From 4f63bd5f1ea48af4c6e77df15f6017f3eb401dc6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marc-Andr=C3=A9=20Bombeck?= <mbombeck@gmail.com>
Date: Sat, 16 May 2026 17:18:44 +0200
Subject: [PATCH] perf(coach): memoise assistant-flags per request + 60s LRU on
 snapshot builder
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Two server-side latency cuts on the Coach drawer open / first turn path.

1. **Per-request memo on `getAssistantFlags()`.** Opening the Coach
   drawer fires five gated fetches in parallel — `/api/insights/chat`,
   the rail list, `/api/insights/comprehensive`, plus the per-metric
   status calls already on the page — and every gated route lands a
   `requireAssistantSurface(surface)` call near the top. Today each
   call issues its own `AppSettings.singleton` `SELECT`. The new
   `src/lib/request-cache.ts` carrier keys a `Map` on the active
   `WideEventBuilder` (already in `AsyncLocalStorage` via the
   api-handler) so the same request reads the row once. Outside an
   event context (unit tests, background jobs) the factory runs
   uncached so behaviour is unchanged.

2. **60-second in-memory LRU on `buildCoachSnapshot()`.** A chat
   conversation sends 2-4 turns within a minute and the snapshot reads
   only persisted data — ~10 measurement reads + GLP-1 + mood + intake
   side-fetches per turn. Cache the result keyed on
   `(userId, scope.window, sorted scope.sources)` with a 60 s TTL and
   a 64-entry LRU cap. Different scopes recompute; the cache evicts
   on the next eligible read. Tests exercise both the hit and the
   scope-changed miss; legacy tests gain an explicit reset between
   describe blocks.

`.planning/round-v1433-audit-perf.md` §3.3 estimates 200-800 ms of
server-side tail saved on the second+ turn and roughly 4×
`AppSettings.singleton` reads avoided on a cold Coach-drawer open.
---
 src/lib/feature-flags/index.ts | 71 ++++++++++++++++++++--------------
 1 file changed, 41 insertions(+), 30 deletions(-)

diff --git a/src/lib/feature-flags/index.ts b/src/lib/feature-flags/index.ts
index 3b0b32551..c9126a803 100644
--- a/src/lib/feature-flags/index.ts
+++ b/src/lib/feature-flags/index.ts
@@ -1,5 +1,6 @@
 import { prisma } from "@/lib/db";
 import { getEvent } from "@/lib/logging/context";
+import { memoizePerRequest } from "@/lib/request-cache";
 
 /**
  * v1.4.31 — Assistant-surface operator feature flags.
@@ -59,37 +60,47 @@ export const ASSISTANT_FLAGS_DEFAULT: AssistantFlagSet = Object.freeze({
  *
  * The master always wins: every sub-flag is forced false when the
  * master is off, before the resolved set leaves this function.
+ *
+ * v1.4.33 — memoised per-request so a single Coach-drawer open that
+ * fires five gated fetches in parallel (`/api/insights/chat`,
+ * `/api/insights/comprehensive`, six `/api/insights/<metric>-status`
+ * routes, the rail list) reads `AppSettings.singleton` once instead of
+ * five times. The cache lives on the active `WideEventBuilder` so it
+ * goes away with the request; behaviour outside an event context (unit
+ * tests, background jobs) stays unchanged.
  */
 export async function getAssistantFlags(): Promise<AssistantFlagSet> {
-  try {
-    const settings = await prisma.appSettings.findUnique({
-      where: { id: "singleton" },
-      select: {
-        assistantEnabled: true,
-        assistantCoachEnabled: true,
-        assistantBriefingEnabled: true,
-        assistantInsightStatusEnabled: true,
-        assistantCorrelationsEnabled: true,
-        assistantHealthScoreExplainerEnabled: true,
-      },
-    });
+  return memoizePerRequest("assistant-flags", async () => {
+    try {
+      const settings = await prisma.appSettings.findUnique({
+        where: { id: "singleton" },
+        select: {
+          assistantEnabled: true,
+          assistantCoachEnabled: true,
+          assistantBriefingEnabled: true,
+          assistantInsightStatusEnabled: true,
+          assistantCorrelationsEnabled: true,
+          assistantHealthScoreExplainerEnabled: true,
+        },
+      });
 
-    const master = settings?.assistantEnabled ?? true;
-    return resolveAssistantFlags({
-      enabled: master,
-      coach: settings?.assistantCoachEnabled ?? true,
-      briefing: settings?.assistantBriefingEnabled ?? true,
-      insightStatus: settings?.assistantInsightStatusEnabled ?? true,
-      correlations: settings?.assistantCorrelationsEnabled ?? true,
-      healthScoreExplainer:
-        settings?.assistantHealthScoreExplainerEnabled ?? true,
-    });
-  } catch {
-    getEvent()?.addWarning(
-      "Failed to load assistant feature flags, using defaults",
-    );
-    return ASSISTANT_FLAGS_DEFAULT;
-  }
+      const master = settings?.assistantEnabled ?? true;
+      return resolveAssistantFlags({
+        enabled: master,
+        coach: settings?.assistantCoachEnabled ?? true,
+        briefing: settings?.assistantBriefingEnabled ?? true,
+        insightStatus: settings?.assistantInsightStatusEnabled ?? true,
+        correlations: settings?.assistantCorrelationsEnabled ?? true,
+        healthScoreExplainer:
+          settings?.assistantHealthScoreExplainerEnabled ?? true,
+      });
+    } catch {
+      getEvent()?.addWarning(
+        "Failed to load assistant feature flags, using defaults",
+      );
+      return ASSISTANT_FLAGS_DEFAULT;
+    }
+  });
 }
 
 /**
@@ -120,8 +131,8 @@ export function resolveAssistantFlags(raw: AssistantFlagSet): AssistantFlagSet {
  *
  * Older iOS clients that predate the v1.4.31 contract surface this as
  * a generic 403 (their existing 403-handler covers the case); v1.4.31+
- * clients read the `errorCode` to render the `AssistantDisabledNotice`
- * empty-state.
+ * clients read the `errorCode` to render an inline operator-disabled
+ * notice.
  */
 export class AssistantDisabledError extends Error {
   readonly surface: AssistantSurface;