study8677 · study8677 · May 21, 2026 · May 21, 2026
diff --git a/frontend/index.html b/frontend/index.html
@@ -13,6 +13,15 @@
       content="OpenCMO, AI CMO, SEO audit, GEO audit, AI search visibility, community visibility, open-source growth tools, self-hosted growth console"
     />
     <meta name="robots" content="index,follow,max-image-preview:large,max-snippet:-1,max-video-preview:-1" />
+    <!--
+      Opt the React-managed app shell out of browser/Google Translate.
+      Chrome Translate replaces text nodes in place, which breaks React's
+      reconciliation and surfaces as `NotFoundError: Failed to execute
+      'insertBefore' on 'Node'` followed by a blank `#root`. The static
+      SEO copy in <main id="static-site-copy"> below opts back in via
+      translate="yes" so crawlers and translated previews still see it.
+    -->
+    <meta name="google" content="notranslate" />
     <meta name="theme-color" content="#0f172a" />
     <link rel="canonical" href="https://www.aidcmo.com/" />
 
@@ -62,8 +71,8 @@
       }
     </script>
   </head>
-  <body class="bg-zinc-50 text-zinc-900 font-sans antialiased selection:bg-indigo-100 selection:text-indigo-900">
-    <main id="static-site-copy">
+  <body class="bg-zinc-50 text-zinc-900 font-sans antialiased selection:bg-indigo-100 selection:text-indigo-900 notranslate" translate="no">
+    <main id="static-site-copy" translate="yes" class="translate">
       <header>
         <p>aidCMO</p>
         <h1>OpenCMO free trial: enter one URL and open your own overseas growth console.</h1>
@@ -99,7 +108,7 @@ <h2>OpenCMO — AI CMO growth console</h2>
         </ul>
       </section>
     </main>
-    <div id="root"></div>
+    <div id="root" translate="no" class="notranslate"></div>
     <script type="module" src="/src/main.tsx"></script>
   </body>
 </html>
diff --git a/frontend/src/components/project/ProjectCommandCenter.tsx b/frontend/src/components/project/ProjectCommandCenter.tsx
@@ -132,7 +132,10 @@ function AgentCard({
 }
 
 function getLatestSerpTimestamp(latest: LatestScans) {
-  const timestamps = latest.serp
+  // Tolerate older/partial payloads where `serp` is missing — without this
+  // fallback the optional-chained `.map` chain blows up and surfaces as a
+  // blank project page.
+  const timestamps = (latest.serp ?? [])
     .map((snapshot) => snapshot.checked_at)
     .filter((value): value is string => Boolean(value))
     .map((value) => utcDate(value).getTime());
@@ -167,7 +170,7 @@ function hasAnyScanData(latest: LatestScans, latestMonitoring?: MonitoringSummar
     latest.seo ||
       latest.geo ||
       latest.community ||
-      latest.serp.length > 0 ||
+      (latest.serp?.length ?? 0) > 0 ||
       latestMonitoring,
   );
 }

diff --git a/tests/test_frontend_translate_optout.py b/tests/test_frontend_translate_optout.py
@@ -0,0 +1,74 @@
+"""Regression tests for issue #26: Chrome Translate blanks the SPA.
+
+Chrome/Google Translate replaces text nodes inside the React-managed DOM,
+which breaks reconciliation and surfaces as
+`NotFoundError: Failed to execute 'insertBefore' on 'Node'` followed by
+`#root` being emptied. We mitigate this by marking the body and `#root`
+with `translate="no"` / `notranslate`, plus a `<meta name="google"
+content="notranslate">` hint. The static SEO copy in
+`<main id="static-site-copy">` opts back in via `translate="yes"` so
+crawlers still see it.
+
+These tests guard those markers so a future edit to `index.html` cannot
+silently regress the fix.
+"""
+
+from __future__ import annotations
+
+import re
+from pathlib import Path
+
+_INDEX_HTML = Path("frontend/index.html")
+
+
+def _read_index_html() -> str:
+    text = _INDEX_HTML.read_text(encoding="utf-8")
+    assert text, "frontend/index.html should not be empty"
+    return text
+
+
+def _strip_html_comments(text: str) -> str:
+    """Drop HTML comments so the example markup inside our own opt-out comment
+    block doesn't accidentally satisfy / shadow the real tag regexes.
+    """
+    return re.sub(r"<!--.*?-->", "", text, flags=re.DOTALL)
+
+
+def test_index_html_has_google_notranslate_meta() -> None:
+    text = _read_index_html()
+    assert re.search(
+        r'<meta\s+name="google"\s+content="notranslate"\s*/?>',
+        text,
+    ), 'expected <meta name="google" content="notranslate"> opt-out hint'
+
+
+def test_body_opts_out_of_translation() -> None:
+    text = _strip_html_comments(_read_index_html())
+    body_match = re.search(r"<body(\s[^>]*)?>", text)
+    assert body_match, "frontend/index.html should declare a <body> tag"
+    attrs = body_match.group(1) or ""
+    assert 'translate="no"' in attrs, "<body> must carry translate=\"no\""
+    assert "notranslate" in attrs, "<body> class list must include 'notranslate'"
+
+
+def test_react_root_opts_out_of_translation() -> None:
+    text = _strip_html_comments(_read_index_html())
+    root_match = re.search(r'<div\s+id="root"(\s[^>]*)?>', text)
+    assert root_match, "frontend/index.html should declare <div id=\"root\"> for React"
+    attrs = root_match.group(1) or ""
+    assert 'translate="no"' in attrs, "#root must carry translate=\"no\""
+    assert "notranslate" in attrs, "#root must include the 'notranslate' class"
+
+
+def test_static_seo_copy_remains_translatable() -> None:
+    """SEO/no-JS fallback copy lives outside #root; it should stay translatable
+    so crawlers and translated previews can still read it.
+    """
+    text = _strip_html_comments(_read_index_html())
+    main_match = re.search(r'<main\s+id="static-site-copy"(\s[^>]*)?>', text)
+    assert main_match, "static SEO copy <main> should be present"
+    attrs = main_match.group(1) or ""
+    assert 'translate="yes"' in attrs, (
+        "static SEO copy must explicitly opt back into translation, "
+        "otherwise the body-level translate=\"no\" would suppress it too"
+    )