Basic LLM Chain issue "Prompt (User Message) is required" even though I filled it in

Questions
1

Hi everyone,
This is probably a dumb beginner question since i’m new to n8n.
I’m using the Basic LLM Chain node with OpenAI GPT 5-mini.

I’m running n8n on my mac with Docker.
I entered a prompt in the “Prompt (User Message)” field, but the node still shows:

  • :warning: Issues: Parameter “Prompt (User Message)” is required
    Same when entering a second “User” and “AI” prompt:
  • :warning: Issues: Parameter “Message” is required.
    And when I close and reopen the node, it also looks like my changes weren’t saved cause the additional “user” and “ai” prompts have disappeared. The workflow still runs correctly.
    I’m probably missing something obvious. Does somebody knows what I might be doing wrong?
    Thanks!

Capture d’écran 2026-07-29 à 11.38.27
Capture d’écran 2026-07-29 à 11.38.272207×1203 475 KB

Edit: Thanks for your responses, i just cleared cache using cmd+shift+R and now it works but I have another issues, the main prompt (user message) shoes an issue saying that it isn’t supported, and now i can’t run my workflow.

2

Hey @nathan_gerber, while you wait for a response, here are some things that might help:

Suggested resources

Automatically matched to your question.

Docs:

Forum:

@mohamed3nan, @barn4k - you’ve helped with similar issues before, can you take a look?

Automatically suggested by n8n’s community bot. It’s a pilot - please share feedback here.

3

Hey, this is a known n8n quirk — the field shows text but doesn’t actually register it until it loses focus. That’s also why your prompts vanish on reopen: the value never saved in the first place.

Try clicking elsewhere in the node before closing it, and type instead of paste (pasted text sometimes carries hidden characters that break it).

4

@nathan_gerber

The Basic LLM Chain node is designed with a legacy schema where the Prompt (User Message) field at the top is marked as “Required” in the UI. Even if you provide a full conversation history in the “Chat Messages” section, the UI validator still checks that top field.

  • The Fix: Enter a simple space or a generic placeholder (e.g., See chat messages below) in the top Prompt (User Message) field. This will silence the warning without affecting the output, as the LLM will prioritize the Chat Messages array.

Looking at your screenshot, your prompts are set to Expression mode (indicated by the fx icon and the red warning triangle).

  • The Root Cause: When a field is in Expression mode, n8n tries to evaluate the result. If the expression is just static text but wrapped in an expression block, or if it refers to data from a previous node that hasn’t been “fetched” in the current editor session, the validator may perceive the result as undefined or empty, triggering the warning.
  • The Fix: If the text is static (not pulling data from a previous node), click the fx icon to switch the field back to Fixed mode. If you are using expressions, ensure the previous node (the Google Sheet node) has been executed recently so the editor has data to validate against.

This is a known UI bug in the Basic LLM Chain node. When you mix Expression and Fixed values within the “Chat Messages” array, the UI sometimes fails to re-render the list upon reopening the node, even though the data is safely stored in the workflow’s JSON.

  • The Fix:
    1. Ensure all prompts in the list are using the same mode (preferably Fixed for static instructions).
    2. Better Alternative: Switch from the Basic LLM Chain node to the AI Agent node.

Switching to the AI Agent node is my recommendation because:

  1. Robust Message Handling: It handles System/User/AI messages much more reliably.
  2. Memory: It allows you to attach a “Window Buffer Memory” node so the AI remembers previous interactions.
  3. Tools: You can give the agent “Tools” (like a Google Search tool or a custom Workflow tool) to perform actions instead of just generating text.
5

Thanks for letting us know about this, We have created CV-52 as the internal dev ticket to look into it.

6

Thanks for letting us know about this, We have created AI-2680 as the internal dev ticket to look into it.

7

It doesn’t sound like you’re doing anything wrong. Since clearing the cache fixed the original validation issue, this points more toward a frontend/UI state problem than a workflow configuration problem.

For the new “Prompt (User Message) isn’t supported” error, I’d check a few things:

  • Make sure you’re using the latest version of the Basic LLM Chain node. Some recent n8n releases changed AI node parameters, and older workflows can show validation errors after an update.
  • Open the OpenAI Chat Model node and confirm the selected model (GPT-5 mini) is supported by your installed n8n version.
  • Try deleting the Basic LLM Chain node and adding a fresh one. If the node schema has changed, recreating it often resolves validation issues.
  • If you’re running n8n in Docker, make sure your Docker image is up to date and perform a hard refresh after upgrading.

Since you mentioned the workflow previously ran correctly, I’d lean toward a node-version mismatch or a UI validation bug rather than an issue with your prompt itself.

Could you also share:

  • Your n8n version
  • The Docker image tag you’re using
  • A screenshot of the current “Prompt (User Message) isn’t supported” error?

That should make it much easier to determine whether this is a known bug or a configuration issue.

8

Thanks for your response, I had simply written the prompt in “Source for prompt (user message)” by clicking on “expression” instead of putting “Define below” to write it below. And clearing the cache seems to have resolved the rest.