gateway: wire channel health monitor into startup with configurable interval

src/config/types.gateway.ts

@@ -312,4 +312,10 @@ export type GatewayConfig = {
   trustedProxies?: string[];
   /** Tool access restrictions for HTTP /tools/invoke endpoint. */
   tools?: GatewayToolsConfig;
+  /**
+   * Channel health monitor interval in minutes.
+   * Periodically checks channel health and restarts unhealthy channels.
+   * Set to 0 to disable. Default: 5.
+   */
+  channelHealthCheckMinutes?: number;
 };

src/gateway/server.impl.ts

@@ -48,6 +48,7 @@ import { createEmptyPluginRegistry } from "../plugins/registry.js";
 import { getTotalQueueSize } from "../process/command-queue.js";
 import { runOnboardingWizard } from "../wizard/onboarding.js";
 import { createAuthRateLimiter, type AuthRateLimiter } from "./auth-rate-limit.js";
+import { startChannelHealthMonitor } from "./channel-health-monitor.js";
 import { startGatewayConfigReloader } from "./config-reload.js";
 import { ExecApprovalManager } from "./exec-approval-manager.js";
 import { NodeRegistry } from "./node-registry.js";
@@ -502,6 +503,15 @@ export async function startGatewayServer(
       }
     : startHeartbeatRunner({ cfg: cfgAtStart });
 
+  const healthCheckMinutes = cfgAtStart.gateway?.channelHealthCheckMinutes;
+  const healthCheckDisabled = healthCheckMinutes === 0;
+  const channelHealthMonitor = healthCheckDisabled
+    ? null
+    : startChannelHealthMonitor({
+        channelManager,
+        checkIntervalMs: (healthCheckMinutes ?? 5) * 60_000,
+      });
+
   if (!minimalTestGateway) {
     void cron.start().catch((err) => logCron.error(`failed to start: ${String(err)}`));
   }
@@ -720,6 +730,7 @@ export async function startGatewayServer(
       }
       skillsChangeUnsub();
       authRateLimiter?.dispose();
+      channelHealthMonitor?.stop();
       await close(opts);
     },
   };