n8n-nodes-secure-ai

Keep customer data out of the AI steps in your n8n workflows.

Secure AI hides names, emails, phone numbers, addresses and account
numbers before text reaches OpenAI, Claude or any other model, and puts them
back in the answer. The model sees believable stand-ins ("Sylvie Nakashima"
for "Sara Whitfield"), so it writes normally.

Install

In n8n: Settings → Community Nodes → Install, and enter
@secure-ai/n8n-nodes-secure-ai.

Credentials

Create a key at secureai.one → Settings → Developer
(the free plan includes 1,000 requests a month), then add a Secure AI API
credential in n8n with that key.

Use

Put Secure AI → Hide Details before your AI step and Secure AI → Put
Details Back after it:

Trigger → Secure AI (Hide Details) → OpenAI / AI Agent → Secure AI (Put Details Back) → Gmail

• Hide Details writes the hidden text to text and the map of stand-ins
  to secureAiMap. Use {{ $json.text }} as the AI step's prompt.
• Put Details Back reads the AI's answer and {{ $json.secureAiMap }},
  and writes the answer with the real details to the field you choose.

Options: hide organisations' names too, and Medical Mode, which also
removes what HIPAA's Safe Harbor lists (dates, ages over 89, towns, ZIP codes,
hospital names, record numbers).

The map stays in your workflow's data. Secure AI keeps no copy.

Prefer one step? Point n8n's own OpenAI credential at Secure AI instead:
see secureai.one/developers/n8n.

Develop

npm install
npm run build
SECURE_AI_KEY=sai_... npm test   # the last test calls the live API