fix(opstack): two cell-sweep pads assumed MinRadius <= MaxRadius; a third already didn't

Second defect from the full 28-op EffectOverBox audit. Three ops roll a radius
as Lerp(MinRadius, MaxRadius, hash01), which lands in [min(A,B), max(A,B)] --
Lerp does not require A <= B. Their EffectOverBox sweeps lattice cells padded by
the largest radius a cell could hold, so a pad below the true maximum means the
cells are never examined and the op reports Identity for a box its own Eval will
fill or carve: no geometry, no collision.

  FGridColumnMod   ~1086  Max(MaxRadius, 0)                        exposed
  FShaftFieldSource ~1436 Max(ShaftMaxRadius, ConnectorRadius)     exposed
  FIslandBlobSource ~1803 Max(IslandMinRadius, IslandMaxRadius)    already correct

The third one is the argument: the concern was met and guarded once in this same
file, and the other two shipped without it. Fixed with FMath::Max3, a spelling
already used here (~2481) and in VoxelCaveMorphology.cpp.

Shipped defaults are ordered correctly (2/5, 2/7, 5/11), so this is a NO-OP at
defaults -- that is its acceptance signal, and any moved number means the diff
did more than intended. It needs a mis-ordered asset value, which nothing
prevents: ClampMin is a per-property floor and UE cannot express "<= that other
property". ColumnMinRadius is also settable per-room via UVoxelTerrainOpDefinition.

Deliberately NOT fixed by normalising the params: swapping the Lerp endpoints
maps the same hash to a different radius, changing geometry and breaking the
eight equivalence tests. Only the bound becomes conservative; Eval is byte-
identical -- verified, no Eval/Roll/GetCells line in the diff.

Also audited and found sound (recorded so they are not re-chased): FPassageCarveOp
and the passage bounding sphere, FOriginSpineOp, FBoundarySealOp, FShaftLedgeMod,
FCaveArchMod, FDomeMod.

Not built -- Jahni builds.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

CODEX-TASK-004-minmax-radius-envelope.md

@@ -0,0 +1,101 @@
+# Codex task 004 — two box verdicts assume `MinRadius ≤ MaxRadius`; a third one already doesn't
+
+**Owner:** Codex (Model Luna, xHigh) · **Orchestrator:** Claude · **Branch:** `experimental`
+**Status:** specified, not started
+**Kind:** ⚠️ **correctness of a box verdict.** Same class as task 003. Small fix, closes a class.
+
+---
+
+## Why this exists
+
+Three operators roll a primitive radius from a hash between two designer-set params:
+
+```cpp
+Out.R   = FMath::Lerp(MinRadius,          MaxRadius,          hash01);   // FGridColumnMod   ~1129
+Out.R   = FMath::Lerp(P.ShaftMinRadius,   P.ShaftMaxRadius,   hash01);   // FShaftFieldSource ~1560
+Out.Rxy = FMath::Lerp(P.IslandMinRadius,  P.IslandMaxRadius,  hash01);   // FIslandBlobSource ~1850
+```
+
+`FMath::Lerp(A, B, t)` with `t ∈ [0,1]` lands anywhere in `[min(A,B), max(A,B)]` — it does **not**
+require `A ≤ B`.
+
+Each op's `EffectOverBox` then sweeps a **range of lattice cells** around the query box, padded by
+the largest radius a cell could hold, and tests each rolled primitive exactly. The pad decides which
+cells are *looked at at all*, so a pad smaller than the true maximum radius means **cells are never
+examined**, their primitives are never tested, and the op reports `Identity` for a box that its own
+`Eval` will carve or fill.
+
+| op | pad used for the cell sweep | correct? |
+|---|---|---|
+| `FGridColumnMod` ~1086 | `FMath::Max(MaxRadius, 0.0f) + ColBlend` | ⛔ **exposed** |
+| `FShaftFieldSource` ~1436 | `FMath::Max(P.ShaftMaxRadius, P.ConnectorRadius) + ExtraReach` | ⛔ **exposed** |
+| `FIslandBlobSource` ~1803 | `const float MaxR = FMath::Max(P.IslandMinRadius, P.IslandMaxRadius);` | ✅ **already correct** |
+
+**The third one is the point.** Someone hit this exact concern while writing the island source and
+guarded it. The other two shipped without the guard. This task makes the three consistent.
+
+## How exploitable, stated honestly
+
+The shipped defaults are correctly ordered (`2/5`, `2/7`, `5/11`), so **nothing is broken out of the
+box.** It needs a mis-ordered asset value — `ColumnMinRadius = 8, ColumnMaxRadius = 4`.
+
+Nothing prevents that. The `UPROPERTY` metas carry `ClampMin = "1.0"`, which is a per-property
+floor; Unreal has no declarative way to say "must be ≤ that other property". And `ColumnMinRadius`
+is *also* settable per-room through `UVoxelTerrainOpDefinition`, so it is not only the strate asset.
+
+What makes it worth the five lines: when it does happen, the failure is **invisible and maddening**.
+`Eval` still draws the fat column perfectly, so every tile that gets meshed looks correct; only the
+tiles the classifier *skipped* are missing — no geometry, no collision, in a world that otherwise
+looks right.
+
+## The fix — five lines, and one thing you must NOT do
+
+Make each pad use the true envelope:
+
+```cpp
+// FGridColumnMod ~1086
+const float Reach = FMath::Max3(MinRadius, MaxRadius, 0.0f) + ColBlend;
+
+// FShaftFieldSource ~1436
+const float Pad = FMath::Max3(P.ShaftMinRadius, P.ShaftMaxRadius, P.ConnectorRadius) + ExtraReach;
+```
+
+(Use whatever spelling is idiomatic here — check that `FMath::Max3` is already used in this codebase
+before reaching for it; nested `FMath::Max` is fine and matches `FIslandBlobSource`'s existing line.)
+
+### ⛔ DO NOT "fix it properly" by normalising the params
+
+The tempting larger fix — swap `Min`/`Max` at resolution time so `Min ≤ Max` always — is **wrong and
+will break the build's tests.** `Eval` computes `Lerp(Min, Max, t)`; swapping the endpoints maps the
+same hash `t` to a *different* radius for the same cell. That changes generated geometry and breaks
+the eight bit-for-bit equivalence tests against the `switch` path.
+
+**Only the BOUND may become conservative. `Eval` stays byte-identical.** This is the same rule as
+task 003 and the same reason.
+
+## ⚠️ Invariants
+
+1. **No `Eval`, `RollColumn`, `RollShaft`, `GetCells`, or `GetCellsAt` body may change.** If your
+   diff touches one, you have the wrong site — stop and say so.
+2. **Do not touch `FIslandBlobSource`.** It is already correct and is the reference for this fix.
+3. The change direction is strictly conservative: a wider sweep examines *more* cells, so a verdict
+   can only move from `Identity` toward `CarveOnly`/`FillOnly`, never the reverse. Do not add any
+   compensating tightening.
+4. Comments are French + English; match the surrounding file. Say **why** the envelope is
+   `max(Min, Max)` and not `Max` — the next reader must not "simplify" it back.
+
+## Acceptance
+
+- `git diff --stat` shows exactly one file: `Source/VoxelForge/Private/VoxelDensityOpStack.cpp`,
+  and a handful of lines.
+- The three ops now agree on the pattern.
+- After the build: **every box-verdict line and every equivalence test is unchanged**, because the
+  shipped defaults are correctly ordered and the envelope only differs when they are not. **A change
+  in any of those numbers means the diff did something it should not have.** That is this task's
+  whole acceptance signal — a *no-op at defaults* is the expected, correct result.
+
+## Notes for the reviewer (Claude)
+
+- Confirm both pads changed and `FIslandBlobSource` did not.
+- Confirm no `Lerp` argument order was touched anywhere — that is the failure mode that would look
+  like a tidy-up and silently change the world.

OPSTACK-PROGRESS.md

@@ -3628,3 +3628,72 @@ The first attempt stopped without editing: my spec wrote `VF_PerlinAbsBound` in
 exactly the behaviour the spec template asks for ("if the code contradicts the spec, stop and say so
 rather than guessing"), and it is worth recording that it works — the cost was one clarification
 instead of a plausible-looking wrong rename.
+
+## 2026-08-16 (f) — the box-verdict audit, finished: 28 ops swept, a second class of bug found
+
+Having found the `ExtraReach` bug by auditing one archetype, I swept **all 28 `EffectOverBox`
+implementations** rather than stopping at the one that paid. Result: one more real defect, three
+false alarms worth recording so nobody re-chases them, and a clean bill for the rest.
+
+### ⛔ FOUND — `Min > Max` radius envelope (`CODEX-TASK-004`)
+
+Three ops roll a primitive radius as `FMath::Lerp(MinRadius, MaxRadius, hash01)`. `Lerp` with
+`t ∈ [0,1]` lands in `[min(A,B), max(A,B)]` — it does **not** require `A ≤ B`. Each op's
+`EffectOverBox` then sweeps a range of lattice cells padded by the largest radius a cell could hold;
+a pad below the true maximum means cells are **never examined**, so the op reports `Identity` for a
+box its own `Eval` will fill or carve.
+
+| op | pad | |
+|---|---|---|
+| `FGridColumnMod` ~1086 | `Max(MaxRadius, 0)` | ⛔ exposed |
+| `FShaftFieldSource` ~1436 | `Max(ShaftMaxRadius, ConnectorRadius)` | ⛔ exposed |
+| `FIslandBlobSource` ~1803 | `FMath::Max(IslandMinRadius, IslandMaxRadius)` | ✅ **already correct** |
+
+**The third is the whole argument.** Someone hit this exact concern writing the island source and
+guarded it; the other two shipped without. Fixed to `FMath::Max3(...)` — a spelling already used in
+this file (~2481) and in `VoxelCaveMorphology.cpp`.
+
+**Honest severity:** shipped defaults are correctly ordered (`2/5`, `2/7`, `5/11`), so nothing is
+broken out of the box, and **this fix is a no-op at defaults — that is its acceptance signal.** It
+needs a mis-ordered asset value, which nothing prevents: `ClampMin` is a per-property floor and
+Unreal cannot express "≤ that other property". `ColumnMinRadius` is also settable per-room via
+`UVoxelTerrainOpDefinition`. What makes five lines worth it is the failure *mode*: `Eval` still draws
+the fat column perfectly, so every meshed tile looks right and only the **skipped** ones are missing
+geometry and collision — invisible, and maddening to diagnose.
+
+⚠️ Recorded in the spec because it is the tempting wrong fix: **do not normalise the params.**
+Swapping the endpoints maps the same hash to a different radius, changing generated geometry and
+breaking the eight equivalence tests. Only the *bound* may become conservative; `Eval` stays
+byte-identical. Same rule as task 003.
+
+### ✅ CHECKED AND SOUND — do not re-audit these
+
+- **`FPassageCarveOp`** — its `Eval` reads `EvaluateModifierSDF` while its box guard asks
+  `AnyPassageNearBox`; the differing nouns look like a set mismatch and are not — **both iterate the
+  same `Passages` array**, "Modifier" is a legacy name. And the guard is the *same function* the
+  hand-written `ClassifyTile` uses, so there is one definition, not two.
+- **The passage bounding sphere** — `BoundRadius = maxDistFromCentre + Radius + 4.0f`, and that `+4`
+  **is** `PASSAGE_BLEND_RADIUS`, so "the bound already includes the blend" is true rather than
+  assumed. Its per-point radii are `RadiusAt(t) = Lerp(Mouth, Mid, Sin(t·π))` with `Sin ∈ [0,1]`, so
+  they never exceed `Passage.Radius = max(Mouth, Mid)` — the one place the `Min > Max` class *could*
+  have bitten a shipped default, and it does not.
+- **`FOriginSpineOp` / `FBoundarySealOp`** — Z-band plus XY-circle tests against named reaches; both
+  fail safe to `CarveOnly`/`FillOnly`.
+- **`FShaftLedgeMod`** — returns `FillOnly` unconditionally whenever ledges are configured, and its
+  `Identity` case is exactly the early-out its `Eval` takes. Conservative by construction.
+- **`FCaveArchMod` / `FDomeMod`** — they *also* roll `Lerp(Min, Max, hash)` radii, but their
+  `Identity` comes only from `VF_NoCaveOverBox` (cave-surface proximity) or the feature being off.
+  **No radius envelope in their verdict ⇒ not exposed.** Checked precisely because they matched the
+  pattern on a grep.
+- `FConstantFieldSource`, `FSlabVoidSource`, `FSurfaceColumnSource` — never return `Identity` at all.
+
+### The shape of both bugs, worth naming
+
+Task 003 and task 004 are the same mistake twice: **a box verdict's bound was derived from the
+parameter that reads like the maximum rather than from the actual supremum of what `Eval` produces.**
+Once it was `sup|FBM|` (assumed 1.0, proved 1.5); once it was `max radius` (assumed `MaxRadius`,
+actually `max(Min, Max)`). Both times a correct instance of the same reasoning existed **elsewhere in
+the same file** — `VF_PerlinAbsBound` for the first, `FIslandBlobSource` for the second.
+
+⇒ When adding a bound, the question is not "what is the max parameter" but **"what is the supremum
+of the thing `Eval` can actually produce, and where in this file has that already been worked out?"**

Source/VoxelForge/Private/VoxelDensityOpStack.cpp

@@ -1081,9 +1081,10 @@ namespace
         {
             if (ColDensity <= 0.0f || Spacing <= 0.0f) { return EVoxelOpEffect::Identity; }
 
-            // Marge : le centre d'une colonne vit dans sa cellule, son influence porte au plus
-            // MaxRadius + ColBlend. Sur-estimer coûte du CPU ; sous-estimer serait un trou.
-            const float Reach = FMath::Max(MaxRadius, 0.0f) + ColBlend;
+            // Marge : l'enveloppe de `Lerp(MinRadius, MaxRadius, t)` est max(MinRadius, MaxRadius),
+            // pas `MaxRadius` seul si l'asset inverse les paramètres. The bound must cover both
+            // endpoints; using `MaxRadius` alone would leave a hole when the asset reverses them.
+            const float Reach = FMath::Max3(MinRadius, MaxRadius, 0.0f) + ColBlend;
 
             const int32 CX0 = FMath::FloorToInt(((float)VoxelBox.Min.X - Reach) / Spacing);
             const int32 CX1 = FMath::FloorToInt(((float)VoxelBox.Max.X + Reach) / Spacing);
@@ -1433,7 +1434,10 @@ namespace
             // La source répond pour la paire source+carve (SIMPLIFICATION DE PHASE 1) : `CarveOnly`
             // si une primitive atteint la boîte, `Identity` sinon. `ExtraReach` couvre la rugosité
             // et le blend en aval — le sous-estimer serait un TROU.
-            const float Pad = FMath::Max(P.ShaftMaxRadius, P.ConnectorRadius) + ExtraReach;
+            // L'enveloppe doit couvrir les deux bornes de `Lerp(ShaftMinRadius, ShaftMaxRadius, t)`,
+            // pas `ShaftMaxRadius` seul si l'asset inverse les paramètres. The bound must cover
+            // both radius endpoints before adding connector and downstream reach.
+            const float Pad = FMath::Max3(P.ShaftMinRadius, P.ShaftMaxRadius, P.ConnectorRadius) + ExtraReach;
             const FBox Padded = VoxelBox.ExpandBy(Pad);
 
             const float Spacing = FMath::Max(P.ShaftSpacing, 1.0f);