fix: use Core PerformAsync for agent monitoring — app stays alive

The monitoring goroutine that waits for agent completion was a raw `go func()`
that Core didn't know about. ServiceShutdown killed it immediately on CLI exit.

Now uses PerformAsync which registers with Core's WaitGroup:
- ServiceShutdown waits for all async tasks to drain
- `core-agent workspace dispatch` blocks until agent completes
- Agent lifecycle properly tracked by the framework

Also whitelist agentic.monitor.* and agentic.complete in entitlement checker.

Co-Authored-By: Virgil <virgil@lethean.io>

pkg/agentic/dispatch.go

@@ -383,11 +383,16 @@ func (s *PrepSubsystem) spawnAgent(agent, prompt, wsDir string) (int, string, er
 	s.broadcastStart(agent, wsDir)
 	s.startIssueTracking(wsDir)
 
-	go func() {
+	// Register a one-shot Action that monitors this agent, then run it via PerformAsync.
+	// PerformAsync tracks it in Core's WaitGroup — ServiceShutdown waits for it.
+	monitorAction := core.Concat("agentic.monitor.", core.PathBase(wsDir))
+	s.Core().Action(monitorAction, func(_ context.Context, _ core.Options) core.Result {
 		<-proc.Done()
 		s.onAgentComplete(agent, wsDir, outputFile,
 			proc.Info().ExitCode, string(proc.Info().Status), proc.Output())
-	}()
+		return core.Result{OK: true}
+	})
+	s.Core().PerformAsync(monitorAction, core.NewOptions())
 
 	return pid, outputFile, nil
 }

pkg/agentic/prep.go

@@ -105,7 +105,10 @@ func (s *PrepSubsystem) OnStartup(ctx context.Context) core.Result {
 		if !core.HasPrefix(action, "agentic.") {
 			return core.Entitlement{Allowed: true, Unlimited: true}
 		}
-		// Read-only actions always allowed
+		// Read-only + internal actions always allowed
+		if core.HasPrefix(action, "agentic.monitor.") || core.HasPrefix(action, "agentic.complete") {
+			return core.Entitlement{Allowed: true, Unlimited: true}
+		}
 		switch action {
 		case "agentic.status", "agentic.scan", "agentic.watch",
 			"agentic.issue.get", "agentic.issue.list", "agentic.pr.get", "agentic.pr.list",