I’m having trouble with the MCP Server Memory in n8n. When I try using the add_observation node tool, I get this error:
Failed to execute operation: Failed to execute tool 'add_observations': MCP error -32603: Cannot read properties of undefined (reading 'includes')
I’ve tried using it as a node instead of a tool, but the error still occurs. I’m not sure if there’s a mistake in my JSON or if it’s a bug with the MCP Client. Has anyone else experienced this issue or have any suggestions?
yo, had this same prob. checked my json n realized i forgot to include a field. double check ur data structure, make sure all fields r there even if theyre empty. if that dont work, try clearing ur workflow cache or restarting n8n. sometimes that fixes weird errors like this
I’ve dealt with this issue before, and it can be frustrating. One thing that worked for me was updating the n8n version. Sometimes, these errors are due to compatibility issues with older versions. Also, make sure you’re using the correct API endpoint for the MCP Server Memory operations. I once spent hours debugging only to realize I was using an outdated endpoint. If you’re still stuck, consider logging the input data just before the problematic node. This can help pinpoint exactly where the undefined property is coming from. Lastly, if all else fails, try recreating the workflow from scratch. It’s time-consuming, but it has solved inexplicable issues for me in the past.
I’ve encountered a similar issue with the MCP Server Memory in n8n. From my experience, this error often stems from an undefined property in the input data structure. Have you double-checked that all required fields are properly defined in your JSON payload?
One thing that helped me was to use the ‘JSON’ node before the MCP Server Memory operation to ensure my data structure was correct. I also found that explicitly defining all expected properties, even if they’re null, can sometimes prevent these undefined property errors.
If you’ve already verified your data structure, it might be worth reaching out to n8n support. They were quite helpful when I had a persistent issue and even provided a workaround until a proper fix was implemented in a later update. Don’t hesitate to open a ticket with them if the problem persists.