§INTEGRATION · N8N

n8n workflows, Gmail triage included.

n8n workflows move from model output to action nodes quickly. We ship a community node that adds a Sentisec control-plane decision before sensitive workflow actions. Drop it in front of Gmail Send, HTTP Request, or database writes and the workflow halts cleanly on injection.

ONE-LINE SETUP

Two lines. One terminal. Done.

The custom node is published at npmjs.com/package/n8n-nodes-sentisec. Configure the Guard with your workspace API key; one environment variable.

bash

$ npm install n8n-nodes-sentisec
$ In n8n → Settings → Community Nodes → install `n8n-nodes-sentisec` · drop SentisecGuard before Gmail Send

PATH OF A TOOL CALL

One hop. No framework rewrite.

Diagram is intentionally kept at the category level — the detail of the checks lives on the developer landing page and under NDA.

Gmail Trigger → AI Agent node
   │
   ▼  proposed action (JSON)
SentisecGuard node
   │   cognitive integrity score · workspace policy
   ▼
Gmail Send / HTTP Request / DB write
   │
   ▼
workflow continues — or halt + alert

§CATCHES

What this catches

Gmail-triage prompt injection: an email whose body convinces the agent to forward credentials or exfiltrate a secret is halted before Gmail Send.
HTTP Request nodes whose destination URL was injected through model output.
Database or webhook writes derived from retrieved email content, when they leave the declared workflow shape.

§LIMITS

What this doesn't (yet)

We gate the action declared to the Guard node — not anything n8n does after. If a later workflow branch ignores the Guard's verdict, it's on you to wire the branch correctly. We ship a sample Gmail-triage workflow.
Ollama-backed AI Agent nodes vary by model. We flag reduced-coverage sessions in the dashboard.

Full limits page →

Try free →Read what we don't catch yet See the 90-second demo →