Stabilize core correctness and sync project plan state

tests/test_retrieval.py

@@ -1,7 +1,7 @@
 """Tests for the retrieval system."""
 
 from atocore.ingestion.pipeline import ingest_file
-from atocore.models.database import init_db
+from atocore.models.database import get_connection, init_db
 from atocore.retrieval.retriever import retrieve
 from atocore.retrieval.vector_store import get_vector_store
 
@@ -39,3 +39,31 @@ def test_vector_store_count(tmp_data_dir, sample_markdown):
     ingest_file(sample_markdown)
     store = get_vector_store()
     assert store.count > 0
+
+
+def test_retrieve_skips_stale_vector_entries(tmp_data_dir, sample_markdown, monkeypatch):
+    """Retriever should ignore vector hits whose chunk rows no longer exist."""
+    init_db()
+    ingest_file(sample_markdown)
+
+    with get_connection() as conn:
+        chunk_ids = [row["id"] for row in conn.execute("SELECT id FROM source_chunks").fetchall()]
+
+    class FakeStore:
+        def query(self, query_embedding, top_k=10, where=None):
+            return {
+                "ids": [[chunk_ids[0], "missing-chunk"]],
+                "documents": [["valid doc", "stale doc"]],
+                "metadatas": [[
+                    {"heading_path": "Overview", "source_file": "valid.md", "tags": "[]", "title": "Valid", "document_id": "doc-1"},
+                    {"heading_path": "Ghost", "source_file": "ghost.md", "tags": "[]", "title": "Ghost", "document_id": "doc-2"},
+                ]],
+                "distances": [[0.1, 0.2]],
+            }
+
+    monkeypatch.setattr("atocore.retrieval.retriever.get_vector_store", lambda: FakeStore())
+    monkeypatch.setattr("atocore.retrieval.retriever.embed_query", lambda query: [0.0, 0.1])
+
+    results = retrieve("overview", top_k=2)
+    assert len(results) == 1
+    assert results[0].chunk_id == chunk_ids[0]