feat: add sticky round-robin routing strategy

Implements a "sticky" round-robin strategy that uses the same provider
account for a configurable number of consecutive calls (default 3)
before switching to the next one. This optimizes for prompt caching
by reducing organization/account rotation. Adds a configuration input
to the Profile settings page.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

src/app/(dashboard)/dashboard/profile/page.js

@@ -76,6 +76,24 @@ export default function ProfilePage() {
     }
   };
 
+  const updateStickyLimit = async (limit) => {
+    const numLimit = parseInt(limit);
+    if (isNaN(numLimit) || numLimit < 1) return;
+
+    try {
+      const res = await fetch("/api/settings", {
+        method: "PATCH",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ stickyRoundRobinLimit: numLimit }),
+      });
+      if (res.ok) {
+        setSettings(prev => ({ ...prev, stickyRoundRobinLimit: numLimit }));
+      }
+    } catch (err) {
+      console.error("Failed to update sticky limit:", err);
+    }
+  };
+
   return (
     <div className="max-w-2xl mx-auto">
       <div className="flex flex-col gap-6">
@@ -165,9 +183,31 @@ export default function ProfilePage() {
                 disabled={loading}
               />
             </div>
+
+            {/* Sticky Round Robin Limit */}
+            {settings.fallbackStrategy === "round-robin" && (
+              <div className="flex items-center justify-between pt-2 border-t border-border/50">
+                <div>
+                  <p className="font-medium">Sticky Limit</p>
+                  <p className="text-sm text-text-muted">
+                    Calls per account before switching
+                  </p>
+                </div>
+                <Input
+                  type="number"
+                  min="1"
+                  max="10"
+                  value={settings.stickyRoundRobinLimit || 3}
+                  onChange={(e) => updateStickyLimit(e.target.value)}
+                  disabled={loading}
+                  className="w-20 text-center"
+                />
+              </div>
+            )}
+
             <p className="text-xs text-text-muted italic pt-2 border-t border-border/50">
               {settings.fallbackStrategy === "round-robin"
-                ? "Currently distributing requests across all available accounts."
+                ? `Currently distributing requests across all available accounts with ${settings.stickyRoundRobinLimit || 3} calls per account.`
                 : "Currently using accounts in priority order (Fill First)."}
             </p>
           </div>

src/lib/localDb.js

@@ -39,7 +39,8 @@ const defaultData = {
   combos: [],
   apiKeys: [],
   settings: {
-    cloudEnabled: false
+    cloudEnabled: false,
+    stickyRoundRobinLimit: 3
   }
 };
 
@@ -178,7 +179,8 @@ export async function createProviderConnection(data) {
     "displayName", "email", "globalPriority", "defaultModel",
     "accessToken", "refreshToken", "expiresAt", "tokenType",
     "scope", "idToken", "projectId", "apiKey", "testStatus",
-    "lastTested", "lastError", "lastErrorAt", "rateLimitedUntil", "expiresIn", "errorCode"
+    "lastTested", "lastError", "lastErrorAt", "rateLimitedUntil", "expiresIn", "errorCode",
+    "consecutiveUseCount"
   ];
   
   for (const field of optionalFields) {
@@ -470,7 +472,8 @@ export async function cleanupProviderConnections() {
     "displayName", "email", "globalPriority", "defaultModel",
     "accessToken", "refreshToken", "expiresAt", "tokenType",
     "scope", "idToken", "projectId", "apiKey", "testStatus",
-    "lastTested", "lastError", "lastErrorAt", "rateLimitedUntil", "expiresIn"
+    "lastTested", "lastError", "lastErrorAt", "rateLimitedUntil", "expiresIn",
+    "consecutiveUseCount"
   ];
 
   let cleaned = 0;

src/sse/services/auth.js

@@ -33,17 +33,44 @@ export async function getProviderCredentials(provider, excludeConnectionId = nul
 
   let connection;
   if (strategy === "round-robin") {
-    // Sort by lastUsed (nulls first) to pick the least recently used
+    const stickyLimit = settings.stickyRoundRobinLimit || 3;
-    const sorted = [...availableConnections].sort((a, b) => {
+
+    // Sort by lastUsed (most recent first) to find current candidate
+    const byRecency = [...availableConnections].sort((a, b) => {
+      if (!a.lastUsedAt && !b.lastUsedAt) return (a.priority || 999) - (b.priority || 999);
+      if (!a.lastUsedAt) return 1;
+      if (!b.lastUsedAt) return -1;
+      return new Date(b.lastUsedAt) - new Date(a.lastUsedAt);
+    });
+
+    const current = byRecency[0];
+    const currentCount = current?.consecutiveUseCount || 0;
+
+    if (current && current.lastUsedAt && currentCount < stickyLimit) {
+      // Stay with current account
+      connection = current;
+      // Update lastUsedAt and increment count
+      updateProviderConnection(connection.id, {
+        lastUsedAt: new Date().toISOString(),
+        consecutiveUseCount: (connection.consecutiveUseCount || 0) + 1
+      }).catch(() => {});
+    } else {
+      // Pick the least recently used (excluding current if possible)
+      const sortedByOldest = [...availableConnections].sort((a, b) => {
         if (!a.lastUsedAt && !b.lastUsedAt) return (a.priority || 999) - (b.priority || 999);
         if (!a.lastUsedAt) return -1;
         if (!b.lastUsedAt) return 1;
         return new Date(a.lastUsedAt) - new Date(b.lastUsedAt);
       });
-    connection = sorted[0];
 
-    // Update lastUsedAt asynchronously
+      connection = sortedByOldest[0];
-    updateProviderConnection(connection.id, { lastUsedAt: new Date().toISOString() }).catch(() => {});
+
+      // Update lastUsedAt and reset count to 1
+      updateProviderConnection(connection.id, {
+        lastUsedAt: new Date().toISOString(),
+        consecutiveUseCount: 1
+      }).catch(() => {});
+    }
   } else {
     // Default: fill-first (already sorted by priority in getProviderConnections)
     connection = availableConnections[0];