diff --git a/plugin-designs/my-upg-100-list/poster-wall.vue b/plugin-designs/my-upg-100-list/poster-wall.vue
index 2879c81e7..92e20aa25 100644
--- a/plugin-designs/my-upg-100-list/poster-wall.vue
+++ b/plugin-designs/my-upg-100-list/poster-wall.vue
@@ -192,7 +192,46 @@ async function fetchListRecords(list) {
 // Product call taken by dx-shepherd (UPGHK, 2026-09-27) under the Driver's standing goal that the
 // production maps be localized.
 const t = useOptionalT('')
-const intro   = computed(() => t('messages.upg100WallIntro'))
+
+// ★ THE MARKUP IS COMPOSED HERE, so the CATALOGUE holds plain text (dx-shepherd UPGHTML, 2026-09-27).
+// vue-i18n warns "[intlify] Detected HTML in '…'" twice per tab pass, and it is right to: a translated
+// string carrying <strong> is the only one in this tree that does — I measured every en catalogue area
+// and upg100WallIntro is 1 of 1, so this was not a convention, it was my outlier.
+//
+// ⚠ WHY THE OBVIOUS FIXES DO NOT APPLY HERE. The shepherd's suggestion — markup in the template, or
+// <i18n-t> interpolation — assumes this profile renders the sentence. It does not: `intro` is a STRING
+// PROP on the library PosterWall, rendered there with v-html (PosterWall.vue:955), and that component
+// has no slots at all. So the markup has to arrive inside the string, and the only question is who
+// builds the string. Building it HERE keeps the catalogue clean without a library change.
+//
+// ★ AND IT IS STRICTLY SAFER THAN WHAT IT REPLACES, which is the part worth keeping. v-html on a whole
+// catalogue string TRUSTS THE CATALOGUE: any tag a translator (or a bad merge) puts in any of the 11
+// locale files renders as markup. Composing here escapes each part and emits exactly one tag I control,
+// so the trusted surface shrinks from "every translation of this key" to "this line".
+//
+// THREE PARTS, NOT A WRAPPER AROUND ONE, because the emphasis is not in the same place in every
+// language. Measured across all 11 locales: each has exactly one <strong> and no other tags, and de, hi
+// and zh carry real text AFTER the emphasised run (" beten.", " प्रार्थना करते हुए।", "為他們禱告。")
+// where the others end on a full stop. A head+emphasis wrapper would have silently dropped those tails.
+const escHtml = (s) => String(s)
+  .replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;')
+const intro = computed(() => {
+  // The three plain-text keys studio adds (rule 5B — the catalogue is theirs). Until they land, the
+  // single legacy key is still the source of truth, so this renders identically before AND after: no
+  // flag day, and my landing does not have to be ordered against theirs.
+  const a = t('messages.upg100WallIntroA')
+  const b = t('messages.upg100WallIntroEmph')
+  const c = t('messages.upg100WallIntroB')
+  const missing = (v, key) => !v || v === key
+  if (missing(a, 'messages.upg100WallIntroA') || missing(b, 'messages.upg100WallIntroEmph')) {
+    // ⚠ THE LEGACY KEY STILL CONTAINS THE MARKUP, so it is passed through unescaped exactly as before.
+    // Escaping it here would print a literal "<strong>" to the reader — a visible regression traded for
+    // a console warning, which is the wrong way round.
+    return t('messages.upg100WallIntro')
+  }
+  const tail = missing(c, 'messages.upg100WallIntroB') ? '' : escHtml(c)
+  return `${escHtml(a)}<strong>${escHtml(b)}</strong>${tail}`
+})
 const heading = computed(() => t('messages.upg100WallHeading'))
 const wallTag = computed(() => t('messages.upg100WallTag'))