Show real Codex Live loop evidence

Replace stale startup and synthetic claims with production browser timing, matched architecture comparisons, Accept latency, and honest run counts.\n\nAI-assisted: OpenAI Codex.

scripts/lib/live-benchmark.mjs

@@ -182,6 +182,7 @@ export function summarizeSetup(events) {
   const stages = [
     ['dependencies', 'setup.install.start', 'setup.install.end'],
     ['liveServer', 'setup.live_server.start', 'setup.live_server.end'],
+    ['codexWorker', 'setup.worker.start', 'setup.worker.end'],
     ['injection', 'setup.inject.start', 'setup.inject.end'],
     ['devServer', 'setup.dev_server.start', 'setup.dev_server.end'],
     ['pageLoad', 'setup.page_load.start', 'setup.page_load.end'],

site/data/live-codex-worker-result.ts

@@ -1,12 +1,12 @@
 export const liveCodexWorkerResult = {
-  generatedAt: '2026-07-13T03:03:13.899Z',
+  generatedAt: '2026-07-13T03:57:44.263Z',
   codexVersion: '0.144.0-alpha.4',
   transport: 'stdio JSONL',
   model: 'GPT-5.6-Sol',
   effort: 'medium',
   profile: 'quality',
   defaultInCodex: true,
-  coldWorkerReadyMs: 5600,
+  coldWorkerReadyMs: 656.82,
   timings: {
     coldHandshakeMs: 36.83,
     coldThreadStartMs: 400.19,
@@ -31,7 +31,30 @@ export const liveCodexWorkerResult = {
     sparkControlRuns: 2,
     sparkMedianGenerationMs: 16470.5,
   },
+  architectureComparison: {
+    directFirstUsableMs: 64649.73,
+    coldAppServerFirstUsableMs: 35591.71,
+    warmAppServerFirstUsableMs: 34931.08,
+    directQualityPassed: 4,
+    directQualityRuns: 4,
+    warmQualityPassed: 4,
+    warmQualityRuns: 4,
+    annotatedQualityPassed: 3,
+    annotatedQualityRuns: 3,
+    coldInputTokens: 41066,
+    coldOutputTokens: 2148,
+  },
+  productionBrowser: {
+    runs: 1,
+    firstVariantMs: 9868.3,
+    allVariantsMs: 31151.17,
+    acceptToResetMs: 298.71,
+    acceptToNextGoDispatchMs: 902.72,
+    acceptToNextFirstVariantMs: 16668,
+    sourcePreflightMs: 52,
+    validationMs: 10,
+  },
   result: '4/4 full-task quality gate',
   passed: true,
-  architecture: 'A dedicated persistent Live app-server connection and worker thread with native Impeccable skill, sub-command, project context, and annotation-image inputs. Do not attach a second app-server to the desktop task.',
+  architecture: 'A dedicated persistent Live app-server connection and worker thread with native Impeccable skill, sub-command, project context, and annotation-image inputs. Accept interrupts unfinished variants; cleanup stays on the foreground control lane while the next generation can start independently.',
 } as const;

site/pages/live-lab/index.astro

@@ -406,7 +406,7 @@ const providerColumns = [
       <div class="live-performance-finding live-finding-card--quiet" role="note">
         <strong>Dedicated Codex worker</strong>
         <p>
-          Codex now defaults to {liveCodexWorkerResult.model} at {liveCodexWorkerResult.effort} effort in a separate Live-owned thread. It passed {liveCodexWorkerResult.quality.passed}/{liveCodexWorkerResult.quality.runs} full <code>/bolder</code> and <code>/polish</code> tasks at a {displayMs(liveCodexWorkerResult.quality.medianGenerationMs)} median; the Spark control passed {liveCodexWorkerResult.quality.sparkControlPassed}/{liveCodexWorkerResult.quality.sparkControlRuns}. Protocol wake is {displayMs(liveCodexWorkerResult.timings.wakeToTurnStartedMs)}; the measured cold worker reached ready in {displayMs(liveCodexWorkerResult.coldWorkerReadyMs)} during Live initialization.
+          A real Pick → <code>/bolder</code> → Go run produced variant 1 in {displayMs(liveCodexWorkerResult.productionBrowser.firstVariantMs)} and all three in {displayMs(liveCodexWorkerResult.productionBrowser.allVariantsMs)}. Accept returned the page in {displayMs(liveCodexWorkerResult.productionBrowser.acceptToResetMs)}; the next Go reached the independent worker {displayMs(liveCodexWorkerResult.productionBrowser.acceptToNextGoDispatchMs)} after Accept. In the matched full-task benchmark, warm app-server reached usable output in {displayMs(liveCodexWorkerResult.architectureComparison.warmAppServerFirstUsableMs)} versus {displayMs(liveCodexWorkerResult.architectureComparison.directFirstUsableMs)} direct; both passed 4/4. Production browser timing is one run.
         </p>
       </div>
     </section>
@@ -441,6 +441,7 @@ const providerColumns = [
       <p>
         <code>bun run bench:live</code> boots a real framework fixture and Chromium, drives Pick → Go → Cycle,
         and records browser preparation, server pickup, scaffold, generation, write, and render boundaries.
+        Add <code>--agent=codex --action=bolder</code> to run the production app-server worker, or <code>--acceptFirst</code> to measure Accept → next Go.
         <code>node scripts/benchmark-live-init.mjs --iterations 10</code> measures configured cold and warm helper initialization separately.
         The deterministic agent makes Impeccable overhead visible; model-backed runs remain opt-in because they send fixture context to an external provider.
       </p>
@@ -469,8 +470,11 @@ const providerColumns = [
             <section class="live-lab-diagnostic-panel" data-lab-context="harness" hidden>
               <header><h2>Dedicated Codex worker</h2></header>
               <dl>
-                <div><dt>Wake</dt><dd>{displayMs(liveCodexWorkerResult.timings.wakeToTurnStartedMs)}</dd></div>
-                <div><dt>Full-task median</dt><dd>{displayMs(liveCodexWorkerResult.quality.medianGenerationMs)}</dd></div>
+                <div><dt>First review</dt><dd>{displayMs(liveCodexWorkerResult.productionBrowser.firstVariantMs)}</dd></div>
+                <div><dt>All variants</dt><dd>{displayMs(liveCodexWorkerResult.productionBrowser.allVariantsMs)}</dd></div>
+                <div><dt>Accept → page ready</dt><dd>{displayMs(liveCodexWorkerResult.productionBrowser.acceptToResetMs)}</dd></div>
+                <div><dt>Accept → next Go</dt><dd>{displayMs(liveCodexWorkerResult.productionBrowser.acceptToNextGoDispatchMs)}</dd></div>
+                <div><dt>Cold worker ready</dt><dd>{displayMs(liveCodexWorkerResult.coldWorkerReadyMs)}</dd></div>
                 <div><dt>Quality gate</dt><dd>{liveCodexWorkerResult.quality.passed}/{liveCodexWorkerResult.quality.runs}</dd></div>
               </dl>
               <p>Default in Codex; other harnesses keep the foreground path.</p>