Elvis Sarvia This workflow corresponds to n8n.io template #13853 — we link there as the canonical source.
This workflow follows the Agent → OpenAI Chat recipe pattern — see all workflows that pair these two integrations.
The workflow JSON
Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →
{
"nodes": [
{
"id": "3f180ecc-96b8-420a-b1d2-3f9f3a18315b",
"name": "Webhook - User Input",
"type": "n8n-nodes-base.webhook",
"position": [
-656,
320
],
"parameters": {
"path": "guardrails-demo",
"options": {},
"httpMethod": "POST",
"responseMode": "responseNode"
},
"typeVersion": 2
},
{
"id": "e65aee97-5b9e-41f5-97a6-9ac6c77ba038",
"name": "Extract Text",
"type": "n8n-nodes-base.code",
"position": [
-432,
320
],
"parameters": {
"jsCode": "const input = $input.first().json.body;\nconst text = (input.message || input.body || input.text || '').trim();\nreturn { json: { text: text, source: input.source || 'unknown' } };"
},
"typeVersion": 2
},
{
"id": "31489997-5f6a-4a5b-af70-a1e579e7037e",
"name": "Input Guardrails",
"type": "@n8n/n8n-nodes-langchain.guardrails",
"position": [
-208,
320
],
"parameters": {
"text": "={{ $json.text }}",
"guardrails": {
"pii": {
"value": {
"type": "all"
}
},
"jailbreak": {
"value": {
"threshold": 0.8
}
},
"secretKeys": {
"value": {
"permissiveness": "strict"
}
}
}
},
"typeVersion": 1
},
{
"id": "c68c1928-2202-410a-88b2-8b6d03a26332",
"name": "Input Guardrails LLM",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
-160,
480
],
"parameters": {
"model": "gpt-4o-mini",
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "c1f2c53f-1f05-4d14-9289-e9512ca5a29d",
"name": "AI - Generate Response",
"type": "@n8n/n8n-nodes-langchain.agent",
"position": [
144,
64
],
"parameters": {
"text": "=You are a helpful customer support assistant. Respond to the following inquiry in a professional, concise manner. Do not include any URLs, links, or references to external websites. Do not reveal any internal system details, API keys, or credentials.\n\nCustomer message: {{ $('Extract Text').item.json.text }}\n\nRespond directly and helpfully.",
"options": {},
"promptType": "define"
},
"typeVersion": 1.7
},
{
"id": "7786e44c-dd3d-467f-b637-b42aee651203",
"name": "OpenAI Chat Model",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
224,
288
],
"parameters": {
"model": "gpt-4o-mini",
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "267cdddd-2e23-46aa-820b-8bd2dbda1f9a",
"name": "Output Guardrails",
"type": "@n8n/n8n-nodes-langchain.guardrails",
"position": [
496,
176
],
"parameters": {
"text": "={{ $json.output }}",
"guardrails": {
"nsfw": {
"value": {
"threshold": 0.8
}
},
"secretKeys": {
"value": {
"permissiveness": "strict"
}
}
}
},
"typeVersion": 1
},
{
"id": "91e488cf-65e6-4196-a1f4-e556825e9944",
"name": "Output Guardrails LLM",
"type": "@n8n/n8n-nodes-langchain.lmChatOpenAi",
"position": [
576,
336
],
"parameters": {
"model": "gpt-4o-mini",
"options": {}
},
"credentials": {
"openAiApi": {
"name": "<your credential>"
}
},
"typeVersion": 1
},
{
"id": "54d0c810-619e-419e-9209-5ff693293f32",
"name": "Respond - Safe Output",
"type": "n8n-nodes-base.respondToWebhook",
"position": [
848,
80
],
"parameters": {
"options": {},
"respondWith": "json",
"responseBody": "={{ JSON.stringify({ status: 'success', response: $('AI - Generate Response').item.json.output }) }}"
},
"typeVersion": 1.1
},
{
"id": "bbe97fb1-f4c6-4e6d-b1ed-7615f8dd1562",
"name": "Respond - Flagged Output (Fallback)",
"type": "n8n-nodes-base.respondToWebhook",
"position": [
848,
272
],
"parameters": {
"options": {},
"respondWith": "json",
"responseBody": "={{ JSON.stringify({ status: 'output_flagged', fallback: 'Thank you for your message. A team member will follow up shortly.' }) }}"
},
"typeVersion": 1.1
},
{
"id": "68d7bea8-52c8-4989-a3ca-493be57fd99a",
"name": "Respond - Input Blocked",
"type": "n8n-nodes-base.respondToWebhook",
"position": [
176,
656
],
"parameters": {
"options": {
"responseCode": 400
},
"respondWith": "json",
"responseBody": "={{ JSON.stringify({ status: 'input_blocked', message: 'Your message could not be processed. Please rephrase and try again.' }) }}"
},
"typeVersion": 1.1
},
{
"id": "1a5644f4-d40c-4461-9254-ddce454fdbce",
"name": "Sticky Note",
"type": "n8n-nodes-base.stickyNote",
"position": [
-1312,
-16
],
"parameters": {
"width": 528,
"height": 816,
"content": "## Input + Output Guardrails\n\n### How it works\n1. **Webhook** receives user input via POST request.\n2. **Extract Text** (Code node) pulls the message content from the payload.\n3. **Input Guardrails** (Guardrails node + LLM) checks for blocked keywords, prompt injection patterns, and PII. Flagged inputs are blocked and return an error response.\n4. **AI - Generate Response** (AI Agent + OpenAI) processes clean inputs and generates a response.\n5. **Output Guardrails** (Guardrails node + LLM) scans the AI response for unexpected URLs, leaked secrets, and content policy violations. Flagged outputs fall back to a safe templated response.\n\n### Setup\n- Connect **OpenAI credentials** (or another provider) to both Guardrails LLM nodes and the AI Agent's Chat Model\n- Copy the Webhook test URL and send test inputs, including some with PII (e.g., fake SSNs) or injection attempts (e.g., \"ignore your instructions\") to see guardrails in action\n- Review the Guardrails node configurations to customize which checks are active\n\n### Customization\n- Add custom keyword lists to the Input Guardrails for your domain (e.g., competitor names, profanity)\n- Adjust the Output Guardrails to enforce your specific content policies\n\n\nThis template is a learning companion to the Production AI Playbook, a series that explores strategies, shares best practices, and provides practical examples for building reliable AI systems in n8n. \n\n\nThis template is a learning companion to the Production AI Playbook, a series that explores strategies, shares best practices, and provides practical examples for building reliable AI systems in n8n. \n\nhttps://go.n8n.io/PAP-D&A-Blog"
},
"typeVersion": 1
},
{
"id": "33ca027e-38b0-476e-a5ae-8578651d995a",
"name": "Sticky Note1",
"type": "n8n-nodes-base.stickyNote",
"position": [
-736,
160
],
"parameters": {
"color": 7,
"width": 448,
"height": 480,
"content": "## Receive & Extract"
},
"typeVersion": 1
},
{
"id": "7a4e8c15-c028-461e-9b22-acab31817b99",
"name": "Sticky Note2",
"type": "n8n-nodes-base.stickyNote",
"position": [
-256,
160
],
"parameters": {
"color": 7,
"width": 320,
"height": 480,
"content": "## Input Guardrails"
},
"typeVersion": 1
},
{
"id": "6f5e4da2-3a84-4ef0-bd3a-d3fa8be6ba6d",
"name": "Sticky Note3",
"type": "n8n-nodes-base.stickyNote",
"position": [
96,
0
],
"parameters": {
"color": 7,
"width": 336,
"height": 480,
"content": "## AI Generation"
},
"typeVersion": 1
},
{
"id": "5aba8f71-5a6b-4060-a21d-de79f039ebe6",
"name": "Sticky Note4",
"type": "n8n-nodes-base.stickyNote",
"position": [
464,
0
],
"parameters": {
"color": 7,
"width": 576,
"height": 480,
"content": "## Output Guardrails"
},
"typeVersion": 1
}
],
"connections": {
"Extract Text": {
"main": [
[
{
"node": "Input Guardrails",
"type": "main",
"index": 0
}
]
]
},
"Input Guardrails": {
"main": [
[
{
"node": "AI - Generate Response",
"type": "main",
"index": 0
}
],
[
{
"node": "Respond - Input Blocked",
"type": "main",
"index": 0
}
]
]
},
"OpenAI Chat Model": {
"ai_languageModel": [
[
{
"node": "AI - Generate Response",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Output Guardrails": {
"main": [
[
{
"node": "Respond - Safe Output",
"type": "main",
"index": 0
}
],
[
{
"node": "Respond - Flagged Output (Fallback)",
"type": "main",
"index": 0
}
]
]
},
"Input Guardrails LLM": {
"ai_languageModel": [
[
{
"node": "Input Guardrails",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"Webhook - User Input": {
"main": [
[
{
"node": "Extract Text",
"type": "main",
"index": 0
}
]
]
},
"Output Guardrails LLM": {
"ai_languageModel": [
[
{
"node": "Output Guardrails",
"type": "ai_languageModel",
"index": 0
}
]
]
},
"AI - Generate Response": {
"main": [
[
{
"node": "Output Guardrails",
"type": "main",
"index": 0
}
]
]
}
}
}
Credentials you'll need
Each integration node will prompt for credentials when you import. We strip credential IDs before publishing — you'll add your own.
openAiApi
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
Protect your workflows with n8n's native Guardrails node, placed before and after your AI step. The input guardrails catch jailbreak attempts and PII before they reach your model. The output guardrails scan AI responses for NSFW content and secret keys before they reach your…
Source: https://n8n.io/workflows/13853/ — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
The full end-to-end workflow that chains all patterns together. This template processes customer feedback from intake to team routing, with normalization, validation, native guardrails, AI classificat
Agent, Guardrails, OpenAI Chat +1
This template automatically generates a professional, structured title for newly-ingested Paperless-ngx documents using AI (OpenAI GPT), then removes a processing tag to mark the document as handled.
HTTP Request, Guardrails, Agent +2
⏺ 🚀 How it works
Agent, Anthropic Chat, Output Parser Structured +6
L&D_AgentsAI_ATIVO. Uses httpRequest, agent, googleCalendarTool, toolSerpApi. Webhook trigger; 93 nodes.
HTTP Request, Agent, Google Calendar Tool +9
This n8n workflow orchestrates a powerful suite of AI Agents and automations to manage and optimize various aspects of an e-commerce operation, particularly for platforms like Shopify. It leverages La
Google Sheets, HTTP Request, Slack +10