run 22

docs/discovery-notes.md

@@ -5066,3 +5066,83 @@ Best current model of the `E8` branch:
   branch.
 - `07 80 FA 50 26 51` and `07 C0 7A 50 A6 11` remain side-family observations,
   but neither has overtaken `7A` as the strongest lead.
+
+### HE12: Nearby Host-Shaped `E8` Neighbors
+
+Test whether the `7A` branch is specific to `E8`, or whether nearby host-shaped
+commands in the same region also provoke structured responses.
+
+Checksums:
+
+- `00 00 E7 40 30 CD`
+- `00 00 E9 40 30 C3`
+- `00 00 EA 40 30 C0`
+
+#### HE12a: Host-Shaped `E7`
+
+```powershell
+python scripts/serial_sequence_probe.py --port COM5 --prompt --frame "00 00 A0 00 80 7A" --frame "00 00 E7 40 30 CD" --frame "00 00 00 00 80 DA" --frame "00 00 00 00 80 DA" --repeat 2 --frame-interval 0.20 --read-after-frame 0.30 --read-after-group 0.8 --log captures/rcp-heartbeat-neighbor-e7.txt
+```
+
+#### HE12b: Host-Shaped `E9`
+
+```powershell
+python scripts/serial_sequence_probe.py --port COM5 --prompt --frame "00 00 A0 00 80 7A" --frame "00 00 E9 40 30 C3" --frame "00 00 00 00 80 DA" --frame "00 00 00 00 80 DA" --repeat 2 --frame-interval 0.20 --read-after-frame 0.30 --read-after-group 0.8 --log captures/rcp-heartbeat-neighbor-e9.txt
+```
+
+#### HE12c: Host-Shaped `EA`
+
+```powershell
+python scripts/serial_sequence_probe.py --port COM5 --prompt --frame "00 00 A0 00 80 7A" --frame "00 00 EA 40 30 C0" --frame "00 00 00 00 80 DA" --frame "00 00 00 00 80 DA" --repeat 2 --frame-interval 0.20 --read-after-frame 0.30 --read-after-group 0.8 --log captures/rcp-heartbeat-neighbor-ea.txt
+```
+
+What would count as a hit:
+
+- any non-heartbeat response after `E7`, `E9`, or `EA`
+- recurrence of `07 80 7A 50 26 D1`
+- recurrence of `07 80 FA 50 26 51` or `07 C0 7A 50 A6 11`
+- any new stable family that suggests `E8` belongs to a broader selector block
+
+### 2026-05-13 `E8` Neighbor Result
+
+Captures:
+
+- `captures/rcp-heartbeat-neighbor-e7.txt`
+- `captures/rcp-heartbeat-neighbor-e9.txt`
+- `captures/rcp-heartbeat-neighbor-ea.txt`
+
+Result summary:
+
+- `E7`, `E9`, and `EA` do not behave the same way, so this is not a simple
+  "everything around `E8` lands in the same branch" region.
+- All three effects only appeared in group 1. Group 2 was heartbeat-only in
+  every run.
+
+Observed outputs:
+
+| Host step | Structured result |
+| --- | --- |
+| `00 00 E7 40 30 CD` | later heartbeat provoked `07 80 40 40 30 ED` |
+| `00 00 E9 40 30 C3` | produced `07 80 7A 28 D3 5C` |
+| `00 00 EA 40 30 C0` | later heartbeat provoked `07 80 C0 40 30 6D` |
+
+Interpretation:
+
+- `E8` still appears special, because it is the one that most cleanly enters the
+  reproducible `07 80 7A 50 26 D1` branch.
+- `E9` is the first nearby neighbor to produce a clearly related but distinct
+  `7A`-family frame:
+  - `07 80 7A 28 D3 5C`
+- `E7` and `EA` look more like they bias the panel toward known heartbeat-family
+  transients (`07 80 40 40 30 ED` and `07 80 C0 40 30 6D`) rather than the
+  `7A` / `FA` branch.
+- So the emerging picture is not a broad contiguous "E* equals same page"
+  block, but a mixed neighborhood where nearby host-shaped values steer the
+  panel into different transient/readable families.
+
+Best current follow-up:
+
+- treat `E9` as the strongest neighbor lead
+- test exact and host-shaped handling of `07 80 7A 28 D3 5C`
+- optionally compare `E8` vs `E9` timing to see whether the differing `7A`
+  payloads reflect selector choice or timing/context