diff --git a/.jules/bolt.md b/.jules/bolt.md
index 7b58f99..64cf4d1 100644
--- a/.jules/bolt.md
+++ b/.jules/bolt.md
@@ -2,14 +2,14 @@
 **Learning:** Found nested loops inside the `OverviewTab` render cycle where `.find()` was being used inside a `.map()` to lookup nodes from a graph diff, turning it into an O(N*M) lookup.
 **Action:** Always verify if `.find()` or `.filter()` inside loops can be hoisted or converted into an O(1) hash map or `Map` using `useMemo` when looking up state in lists.
 
-## 2024-03-30 - Deferred Value for Tree Filtering
-**Learning:** In highly nested or complex recursive component structures (like `SmartFileExplorer` which recursively filters a repository file tree), real-time filtering tied to user input blocks the main React rendering thread, leading to noticeable typing latency.
-**Action:** Use React's `useDeferredValue` hook on the search input query. This allows React to process the keystroke updates synchronously while rendering the expensive filtered list at a lower priority, keeping UI interactions snappy.
-
 ## 2023-11-10 - [O(N*M) Loop Optimization in AnalysisTab]
 **Learning:** Found nested loops inside the `AnalysisTab` render cycle where `.find()` was being used inside a `.map()` to lookup nodes from a graph diff, turning it into an O(N*M) lookup.
 **Action:** Replaced `.find()` inside loops with an O(1) `Map` lookup using `useMemo` to create a `nodeById` map mapping node IDs to nodes. This is a recurring pattern in the visualization codebase.
 
+## 2024-03-30 - Deferred Value for Tree Filtering
+**Learning:** In highly nested or complex recursive component structures (like `SmartFileExplorer` which recursively filters a repository file tree), real-time filtering tied to user input blocks the main React rendering thread, leading to noticeable typing latency.
+**Action:** Use React's `useDeferredValue` hook on the search input query. This allows React to process the keystroke updates synchronously while rendering the expensive filtered list at a lower priority, keeping UI interactions snappy.
+
 ## 2024-11-20 - [O(N*M) Loop Optimization in LearningPathTab]
 **Learning:** Found nested loops inside the `LearningPathTab` render cycle event handlers (`onKeyDown`, `onClick`) where array `.find()` was being used inside a `.map()` to lookup target node IDs, turning it into an O(N*M) time complexity bottleneck for large codebases.
 **Action:** Replaced `.find()` inside the `.map()` loop with an O(1) `Map` lookup by pre-computing a `nodeById` map mapping node IDs to node IDs using `useMemo` at the component level. This reinforces the pattern of using memoized Hash Maps for list lookups in React components.
@@ -26,3 +26,7 @@
 **Action:** Replace multiple `.filter()` passes with a single `for` loop pass over the array utilizing a mutable tally counter struct, reducing iteration overhead and allocations to $O(N)$ exactly.
 
 - In backend data processing pipelines, avoid nested O(N) array operations (like `.find()` inside `.filter()`). Pre-compute a `Map` keyed by unique identifiers (e.g., `path`) for O(1) lookups to optimize performance and prevent O(N*M) bottlenecks. For example, converting array lookups inside incremental indexing mapping logic.
+
+## 2024-11-20 - [O(N*M) Loop Optimization in groupDependencies]
+**Learning:** Found nested loops inside the `groupDependencies` algorithm where array `.filter()` was being used inside a `for` loop to match `scopedDependencies` to `dependencies` based on a composite string key, creating an O(N*M) time complexity bottleneck for large codebases and dependency trees.
+**Action:** Replaced `.filter()` inside the loop with an O(1) `Map` lookup by pre-computing a map grouping `scopedDependencies` by their keys. This eliminates the inner O(M) loop, speeding it up significantly (reducing time complexity to O(N+M)).
diff --git a/server/services/dependency-health.ts b/server/services/dependency-health.ts
index a4c2593..945e494 100644
--- a/server/services/dependency-health.ts
+++ b/server/services/dependency-health.ts
@@ -53,6 +53,21 @@ function groupDependencies(
 ): readonly DependencyGroup[] {
   const grouped = new Map<DependencyKey, DependencyGroup>();
 
+  // Performance Optimization (Bolt):
+  // Pre-compute a Map of scoped dependencies grouped by dependencyKey.
+  // This eliminates the need for an O(M) .filter() array lookup inside the
+  // O(N) loop below, reducing time complexity from O(N*M) to O(N+M).
+  const scopedByDepKey = new Map<DependencyKey, ScopedDependency[]>();
+  for (const scoped of scopedDependencies) {
+    const key = dependencyKey(scoped);
+    let arr = scopedByDepKey.get(key);
+    if (!arr) {
+      arr = [];
+      scopedByDepKey.set(key, arr);
+    }
+    arr.push(scoped);
+  }
+
   for (const dependency of dependencies) {
     const key = dependencyKey(dependency);
     const existing = grouped.get(key);
@@ -60,7 +75,7 @@ function groupDependencies(
 
     grouped.set(key, {
       dependency,
-      scopedDependencies: scopedDependencies.filter((scopedDependency) => dependencyKey(scopedDependency) === key),
+      scopedDependencies: scopedByDepKey.get(key) || [],
     });
   }