Fixing the AI Agent with Simple Memory in n8n

I see you're encountering the same issue with the AI Agent node that's using the Simple Memory sub-node. This is a common configuration challenge when setting up AI workflows in n8n.

Understanding the Connection

The error occurs because:

1. The AI Agent node uses Simple Memory as a sub-node to maintain conversation context
2. The Simple Memory sub-node can't find a sessionId in the incoming data
3. The {{ $json.chatInput }} expression is undefined, indicating data structure issues

Complete Solution for AI Agent + Simple Memory

Let's create a comprehensive solution that fixes both nodes:

1. Modify Your Frontend to Include Required Fields

2. Configure Your n8n Workflow Correctly

Here's how to set up your n8n workflow to work with the AI Agent and Simple Memory:

Step 1: Add a Function Node Before AI Agent

Add a Function node to ensure all required fields are present:

Step 2: Configure the AI Agent Node

1. Set "Source for Prompt (User Message)" to "Expression"
2. Enter this expression: {{ $json.chatInput }}

3. Under the AI Agent settings, make sure Simple Memory is properly configured:

   • Set "Session ID" to "Expression"
   • Enter this expression: {{ $json.sessionId }}

Step 3: Add a Response Handler

Add a "Set" node after the AI Agent to format the response:

Then connect this to a "Respond to Webhook" node.

Testing Your Solution

1. Deploy the updated frontend
2. Configure your n8n workflow as described
3. Send a test message from the frontend
4. Check if both the Simple Memory and AI Agent nodes execute without errors

Troubleshooting Tips

If you still encounter issues:

1. Check Data Structure: Use a "Debug" node after the webhook to see exactly what data is coming in
2. Verify Expressions: Make sure all expressions in n8n are correctly referencing the right fields
3. Session ID Consistency: Ensure the same sessionId is used across all messages in a conversation
4. Memory Context: Check if the Context Window Length in Simple Memory is appropriate (5-10 is usually good)

This solution provides a complete chat interface that works with n8n's AI Agent and Simple Memory nodes, ensuring all the required fields are present and properly formatted.

Fixing the Response Stream Error

The error occurs because we're trying to read the response body stream twice. Let's fix the sendToWebhook function to properly handle the response:

Explanation of the Fix

The error "body stream already read" occurs because we were trying to read the response body twice:

1. First with response.json()
2. Then with response.text() if the JSON parsing failed

In JavaScript, response streams from fetch can only be consumed once. The fix:

1. We now read the response as text first with await response.text()
2. Then we try to parse that text as JSON using JSON.parse()
3. If parsing succeeds, we extract the AI response from the JSON object
4. If parsing fails, we use the raw text as the response

This approach ensures we only read the response stream once, avoiding the "body stream already read" error.

Testing the Solution

To test this solution:

1. Make sure your n8n workflow is running
2. Enter a valid webhook URL in the form
3. Send a message
4. The response should now be properly processed without errors

Additional Improvements

For even more robust error handling, you might want to:

1. Add logging to help debug issues
2. Implement retry logic for failed requests
3. Add timeout handling for long-running requests

Here's a simple test you can run to verify your webhook is working correctly:

This fix should resolve the "body stream already read" error and allow your application to properly communicate with the n8n webhook.