Deduplication

tests/test_registry_service.py

@@ -641,6 +641,43 @@ def test_analyze_repository_can_disable_optional_llm_extractor(tmp_path):
     assert all(decision.action != "llm_extraction_used" for decision in decisions)
 
 
+def test_analyze_repository_normalizes_duplicate_llm_candidates(tmp_path):
+    source = tmp_path / "repo"
+    source.mkdir()
+    (source / "README.md").write_text(
+        "# LLM Connect\nSupports OpenRouter providers.\n",
+        encoding="utf-8",
+    )
+    store = RegistryStore(tmp_path / "registry.sqlite3")
+    store.initialize()
+    extractor = FakeLLMExtractor(
+        [
+            ExtractedAbility(
+                name="LLM Provider Integration",
+                description="Connects to model providers.",
+                source_paths=["README.md"],
+            ),
+            ExtractedAbility(
+                name="LLM Provider Integrations",
+                description="Connects prompts to OpenRouter providers.",
+                source_paths=["README.md"],
+            ),
+        ]
+    )
+    service = RegistryService(
+        store,
+        ingestion=GitIngestionService(tmp_path / "checkouts"),
+        llm_extractor=extractor,
+    )
+    repository = service.register_repository(name="LLM Connect", url=str(source))
+
+    summary = service.analyze_repository(repository.id)
+    graph = service.candidate_graph(repository.id, summary.analysis_run.id)
+
+    assert len(graph.abilities) == 1
+    assert graph.abilities[0].name == "LLM Provider Integrations"
+
+
 def test_analyze_repository_falls_back_when_optional_llm_extractor_returns_no_candidates(tmp_path):
     source = tmp_path / "repo"
     source.mkdir()