I hope that someone can help as I am noob to this.
I have been following NetworkChuck - n8n Now runs mu entire homelab, but I have ran into a wall, I am 10:46 mins when he is talking about adding a workflow to link up to a second flow
within here you choose a subflow call SSH_n8n_server, and it then should show you the commands within that flow but all I am getting is
This sub-workflow is set up to receive all input data, without specific inputs the Agent will not be able to pass data to this tool. You can define specific inputs in the sub-workflow’s trigger.
I have gone back through the video to check that I havn’t missed anything and I have even delete it and recreated it but I am still getting the same error.
Your trigger in the sub-workflow is set to receive all data, which isn’t the case in the video. You have to define the parameter manually first, then you can see it when you add Call n8n Workflow tool.
Navigate to your sub-workflow, enter the trigger node, set th input data mode as Define using fields below and define the command parameter just as shown in the image and save the workflow.
and when I run this I get a command of null, and then wilthin the excute a command I have drag over the input vaiable “command” and drop it into the command box.
If it says null then you may have left it empty in Call n8n Workflow tool, or your agent did not fill this field.
First scenario → Click on the button right next to the parameter on this tool. This lets the AI decide what to fill it with. Make sure to type a good description and mention this parameter in the system prompt of your agent.
Second scenario → Most likely a bad description/unclear system prompt that confused your agent and led to this. Mention exact tool names and parameters(command, Call n8n Workflow) in your system prompt.
ok I have found out one problem within the DOCKER TOOL - and why it was not showing my any resorts and that was due to permissions, so for my installation I have to run everything with the word sudo in front of it, so to over come this issues, I ran the follow command and not the DOCKER TOOL returns the current dockers information
the command is o usermod -aG docker YOURROOTUSERNAME
as this needs to be a member of the docker group.
the other issue, that my AI Agent is not calling the DOCKER TOOL,
Hi, I am still having issues, I think the video was record with a old version of n8n, as everything I have tryied and google, I have ran into issues, where the suggesting is no longer apart of n8n.
I have even tried to link my proxmox to it, and again ran into issues, as the guide says one thing, but the software says something else. if I find a solution then I will post it
I’m having the same issue with the most current version of n8n. When attempting to call the sub workflow there is an error that is appears to show a bug:
The text: “This sub-workflow is set up to receive all input data, without specific inputs the Agent will not be able to pass data to this tool. You can define specific inputs in the sub-workflow’s trigger.” Doesn;t make sense, because the sub-workflows trigger is not in fact not set to “all input” but rather “Define using fields below” which is exactly what Chuck shows in his video.
Either way, regardless of the video - not able to trigger AI ssh commands which effectively halts my use of n8n until that is resolved. Love the idea of automating sysadmin tasks - but that starts with ssh access where we’re blocked.
I have noticed you can work around it by creating a new workflow, and then attach it onto the AI agent tool connection on the first workflow. Just make sure you publish the second workflow.
You need to publish the changes you made to the initial sub-workflow (SSH_n8n_server). Try creating “Call n8n Workflow Tool“ and adding your workflow from the list. It should show command now.
Initial sub-workflow is published when you create it from the canvas. But all changes need to be re-published.
Confirmed! As @mikam6a mentioned, that was exactly the issue. I didn’t realize I had to re-publish the sub-workflow after defining the input fields. Once I did that, the main workflow recognized the inputs immediately. Thanks!
