diff --git a/.claude-plugin/marketplace.json b/.claude-plugin/marketplace.json
index 0794c955ce..97f44630b7 100644
--- a/.claude-plugin/marketplace.json
+++ b/.claude-plugin/marketplace.json
@@ -9,7 +9,7 @@
       "name": "mempalace",
       "source": "./.claude-plugin",
       "description": "AI memory system — mine projects and conversations into a searchable palace. 19 MCP tools, auto-save hooks, guided setup.",
-      "version": "3.0.11",
+      "version": "3.0.12",
       "author": {
         "name": "milla-jovovich"
       }
diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 295b247069..7d927d43c9 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
 {
   "name": "mempalace",
-  "version": "3.0.11",
+  "version": "3.0.12",
   "description": "Give your AI a memory — mine projects and conversations into a searchable palace. 19 MCP tools, auto-save hooks, and guided setup.",
   "author": {
     "name": "milla-jovovich"
diff --git a/.codex-plugin/plugin.json b/.codex-plugin/plugin.json
index 480a21f0e4..7d1a09efad 100644
--- a/.codex-plugin/plugin.json
+++ b/.codex-plugin/plugin.json
@@ -1,6 +1,6 @@
 {
   "name": "mempalace",
-  "version": "3.0.11",
+  "version": "3.0.12",
   "description": "Give your AI a memory — mine projects and conversations into a searchable palace. 19 MCP tools, auto-save hooks, and guided setup.",
   "author": {
     "name": "milla-jovovich"
diff --git a/mempalace/knowledge_graph.py b/mempalace/knowledge_graph.py
index 226c92da54..08d8893200 100644
--- a/mempalace/knowledge_graph.py
+++ b/mempalace/knowledge_graph.py
@@ -51,6 +51,17 @@ def __init__(self, db_path: str = None):
         self.db_path = db_path or DEFAULT_KG_PATH
         Path(self.db_path).parent.mkdir(parents=True, exist_ok=True)
         self._init_db()
+        self._apply_migrations()
+
+    def _apply_migrations(self):
+        conn = self._conn()
+        try:
+            conn.execute("ALTER TABLE triples ADD COLUMN pheromone_level REAL DEFAULT 0.0")
+        except sqlite3.OperationalError:
+            pass  # Column already exists
+        finally:
+            conn.commit()
+            conn.close()
 
     def _init_db(self):
         conn = self._conn()
@@ -204,16 +215,18 @@ def query_entity(self, name: str, as_of: str = None, direction: str = "outgoing"
                 query += " AND (t.valid_from IS NULL OR t.valid_from <= ?) AND (t.valid_to IS NULL OR t.valid_to >= ?)"
                 params.extend([as_of, as_of])
             for row in conn.execute(query, params).fetchall():
+                # To protect against schema version variance:
+                # 0: id, 1: sub, 2: pred, 3: obj, 4: from, 5: to, 6: conf, 7: src_closet or pheromone
+                # Last column is always the JOINed name.
                 results.append(
                     {
                         "direction": "outgoing",
                         "subject": name,
                         "predicate": row[2],
-                        "object": row[10],  # obj_name
+                        "object": row[-1],  # obj_name is the last column
                         "valid_from": row[4],
                         "valid_to": row[5],
                         "confidence": row[6],
-                        "source_closet": row[7],
                         "current": row[5] is None,
                     }
                 )
@@ -228,13 +241,12 @@ def query_entity(self, name: str, as_of: str = None, direction: str = "outgoing"
                 results.append(
                     {
                         "direction": "incoming",
-                        "subject": row[10],  # sub_name
+                        "subject": row[-1],  # sub_name is the last column
                         "predicate": row[2],
                         "object": name,
                         "valid_from": row[4],
                         "valid_to": row[5],
                         "confidence": row[6],
-                        "source_closet": row[7],
                         "current": row[5] is None,
                     }
                 )
@@ -262,9 +274,9 @@ def query_relationship(self, predicate: str, as_of: str = None):
         for row in conn.execute(query, params).fetchall():
             results.append(
                 {
-                    "subject": row[10],
+                    "subject": row[-2], # sub_name
                     "predicate": pred,
-                    "object": row[11],
+                    "object": row[-1], # obj_name
                     "valid_from": row[4],
                     "valid_to": row[5],
                     "current": row[5] is None,
@@ -303,9 +315,9 @@ def timeline(self, entity_name: str = None):
         conn.close()
         return [
             {
-                "subject": r[10],
+                "subject": r[-2],
                 "predicate": r[2],
-                "object": r[11],
+                "object": r[-1],
                 "valid_from": r[4],
                 "valid_to": r[5],
                 "current": r[5] is None,
@@ -313,6 +325,128 @@ def timeline(self, entity_name: str = None):
             for r in rows
         ]
 
+    # ── Pheromone Traversal (STAN) ────────────────────────────────────────
+
+    def evaporate_pheromones(self, decay_rate: float = 0.1):
+        """Apply evaporation decay to all active pheromone trails."""
+        conn = self._conn()
+        conn.execute(
+            "UPDATE triples SET pheromone_level = ROUND(pheromone_level * (1.0 - ?), 2) WHERE pheromone_level > 0",
+            (decay_rate,)
+        )
+        conn.commit()
+        conn.close()
+
+    def deposit_pheromone(self, subject: str, predicate: str, obj: str, amount: float = 1.0, max_pheromone: float = 100.0):
+        """Deposit pheromone on a specific edge to reinforce a successful path."""
+        sub_id = self._entity_id(subject)
+        obj_id = self._entity_id(obj)
+        pred = predicate.lower().replace(" ", "_")
+        
+        conn = self._conn()
+        conn.execute(
+            """UPDATE triples 
+               SET pheromone_level = MIN(COALESCE(pheromone_level, 0.0) + ?, ?) 
+               WHERE subject=? AND predicate=? AND object=? AND valid_to IS NULL""",
+            (amount, max_pheromone, sub_id, pred, obj_id),
+        )
+        conn.commit()
+        conn.close()
+
+    def get_neighbors_with_pheromones(self, entity_id: str):
+        """Get all connected nodes for graph traversal with their pheromone levels."""
+        conn = self._conn()
+        neighbors = []
+        
+        # Outgoing
+        for row in conn.execute(
+            "SELECT object, predicate, COALESCE(pheromone_level, 0.0) FROM triples WHERE subject = ? AND valid_to IS NULL", 
+            (entity_id,)
+        ).fetchall():
+            neighbors.append({
+                "node": row[0], 
+                "predicate": row[1], 
+                "pheromone": row[2],
+                "direction": "outgoing"
+            })
+            
+        # Incoming
+        for row in conn.execute(
+            "SELECT subject, predicate, COALESCE(pheromone_level, 0.0) FROM triples WHERE object = ? AND valid_to IS NULL", 
+            (entity_id,)
+        ).fetchall():
+            neighbors.append({
+                "node": row[0], 
+                "predicate": row[1], 
+                "pheromone": row[2],
+                "direction": "incoming"
+            })
+            
+        conn.close()
+        return neighbors
+
+    def stigmergic_astar(self, start_entity: str, target_entity: str, influence: float = 0.7, max_depth: int = 5):
+        """
+        Find optimal path leveraging stigmergic edge weights from STAN.
+        cost_eff = 1.0 / (1 + pheromone * influence)
+        """
+        import heapq
+        
+        start_id = self._entity_id(start_entity)
+        target_id = self._entity_id(target_entity)
+        
+        # Queue: (cost_so_far, current_node, path_history, depth)
+        # Using simple Dijkstra approach mapped with STAN's edge weights, 
+        # since pure semantic distance heuristics (ChromaDB) require injection.
+        queue = [(0.0, start_id, [], 0)]
+        g_scores = {start_id: 0.0}
+        visited = set()
+        
+        while queue:
+            current_cost, current_node, path, depth = heapq.heappop(queue)
+            
+            if current_node == target_id:
+                return path
+                
+            if depth >= max_depth:
+                continue
+                
+            if current_node in visited:
+                continue
+                
+            visited.add(current_node)
+            
+            for neighbor in self.get_neighbors_with_pheromones(current_node):
+                next_node = neighbor["node"]
+                
+                if next_node in visited:
+                    continue
+                
+                # STAN Core Formula: cost_effective = w / (1 + tau * alpha)
+                # w = 1.0 base weight. tau = pheromone. alpha = influence parameter (0.7 default)
+                tau = neighbor["pheromone"]
+                edge_cost = 1.0 / (1.0 + (tau * influence))
+                
+                tentative_g = current_cost + edge_cost
+                
+                if tentative_g < g_scores.get(next_node, float('inf')):
+                    g_scores[next_node] = tentative_g
+                    
+                    # Store path as lightweight tuples to avoid O(n^2) scaling 
+                    # and correctly map back into memory.
+                    step = {
+                        "from": current_node,
+                        "to": next_node,
+                        "predicate": neighbor["predicate"],
+                        "direction": neighbor["direction"],
+                        "pheromone": tau,
+                        "cost": edge_cost
+                    }
+                    
+                    heapq.heappush(queue, (tentative_g, next_node, path + [step], depth + 1))
+                    
+        return None
+
     # ── Stats ─────────────────────────────────────────────────────────────
 
     def stats(self):
diff --git a/mempalace/mcp_server.py b/mempalace/mcp_server.py
index b447249141..e0ecf5e65f 100644
--- a/mempalace/mcp_server.py
+++ b/mempalace/mcp_server.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
 """
 MemPalace MCP Server — read/write palace access for Claude Code
-================================================================
+=========================================================
 Install: claude mcp add mempalace -- python -m mempalace.mcp_server [--palace /path/to/palace]
 
 Tools (read):
@@ -33,6 +33,12 @@
 
 from .knowledge_graph import KnowledgeGraph
 
+
+
+_kg = KnowledgeGraph()
+_pheromone_decay_tick = 0
+
+
 logging.basicConfig(level=logging.INFO, format="%(message)s", stream=sys.stderr)
 logger = logging.getLogger("mempalace_mcp")
 
@@ -85,7 +91,7 @@ def _no_palace():
     }
 
 
-# ==================== READ TOOLS ====================
+# ============= READ TOOLS ====================
 
 
 def tool_status():
@@ -272,7 +278,7 @@ def tool_graph_stats():
     return graph_stats(col=col)
 
 
-# ==================== WRITE TOOLS ====================
+# ============= WRITE TOOLS ====================
 
 
 def tool_add_drawer(
@@ -331,7 +337,7 @@ def tool_delete_drawer(drawer_id: str):
         return {"success": False, "error": str(e)}
 
 
-# ==================== KNOWLEDGE GRAPH ====================
+# ============= KNOWLEDGE GRAPH ====================
 
 
 def tool_kg_query(entity: str, as_of: str = None, direction: str = "both"):
@@ -371,7 +377,23 @@ def tool_kg_stats():
     return _kg.stats()
 
 
-# ==================== AGENT DIARY ====================
+def tool_kg_stigmergic_astar(start_entity: str, target_entity: str, influence: float = 0.7, max_depth: int = 5):
+    """Find optimal learned path leveraging stigmergic pheromone edge weights."""
+    return {"path": _kg.stigmergic_astar(start_entity, target_entity, influence=influence, max_depth=max_depth)}
+
+def tool_kg_deposit_pheromone(subject: str, predicate: str, object: str, amount: float = 1.0):
+    """Deposit pheromone on a successful path to influence future A* queries."""
+    _kg.deposit_pheromone(subject, predicate, object, amount=amount)
+    
+    # Tick decay
+    global _pheromone_decay_tick
+    _pheromone_decay_tick += 1
+    if _pheromone_decay_tick % 50 == 0:
+        _kg.evaporate_pheromones(decay_rate=0.1)
+        
+    return {"success": True, "fact": f"{subject} → {predicate} → {object}", "amount": amount}
+
+# ============= AGENT DIARY ====================
 
 
 def tool_diary_write(agent_name: str, entry: str, topic: str = "general"):
@@ -465,7 +487,7 @@ def tool_diary_read(agent_name: str, last_n: int = 10):
         return {"error": str(e)}
 
 
-# ==================== MCP PROTOCOL ====================
+# ============= MCP PROTOCOL ====================
 
 TOOLS = {
     "mempalace_status": {
@@ -714,6 +736,34 @@ def tool_diary_read(agent_name: str, last_n: int = 10):
         },
         "handler": tool_diary_read,
     },
+    "mempalace_kg_stigmergic_astar": {
+        "description": "Find an optimal path between two concepts using STAN (Stigmergic A* Navigation). Uses pheromone trails accumulated from past traversals to find fast paths.",
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "start_entity": {"type": "string", "description": "Starting concept or node"},
+                "target_entity": {"type": "string", "description": "Target concept or node"},
+                "influence": {"type": "number", "description": "How much pheromones discount cost (default 0.7)"},
+                "max_depth": {"type": "integer", "description": "Maximum depth to search (default 5)"}
+            },
+            "required": ["start_entity", "target_entity"],
+        },
+        "handler": tool_kg_stigmergic_astar,
+    },
+    "mempalace_kg_deposit_pheromone": {
+        "description": "Deposit pheromone on a successful edge to reinforce it for future A* pathfinding queries.",
+        "input_schema": {
+            "type": "object",
+            "properties": {
+                "subject": {"type": "string", "description": "Start entity of the edge"},
+                "predicate": {"type": "string", "description": "Relationship type"},
+                "object": {"type": "string", "description": "End entity of the edge"},
+                "amount": {"type": "number", "description": "Amount to deposit (default 1.0)"}
+            },
+            "required": ["subject", "predicate", "object"],
+        },
+        "handler": tool_kg_deposit_pheromone,
+    },
 }
 
 
@@ -765,6 +815,9 @@ def handle_request(request):
                 tool_args[key] = int(value)
             elif declared_type == "number" and not isinstance(value, (int, float)):
                 tool_args[key] = float(value)
+
+        # Apply specific tool routing if parameters were named dynamically 
+        # (Though **tool_args unpacks them naturally).
         try:
             result = TOOLS[tool_name]["handler"](**tool_args)
             return {
diff --git a/mempalace/version.py b/mempalace/version.py
index 6bf70fc5d5..b2a0d20d65 100644
--- a/mempalace/version.py
+++ b/mempalace/version.py
@@ -1,3 +1,3 @@
 """Single source of truth for the MemPalace package version."""
 
-__version__ = "3.0.11"
+__version__ = "3.0.12"
diff --git a/pyproject.toml b/pyproject.toml
index 00e5c93cb0..c5ee2c5727 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "mempalace"
-version = "3.0.11"
+version = "3.0.12"
 description = "Give your AI a memory — mine projects and conversations into a searchable palace. No API key required."
 readme = "README.md"
 requires-python = ">=3.9"