perf: skip per-edge git commits during auto-edge creation

Auto-edges called relate() up to 3 times per store, each triggering a
blocking git subprocess (~50-100ms each). Now relate() accepts
skip_commit=True so auto-edges defer to the fire-and-forget git sync,
cutting 150-300ms of latency from every memory_store call.

Closes #2

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

edges.py

@@ -30,6 +30,7 @@ class EdgesMixin:
         strength: float = 0.8,
         context: Optional[str] = None,
         description: Optional[str] = None,
+        skip_commit: bool = False,
     ) -> str:
         """Create a relationship between two memories with an edge file.
 
@@ -137,10 +138,11 @@ class EdgesMixin:
             edge_id, edge_data, f"graph/{EDGES_DIR_NAME}/{edge_filename}"
         )
 
-        self._git_commit(
-            f"relate: {from_id[:8]} --{rel_type}--> {to_id[:8]}",
-            [from_path, to_path, edge_path],
-        )
+        if not skip_commit:
+            self._git_commit(
+                f"relate: {from_id[:8]} --{rel_type}--> {to_id[:8]}",
+                [from_path, to_path, edge_path],
+            )
         return edge_id
 
     def edge_get(self, edge_id: str) -> Optional[Dict[str, Any]]:

mcp_server.py

@@ -632,6 +632,7 @@ def _auto_create_edges(
                 rel_type=rel_type,
                 description=desc,
                 strength=strength,
+                skip_commit=True,  # batched — git sync handles persistence
             )
 
             if edge_id:  # Empty string means duplicate