feat: complete catalog-driven preprocessing

harness/tests/test_preprocess_cli.py

@@ -93,6 +93,39 @@ def test_v2_valid_materialized_corpus_is_validated_before_preprocessing(monkeypa
     assert calls[:2] == [("validate", tmp_path), ("vector", True)]
 
 
+def test_clear_removes_only_derived_paths_and_preserves_memory(monkeypatch, tmp_path):
+    import tht.cli.preprocess_cmd as command
+
+    monkeypatch.setenv("THT_PROFILE", "server")
+    config = _runtime_config(tmp_path)
+    for path in (
+        tmp_path / ".tht-dwh",
+        tmp_path / ".tht-jobs",
+        tmp_path / "corpus",
+        tmp_path / "indexes" / "lsh",
+    ):
+        path.mkdir(parents=True, exist_ok=True)
+        (path / "derived").write_text("generated", encoding="utf-8")
+    physical = tmp_path / "artifacts" / "mschema" / "physical.yaml"
+    physical.parent.mkdir(parents=True)
+    physical.write_text("generated", encoding="utf-8")
+    memory = tmp_path / "memory" / "registry.jsonl"
+    memory.parent.mkdir()
+    memory.write_text("durable", encoding="utf-8")
+
+    class Store:
+        def clear_reference(self):
+            return True
+
+    monkeypatch.setattr("tht.adapters.factory.build_vector_store", lambda *_args, **_kwargs: Store())
+    counts = command.clear_from_config(config)
+
+    assert counts == {"referenceCollections": 1, "derivedPaths": 5}
+    assert memory.read_text(encoding="utf-8") == "durable"
+    assert not (tmp_path / ".tht-dwh").exists()
+    assert not (tmp_path / "indexes" / "lsh").exists()
+
+
 def test_preprocess_evidence_json_is_pristine(monkeypatch, tmp_path):
     import tht.cli.preprocess_cmd as command