MargIQ for n8n

Use the right AI model for each recurring workflow path, then report the result as evidence—without moving provider credentials into MargIQ.

Nodes

• MargIQ Recommend examines the request and returns the selected model.
• MargIQ Observe reports the provider output, token usage, latency, success, and business outcome.

Your LLM node stays between them:

MargIQ Recommend → Existing LLM node → MargIQ Observe

Installation

Until n8n completes Cloud verification, owners of self-hosted instances can install n8n-nodes-margiq from Settings → Community Nodes.

Setup

1. Create a MargIQ API key at getmargiq.com.
2. Add a MargIQ API credential in n8n.
3. Add MargIQ Recommend before your existing LLM node.
4. Map {{$json.margiq.selectedModel}} to the model field of the LLM node.
5. Add MargIQ Observe after the LLM node and map its output and usage fields.

Start with Report Only to inspect recommendations while every request continues using the requested model. Switch to Automatic Routing when the observed workflow paths and quality evidence are satisfactory.

Data handling

Recommend sends the configured model allow-list, requested model, provider, messages, and derived request context to the MargIQ API. Observe sends the decision context, provider output, usage, latency, success, and optional business outcome. Provider credentials are never sent to MargIQ.

Support

Visit getmargiq.com for product documentation and support.

License

MIT