diff --git a/test/project-connection-ownership-claim-lr-768c9e.test.js b/test/project-connection-ownership-claim-lr-768c9e.test.js
index c0587610..907ff58e 100644
--- a/test/project-connection-ownership-claim-lr-768c9e.test.js
+++ b/test/project-connection-ownership-claim-lr-768c9e.test.js
@@ -143,14 +143,13 @@ function makeConnectionCtx(overrides) {
   }, overrides);
 }
 
-// lr-ae85d5: this case still flakes intermittently under full-suite load
-// (~1/19) after the loadUsers ENOENT-vs-corrupt fix landed in this same PR.
-// Root-caused as a SECOND, distinct, pre-existing race unrelated to the
-// module-cache/env-var races documented above -- tracked separately as
-// lr-a7b03e and explicitly out of scope for this CI-wiring PR. Quarantined
-// here rather than chased further; do not remove this skip without lr-a7b03e
-// being resolved first.
-test("lr-768c9e: admin connecting to an evicted, unowned session does not truncate on-disk history", { skip: "flaky under full-suite load -- second distinct race tracked as lr-a7b03e, out of scope for lr-ae85d5" }, function () {
+// lr-a7b03e: re-enabled. The ~1/19 flake this test was quarantined for was
+// measured on a broken harness (node --test --test-force-exit could silently
+// drop whole ESM test files while exiting 0 -- see lr-795882). With that
+// harness defect fixed, this test's flake rate was re-measured empirically
+// under repeated full-suite runs before un-skipping (see PR body for the
+// stress-run evidence); it did not reproduce.
+test("lr-768c9e: admin connecting to an evicted, unowned session does not truncate on-disk history", function () {
   var tmpHome = makeTempHome();
   try {
     var adminId = "admin-user-1";