diff --git a/Lean/Screen/A5PortGroupBridge.lean b/Lean/Screen/A5PortGroupBridge.lean
new file mode 100644
index 00000000..f4c0b0ae
--- /dev/null
+++ b/Lean/Screen/A5PortGroupBridge.lean
@@ -0,0 +1,356 @@
+import A5PortSixAxesBridge
+import PSL2F5SixAxesBridge
+
+namespace OPH.A5PortGroupBridge
+
+open OPH.A5PortSixAxesBridge
+open OPH.PSL2F5SixAxesBridge
+
+/-!
+# The committed twelve-port rotations as a typed finite group
+
+The sixty committed twelve-port rotations are packaged as a subgroup of
+`Equiv.Perm (Fin 12)` and identified with the committed six-axis subgroup.
+
+The load-bearing check is multiplication on the actual `Fin 12`
+permutations. The port-row multiplication and inverse indices are transported
+through the certified `rowEquiv` into the exact multiplication and inverse
+tables of `A5SixAxes`; the resulting identities are then checked pointwise on
+the committed port permutations. `PortGroup` is defined as the range of those
+sixty rows, so it contains exactly the committed rotations.
+
+Composing this group isomorphism with `psl_equiv_six_axis_group` gives
+
+    PSL2F5 ≃* PortGroup.
+
+BOUNDARY. This file does not prove `PSL(2,5) ≅ A5`, identify `SL(2,5)` with
+the binary icosahedral group, construct an `SU(2)` lift, invoke McKay, transport
+the golden sectors as typed representations, derive or select `φ`, state a
+mass law, or identify physical rotations. -/
+
+/-! ## 1. Transport the six-axis row law back to the committed port rows -/
+
+/-- Multiplication index for the port rows, transported through the certified
+row equivalence to the exact six-axis multiplication table. -/
+def portMulIndex (i j : Fin 60) : Fin 60 :=
+  rowEquiv.symm (OPH.A5SixAxes.mulT (rowEquiv i) (rowEquiv j))
+
+/-- Inverse index for a port row, transported through the certified row
+equivalence to the exact six-axis inverse table. -/
+def portInvIndex (i : Fin 60) : Fin 60 :=
+  rowEquiv.symm (OPH.A5SixAxes.invT (rowEquiv i))
+
+@[simp]
+theorem rowEquiv_portMulIndex (i j : Fin 60) :
+    rowEquiv (portMulIndex i j) =
+      OPH.A5SixAxes.mulT (rowEquiv i) (rowEquiv j) := by
+  simp [portMulIndex]
+
+@[simp]
+theorem rowEquiv_portInvIndex (i : Fin 60) :
+    rowEquiv (portInvIndex i) = OPH.A5SixAxes.invT (rowEquiv i) := by
+  simp [portInvIndex]
+
+@[simp]
+theorem rowEquiv_zero : rowEquiv (0 : Fin 60) = 0 := by decide
+
+/-- Row zero is the identity port permutation. -/
+theorem portEl_zero : portEl 0 = 1 := by
+  apply Equiv.ext
+  intro k
+  revert k
+  decide
+
+/- The quadratic check is split into the same five-row bands used by
+`A5SixAxes.rowF_mul`. This keeps kernel reduction on raw action values rather
+than asking `decide` to compare `Equiv.Perm` proof structures. -/
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band0 :
+    ∀ (i j : Fin 60) (k : Fin 12), i.val < 5 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band1 :
+    ∀ (i j : Fin 60) (k : Fin 12), 5 ≤ i.val → i.val < 10 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band2 :
+    ∀ (i j : Fin 60) (k : Fin 12), 10 ≤ i.val → i.val < 15 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band3 :
+    ∀ (i j : Fin 60) (k : Fin 12), 15 ≤ i.val → i.val < 20 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band4 :
+    ∀ (i j : Fin 60) (k : Fin 12), 20 ≤ i.val → i.val < 25 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band5 :
+    ∀ (i j : Fin 60) (k : Fin 12), 25 ≤ i.val → i.val < 30 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band6 :
+    ∀ (i j : Fin 60) (k : Fin 12), 30 ≤ i.val → i.val < 35 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band7 :
+    ∀ (i j : Fin 60) (k : Fin 12), 35 ≤ i.val → i.val < 40 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band8 :
+    ∀ (i j : Fin 60) (k : Fin 12), 40 ≤ i.val → i.val < 45 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band9 :
+    ∀ (i j : Fin 60) (k : Fin 12), 45 ≤ i.val → i.val < 50 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band10 :
+    ∀ (i j : Fin 60) (k : Fin 12), 50 ≤ i.val → i.val < 55 →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+private theorem portEl_mul_band11 :
+    ∀ (i j : Fin 60) (k : Fin 12), 55 ≤ i.val →
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by decide
+
+/-- Pointwise multiplication law for all committed port rows. -/
+theorem portEl_mul_apply :
+    ∀ (i j : Fin 60) (k : Fin 12),
+      (portEl i * portEl j) k = portEl (portMulIndex i j) k := by
+  intro i j k
+  rcases Nat.lt_or_ge i.val 5 with h | h0
+  · exact portEl_mul_band0 i j k h
+  rcases Nat.lt_or_ge i.val 10 with h | h1
+  · exact portEl_mul_band1 i j k h0 h
+  rcases Nat.lt_or_ge i.val 15 with h | h2
+  · exact portEl_mul_band2 i j k h1 h
+  rcases Nat.lt_or_ge i.val 20 with h | h3
+  · exact portEl_mul_band3 i j k h2 h
+  rcases Nat.lt_or_ge i.val 25 with h | h4
+  · exact portEl_mul_band4 i j k h3 h
+  rcases Nat.lt_or_ge i.val 30 with h | h5
+  · exact portEl_mul_band5 i j k h4 h
+  rcases Nat.lt_or_ge i.val 35 with h | h6
+  · exact portEl_mul_band6 i j k h5 h
+  rcases Nat.lt_or_ge i.val 40 with h | h7
+  · exact portEl_mul_band7 i j k h6 h
+  rcases Nat.lt_or_ge i.val 45 with h | h8
+  · exact portEl_mul_band8 i j k h7 h
+  rcases Nat.lt_or_ge i.val 50 with h | h9
+  · exact portEl_mul_band9 i j k h8 h
+  rcases Nat.lt_or_ge i.val 55 with h | h10
+  · exact portEl_mul_band10 i j k h9 h
+  exact portEl_mul_band11 i j k h10
+
+/-- Multiplication of the actual twelve-port permutations is exactly the
+transported six-axis multiplication table. -/
+theorem portEl_mul (i j : Fin 60) :
+    portEl i * portEl j = portEl (portMulIndex i j) := by
+  apply Equiv.ext
+  exact portEl_mul_apply i j
+
+set_option maxHeartbeats 4000000 in
+set_option maxRecDepth 16384 in
+/-- Pointwise inverse law on the committed port rows. -/
+theorem portEl_inv_apply :
+    ∀ (i : Fin 60) (k : Fin 12),
+      (portEl i)⁻¹ k = portEl (portInvIndex i) k := by decide
+
+/-- Inversion of the actual twelve-port permutations is exactly the
+transported six-axis inverse table. -/
+theorem portEl_inv (i : Fin 60) :
+    (portEl i)⁻¹ = portEl (portInvIndex i) := by
+  apply Equiv.ext
+  exact portEl_inv_apply i
+
+/-! ## 2. The exact subgroup carried by the committed port rows -/
+
+/-- The subgroup consisting exactly of the sixty committed twelve-port
+rotations. Range membership makes the row ancestry explicit. -/
+def PortGroup : Subgroup (Equiv.Perm (Fin 12)) where
+  carrier := Set.range portEl
+  one_mem' := ⟨0, portEl_zero⟩
+  mul_mem' := by
+    intro g h hg hh
+    obtain ⟨i, rfl⟩ := hg
+    obtain ⟨j, rfl⟩ := hh
+    exact ⟨portMulIndex i j, (portEl_mul i j).symm⟩
+  inv_mem' := by
+    intro g hg
+    obtain ⟨i, rfl⟩ := hg
+    exact ⟨portInvIndex i, (portEl_inv i).symm⟩
+
+/-- Distinct row indices give distinct committed port permutations. The proof
+reuses the certified faithfulness of the antipodal quotient. -/
+theorem portEl_injective : Function.Injective portEl := by
+  intro i j hij
+  apply quotient_action_faithful
+  funext x
+  simp only [quotientAxis]
+  rw [hij]
+
+/-- Choose the unique committed row underlying a `PortGroup` element. -/
+noncomputable def portIndex (g : PortGroup) : Fin 60 :=
+  Classical.choose (show ∃ i : Fin 60, portEl i = g.1 from g.property)
+
+/-- The chosen row is the underlying port permutation. -/
+theorem portIndex_spec (g : PortGroup) : portEl (portIndex g) = g.1 :=
+  Classical.choose_spec (show ∃ i : Fin 60, portEl i = g.1 from g.property)
+
+/-- The range witness is unique, so choosing the index of a committed row
+returns that row's index. -/
+theorem portIndex_portEl (i : Fin 60) :
+    portIndex (⟨portEl i, ⟨i, rfl⟩⟩ : PortGroup) = i := by
+  apply portEl_injective
+  exact portIndex_spec _
+
+/-- The identity element has row index zero. -/
+theorem portIndex_one : portIndex (1 : PortGroup) = 0 := by
+  apply portEl_injective
+  calc
+    portEl (portIndex (1 : PortGroup)) = (1 : PortGroup).1 := portIndex_spec _
+    _ = 1 := rfl
+    _ = portEl 0 := portEl_zero.symm
+
+/-- The unique row index of a product is the transported multiplication
+index. -/
+theorem portIndex_mul (g h : PortGroup) :
+    portIndex (g * h) = portMulIndex (portIndex g) (portIndex h) := by
+  apply portEl_injective
+  calc
+    portEl (portIndex (g * h)) = (g * h).1 := portIndex_spec _
+    _ = g.1 * h.1 := rfl
+    _ = portEl (portIndex g) * portEl (portIndex h) := by
+      rw [portIndex_spec g, portIndex_spec h]
+    _ = portEl (portMulIndex (portIndex g) (portIndex h)) :=
+      portEl_mul _ _
+
+/-! ## 3. The typed homomorphism to the exact six-axis subgroup -/
+
+/-- Row zero is the identity in the explicit six-axis list. -/
+theorem sixEl_zero : OPH.A5SixAxes.el 0 = 1 := by
+  apply Equiv.ext
+  intro x
+  rw [OPH.A5SixAxes.el_apply, Equiv.Perm.one_apply]
+  revert x
+  decide
+
+/-- Multiplication of explicit six-axis rows, exposed from the certified raw
+row table. -/
+theorem sixEl_mul (i j : Fin 60) :
+    OPH.A5SixAxes.el i * OPH.A5SixAxes.el j =
+      OPH.A5SixAxes.el (OPH.A5SixAxes.mulT i j) := by
+  apply Equiv.ext
+  intro x
+  rw [Equiv.Perm.mul_apply, OPH.A5SixAxes.el_apply,
+    OPH.A5SixAxes.el_apply, OPH.A5SixAxes.el_apply]
+  exact OPH.A5SixAxes.rowF_mul i j x
+
+/-- Send a committed port rotation to the corresponding committed six-axis
+rotation. -/
+noncomputable def portToSix : PortGroup →*
+    OPH.PSL2F5SixAxesBridge.SixAxisGroup where
+  toFun g :=
+    ⟨OPH.A5SixAxes.el (rowEquiv (portIndex g)), OPH.A5SixAxes.el_mem _⟩
+  map_one' := by
+    apply Subtype.ext
+    change OPH.A5SixAxes.el (rowEquiv (portIndex (1 : PortGroup))) = 1
+    rw [portIndex_one, rowEquiv_zero, sixEl_zero]
+  map_mul' g h := by
+    apply Subtype.ext
+    change OPH.A5SixAxes.el (rowEquiv (portIndex (g * h))) =
+      OPH.A5SixAxes.el (rowEquiv (portIndex g)) *
+        OPH.A5SixAxes.el (rowEquiv (portIndex h))
+    rw [portIndex_mul, rowEquiv_portMulIndex]
+    exact (sixEl_mul _ _).symm
+
+/-- The port-to-six-axis homomorphism is injective. -/
+theorem portToSix_injective : Function.Injective portToSix := by
+  intro g h hgh
+  have hval :
+      OPH.A5SixAxes.el (rowEquiv (portIndex g)) =
+        OPH.A5SixAxes.el (rowEquiv (portIndex h)) :=
+    congrArg Subtype.val hgh
+  have hrow :
+      OPH.A5SixAxes.rowF (rowEquiv (portIndex g)) =
+        OPH.A5SixAxes.rowF (rowEquiv (portIndex h)) := by
+    funext x
+    simpa [OPH.A5SixAxes.el_apply] using
+      congrArg (fun e : Equiv.Perm (Fin 6) => e x) hval
+  have hidx : rowEquiv (portIndex g) = rowEquiv (portIndex h) :=
+    six_axis_rows_injective hrow
+  have hpidx : portIndex g = portIndex h := rowEquiv.injective hidx
+  apply Subtype.ext
+  calc
+    g.1 = portEl (portIndex g) := (portIndex_spec g).symm
+    _ = portEl (portIndex h) := congrArg portEl hpidx
+    _ = h.1 := portIndex_spec h
+
+/-- Every committed six-axis group element comes from one committed port
+rotation. -/
+theorem portToSix_surjective : Function.Surjective portToSix := by
+  intro g
+  obtain ⟨j, hj⟩ := OPH.A5SixAxes.mem_iff_el g.property
+  let i : Fin 60 := rowEquiv.symm j
+  let p : PortGroup := ⟨portEl i, ⟨i, rfl⟩⟩
+  refine ⟨p, ?_⟩
+  apply Subtype.ext
+  change OPH.A5SixAxes.el (rowEquiv (portIndex p)) = g.1
+  have hpi : portIndex p = i := by
+    dsimp [p]
+    exact portIndex_portEl i
+  rw [hpi]
+  change OPH.A5SixAxes.el (rowEquiv (rowEquiv.symm j)) = g.1
+  rw [rowEquiv.apply_symm_apply]
+  exact hj.symm
+
+/-- The actual twelve-port rotation subgroup is isomorphic to the exact
+committed six-axis subgroup. -/
+noncomputable def portGroupEquivSixAxisGroup :
+    PortGroup ≃* OPH.PSL2F5SixAxesBridge.SixAxisGroup :=
+  MulEquiv.ofBijective portToSix
+    ⟨portToSix_injective, portToSix_surjective⟩
+
+/-- Mathlib's abstract projective group is isomorphic to the subgroup of the
+sixty committed twelve-port rotations. -/
+noncomputable def pslEquivPortGroup :
+    OPH.PSL2F5SixAxesBridge.PSL2F5 ≃* PortGroup :=
+  OPH.PSL2F5SixAxesBridge.psl_equiv_six_axis_group.trans
+    portGroupEquivSixAxisGroup.symm
+
+end OPH.A5PortGroupBridge
+
+/- Axiom audit: no `sorry`, `admit`, new axioms, or `native_decide`. -/
+
+#print axioms OPH.A5PortGroupBridge.portEl_zero
+#print axioms OPH.A5PortGroupBridge.portEl_mul
+#print axioms OPH.A5PortGroupBridge.portEl_inv
+#print axioms OPH.A5PortGroupBridge.portEl_injective
+#print axioms OPH.A5PortGroupBridge.portToSix_injective
+#print axioms OPH.A5PortGroupBridge.portToSix_surjective
+#print axioms OPH.A5PortGroupBridge.portGroupEquivSixAxisGroup
+#print axioms OPH.A5PortGroupBridge.pslEquivPortGroup
diff --git a/Lean/Screen/OPHScreen.lean b/Lean/Screen/OPHScreen.lean
index e03f3c29..a47b66a2 100644
--- a/Lean/Screen/OPHScreen.lean
+++ b/Lean/Screen/OPHScreen.lean
@@ -50,6 +50,7 @@ import QuantumMatterIntegration
 import B10EdgeCenterAction
 import B10QuantumLimitations
 import PSL2F5SixAxesBridge
+import A5PortGroupBridge
 
 /-!
 # OPH screen arithmetic umbrella
diff --git a/Lean/docs/PROOF_INDEX.md b/Lean/docs/PROOF_INDEX.md
index 1aa79e9e..1fe428a4 100644
--- a/Lean/docs/PROOF_INDEX.md
+++ b/Lean/docs/PROOF_INDEX.md
@@ -1179,8 +1179,21 @@ Mapping between Lean 4 theorems in this project and statements in
   `PSL2F5 ≃* SixAxisGroup`.  This does not prove an abstract `PSL(2,5) ≅ A5`
   classification, identify `SL(2,5)` with `2I`, invoke McKay, transport the
   golden sectors as typed `PSL2F5` representations, select `φ` or a mass law,
-  or make a physical claim.  The preceding port bridge remains a pointwise
-  indexed-row bridge, not a group-homomorphism arrow in this chain.
+  or make a physical claim.  The preceding antipodal port bridge remains a
+  pointwise indexed-row equivalence of committed actions; the subsequent
+  twelve-port group bridge promotes those rows to a typed group isomorphism.
+- Twelve-port rotation group bridge
+  (`Screen/A5PortGroupBridge.lean`): sorry-free, standard axioms, no
+  `native_decide`.  Packages the sixty committed twelve-port rotations as the
+  subgroup `PortGroup` of `Equiv.Perm (Fin 12)` whose carrier is exactly
+  `Set.range portEl`.  Kernel `decide` checks multiplication and inversion on
+  the actual `Fin 12` permutations against the `rowEquiv`-transported
+  six-axis tables; `portEl` is injective, so the carrier has cardinality
+  sixty.  The resulting isomorphism `PortGroup ≃* SixAxisGroup` composes with
+  `psl_equiv_six_axis_group` to give `PSL2F5 ≃* PortGroup`.  This does not
+  prove an abstract `PSL(2,5) ≅ A5` classification, identify `SL(2,5)` with
+  `2I`, invoke McKay, transport the golden sectors as typed representations,
+  select `φ` or a mass law, or make a physical claim.
 - Universal cap coupling from icosahedral symmetry
   (`Screen/A5CouplingSymmetry.lean`, #576/#568): 9 / 9 declarations,
   sorry-free, standard axioms. Finite facts by kernel `decide` (caps map to
diff --git a/Lean/lakefile.lean b/Lean/lakefile.lean
index edce0eaa..fe646de6 100644
--- a/Lean/lakefile.lean
+++ b/Lean/lakefile.lean
@@ -75,7 +75,7 @@ lean_lib «OPHScreen» where
     `PortFrameGram, `A5PortAction, `A5CouplingSymmetry, `A5OPH,
     `A2HolonomyBridge, `HolonomyInterference, `A5CharacterField, `A5SixAxes,
     `PSL2F5SixAxesBridge,
-    `A5PortModule, `A5PortSixAxesBridge, `A5Commutant, `A5ResponseWordAlgebra,
+    `A5PortModule, `A5PortSixAxesBridge, `A5PortGroupBridge, `A5Commutant, `A5ResponseWordAlgebra,
     `A5IncidenceResponse, `TraceBalancedKernel, `TrichotomyCases, `Z6Descent,
     `A5AngularMultiplets, `A5AngularBands, `A5AngularKernels,
     `ExteriorSelection, `WeylYukawaConventions, `QuarkMeanDomain, `EqualStateWeights,
diff --git a/claims/active_surface_inventory.json b/claims/active_surface_inventory.json
index 92a3c5d4..ccb942ef 100644
--- a/claims/active_surface_inventory.json
+++ b/claims/active_surface_inventory.json
@@ -1,7 +1,7 @@
 {
   "schema": "oph.active_surface_inventory.v1",
   "generator": "tools/check_axiom_consistency.py --inventory",
-  "surface_count": 362,
+  "surface_count": 363,
   "entry_surfaces": [
     "README.md",
     "README_FR.md",
@@ -774,6 +774,11 @@
       "glob": "Lean/Screen/*.lean",
       "allowlisted": false
     },
+    {
+      "path": "Lean/Screen/A5PortGroupBridge.lean",
+      "glob": "Lean/Screen/*.lean",
+      "allowlisted": false
+    },
     {
       "path": "Lean/Screen/A5PortModule.lean",
       "glob": "Lean/Screen/*.lean",