Skip to main content

​
Overview

PromptGuard supports Azure OpenAI in passthrough mode, allowing you to use your own Azure OpenAI deployments with full security protection.
No configuration needed on our end - simply provide your Azure credentials in each request.

​
Quick Start

To use Azure OpenAI with PromptGuard, you need three things:
  1. Model name: Format as azure/{deployment-name}
  2. X-Azure-Resource header: Your Azure resource name
  3. Authorization header: Your Azure API key

​
Example Request

curl https://api.promptguard.co/api/v1/chat/completions \
  -H "X-API-Key: pg_live_xxx" \
  -H "X-Azure-Resource: my-azure-resource" \
  -H "Authorization: Bearer <your-azure-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "azure/my-gpt4-deployment",
    "messages": [
      {"role": "user", "content": "What is 2+2?"}
    ]
  }'

​
Finding Your Azure Details

​
Resource Name

  1. Go to Azure Portal → Your Azure OpenAI resource
  2. Under Keys and Endpoint, find your endpoint URL
  3. Format: https://{resource-name}.openai.azure.com/
  4. Your resource name is everything before .openai.azure.com
Example:
  • Endpoint: https://my-company-openai.openai.azure.com/
  • Resource name: my-company-openai

​
Deployment Name

  1. Go to Azure AI Studio → Your resource
  2. Click Deployments in the left sidebar
  3. Your deployment name is in the Deployment Name column
Common deployment names:
  • gpt-4-deployment
  • gpt-4o-mini
  • my-custom-gpt4

​
API Key

  1. Go to Azure Portal → Your Azure OpenAI resource
  2. Under Keys and Endpoint, copy Key 1 or Key 2

​
SDK Integration

​
Python (OpenAI SDK)

import openai

client = openai.OpenAI(
    base_url="https://api.promptguard.co/api/v1",
    api_key="<your-azure-api-key>",
    default_headers={
        "X-API-Key": "pg_live_xxx",
        "X-Azure-Resource": "my-azure-resource",
    }
)

response = client.chat.completions.create(
    model="azure/my-gpt4-deployment",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

​
Node.js (OpenAI SDK)

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.promptguard.co/api/v1',
  apiKey: '<your-azure-api-key>',
  defaultHeaders: {
    'X-API-Key': 'pg_live_xxx',
    'X-Azure-Resource': 'my-azure-resource',
  }
});

const response = await client.chat.completions.create({
  model: 'azure/my-gpt4-deployment',
  messages: [{ role: 'user', content: 'Hello!' }]
});

console.log(response.choices[0].message.content);

​
TypeScript

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.promptguard.co/api/v1',
  apiKey: process.env.AZURE_API_KEY,
  defaultHeaders: {
    'X-API-Key': process.env.PROMPTGUARD_API_KEY,
    'X-Azure-Resource': process.env.AZURE_RESOURCE_NAME,
  }
});

const response = await client.chat.completions.create({
  model: `azure/${process.env.AZURE_DEPLOYMENT_NAME}`,
  messages: [{ role: 'user', content: 'Hello!' }]
});

​
Supported Models

PromptGuard supports all Azure OpenAI models through passthrough:
Model FamilyExample
GPT-4oazure/my-gpt4o-deployment
GPT-4 Turboazure/my-gpt4-turbo
GPT-4azure/my-gpt4
GPT-3.5 Turboazure/my-gpt35-turbo
Embeddingsazure/text-embedding-ada-002
DALL-Eazure/my-dalle3
The model name is always: azure/{your-deployment-name}

​
Security Features

All PromptGuard security features work seamlessly with Azure OpenAI:

Prompt Injection

Blocks jailbreak attempts and instruction hijacking

PII Detection

Redacts emails, SSNs, phone numbers automatically

Credit Cards

Luhn-validated credit card redaction

Data Exfiltration

Prevents “send to external email” attacks

​
Environment Variables

For cleaner code, use environment variables: 
# .env
PROMPTGUARD_API_KEY=pg_live_xxx
AZURE_RESOURCE_NAME=my-azure-resource
AZURE_DEPLOYMENT_NAME=my-gpt4-deployment
AZURE_API_KEY=your-azure-api-key
Then in your code: 
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.promptguard.co/api/v1",
    api_key=os.getenv("AZURE_API_KEY"),
    default_headers={
        "X-API-Key": os.getenv("PROMPTGUARD_API_KEY"),
        "X-Azure-Resource": os.getenv("AZURE_RESOURCE_NAME"),
    }
)

response = client.chat.completions.create(
    model=f"azure/{os.getenv('AZURE_DEPLOYMENT_NAME')}",
    messages=[{"role": "user", "content": "Hello!"}]
)

​
Troubleshooting

​
Error: “Azure OpenAI passthrough requires X-Azure-Resource header”

Solution: Add the X-Azure-Resource header with your Azure resource name. 
# ❌ Missing header
client = OpenAI(base_url="...", api_key="...")

# ✅ Correct
client = OpenAI(
    base_url="...",
    api_key="...",
    default_headers={"X-Azure-Resource": "my-resource"}
)

​
Error: “No provider found for model”

Solution: Make sure your model name starts with azure/ prefix. 
# ❌ Wrong format
model="my-gpt4-deployment"

# ✅ Correct format
model="azure/my-gpt4-deployment"

​
Error: “Deployment not found”

Solution: Verify your deployment name in Azure AI Studio matches exactly.
  1. Go to Azure AI Studio → Deployments
  2. Copy the exact deployment name
  3. Use it after azure/ prefix

​
Error: “Invalid credentials”

Solution: Make sure you’re using your Azure API key, not your Azure password.
  1. Azure Portal → Your OpenAI resource
  2. Keys and Endpoint → Copy Key 1 or Key 2
  3. This is your API key (not your account password)

​
Next Steps

Security Policies

Configure custom security rules

Monitoring

Track usage and detect threats

Python Guide

Full Python integration guide

Node.js Guide

Full Node.js integration guide

​
Support

Need help with Azure OpenAI integration?