diff --git a/python/lsst/pipe/base/quantum_graph_builder.py b/python/lsst/pipe/base/quantum_graph_builder.py
index 8961282da..5884b94f4 100644
--- a/python/lsst/pipe/base/quantum_graph_builder.py
+++ b/python/lsst/pipe/base/quantum_graph_builder.py
@@ -180,7 +180,6 @@ def __init__(
         self.metadata = TaskMetadata()
         self._pipeline_graph = pipeline_graph
         self.butler = butler
-        self._pipeline_graph.resolve(self.butler.registry)
         if input_collections is None:
             input_collections = butler.collections
         if not input_collections:
@@ -233,11 +232,13 @@ def __init__(
                 packages_storage_class,
             )
         }
-        self._find_empty_dimension_datasets()
-        self.prerequisite_info = {
-            task_node.label: PrerequisiteInfo(task_node, self._pipeline_graph)
-            for task_node in pipeline_graph.tasks.values()
-        }
+        with self.butler.registry.caching_context():
+            self._pipeline_graph.resolve(self.butler.registry)
+            self._find_empty_dimension_datasets()
+            self.prerequisite_info = {
+                task_node.label: PrerequisiteInfo(task_node, self._pipeline_graph)
+                for task_node in pipeline_graph.tasks.values()
+            }
 
     log: LsstLogAdapter
     """Logger to use for all quantum-graph generation messages.
@@ -339,42 +340,44 @@ def build(self, metadata: Mapping[str, Any] | None = None) -> QuantumGraph:
         call this method exactly once.  See class documentation for details on
         what it does.
         """
-        full_skeleton = QuantumGraphSkeleton(self._pipeline_graph.tasks)
-        subgraphs = list(self._pipeline_graph.split_independent())
-        for i, subgraph in enumerate(subgraphs):
-            self.log.info(
-                "Processing pipeline subgraph %d of %d with %d task(s).",
-                i + 1,
-                len(subgraphs),
-                len(subgraph.tasks),
-            )
-            self.log.verbose("Subgraph tasks: [%s]", ", ".join(label for label in subgraph.tasks))
-            subgraph_skeleton = self.process_subgraph(subgraph)
-            full_skeleton.update(subgraph_skeleton)
-        # Loop over tasks.  The pipeline graph must be topologically sorted,
-        # so a quantum is only processed after any quantum that provides its
-        # inputs has been processed.
-        for task_node in self._pipeline_graph.tasks.values():
-            self._resolve_task_quanta(task_node, full_skeleton)
-        # Add global init-outputs to the skeleton.
-        for dataset_type in self._global_init_output_types.values():
-            dataset_key = full_skeleton.add_dataset_node(
-                dataset_type.name, self.empty_data_id, is_global_init_output=True
-            )
-            ref = self.existing_datasets.outputs_in_the_way.get(dataset_key)
-            if ref is None:
-                ref = DatasetRef(dataset_type, self.empty_data_id, run=self.output_run)
-            full_skeleton[dataset_key]["ref"] = ref
-        # Remove dataset nodes with no edges that are not global init outputs,
-        # which are generally overall-inputs whose original quanta end up
-        # skipped or with no work to do (we can't remove these along with the
-        # quanta because no quantum knows if its the only consumer).
-        full_skeleton.remove_orphan_datasets()
-        self._attach_datastore_records(full_skeleton)
-        # TODO initialize most metadata here instead of in ctrl_mpexec.
-        if metadata is None:
-            metadata = {}
-        return self._construct_quantum_graph(full_skeleton, metadata)
+        with self.butler.registry.caching_context():
+            full_skeleton = QuantumGraphSkeleton(self._pipeline_graph.tasks)
+            subgraphs = list(self._pipeline_graph.split_independent())
+            for i, subgraph in enumerate(subgraphs):
+                self.log.info(
+                    "Processing pipeline subgraph %d of %d with %d task(s).",
+                    i + 1,
+                    len(subgraphs),
+                    len(subgraph.tasks),
+                )
+                self.log.verbose("Subgraph tasks: [%s]", ", ".join(label for label in subgraph.tasks))
+                subgraph_skeleton = self.process_subgraph(subgraph)
+                full_skeleton.update(subgraph_skeleton)
+            # Loop over tasks.  The pipeline graph must be topologically
+            # sorted, so a quantum is only processed after any quantum that
+            # provides its inputs has been processed.
+            for task_node in self._pipeline_graph.tasks.values():
+                self._resolve_task_quanta(task_node, full_skeleton)
+            # Add global init-outputs to the skeleton.
+            for dataset_type in self._global_init_output_types.values():
+                dataset_key = full_skeleton.add_dataset_node(
+                    dataset_type.name, self.empty_data_id, is_global_init_output=True
+                )
+                ref = self.existing_datasets.outputs_in_the_way.get(dataset_key)
+                if ref is None:
+                    ref = DatasetRef(dataset_type, self.empty_data_id, run=self.output_run)
+                full_skeleton[dataset_key]["ref"] = ref
+            # Remove dataset nodes with no edges that are not global init
+            # outputs, which are generally overall-inputs whose original quanta
+            # end up skipped or with no work to do (we can't remove these along
+            # with the quanta because no quantum knows if its the only
+            # consumer).
+            full_skeleton.remove_orphan_datasets()
+            self._attach_datastore_records(full_skeleton)
+            # TODO initialize most metadata here instead of in ctrl_mpexec.
+            if metadata is None:
+                metadata = {}
+            return self._construct_quantum_graph(full_skeleton, metadata)
 
     @abstractmethod
     def process_subgraph(self, subgraph: PipelineGraph) -> QuantumGraphSkeleton: