Enterprise PII Guardrails Node for n8n
Connect sensitive business workflows to LLMs (OpenAI, Anthropic Claude, Google Gemini, Ollama) with zero PII leakage risk. Automatically mask private data before sending to models, and restore original values downstream.
Zero-Leakage AI Architecture
Add the Node to Your Workflow
Officially Verified on n8n Cloud Native In-App Search
No manual installation required! In n8n Cloud and verified self-hosted instances, press + or Tab directly on your workflow canvas, search for Enterprise PII Guardrails, and drag the node into your flow with the Verified Shield badge 🛡️.
- In your self-hosted n8n instance, navigate to Settings > Community Nodes.
- Click Install a community node, enter
n8n-nodes-piiguardrails, and click Install.
npm install n8n-nodes-piiguardrails
Configure Credentials in n8n
When adding the node to your canvas, n8n will prompt you to create an "Enterprise PII Guardrails API" credential. Choose either of the two setup options below:
⚡ Option A: Instant Free Sandbox No Installation Needed
Perfect for immediate verification and zero-friction testing in less than 60 seconds.
🏢 Option B: Production On-Premises Docker Air-Gapped / HIPAA / GDPR
Runs 100% locally inside your corporate infrastructure. Data never leaves your network.
-e ACCEPT_LICENSE=true \
piiguardrails/enterprise-pii-guardrail:latest
http://host.containers.internal:8000
How to Use the Node in Workflows
Mask Text (Pre-LLM)
Scrubs Names, Emails, Phone Numbers, SSNs, Credit Cards, Secrets, and Medical Record Numbers.
Unmask Text (Post-LLM)
Flawlessly swaps synthetic surrogate tokens (<PERSON_1>) back with the original sensitive customer values.
Mapping: {{ $('Mask Node').item.json.mapping }}
Scan / Audit (DLP Only)
Non-mutating data loss prevention inspection. Audits incoming or outgoing payloads for regulatory violations.
Pre-Built Workflow Templates
Redact sensitive PII in AI prompts and restore responses with Enterprise PII Guardrails
Officially published and approved on the n8n Workflow Library. Install with one click directly inside your n8n workspace or browse on n8n.io.
OpenAI PII-Safe Dispute Pipeline
Complete end-to-end customer support automation with live OpenAI inference. Ingests raw banking disputes, scrubs 9 distinct PII categories (<PERSON_1>, <CREDIT_CARD_1>, <SSN_1>, <BANK_ACCOUNT_1>, <EMAIL_1>, <PHONE_1>) in <25ms, passes zero private data to OpenAI, and restores all original values downstream.
https://raw.githubusercontent.com/piiguardrails/piiguardrails/main/workflows/Local_PII_Safe_LLM_Pipeline_OpenAI_demo.json
Self-Contained Evaluation Pipeline
Ideal for evaluation and CI/CD smoke tests without needing an external LLM vendor account. Features simulated customer inquiries, deterministic entity tokenization, mock model reasoning, and client perimeter unmasking verification.
https://raw.githubusercontent.com/piiguardrails/piiguardrails/main/workflows/Local_PII_Safe_LLM_Pipeline.json
How to import: Inside your n8n workspace, click the canvas options menu (...) in the top right > select Import from URL > paste either template URL above. The entire flow, nodes, and test data populate instantly.
Frequently Asked Questions
Q: I get "ECONNREFUSED" when testing credentials with localhost:8000
If your n8n instance is running inside a Docker or Podman container, localhost refers to the n8n container itself. Change the Base URL to http://host.containers.internal:8000 (Podman) or http://host.docker.internal:8000 (Docker) so n8n can reach the host machine.
Q: Does Enterprise PII Guardrails store or retain our prompts?
Zero Retention Guaranteed. The engine processes tokens strictly in ephemeral memory. No prompts, customer data, or token mappings are ever saved to disk or transmitted outside your local environment.
Q: How do I handle quota limits on the free sandbox?
The hosted sandbox at https://demo.piiguardrails.com provides 50 free requests per browser session for evaluation. For production workloads or unlimited requests, launch your own on-premise container using Docker or contact us.