMatchObject({
       actions: "read",

test/scripts/plugin-gateway-gauntlet.test.ts

@@ -150,15 +150,6 @@ describe("plugin gateway gauntlet helpers", () => {
     );
   });
 
-  it("rejects duplicate single-value controls", () => {
-    expect(() =>
-      parseArgs(["--output-dir", ".artifacts/one", "--output-dir", ".artifacts/two"]),
-    ).toThrow("--output-dir was provided more than once");
-    expect(() => parseArgs(["--shard-total", "2", "--shard-total", "3"])).toThrow(
-      "--shard-total was provided more than once",
-    );
-  });
-
   it("rejects valued flags followed by another option", () => {
     for (const flag of [
       "--repo-root",