I am getting an error on LLM modules in my AI Agent for whichever model I use, Google’s Gemini or OpenAI’s ChatGPT4.1.
I am sending it a message through slack and part hardcoded and I am sure that it gets it, but it still says that it received a null. This only happens if u use a simple memory module.
The message Cannot read properties of undefined (reading 'content') typically happens when memory node returns a null or malformed content object that the LLM model expects
1. Check the input format
Make sure that the message being passed to the AI Agent (from Slack or any other source) is a simple string or has this structure:
{
"content": "your actual message here"
}
If the message is nested or missing the content field, the model can’t find what to read — and you’ll get the undefined error.
2. Try bypassing memory temporarily
As a test disconnect the Simple Memory node from the AI Agent and try running the Agent with a basic message
If it works without memory, the issue might be how the memory is storing or recalling the data.
forgive me I probably miss a step, but what is the trigger for this flow?
Because in the prompt it seems that a variable is missing to give context to the agent.
when memory returns unexpected or empty structures, some LLMs (especially Gemini) throw errors like this.
Let’s try to do a test if you don’t mind and reconnect only one model and only one memory.
Please remember to ser a session ID for simple memory tool.
from your screenshot you can see what inputs and outputs exist so it would be understandable if you have more memories in the flow.
Outputs show that at least one memory returns valid chatHistory, but thereis probably more than one, and one of these is empty
