diff --git a/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-recipe-runtime-note.md b/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-recipe-runtime-note.md
index a9bdb45..8a123be 100644
--- a/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-recipe-runtime-note.md
+++ b/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-recipe-runtime-note.md
@@ -58,6 +58,19 @@ This supports the current Tier 2 bias:
   recipe rebuild, candidate runtime-record rebuild, and later availability-latch refresh
 - not in one isolated helper or one isolated named-availability bit
 
+One upstream pairwise check now supports that same read:
+
+- `compare-setup-payload-core Louisiana.gmp Dutchlantis.gmp`
+  already differs at:
+  - `payload_word_0x14`: `1870` vs `2025`
+  - `payload_byte_0x20`: `0x3a` vs `0xfd`
+  - `payload_word_0x3b2`: `2` vs `1`
+  - candidate-header words:
+    `0xcdcdcdcd / 0xcdcdcdcd` vs `0x00000000 / 0x00000000`
+
+So the current evidence no longer stops at the recipe-book lines themselves; there are already
+setup-payload-core differences upstream of the coupled Tier 2 rebuild strip.
+
 So the next recovery pass should target the internal sequencing and data handoff across:
 
 - `0x00435630`
diff --git a/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-setup-core-note.md b/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-setup-core-note.md
new file mode 100644
index 0000000..b3663e2
--- /dev/null
+++ b/artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-setup-core-note.md
@@ -0,0 +1,67 @@
+# Runtime Effect Kind-8 Tier2 Setup Core Note
+
+This note records the current setup-payload-core comparison across the six bundled add-building
+carrier maps:
+
+- `Alternate USA.gmp`
+- `Chicago to New York.gmp`
+- `Louisiana.gmp`
+- `Pacific Coastal.gmp`
+- `Rhodes Unfinished.gmp`
+- `Texas Tea.gmp`
+
+The comparison uses the checked `compare-setup-payload-core` surface.
+
+## Immediate result
+
+The upstream setup payload already differs across the carrier set.
+
+Observed differing fields:
+
+- `payload_word_0x14`
+- `payload_byte_0x20`
+- `payload_word_0x3b2`
+- `candidate_header_word_0_hex`
+- `candidate_header_word_1_hex`
+
+## Carrier-set shape
+
+Current `payload_word_0x3b2` values:
+
+- `Alternate USA.gmp = 1`
+- `Chicago to New York.gmp = 2`
+- `Louisiana.gmp = 2`
+- `Pacific Coastal.gmp = 1`
+- `Rhodes Unfinished.gmp = 1`
+- `Texas Tea.gmp = 3`
+
+That means `Louisiana.gmp` is **not** unique on `payload_word_0x3b2`.
+
+Current candidate-header word pairs:
+
+- `Alternate USA.gmp = 0x10000000 / 0x00009000`
+- `Chicago to New York.gmp = 0x00000000 / 0x00000000`
+- `Louisiana.gmp = 0xcdcdcdcd / 0xcdcdcdcd`
+- `Pacific Coastal.gmp = 0x00000000 / 0x00000000`
+- `Rhodes Unfinished.gmp = 0x00000000 / 0x00000000`
+- `Texas Tea.gmp = 0x00000000 / 0x00000000`
+
+So `Louisiana.gmp` is currently unique among the shipped carrier set on the candidate-header
+sentinel pair, while `Alternate USA.gmp` stays separately unique on the recognized
+`rt3-105-map-container-v1` header pair.
+
+## Current implication
+
+The upstream side of the Tier 2 strip is now bounded more tightly:
+
+- the current `Louisiana.gmp` recipe/runtime outlier status is not explained by `payload_word_0x3b2`
+  alone, because `Chicago to New York.gmp` also carries `2`
+- but `Louisiana.gmp` does stand out on the candidate-header sentinel pair
+
+So the next Tier 2 upstream question is narrower:
+
+- whether the unique `0xcdcdcdcd / 0xcdcdcdcd` candidate-header state in `Louisiana.gmp` is
+  relevant to the coupled
+  `0x00435630 / 0x00412d70 / 0x00412fb0 / 0x00412c10`
+  rebuild strip, or whether it is only another container-side artifact unrelated to the shipped
+  `Add Building Warehouse05` row.
diff --git a/docs/rehost-queue.md b/docs/rehost-queue.md
index d0a8cc4..9837f89 100644
--- a/docs/rehost-queue.md
+++ b/docs/rehost-queue.md
@@ -542,6 +542,21 @@ Working rule:
     `0x004120b0 -> 0x00412d70 -> 0x00412ab0 -> 0x00412c10`.
     That keeps the next pass focused on the internal sequencing and handoff across the coupled
     recipe/runtime/availability rebuild strip, not on one isolated helper.
+    The same note now carries the first upstream feed-in difference too:
+    `compare-setup-payload-core Louisiana.gmp Dutchlantis.gmp` already differs on
+    `payload_word_0x14` (`1870` vs `2025`), `payload_byte_0x20` (`0x3a` vs `0xfd`),
+    `payload_word_0x3b2` (`2` vs `1`), and the candidate-header words
+    (`0xcdcdcdcd` vs `0x00000000`). So the next pass can now work on both sides of the coupled
+    Tier 2 strip: the upstream setup payload core and the downstream recipe/runtime rebuild loop.
+  - the carrier-set setup-core comparison is checked in now too:
+    `artifacts/exports/rt3-1.06/runtime-effect-kind8-tier2-setup-core-note.md`
+    widens that upstream comparison across all six bundled add-building carriers. It shows
+    `Louisiana.gmp` is not unique on `payload_word_0x3b2` (`Chicago to New York.gmp` also has `2`)
+    but is unique among the shipped carrier set on the candidate-header sentinel pair
+    `0xcdcdcdcd / 0xcdcdcdcd`, while `Alternate USA.gmp` stays separately unique on the
+    recognized `rt3-105-map-container-v1` header pair. That narrows the next upstream Tier 2
+    question to whether the `Louisiana` candidate-header state actually matters to the coupled
+    rebuild strip or is just another container-side artifact.
     kinds”; it is the smaller set of scenario-specific records where that sweep explicitly writes
     `[event+0x7ef]` itself or a still-later owner does.
   - two explicit trigger-kind materializations are now grounded inside that retagger: