(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
So, you could categorise the http nodes into block say like 5 each, attach these to another ai agent or mcp server, and then have a main flow like above, which calls each mcp server / agent.
I guess a multilayer ai agent, which the main ai agent calls the most relevant ai agent which has 5 tools ie HTTP nodes?
This might work, I’m yet to experiment with it though.
How many https requests are you planning to add? Are you seeing any performance issues with the current layout? @Blessing
I made a further post here about it too, Exploring Multi-Agent Patterns in n8n Using MCP Triggers & Clients (Without Webhooks) This might help further.
Hey Blessing,
It’s not completing the flow here
you can either removed the connection to the edit node (for testing the first ai agent, then should be able to connect up the rest) I think once it passes that switch, it should reply after further processing by other ai agents.
Hope this helps
how will i connect the ai agent
can we fix it over a google meet call
basically when i disconnect the switch node from the edit field, the issue is that the ai agent will have limited information to answer question
Correct — but you need to get it to pass through the Switch node properly.
**redacted
That said, I’m still happy to help where I can. Let me know if you want to troubleshoot further or hop on a call.
really false accusation just because you are helping people, it is well, i hope things change
How can i fix that please, i’ve all i could
or are you avaliable to hop on a can rn
Give me like 1hour or so, i’ll send a reply or direct message.
I would love to rant about it, but frankly, I’ve replied stating my stance, that I even had recent interviews with the team, and I felt that someone else got it because I wasn’t active on the community, so here I am.
sorry about that man. i will be expecting your feedback in an hr from now
Hey, so am thinking, if u want to share the workflow, I can get running for you or am thinking maybe discord? I normally meet on google etc, but trying to avoid any potential issues, so what I’m proposing, u can try upload workflow here, or send via a private message.
I work on it, to get it responding, I will provide details of what I did and share back the workflow if acceptable to you?
Let me know, as I should be free to next few hours etc, then will focusing on my workflows 
@Blessing
From what I can see, the main part is adding output parser on ure ai agent, ensure the flow in continous and filters correctly through sub agent processing, and return once fully completed to the chat.
how do i share my workflow
how do i send you a private text
Hey,
So I created this to help explain,
This is my main agent, which has access to subworkflows with ai agents in them. it acts like the brain and decides what sub-agent should handle the request.
next I connect the other agents via tools on the main agent, each call goes to these:
"Run the GitHub backup workflow in agent confirm""Analyze this invoice and return the total due and due date.""Answer: What company sent this invoice, and what is the payment method listed?"
all sub-agents only act based on clear input passed to them — no assumptions.
the main agent always gives specific context and goal for each task.
they each live in their own subworkflow, with their own langchain agent + tool setup.
there kick off different things like:
the main agent just delegates → waits → replies.
never acts alone. just routes tasks to the right tool and returns the output once it’s all done.
This isn’t complete but works to show how you can structure them also
Hope this helps.
hi, thanks for that
how can i share my workflow for you to see
It’s good practice to have a “default” set for a switch using the fallback option, so that when none of the conditions match, it still follows a default path and does not stop the workflow
you can share quickly via google drive link, and direct message me it if you’d like.