diff --git a/context/open-questions.md b/context/open-questions.md
index f0afa29..546fded 100644
--- a/context/open-questions.md
+++ b/context/open-questions.md
@@ -1174,3 +1174,19 @@ Gitea #116 and `experience-levels.md`.
 - **des_night-family pages have subtle fog deltas** (both sets near-black); arena/polar night
   pages carry the dramatic 5↔40 near-plane swing. If testers report "searchlight does nothing"
   on some maps, it's page authoring, not the mechanism.
+
+
+## Replicant-side TakeDamage delivery -- mechanism uncharacterized (2026-08-11, from the #162 audit) [T4]
+
+Replicant mech copies demonstrably EXECUTE `Mech::TakeDamageMessageHandler` (matchlog `DMG
+inst=R` rows; the #150 bench observed replicant-side type-2 arms) -- yet every damage producer
+uses `Entity::Dispatch`, and ENTITY.cpp:235-272 reroutes a replicant target's dispatch to the
+wire WITHOUT local execution.  Something delivers damage to the local replicant copy anyway
+(wire-seam double-delivery? interest/watcher path? a port-side mirror?), and the 2026-07-25
+corpus saw 0 of 18,818 DMG rows inst=R -- so the behavior APPEARED between then and 2026-08-09.
+The #162 fix GATES every scoring/death consequence of it (reports + death tail are master-only),
+so the scoreboard is safe regardless -- but the delivery mechanism itself is unmapped, and
+replicant-side execution may have OTHER un-audited side effects (zone/subsystem state divergence
+on peer copies beyond the intended visual application).  To close: instrument the entry with the
+delivery path (who posted it, from which socket/queue), find the first build where inst=R rows
+appear, and decide whether the delivery is intended (visual-damage design) or a seam bug.