Package Information
Available Nodes
Documentation generating…
Documentation
n8n-nodes-halu
An n8n community node for Komplex AI —
a hallucination detector for LLM output. Drop a "Detect Hallucination" step into any
n8n workflow to score AI-generated text for hallucination risk.
Install
In self-hosted n8n: Settings → Community Nodes → Install → n8n-nodes-halu.
Credentials
Create a Komplex AI (halu) API credential with your API key (sk_...) — get one free
at https://detector.komplexai.io/account/keys.
Node: Komplex AI Hallucination Detector
Operation: Detect Hallucination.
| Field | Description |
|---|---|
| Response Text | The LLM output to score (English, ≤ 2,048 chars). Required. |
| Prompt (optional) | The prompt the model saw — improves accuracy on context-dependent claims. |
| Task | multiclass (default; full regime breakdown) or binary. |
Returns the API JSON: p_hallucination, flag, top_regime, regime_scores, etc.
Wire the flag / p_hallucination into an IF node to gate, annotate, or route.
Scope & limits
English natural-language responses, ≤ 2,048 chars; probabilistic signal, not a fact-checker;
first call after idle can cold-start ~10–30s. See
Performance and the
Guide.
License
Apache-2.0