diff --git a/.changeset/contains-fold-prose-6682.md b/.changeset/contains-fold-prose-6682.md
new file mode 100644
index 0000000000..e6cc078b88
--- /dev/null
+++ b/.changeset/contains-fold-prose-6682.md
@@ -0,0 +1,29 @@
+---
+"@objectstack/spec": patch
+---
+
+docs(spec): close the stale `$contains` fold pointers in the `filter.zod.ts` per-backend history table and the `filter-text-conformance.ts` `@see` line
+
+#7723 took the `i` flag off `driver-memory`'s `filterSubstringPattern`, closing
+the #6682 tracking issue. Two docblock spots in `packages/spec` still pointed
+forward at that landing as if it hadn't happened yet:
+
+- `filter.zod.ts`'s per-backend history table (explicitly headed "measured …
+  BEFORE the ruling" — kept as history per #7625) had a `driver-memory` cell
+  reading "the last one standing, until #6682". The cell is historically
+  accurate (it WAS the last case-folding face when written) but the "until
+  #6682" tail described a future event that has since landed; it now reads
+  "the last one standing; #6682 closed it (see 'Implementation status'
+  below)". The table's other rows, including `driver-mongodb`'s, are left
+  untouched — they are correct as pre-ruling history and #7625 already judged
+  them that way.
+- `filter-text-conformance.ts`'s `@see #6682` line read "mongodb landed,
+  memory open", contradicting the same file's own header 37 lines earlier
+  (updated correctly by #7723), which already says the memory half landed
+  too. It now reads "mongodb and memory both landed".
+
+No behavior, schema, or generated baseline changes — these are TSDoc comments
+attached to exported schemas, which do reach `@objectstack/spec`'s published
+`dist/*.d.ts` (verified: `grep` on the rebuilt dist shows the edited sentence
+in `dist/filter.zod-*.d.ts`), so this ships as a patch changeset rather than
+`skip-changeset`.
diff --git a/packages/spec/src/data/filter-text-conformance.ts b/packages/spec/src/data/filter-text-conformance.ts
index 62956cfbad..3657837ca9 100644
--- a/packages/spec/src/data/filter-text-conformance.ts
+++ b/packages/spec/src/data/filter-text-conformance.ts
@@ -92,7 +92,7 @@
  * @see https://github.com/objectstack-ai/objectstack/issues/5701 (this table)
  * @see https://github.com/objectstack-ai/objectstack/issues/5702 (the SQL family — landed)
  * @see https://github.com/objectstack-ai/objectstack/issues/6520 ($icontains on the JS faces — landed)
- * @see https://github.com/objectstack-ai/objectstack/issues/6682 (the $contains family — mongodb landed, memory open)
+ * @see https://github.com/objectstack-ai/objectstack/issues/6682 (the $contains family — mongodb and memory both landed)
  */
 
 import type { FilterCondition } from './filter.zod';
diff --git a/packages/spec/src/data/filter.zod.ts b/packages/spec/src/data/filter.zod.ts
index 17eee7061e..9ed7092e88 100644
--- a/packages/spec/src/data/filter.zod.ts
+++ b/packages/spec/src/data/filter.zod.ts
@@ -518,7 +518,7 @@ export const RangeOperatorSchema = lazySchema(() => z.object({
  * | surface | `$contains` case behaviour | mechanism |
  * |---|---|---|
  * | `formula` `matchesFilterCondition` | SENSITIVE | `actual.includes(v)` |
- * | `driver-memory` — query path and analytics face | INSENSITIVE, full Unicode | `new RegExp(escapeRegex(v), 'i')` — the last one standing, until #6682 |
+ * | `driver-memory` — query path and analytics face | INSENSITIVE, full Unicode | `new RegExp(escapeRegex(v), 'i')` — the last one standing; #6682 closed it (see "Implementation status" below) |
  * | `driver-memory` — reference matcher (`memory-matcher`) | SENSITIVE | `value.includes(target)` |
  * | `driver-mongodb` | INSENSITIVE, full Unicode | hardcoded `$options: 'i'` |
  * | `driver-sql` family | the DIALECT's | `LIKE '%v%'` — ASCII-insensitive on SQLite (so also turso and sqlite-wasm), sensitive on Postgres, collation-dependent on MySQL |