Fix CLI errors and Postgres persistence

memory_thread/cli.py

@@ -33,6 +33,9 @@
 from rich.panel import Panel
 from rich.text import Text
 from rich import print as rprint
+from dotenv import load_dotenv
+
+load_dotenv()
 
 # ═══════════════════════════════════════════════════════════════════════════════
 # WINDOWS UTF-8 FIX — no more Wakandan runes
@@ -216,6 +219,26 @@ def _enter_chat():
             console.print("[dim]Goodbye.[/dim]")
             break
 
+        # Slash commands (e.g., /provider use groq)
+        if raw.startswith("/"):
+            import shlex
+            import subprocess
+            try:
+                args = shlex.split(raw[1:])
+                if not args:
+                    continue
+
+                # Construct command to invoke same entry point
+                cmd_args = [sys.executable, sys.argv[0]] if sys.argv[0].endswith(".py") else [sys.argv[0]]
+                cmd_args.extend(args)
+
+                subprocess.run(cmd_args)
+                console.print()
+                continue
+            except Exception as e:
+                console.print(f"[red]Command failed: {e}[/red]")
+                continue
+
         # Chat handles EVERYTHING: remember, extract, contradict, respond
         try:
             response = client.chat(raw)

memory_thread/db/qdrant_client.py

@@ -48,11 +48,21 @@ def upsert(self, collection_name: str, points: List[Dict[str, Any]]):
     def search(self, collection_name: str, query_vector: List[float], limit: int = 5) -> List[Any]:
         """Search for similar vectors."""
         try:
-            return self.client.search(
-                collection_name=collection_name,
-                query_vector=query_vector,
-                limit=limit
-            )
+            # Fallback for deprecated 'search' if needed, or use 'query_points' for future compat
+            if hasattr(self.client, "search"):
+                return self.client.search(
+                    collection_name=collection_name,
+                    query_vector=query_vector,
+                    limit=limit
+                )
+            elif hasattr(self.client, "query_points"):
+                return self.client.query_points(
+                    collection_name=collection_name,
+                    query=query_vector,
+                    limit=limit
+                ).points
+            else:
+                raise AttributeError("QdrantClient has no search method")
         except Exception as e:
             from memory_thread.utils.logger import get_logger
             log = get_logger(__name__)

memory_thread/sdk.py

@@ -170,9 +170,9 @@ def _generate_embedding(self, text: str) -> List[float]:
         try:
             from memory_thread.utils.embeddings import generate_embeddings
             return generate_embeddings(tuple([text]))[0]
-        except ImportError:
+        except ImportError as e:
             # sentence-transformers not installed
-            log.debug("sentence-transformers not installed, using zero vector")
+            log.warning(f"sentence-transformers failed: {e}")
             return [0.0] * settings.EMBEDDING_DIMENSION
         except Exception as e:
             # Return zeros if embedding fails
@@ -188,8 +188,8 @@ def _extract_entities(self, text: str) -> List[Dict]:
         try:
             from memory_thread.services.hybrid_ner_service import extract_entities
             return extract_entities(text)
-        except ImportError:
-            log.warning("hybrid_ner_service not available, skipping entity extraction")
+        except ImportError as e:
+            log.warning(f"hybrid_ner_service not available (ImportError: {e}), skipping entity extraction")
             return []
         except Exception as e:
             log.warning(f"Entity extraction failed: {e}")
@@ -439,6 +439,9 @@ def remember(
     def _persist_to_postgres(self, entity_id: uuid.UUID, content: str, 
                               memory_type: str, state: EntityState, event: Event):
         """Persist memory to PostgreSQL."""
+        # Ensure event is persisted first to satisfy foreign key constraint
+        self.tms.persist_event(event)
+
         with self._pg.get_cursor() as cur:
             # Upsert into memories table (or entity_state)
             cur.execute("""
@@ -1130,11 +1133,13 @@ def chat(
 
         # Check explicit env override first
         env_provider = os.environ.get("MT_PROVIDER")
-        if env_provider:
+        if env_provider and env_provider != "auto":
             active_provider = env_provider.lower()
+            log.info(f"Using provider from ENV: {active_provider}")
         else:
             # Ensure we respect the vault's setting (which includes env vars now)
             active_provider = vault.get_active_provider(user_id)
+            log.info(f"Using provider from Vault: {active_provider}")
 
 
         log.debug(f"Chat request - Provider: {active_provider}, User: {user_id}")
@@ -1296,6 +1301,8 @@ def _generate_cloud(self, prompt: str, provider: str = "auto") -> str:
                 key = creds.get("api_key") if creds else os.environ.get("GROQ_API_KEY")
                 model = creds.get("model") if creds else os.environ.get("GROQ_MODEL", "llama-3.3-70b-versatile")
 
+                log.info(f"Groq API Key Check: Vault={'Found' if creds else 'Missing'}, Env={'Found' if os.environ.get('GROQ_API_KEY') else 'Missing'}")
+
                 if key:
                     log.info(f"Using Groq ({model})")
                     response = requests.post(

memory_thread/services/hybrid_ner_service.py

@@ -48,8 +48,17 @@ def extract_entities(text: str):
         return ner_cache.get(text_hash)
 
     # Run all stages
-    regex_entities = regex_ner(text)
-    spacy_entities, confidence = spacy_ner(text)
+    try:
+        regex_entities = regex_ner(text)
+    except Exception as e:
+        print(f"Regex NER failed: {e}")
+        regex_entities = []
+
+    try:
+        spacy_entities, confidence = spacy_ner(text)
+    except Exception as e:
+        print(f"Spacy NER failed: {e}")
+        spacy_entities, confidence = [], 0.0
 
     final_entities = regex_entities + spacy_entities
 

memory_thread/utils/embeddings.py

@@ -38,7 +38,7 @@ def _get_local_model():
         log.warning("sentence-transformers not installed. pip install sentence-transformers")
         return None
     except Exception as e:
-        log.warning(f"Failed to load embedding model: {e}")
+        log.warning(f"Failed to load embedding model: {e}", exc_info=True)
         return None
 
 

requirements.txt

@@ -20,3 +20,4 @@ prometheus-client
 fastapi
 uvicorn
 zmq
+sentence-transformers