[flight-booking-agent] Multi-turn chat and message injection (variant A)

flight-booking-app/app/api/chat/[id]/route.ts

@@ -0,0 +1,30 @@
+import { chatMessageHook } from '@/workflows/chat/hooks/chat-message';
+
+/**
+ * POST /api/chat/[id]
+ *
+ * Sends a follow-up message to an existing chat session.
+ * The [id] parameter is the workflow run ID returned when starting the session.
+ */
+export async function POST(
+  req: Request,
+  { params }: { params: Promise<{ id: string }> }
+) {
+  const { id: runId } = await params;
+  const { message } = await req.json();
+
+  console.log(`Follow-up for workflow: ${runId} message: ${message}`);
+
+  try {
+    // Resume the hook using the run ID as the token
+    // This injects the message into the workflow
+    await chatMessageHook.resume(runId, { message });
+    return Response.json({ success: true });
+  } catch (error) {
+    console.error(`Error resuming hook: ${runId}`, error);
+    return Response.json(
+      { error: 'Failed to send message', details: String(error) },
+      { status: 500 }
+    );
+  }
+}

flight-booking-app/app/api/chat/route.ts

@@ -2,11 +2,20 @@ import { createUIMessageStreamResponse, type UIMessage } from 'ai';
 import { start } from 'workflow/api';
 import { chat } from '@/workflows/chat';
 
-// Uncomment to simulate a long running Vercel Function timing
-// out due to a long running agent. The client-side will
-// automatically reconnect to the stream.
-//export const maxDuration = 8;
+/**
+ * Uncomment the following line to simulate a long running
+ * Vercel Function timing out due to a long running agent.
+ * The client-side will automatically reconnect to the stream.
+ */
+// export const maxDuration = 8;
 
+/**
+ * POST /api/chat
+ *
+ * Starts a new multi-turn chat session. The workflow will handle all subsequent
+ * turns via the message hook - the client should use the run ID returned in
+ * the `x-workflow-run-id` header to send follow-up messages.
+ */
 export async function POST(req: Request) {
   const { messages }: { messages: UIMessage[] } = await req.json();
 
@@ -16,8 +25,9 @@ export async function POST(req: Request) {
   return createUIMessageStreamResponse({
     stream: workflowStream,
     headers: {
-      // The workflow run ID is stored into `localStorage` on the client side,
-      // which influences the `resume` flag in the `useChat` hook.
+      // The workflow run ID is used by the client to:
+      // 1. Send follow-up messages via POST /api/chat/[id]
+      // 2. Reconnect to the stream via GET /api/chat/[id]/stream
       'x-workflow-run-id': run.runId,
     },
   });