diff --git a/CLAUDE.md b/CLAUDE.md
index 7835246f..6b958af4 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -1253,6 +1253,22 @@ consistency check explicitly printed *"No duplicate device assignments"*. The do
 summary. **Instrument, don't infer.**
 
 ## Next steps (proposed)
+- [ ] (Parked, diagnostics-only, cannot affect a real pod) Two gaps found while testing `-tident`
+  on a **single-monitor VM** (2026-08-05). Both only occur when one physical monitor is exposed as
+  two DirectDraw devices, which happens because the NULL-device merge is gated on
+  `NumMonitors>=2` (`VideoCard.cpp`) — by design, it is a multi-monitor fixup, so a single-monitor
+  box keeps the alias in slot 0.
+  1. **`-tident` overpaints itself.** With device 0 = NULL alias and device 1 = the real output,
+     both resolve to the same screen, so it paints red/"1" then immediately green/"2" over the top.
+     Looks like a glitch; is actually correct. Should detect devices sharing an `HMONITOR` and
+     label them together (`1 & 2 - same monitor`) instead of painting sequentially.
+  2. **The clash check compares device INDICES, not resolved monitors.** On that VM it reported
+     *"No duplicate device assignments"* while `main`→device 0 (primary alias) and `radar`→device 1
+     (`\\.\DISPLAY49`) were the **same physical monitor** — i.e. exactly the collision the check
+     exists to predict, reported as all-clear. Fix: treat a device with no `HMONITOR` as the
+     Windows primary, then compare resolved `HMONITOR`s rather than indices.
+  Neither can occur on a multi-monitor pod. Worth doing only if bench/VM testing starts relying on
+  the check being trustworthy.
 - [ ] (Researched, NOT started) Make `-window` work for **all** modes — full findings in
   **`WINDOWED-MODE.md`** (repo root, 2026-08-05). Headline: `-window` does not fail for MFD/
   cameraship modes, it **silently disables** them, because `use_shgui` is derived from