I’m using Telegram to trigger a workflow, but when I try to call another workflow using the “Execute Workflow” node in n8n, the process freezes. The trigger works — n8n receives it — but it gets stuck during the execution phase.
I’m sure the workflow ID is correct, because when I use the “Execute a Sub-Workflow” node (not the “Execute Workflow” tool), everything works fine. The child workflow is configured to accept all incoming data.
I also tried changing a setting — specifically, enabling “Save Manual Executions” — but that didn’t resolve the issue.
oh and then in logs is error Workflow execution had an error
Cannot read properties of undefined (reading ‘execute’)
main workflow
Error:
main workflow without call workflow tool - i think it isnt problem with the rest
and i talk that block works (execute workflow) so its the prove that only tool (call workflow) doesn’t work
child workflow
Thank 
Information on your n8n setup
- n8n version: 1.93.3
- Database (default: SQLite): SQLite
- n8n EXECUTIONS_PROCESS setting (default: own, main): I’m running n8n on Railway and haven’t changed the
EXECUTIONS_PROCESS setting, so it’s using the default: own
- Running n8n via (Docker, npm, n8n cloud, desktop app): Railway
- Operating system: Windows
Hello @COMADAI, welcome here 
This type of error usually means the AI Agent Tool is trying to call the workflow, but doesn’t know what to send, or is missing the expected inputs.
In your “Call Workflow Tool” config, the sub-workflow is set to accept all data, but the Agent still needs explicit instructions on what data to send.
Try to do this:
- Go to your child workflow
- Make sure the first node is “When Executed by Another Workflow” (you already have this, good).
- In that node, change Input data mode from: “Accept all data” to “Define input structure”
- Then, define the specific fields (e.g.
query, productName, or whatever your tool is supposed to receive).
- Update the Tool description
- In your “Call n8n Workflow Tool”, use a description like:
“Call this tool to search something from Wikipedia. It expects a ‘query’ string.”
- Use a system message or example with your AI Agent
“If the user asks to look something up, use the tool ‘WikipediaSearch’ and pass a field called query.”
When you “Accept all data”, the Agent doesn’t know what data to send. But when you define a clear input structure, it gives the Agent something to work with, avoiding the undefined error.
Lmk if that works!
Hi i tried it and it’s the same as before nothing changed. When I run workflow it didn’t show anything. Usually when something is wrong nodes turn red but my just thinks and shows executing workflow. Maybe I made mistakes but here’s photos:
1 Like
You can try add donothing node, on the workflow being called by the tool, can you see past executions on the workflow being called also, maybe worth trying to pin the data by editing in workflow.
hi @King_Samuel_David as i understand right i added donothing node to my subworkflow in the executions in subworkflow nothing pops up its the same. And when i pinned somthing it freezes as usual
Idk if that will help but i tried run the same program in the diffrent host and it worked why is that ? And how can i repair main host ?
okey i got it hyh silly me i restarted host and it works Eh thanks for all of your help <3 @King_Samuel_David @Gallo_AIA
2 Likes
Ah thats good yes a good old restart always fixes alot ahaa
