diff --git a/docs/architecture/tmdb-metadata-enrichment.md b/docs/architecture/tmdb-metadata-enrichment.md
index 99abf701a..94f64cea8 100644
--- a/docs/architecture/tmdb-metadata-enrichment.md
+++ b/docs/architecture/tmdb-metadata-enrichment.md
@@ -537,8 +537,9 @@ identity is the query with only its case removed (`searchQueryIdentity`):
 variants are deduplicated by it and every attempted variant is cached under
 its own key (`title:<identity>|year:<y>|v4`, in the language that variant
 was searched in), never by the folded key and never only under the first
-variant — "Леика" and "Лейка" fold to one key but are different searches with
-different answers, so a verdict for one must not be read back for the other;
+variant — "Леика" and "Лейка", an illustrative pair, fold to one key while
+staying two different searches that can get different answers, so a verdict
+for one must not be read back for the other;
 a misspelled original title must not swallow the display title that TMDB
 actually knows; and two items that share an original title but not a display
 title walk different variant lists, so a row keyed on the first variant alone
@@ -547,11 +548,13 @@ splits Cyrillic "й" into "и" + a combining breve and "ё" into "е" + a
 diaeresis, and Arabic hamza forms ("أ") into a bare alef + a combining hamza
 that the punctuation step then turns into a space inside the word. The key
 drops or splits on those marks, and TMDB's `/search` does not fold them the
-same way — a query of `леика` returns zero results while `Лейка` returns the
-show. Under the old single-form design every Russian title with "й"/"ё"
-("Лейка (10 серий)", "Тестовый Сериал", "Пробный Выпуск") was searched
-folded, missed, and cached as missing for the 7-day negative TTL. Compare
-results only through `normalized`; never send it over the wire.
+same way, so a folded query matches nothing there. Under the old single-form
+design that hit every Russian title carrying "й" or "ё" and every Arabic
+title carrying a hamza form: each was searched folded, missed, and cached as
+missing for the 7-day negative TTL. The Cyrillic and Arabic strings used
+throughout this section are illustrative stand-ins chosen to fold the same
+way, not the titles the failures were observed on. Compare results only
+through `normalized`; never send it over the wire.
 
 Electron IPC path (follows the standard DB worker contract, see
 [SQLite DB Worker](./sqlite-db-worker.md)):
diff --git a/libs/services/src/lib/tmdb/tmdb-id-resolver.service.spec.ts b/libs/services/src/lib/tmdb/tmdb-id-resolver.service.spec.ts
index 3669196ad..fec3034e9 100644
--- a/libs/services/src/lib/tmdb/tmdb-id-resolver.service.spec.ts
+++ b/libs/services/src/lib/tmdb/tmdb-id-resolver.service.spec.ts
@@ -9,7 +9,8 @@ import { TmdbSearchResult } from './tmdb.types';
  * Regression coverage for the search wire format. The comparison key folds
  * diacritics — Cyrillic "й" becomes "и" — and that key used to be sent as
  * the TMDB query, which matched nothing for any title carrying "й"/"ё" and
- * cached the miss for a week ("Лейка (10 серий)", "Тестовый Сериал").
+ * cached the miss for a week. The titles below are illustrative stand-ins
+ * that fold the same way, not the ones the failures were observed on.
  */
 describe('TmdbIdResolverService.resolveBySearch', () => {
     const series2026: TmdbSearchResult = {
diff --git a/libs/services/src/lib/tmdb/tmdb-matcher.spec.ts b/libs/services/src/lib/tmdb/tmdb-matcher.spec.ts
index 49159bf70..7a1ac086f 100644
--- a/libs/services/src/lib/tmdb/tmdb-matcher.spec.ts
+++ b/libs/services/src/lib/tmdb/tmdb-matcher.spec.ts
@@ -160,9 +160,9 @@ describe('buildSearchTitleVariants', () => {
     });
 
     it('keeps spellings that fold to one key but differ on the wire', () => {
-        // A misspelled original title must not swallow the display title:
-        // TMDB knows "Лейка" and not "Леика", and only the second variant
-        // would find it.
+        // A misspelled original title must not swallow the display title.
+        // Take a pair where only the display spelling is the one TMDB
+        // indexes: only the second variant can find it.
         expect(buildSearchTitleVariants('Лейка', 'Леика')).toEqual([
             { query: 'Леика', normalized: 'леика' },
             { query: 'Лейка', normalized: 'леика' },
diff --git a/libs/services/src/lib/tmdb/tmdb-matcher.ts b/libs/services/src/lib/tmdb/tmdb-matcher.ts
index 57ef57088..e7c862292 100644
--- a/libs/services/src/lib/tmdb/tmdb-matcher.ts
+++ b/libs/services/src/lib/tmdb/tmdb-matcher.ts
@@ -53,7 +53,8 @@ export interface SearchTitleVariant {
 /**
  * The identity of one search on the wire: the query with only the case
  * removed, since TMDB matches case-insensitively and nothing else about
- * the spelling may be folded away — "Леика" and "Лейка" are different
+ * the spelling may be folded away — "Леика" and "Лейка" (illustrative) are
+ * different
  * searches with different answers, however alike their comparison keys.
  * Both the variant deduplication and the cache row use this, so a cached
  * verdict can never be read back for a search that was never sent.
@@ -116,11 +117,11 @@ export function buildSearchLookupKey(
     // v2: normalizeTitleKeys learned to strip appended language/quality
     // tags; the version suffix invalidates cached (incl. negative) match
     // resolutions keyed on the old polluted titles.
-    // v3: the search query stopped being the folded key ("леика" for
-    // "Лейка"), which TMDB answered with nothing; every negative row recorded
-    // under v2 for a title with "й"/"ё" is that bug, not a missing title, and
-    // must not block the retry for its 7-day TTL. Rows are keyed by the
-    // query since then.
+    // v3: the search query stopped being the folded key — the fold rewrites
+    // "й" as "и" ("леика" for "Лейка", to illustrate) and TMDB answers that
+    // spelling with nothing; every negative row recorded under v2 for a title
+    // with "й"/"ё" is that bug, not a missing title, and must not block the
+    // retry for its 7-day TTL. Rows are keyed by the query since then.
     // v4: year evidence is tiered (see `yearEvidenceTier`), so every v3 row
     // resolved by popularity across tiers may name the wrong show — and a
     // positive row stays fresh for 30 days.
diff --git a/libs/shared/database/src/lib/connection.ts b/libs/shared/database/src/lib/connection.ts
index f19806d3d..efc7a43f7 100644
--- a/libs/shared/database/src/lib/connection.ts
+++ b/libs/shared/database/src/lib/connection.ts
@@ -976,9 +976,9 @@ function widenTmdbMetadataMediaTypeCheck(sqliteDb: Database.Database): void {
  * - unversioned → v2: title normalization learned to strip appended
  *   language/quality tags.
  * - v2 → v3: the search query stopped being the folded comparison key. Under
- *   v2 every title with a Cyrillic "й"/"ё" was searched folded ("леика" for
- *   "Лейка", "ежик" for "Ёжик"), got no answer, and was cached as missing for
- *   7 days.
+ *   v2 every title with a Cyrillic "й"/"ё" was searched folded — the fold
+ *   spells them "и" and "е", as in the illustrative "леика" for "Лейка" —
+ *   got no answer, and was cached as missing for 7 days.
  * - v3 → v4: year evidence became tiered. Under v3 a series admitted only by
  *   the "premiered earlier" tolerance competed with an exact-year match on
  *   popularity alone, so a new series resolved to its older, better-known
diff --git a/libs/shared/interfaces/src/lib/title-normalization.util.spec.ts b/libs/shared/interfaces/src/lib/title-normalization.util.spec.ts
index 08e9c9a47..40a96cab0 100644
--- a/libs/shared/interfaces/src/lib/title-normalization.util.spec.ts
+++ b/libs/shared/interfaces/src/lib/title-normalization.util.spec.ts
@@ -434,8 +434,9 @@ describe('titleYearsCompatible', () => {
 describe('cleanTitleForSearch', () => {
     it('keeps Cyrillic letters that folding would rewrite', () => {
         // NFD splits "й" into "и" + a breve and "ё" into "е" + a diaeresis;
-        // the comparison key drops both marks, and TMDB answers the folded
-        // spelling with nothing (issue: "Лейка (10 серий)" never matched).
+        // the comparison key drops both marks, and TMDB answers that folded
+        // spelling with nothing. The titles here are illustrative stand-ins
+        // chosen to fold the same way.
         expect(normalizeTitle('Лейка (10 серий)')).toBe('леика');
         expect(cleanTitleForSearch('Лейка (10 серий)')).toBe('Лейка');
         expect(cleanTitleForSearch('Ёжик 2010')).toBe('Ёжик');
@@ -446,8 +447,8 @@ describe('cleanTitleForSearch', () => {
 
     it('keeps Arabic hamza forms that folding splits into two words', () => {
         // "أ" decomposes into a bare alef + U+0654, which is outside the
-        // stripped mark range and so becomes a SPACE in the key. TMDB finds
-        // "أمثلة تجريبية" and nothing for "ا مثلة تجريبية".
+        // stripped mark range and so becomes a SPACE in the key — splitting
+        // the word in two, which is not a spelling TMDB indexes.
         expect(normalizeTitle('AR| أمثلة تجريبية')).toBe('ا مثلة تجريبية');
         expect(cleanTitleForSearch('AR| أمثلة تجريبية')).toBe('أمثلة تجريبية');
     });
diff --git a/libs/shared/interfaces/src/lib/title-normalization.util.ts b/libs/shared/interfaces/src/lib/title-normalization.util.ts
index 19ca113b5..0c3759376 100644
--- a/libs/shared/interfaces/src/lib/title-normalization.util.ts
+++ b/libs/shared/interfaces/src/lib/title-normalization.util.ts
@@ -520,7 +520,8 @@ export function normalizeTitleKeys(
  *
  * Comparison keys must fold so two spellings of one film meet; a search
  * query must not, because the search engine folds by its own rules and a
- * pre-folded Cyrillic query ("леика" for "Лейка") matches nothing there.
+ * pre-folded Cyrillic query matches nothing there ("леика" for "Лейка"
+ * illustrates the rewrite).
  * Compare the results with `normalizeTitle`, never with this.
  */
 export function cleanTitleForSearch(raw: string | null | undefined): string {