Update app.js

app.js

@@ -67,16 +67,15 @@ app.post('/api/generate', async (req, res) => {
     }
 });
 
+// --- STREAMING ENDPOINT WITH THINKING SUPPORT ---
 app.post('/api/stream', async (req, res) => {
     const { model, prompt, system_prompt } = req.body;
     console.log(`[STREAM] Request for ${model}`);
 
-    // Headers to force immediate flushing and disable buffering on Nginx/Cloudflare
     res.setHeader('Content-Type', 'text/plain; charset=utf-8');
     res.setHeader('Transfer-Encoding', 'chunked');
     res.setHeader('X-Accel-Buffering', 'no'); 
-    res.setHeader('Cache-Control', 'no-cache, no-transform');
-    res.flushHeaders(); // Ensure headers are sent immediately
+    res.flushHeaders();
 
     try {
         if (model === "claude") {
@@ -92,18 +91,25 @@ app.post('/api/stream', async (req, res) => {
             });
 
             const response = await bedrockClient.send(command);
+            
             for await (const chunk of response.stream) {
                 if (chunk.contentBlockDelta) {
-                    const text = chunk.contentBlockDelta.delta?.text;
-                    if (text) {
-                        res.write(text);
-                        // In some Node setups with compression, we might need manual flushing, 
-                        // but usually res.write works with disabled buffering headers.
+                    const delta = chunk.contentBlockDelta.delta;
+                    
+                    // Handle Bedrock "Thinking" (Reasoning Content)
+                    if (delta.reasoningContent && delta.reasoningContent.text) {
+                        res.write(`__THINK__${delta.reasoningContent.text}`);
+                    }
+                    // Handle Bedrock "Final Output" (Text)
+                    else if (delta.text) {
+                        res.write(delta.text);
                     }
                 }
             }
             res.end();
+
         } else {
+            // Azure/OpenAI Streaming
             const stream = await azureOpenAI.chat.completions.create({
                 model: "gpt-5-mini",
                 messages: [
@@ -115,8 +121,16 @@ app.post('/api/stream', async (req, res) => {
             });
 
             for await (const chunk of stream) {
-                const text = chunk.choices[0]?.delta?.content || "";
-                if (text) res.write(text);
+                const delta = chunk.choices[0]?.delta;
+                
+                // Handle OpenAI "Thinking"
+                if (delta?.reasoning_content) {
+                    res.write(`__THINK__${delta.reasoning_content}`);
+                } 
+                // Handle OpenAI "Final Output"
+                else if (delta?.content) {
+                    res.write(delta.content);
+                }
             }
             res.end();
         }
@@ -126,8 +140,9 @@ app.post('/api/stream', async (req, res) => {
         res.end();
     }
 });
+
 app.post('/', async (req, res) => {
-  res.json({ success: true, status: 200 });
+res.json({ success: true, data: response.choices[0].message.content });
 });
 
-app.listen(PORT, '0.0.0.0', () => console.log(`Main AI Agent live on port ${PORT}`));
+app.listen(PORT, '0.0.0.0', () => console.log(`Main AI Agent live on port ${PORT}`));