feat(mnemosyne): add graph_data() for 3D constellation export

Returns {nodes, edges} with live link weights. Supports topic_filter
for subgraph extraction. Edges are deduplicated (bidirectional links
become single undirected edges).

Closes #1232

nexus/mnemosyne/archive.py

@@ -212,6 +212,65 @@ class MnemosyneArchive:
     def count(self) -> int:
         return len(self._entries)
 
+    def graph_data(
+        self,
+        topic_filter: Optional[str] = None,
+    ) -> dict:
+        """Export the full connection graph for 3D constellation visualization.
+
+        Returns a dict with:
+        - nodes: list of {id, title, topics, source, created_at}
+        - edges: list of {source, target, weight} from holographic links
+
+        Args:
+            topic_filter: If set, only include entries matching this topic
+                and edges between them.
+        """
+        entries = list(self._entries.values())
+
+        if topic_filter:
+            topic_lower = topic_filter.lower()
+            entries = [
+                e for e in entries
+                if topic_lower in [t.lower() for t in e.topics]
+            ]
+
+        entry_ids = {e.id for e in entries}
+
+        nodes = [
+            {
+                "id": e.id,
+                "title": e.title,
+                "topics": e.topics,
+                "source": e.source,
+                "created_at": e.created_at,
+            }
+            for e in entries
+        ]
+
+        # Build edges from links, dedup (A→B and B→A become one edge)
+        seen_edges: set[tuple[str, str]] = set()
+        edges = []
+        for e in entries:
+            for linked_id in e.links:
+                if linked_id not in entry_ids:
+                    continue
+                pair = (min(e.id, linked_id), max(e.id, linked_id))
+                if pair in seen_edges:
+                    continue
+                seen_edges.add(pair)
+                # Compute weight via linker for live similarity score
+                linked = self._entries.get(linked_id)
+                if linked:
+                    weight = self.linker.compute_similarity(e, linked)
+                    edges.append({
+                        "source": pair[0],
+                        "target": pair[1],
+                        "weight": round(weight, 4),
+                    })
+
+        return {"nodes": nodes, "edges": edges}
+
     def stats(self) -> dict:
         entries = list(self._entries.values())
         total_links = sum(len(e.links) for e in entries)