n8n-nodes-super-agent

Better AI agents for n8n. Because the default ones were driving me crazy.

What is this?

I got tired of n8n's AI limitations, so I built my own nodes that actually work how you'd expect them to.

Super Agent - An AI agent that can use multiple system prompts and multiple memories. Finally.

Super Prompts - Routes your messages to different prompts/models based on what you're asking. One agent, multiple personalities.

Super MCP - Connect all your MCP tools with one node. And they actually work.

Why I built this

The standard n8n AI agent forces you to:

• Use ONE system prompt (seriously?)
• Connect ONE memory (what if I need short-term AND long-term?)
• Create separate nodes for each MCP tool
• Build separate workflows for different types of requests

That's dumb. So I fixed it.

What's different

Multiple System Prompts

Instead of one agent = one prompt, you can have:

• Technical prompt for coding questions → Claude 4 Sonnet
• Creative prompt for writing → Claude 4 Sonnet
• Data analysis → Gemini 2.5 Pro
• Quick responses → GPT-4.1-mini
• All in ONE workflow

The router (GPT-4.1-mini) figures out which prompt to use based on your message.

Multiple Memories

Connect as many as you want:

• Window memory for recent context
• Buffer memory for full history
• Different memories for different purposes
• They all work together

MCP That Works

• One node for all your MCP servers
• Actually follows the MCP spec
• No more debugging why tools won't load

Installation

In your n8n instance:

1. Go to Settings > Community nodes
2. Click "Install a community node"
3. Paste in @kenkaiii/n8n-nodes-super-agent
4. Click Install

Basic Setup

1. Add Super Agent node
2. Connect Super Prompts with your different system prompts
3. Add Super MCP if you need tools
4. Connect whatever memories you want
5. Done

Example

Let's say you want an agent that can:

• Write code
• Create marketing copy
• Analyze data

Old way: 3 different workflows or a really confused agent

My way:

• One Super Agent
• Three prompts in Super Prompts
• Each with the right model for the job
• Router figures out which to use

Who's this for?

Anyone who's tried to build a serious AI workflow in n8n and thought "why is this so limited?"

Not a magic fix

This doesn't make your prompts better or your workflow faster. It just removes the artificial limitations that were annoying me (and probably you).

Built by

Ken Kai does AI

License

MIT