╔═╗
  ║Q║ Claude Azure
  ╚═╝

Claude Azure

Run Claude Code with Azure OpenAI, OpenAI, or Anthropic

Features • Quick Start • Configuration • Deployment Modes • Troubleshooting

Overview

Claude Azure is a fork of Claude Code that adds native support for Azure OpenAI and OpenAI alongside the standard Anthropic API. It provides a seamless local proxy that translates Anthropic API calls to Azure/OpenAI formats, letting you use Claude Code with your existing Azure deployments.

Features

✨ Multi-Provider Support

• Azure OpenAI (Tiered or Model Router modes)
• OpenAI API
• Anthropic API (native passthrough)

🔧 Flexible Deployment Options

• Tiered Mode: Map model sizes to separate Azure deployments
• Model Router Mode: Single endpoint with Azure APIM routing

🚀 Zero-Friction Setup

• Interactive setup wizard
• Automatic proxy configuration
• Seamless Claude Code integration

🔒 Secure Configuration

• Local config storage (~/.claude-azure/config.json)
• No cloud credentials required beyond provider API keys

Quick Start

Prerequisites

1. Install Claude Code (required):

   curl -fsSL https://claude.ai/install.sh | bash

2. Azure OpenAI resource (if using Azure) with deployed models

Installation

Option 1: npm (Recommended)

# Install globally from npm
npm install -g claude-azure

# Run setup wizard
claude-azure

Option 2: From Source

# Clone and install
git clone https://github.com/schwarztim/claude-azure.git
cd claude-azure
npm install
npm run build
npm link

# Run setup wizard
claude-azure

Windows Users: Works on PowerShell, Command Prompt, and WSL. Claude Code must be installed first.

First Run

The setup wizard will guide you through configuration:

  ╔═╗
  ║Q║ Claude Azure
  ╚═╝

  Claude Azure Setup
  ─────────────────────────────────────

? Select your AI provider:
❯ Azure OpenAI - Use Azure-hosted models
  OpenAI - Use OpenAI API directly
  Anthropic - Use Anthropic API directly

Azure OpenAI Configuration

Azure OpenAI Configuration
Get these from Azure Portal → Azure OpenAI → Keys and Endpoint

? Azure OpenAI Endpoint: https://myresource.openai.azure.com
? Azure OpenAI API Key: ********
? API Version: 2024-12-01-preview

Deployment Mode
? Select deployment mode:
  ❯ Tiered - Separate deployments for different model sizes
    Model Router - Single deployment for all models

# For Tiered mode:
? Opus/Large model deployment: gpt-4o
? Sonnet/Medium model deployment: gpt-4o
? Haiku/Small model deployment: gpt-4o-mini

# For Model Router mode:
? Model Router deployment name: model-router

✔ Testing Azure connection...
✓ Configuration saved!

Usage

# First run - interactive setup
claude-azure

# Reconfigure provider/settings
claude-azure --setup

# Reset all configuration
claude-azure --reset

# Show proxy debug logs
claude-azure --verbose

# Pass arguments to Claude Code
claude-azure -p "explain this codebase"
claude-azure --model sonnet

Configuration

Config File Location

~/.claude-azure/config.json

Example Configuration (Tiered Mode)

{
  "provider": "azure",
  "azure": {
    "endpoint": "https://myresource.openai.azure.com",
    "apiKey": "sk-...",
    "apiVersion": "2024-12-01-preview",
    "deployments": {
      "opus": "gpt-4o",
      "sonnet": "gpt-4o",
      "haiku": "gpt-4o-mini"
    }
  }
}

Example Configuration (Model Router Mode)

{
  "provider": "azure",
  "azure": {
    "endpoint": "https://myresource.openai.azure.com",
    "apiKey": "sk-...",
    "apiVersion": "2024-12-01-preview",
    "modelRouter": "model-router"
  }
}

Deployment Modes

Tiered Mode (Traditional)

When to use:

• Fine-grained control over model deployments
• Cost optimization by model size
• Different models for different tiers

How it works:

• Separate Azure deployments for each model class
• Maps Claude model names to your deployments:
  • claude-opus-* → deployments.opus (e.g., gpt-4o)
  • claude-sonnet-* → deployments.sonnet (e.g., gpt-4o)
  • claude-haiku-* → deployments.haiku (e.g., gpt-4o-mini)

Model Router Mode (Azure APIM)

When to use:

• Azure API Management with routing logic
• Dynamic model selection
• Simplified deployment management

How it works:

• Single deployment endpoint
• Passes original Claude model name (e.g., claude-sonnet-4) to the router
• Your APIM policy routes to appropriate backend

How It Works

┌─────────────────┐    ┌──────────────┐    ┌─────────────────┐
│   Claude Code   │───▶│ Local Proxy  │───▶│  Azure OpenAI   │
│                 │    │ (auto port)  │    │                 │
│ ANTHROPIC_BASE  │    │ Translates:  │    │ /chat/complete  │
│ _URL=localhost  │    │ Anthropic→   │    │                 │
│                 │    │ OpenAI       │    │                 │
└─────────────────┘    └──────────────┘    └─────────────────┘

Key Components:

1. Claude Code - The official Anthropic CLI tool
2. Local Proxy - Translates API formats in real-time
3. Azure/OpenAI - Your cloud AI provider

For Anthropic provider, no proxy is used (direct passthrough).

Keeping Up to Date

Stay in sync with upstream Claude Code:

cd ~/Scripts/claude-azure
git fetch upstream
git merge upstream/main
npm install
npm run build

Troubleshooting

| Issue | Solution | |-------|----------| | claude-azure: command not found | Run npm link from project directory | | Claude Code not found | Install from https://claude.ai/code | | Connection errors | Verify endpoint/key in ~/.claude-azure/config.json | | Tool calls failing | Run with --verbose to see proxy request/response logs | | Model not found | Check deployment names match your Azure resource | | Rate limiting | Azure quotas apply - check Azure Portal metrics |

Verbose Logging

claude-azure --verbose

Shows:

• Proxy startup details
• Request/response translation
• Azure API calls
• Error details

Development

# Clone and install
git clone https://github.com/schwarztim/claude-azure.git
cd claude-azure
npm install

# Watch mode
npm run dev

# Build
npm run build

# Test locally
npm link
claude-azure --setup

Architecture

src/
├── cli.ts           # Main entry point, setup wizard
├── config.ts        # Configuration management
├── proxy.ts         # API translation proxy
└── index.ts         # Exports

dist/                # Compiled JavaScript
~/.claude-azure/     # User configuration
  └── config.json

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Related Projects

• Claude Code - The official Anthropic CLI tool
• Azure OpenAI - Microsoft's Azure OpenAI service

License

MIT License - See LICENSE for details

Azure wrapper code is MIT licensed. Original Claude Code license applies to upstream code.

Acknowledgments

• Built on top of Claude Code by Anthropic
• Inspired by the need for Azure OpenAI enterprise deployment support

Made with ❤️ for Azure OpenAI users

Report Bug • Request Feature