Recover repo-scoping native candidate families

tests/test_candidate_graph.py

@@ -561,7 +561,94 @@ def test_candidate_generator_does_not_promote_owned_provider_vocabulary_to_capab
 
     capability_names = {capability.name for capability in graph[0].capabilities}
     assert "Route LLM Requests Across Providers" not in capability_names
-    assert "Expose Repository Interface" in capability_names
+    assert "Scan Repositories Into Observed Facts" in capability_names
+
+
+def test_candidate_generator_recovers_repo_scoping_native_candidate_families():
+    repository = Repository(
+        id=1,
+        name="repo-scoping",
+        url="/tmp/repo-scoping",
+        description="Maps repositories into reviewable capability graphs.",
+        branch="main",
+        status="analyzed",
+    )
+    facts = [
+        fact(1, "documentation", "README", "README.md"),
+        fact(2, "documentation", "api-contract.md", "docs/api-contract.md"),
+        fact(
+            3,
+            "documentation",
+            "characteristic-evidence-model.md",
+            "docs/characteristic-evidence-model.md",
+        ),
+        fact(4, "documentation", "scope-md-spec.md", "docs/scope-md-spec.md"),
+        fact(
+            5,
+            "documentation",
+            "dependency-aware-scope-propagation.md",
+            "docs/dependency-aware-scope-propagation.md",
+        ),
+        fact(
+            6,
+            "documentation",
+            "repo-scope-context-response.json",
+            "docs/schemas/repo-scope-context-response.json",
+        ),
+        fact(7, "test", "test_git_ingestion.py", "tests/test_git_ingestion.py"),
+        fact(
+            8,
+            "test",
+            "test_repository_metadata.py",
+            "tests/test_repository_metadata.py",
+        ),
+        fact(
+            9,
+            "test",
+            "test_repository_scanner.py",
+            "tests/test_repository_scanner.py",
+        ),
+        fact(10, "test", "test_content_indexing.py", "tests/test_content_indexing.py"),
+        fact(11, "test", "test_candidate_graph.py", "tests/test_candidate_graph.py"),
+        fact(12, "test", "test_llm_extraction.py", "tests/test_llm_extraction.py"),
+        fact(13, "test", "test_registry_service.py", "tests/test_registry_service.py"),
+        fact(14, "test", "test_scope_generator.py", "tests/test_scope_generator.py"),
+        fact(15, "test", "test_web_api.py", "tests/test_web_api.py"),
+        fact(16, "test", "test_scope_context_api.py", "tests/test_scope_context_api.py"),
+        fact(
+            17,
+            "interface",
+            "python route decorator",
+            "src/repo_registry/web_api/app.py",
+            '@app.post("/repos")',
+        ),
+    ]
+
+    graph = CandidateGraphGenerator().generate(repository, facts)
+
+    capability_names = {capability.name for capability in graph[0].capabilities}
+    assert {
+        "Register And Track Repositories",
+        "Scan Repositories Into Observed Facts",
+        "Index Source Content With Provenance",
+        "Generate Reviewable Candidate Characteristics",
+        "Review And Approve Candidate Characteristics",
+        "Search Compare And Export Approved Profiles",
+        "Generate And Maintain SCOPE.md",
+        "Explore Dependency And Impact Graphs",
+        "Provide Scope Context To Downstream Agents",
+    } <= capability_names
+    assert "Route LLM Requests Across Providers" not in capability_names
+    scanning = next(
+        capability
+        for capability in graph[0].capabilities
+        if capability.name == "Scan Repositories Into Observed Facts"
+    )
+    assert scanning.primary_class == "analysis"
+    assert {"deterministic", "facts", "provenance", "utility-owned"} <= set(
+        scanning.attributes
+    )
+    assert all(ref.path.startswith(("docs/", "tests/", "src/")) for ref in scanning.source_refs)
 
 
 def test_candidate_generator_excludes_mention_only_providers_from_promoted_capability():