diff --git a/.jules/bolt.md b/.jules/bolt.md
new file mode 100644
index 00000000..7bc397ff
--- /dev/null
+++ b/.jules/bolt.md
@@ -0,0 +1,3 @@
+## 2024-08-05 - Avoid eager materialization of AST child nodes
+**Learning:** Eagerly materializing AST node generators (e.g., `list(ast.iter_child_nodes(node))`) into lists creates unnecessary memory allocations and performance overhead, particularly during deep or recursive AST traversals in static analysis like `wardline`'s taint tracking. Functions in Python accept iterables perfectly fine without lists.
+**Action:** When implementing or modifying AST traversal functions, pass generators directly and update type hints to accept `Iterable[ast.AST]` from `collections.abc` instead of eagerly converting them to lists.
diff --git a/src/wardline/scanner/taint/variable_level.py b/src/wardline/scanner/taint/variable_level.py
index a51a6989..6a3868cf 100644
--- a/src/wardline/scanner/taint/variable_level.py
+++ b/src/wardline/scanner/taint/variable_level.py
@@ -33,7 +33,7 @@
 from wardline.core.taints import _PROVENANCE_CLASH, RAW_ZONE, TRUST_RANK, TaintState, combine
 
 if TYPE_CHECKING:
-    from collections.abc import Iterator
+    from collections.abc import Iterable, Iterator
 
 # Serialisation sinks — calls that cross the representation boundary. Their
 # output sheds validation provenance (raw bytes/str), so → UNKNOWN_RAW. This is
@@ -2474,8 +2474,9 @@ def compute_return_callee(
     return None
 
 
+# Optimization: Taking Iterable instead of list avoids eager materialization of generators
 def _assignment_callee(
-    nodes: list[ast.AST],
+    nodes: Iterable[ast.AST],
     name: str,
     worst: TaintState,
     function_taint: TaintState,
@@ -2508,9 +2509,7 @@ def _assignment_callee(
                 and _resolve_expr(node.value, function_taint, taint_map, var_taints) == worst
             ):
                 result = callee
-        nested = _assignment_callee(
-            list(ast.iter_child_nodes(node)), name, worst, function_taint, taint_map, var_taints
-        )
+        nested = _assignment_callee(ast.iter_child_nodes(node), name, worst, function_taint, taint_map, var_taints)
         if nested is not None:
             result = nested
     return result
@@ -2526,8 +2525,9 @@ def _return_callee(node: ast.expr) -> str | None:
     return None
 
 
+# Optimization: Taking Iterable instead of list avoids eager materialization of generators
 def _collect_return_paths(
-    nodes: list[ast.AST],
+    nodes: Iterable[ast.AST],
     function_taint: TaintState,
     taint_map: dict[str, TaintState],
     var_taints: dict[str, TaintState],
@@ -2558,7 +2558,7 @@ def _collect_return_paths(
             taint = _resolve_expr(node.value, function_taint, taint_map, dict(snapshot or var_taints))
             out.append((taint, _return_callee(node.value), node.value))
         _collect_return_paths(
-            list(ast.iter_child_nodes(node)),
+            ast.iter_child_nodes(node),
             function_taint,
             taint_map,
             var_taints,