Fix manager install URL normalization

tests/test_generate_popular_node_signatures.py

@@ -5563,6 +5563,39 @@ class ManagerIngestionTests(unittest.TestCase):
         self.assertEqual(42, entries[0]["metrics"]["downloads"])
         self.assertEqual("https://github.com/example/reference-nodes.git", entries[1]["repository"])
 
+    def test_normalise_manager_entries_prefers_git_install_files_over_reference(self):
+        manager_data = {
+            "custom_nodes": [
+                {
+                    "id": "stale-reference",
+                    "title": "Stale Reference",
+                    "reference": "https://github.com/example/stale-or-docs",
+                    "files": ["https://github.com/example/actual-install-repo"],
+                    "install_type": "git-clone",
+                },
+            ]
+        }
+
+        entries = normalise_manager_entries(manager_data)
+
+        self.assertEqual(1, len(entries))
+        self.assertEqual("https://github.com/example/actual-install-repo", entries[0]["repository"])
+
+    def test_normalise_manager_entries_skips_copy_installs_even_with_github_reference(self):
+        manager_data = {
+            "custom_nodes": [
+                {
+                    "id": "copy-only",
+                    "title": "Copy Only",
+                    "reference": "https://github.com/example/copy-only-docs",
+                    "files": ["https://raw.githubusercontent.com/example/copy-only/main/node.py"],
+                    "install_type": "copy",
+                },
+            ]
+        }
+
+        self.assertEqual([], normalise_manager_entries(manager_data))
+
     def test_rank_packs_uses_popularity_metrics_then_stable_fallbacks(self):
         packs = [
             {

tools/generate_popular_node_signatures.py

@@ -169,20 +169,42 @@ def _repository_candidates(item):
             yield candidate
 
 
+def _file_candidates(item):
+    files = item.get("files")
+    if isinstance(files, str):
+        yield files
+    elif isinstance(files, list):
+        for candidate in files:
+            yield candidate
+
+
+def _fallback_repository_candidates(item):
+    for key in ("repository", "repo", "git", "git_url", "url", "reference"):
+        value = item.get(key)
+        if isinstance(value, str):
+            yield value
+        elif isinstance(value, list):
+            for candidate in value:
+                yield candidate
+
+
+def _first_normalised_repository(candidates):
+    for candidate in candidates:
+        repository = _normalise_repository_url(candidate)
+        if repository:
+            return repository
+    return None
+
+
 def _manager_entry_repository(item):
     install_type = str(item.get("install_type") or item.get("installType") or "").lower()
-    candidates = list(_repository_candidates(item))
     if "git" in install_type:
-        for candidate in candidates:
-            repository = _normalise_repository_url(candidate)
-            if repository:
-                return repository
-        return None
-    for candidate in candidates:
-        repository = _normalise_repository_url(candidate)
-        if repository:
-            return repository
+        return _first_normalised_repository(_file_candidates(item)) or _first_normalised_repository(
+            _fallback_repository_candidates(item)
+        )
+    if install_type:
         return None
+    return _first_normalised_repository(_repository_candidates(item))
 
 
 def _entry_metrics(item):