@@ -696,27 +696,6 @@ jobs:
             "${{ env.CUDA_CORE_ARTIFACTS_DIR }}"/cu"${BUILD_PREV_CUDA_MAJOR}"/cuda_core*.whl \
             --output-dir "${{ env.CUDA_CORE_ARTIFACTS_DIR }}"
 
-      - name: Check merged cuda.core wheel layout
-        if: ${{ env.BUILD_CORE == 'true' }}
-        run: |
-          # cuda/core/__init__.py rewrites __path__ to the active cuXX/ tree, so
-          # only the entry points and the two versioned trees belong at the top
-          # level of cuda/core/; anything else is a dead copy.
-          python - "${{ env.CUDA_CORE_ARTIFACTS_DIR }}"/cuda_core*.whl <<'EOF'
-          import os
-          import sys
-          import zipfile
-
-          prefix = "cuda/core/"
-          members = zipfile.ZipFile(sys.argv[1]).namelist()
-          names = {n[len(prefix) :].split("/")[0] for n in members if n.startswith(prefix)}
-          names.discard("")
-          expected = {"__init__.py", "_version.py"}
-          expected |= {f"cu{os.environ[v]}" for v in ("BUILD_CUDA_MAJOR", "BUILD_PREV_CUDA_MAJOR")}
-          assert names == expected, f"unexpected top-level entries under cuda/core/: {sorted(names ^ expected)}"
-          print("merged wheel top level:", sorted(names))
-          EOF
-
       - name: Check cuda.core wheel
         if: ${{ env.BUILD_CORE == 'true' }}
         run: |
 
@@ -133,9 +133,11 @@ def merge_wheels(wheels: list[Path], output_dir: Path, show_wheel_contents: bool
         # Copy version-specific directories from each wheel into versioned subdirectories
         base_dir = Path("cuda") / "core"
 
+        versioned_dirs = set()
         for i, wheel_dir in enumerate(extracted_wheels):
             cuda_version = wheels[i].name.split(".cu")[1].split(".")[0]
             versioned_dir = base_wheel / base_dir / f"cu{cuda_version}"
+            versioned_dirs.add(versioned_dir.name)
 
             # Copy entire directory tree from source wheel to versioned directory
             print(f"  Copying {wheel_dir / base_dir} to {versioned_dir}", file=sys.stderr)
@@ -147,14 +149,9 @@ def merge_wheels(wheels: list[Path], output_dir: Path, show_wheel_contents: bool
         print("\n=== Removing files from cuda/core/ directory ===", file=sys.stderr)
         # Only what cuda/core/__init__.py uses before it rewrites __path__ to the
         # versioned subpackage stays at top level: it imports _version, then
-        # redirects every later import into cu12/ or cu13/. Anything else left at
-        # top level is a dead copy that nothing imports.
-        items_to_keep = (
-            "__init__.py",
-            "_version.py",
-            "cu12",
-            "cu13",
-        )
+        # redirects every later import into the versioned tree. Anything else
+        # left at top level is a dead copy that nothing imports.
+        items_to_keep = {"__init__.py", "_version.py", *versioned_dirs}
         all_items = os.scandir(base_wheel / base_dir)
         removed_count = 0
         for f in all_items:
@@ -169,6 +166,11 @@ def merge_wheels(wheels: list[Path], output_dir: Path, show_wheel_contents: bool
                 os.remove(f_abspath)
             removed_count += 1
         print(f"Removed {removed_count} items from cuda/core/ directory", file=sys.stderr)
+        remaining = {entry.name for entry in os.scandir(base_wheel / base_dir)}
+        if remaining != items_to_keep:
+            raise RuntimeError(
+                f"unexpected top level under cuda/core/: {sorted(remaining)} (expected {sorted(items_to_keep)})"
+            )
 
         # Repack the merged wheel
         output_dir.mkdir(parents=True, exist_ok=True)
 
@@ -188,14 +188,14 @@ def _extension_sources(mod_name):
     cuda/core/_cpp/<stem>/, or the single legacy file cuda/core/_cpp/<stem>.cpp.
     Example: _tensor_map.pyx compiles _cpp/tensor_map.cpp."""
     sources = [f"cuda/core/{mod_name}.pyx"]
-    cpp_stem = os.path.join("cuda", "core", "_cpp", mod_name.lstrip("_"))
-    if os.path.isdir(cpp_stem):
-        cpp_sources = sorted(glob.glob(os.path.join(cpp_stem, "**", "*.cpp"), recursive=True))
+    cpp_stem = Path("cuda", "core", "_cpp", mod_name.lstrip("_"))
+    if cpp_stem.is_dir():
+        cpp_sources = sorted(str(path) for path in cpp_stem.rglob("*.cpp"))
         if not cpp_sources:
             raise RuntimeError(f"{cpp_stem}/ exists but contains no .cpp files")
         sources.extend(cpp_sources)
-    elif os.path.isfile(cpp_stem + ".cpp"):
-        sources.append(cpp_stem + ".cpp")
+    elif cpp_stem.with_suffix(".cpp").is_file():
+        sources.append(str(cpp_stem.with_suffix(".cpp")))
     return sources