Anthropic’s Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. It’s available as a CLI tool and VS Code extension. When you configure Claude Code with Microsoft Foundry, you run the coding agent on Azure infrastructure while keeping your data inside your compliance boundary. This configuration provides enterprise-grade security, private networking, role-based access control, and cost management. In this article, you learn how to:

Install and configure Claude Code CLI for Microsoft Foundry
Set up the Claude Code VS Code extension
Authenticate with Microsoft Entra ID or API keys
Create project context files for better AI assistance
Run Claude Code in GitHub Actions for CI/CD automation

To use Claude models in Microsoft Foundry, you must have a paid Azure subscription with a billing account in a country or region where Anthropic offers the models for purchase. For a list of common subscription-related errors, see Common error messages and solutions. The following subscription types are currently not supported:

Enterprise Accounts located in South Korea
Cloud Solution Provider subscriptions
Azure subscriptions that don’t have an active pay-as-you-go billing method (for example, student, free trial, or startup credit–based accounts)
Sponsored subscriptions that only use Azure credits. Note: If you have an account with a credit card on file, the credit card will be charged instead of Azure Credits.

For a list of supported regions, see supported geographic locations. Note that, Anthropic’s “Supported Regions Policy” may apply for the availability in your region, check supported regions for details.

Prerequisites

An Azure subscription with a valid payment method. If you don’t have an Azure subscription, create a paid Azure account.
Access to Microsoft Foundry with Contributor permissions to create and manage resources.
A Microsoft Foundry project created in one of the supported regions for Claude models. Claude models are currently available in East US 2 and Sweden Central only.
Contributor or Owner role on your Foundry resource group. For more information, see Azure RBAC roles.
Access to Azure Marketplace to deploy Foundry Models from partners.
For Windows, use Git Bash (included with Git for Windows) or install WSL2 (recommended for full Linux compatibility). See Install WSL.
(Optional) Azure CLI installed with az login completed for Microsoft Entra ID authentication.

System requirements

Requirement	Details
Operating system	macOS 13+, Ubuntu 20.04+/Debian 10+, Windows 10+ (native via Git Bash, or WSL)
RAM	4-GB minimum (8-GB recommended)
Git (optional, recommended)	2.23+ for pull request helpers

Deploy a Claude model in Foundry

Before configuring Claude Code, deploy the available Claude models that Claude Code needs. Claude models in Foundry are available for global standard deployment. Claude Code uses different models for different tasks:

Claude models in Microsoft Foundry are currently in preview. Model availability might change. Check the Foundry Models from partners page for the latest list of available models.

Claude Code role	Recommended deployment	Purpose
Primary model	`claude-sonnet-4-6`	General coding — balanced speed and quality
Fast model	`claude-haiku-4-5`	Quick operations — file reads, small edits
Extended thinking	`claude-opus-4-6`	Complex reasoning tasks (optional)

Other Claude models available in Foundry include claude-sonnet-4-5, claude-opus-4-5, and claude-opus-4-1.

Deploy model directly in Foundry

To deploy a model:

Follow the instructions in Deploy Microsoft Foundry Models in the Foundry portal to deploy a Claude model, such as Opus 4.6.
After deployment, select the deployment’s Details tab and note your Target URI and Key. You need these values for configuration.

Alternative: Deploy model using Model Router

Model Router is a Foundry model that intelligently routes each prompt to the best underlying model based on query complexity, cost, and performance. Model Router version 2025-11-18 supports select Claude models (claude-haiku-4-5, claude-opus-4-1, and claude-sonnet-4-5), alongside other Foundry models.

Model Router doesn’t currently support claude-sonnet-4-6 or claude-opus-4-6. If you need these models, deploy and reference them directly instead of using Model Router.

Benefits for Claude Code users:

Automatic model selection: Simple prompts route to faster, cheaper models. Complex coding tasks route to more capable models.
Cost optimization: Use routing profiles to balance quality versus cost while maintaining baseline performance.
Single endpoint: One deployment handles all routing decisions across your model fleet.

To use Model Router with Claude Code, first deploy the supported Claude models, then deploy Model Router and enable them through model subset configuration.

Install Claude Code CLI

Install the Claude Code CLI to work with Claude Code directly in your terminal. Then, verify that claude is in your PATH by running claude --version.

Anthropic has deprecated the npm installation method. Use the native installer or Homebrew instead. If you already installed via npm, run claude install to migrate to the native method.

Native install (recommended)

Native installations automatically update in the background to keep you on the latest version.

Bash / WSL
PowerShell

On macOS or Windows (Git Bash or WSL), run the installer script, which downloads and configures the claude binary:

curl -fsSL https://claude.ai/install.sh | bash
claude --version # verify installation

irm https://claude.ai/install.ps1 | iex
claude --version # verify installation

Installation with Homebrew

If using macOS:

brew install --cask claude-code
claude --version # verify installation

Homebrew installations don’t auto-update. Run brew upgrade claude-code periodically to get the latest features and security fixes.

Troubleshoot install location

If the claude --version command isn’t found, add the install location to your PATH as follows:

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc && source ~/.bashrc

For more installation options, see Claude Code documentation.

Configure Claude Code for Foundry

You use either your Foundry resource name or the base URL to configure Claude Code for Foundry. To find your Foundry resource name from the Foundry portal:

Go to the top left navigation and select the project name > Project details.
Copy the value of Parent resource from the Project details page.

To find your base URL from the Foundry portal:

Go to the home page of the Foundry portal
Find the Project endpoint and copy the part of the URL that comes before /api/projects/<your-project-name>. Your base URL is of the form https://<your-resource-name>.services.ai.azure.com. Claude Code appends /anthropic to this URL automatically when you use ANTHROPIC_FOUNDRY_RESOURCE.

Set environment variables to connect Claude Code to your Microsoft Foundry deployment:

    # Required: Enable Foundry integration
    export CLAUDE_CODE_USE_FOUNDRY=1

    # Azure resource name (replace <your-resource-name> with your resource name)
    export ANTHROPIC_FOUNDRY_RESOURCE=<your-resource-name>
    # Or provide the full base URL:
    # export ANTHROPIC_FOUNDRY_BASE_URL=https://<your-resource-name>.services.ai.azure.com

    # Optional: Specify model deployment names if different from defaults
    export ANTHROPIC_DEFAULT_SONNET_MODEL="claude-sonnet-4-6"
    export ANTHROPIC_DEFAULT_HAIKU_MODEL="claude-haiku-4-5"
    export ANTHROPIC_DEFAULT_OPUS_MODEL="claude-opus-4-6"

The following table describes each variable:

Variable	Description
`CLAUDE_CODE_USE_FOUNDRY`	Set to `1` to enable the Microsoft Foundry integration.
`ANTHROPIC_FOUNDRY_RESOURCE`	Your Foundry resource name. Claude Code constructs the endpoint URL as `https://<resource-name>.services.ai.azure.com/anthropic`.
`ANTHROPIC_DEFAULT_SONNET_MODEL`	The deployment name for the Claude Sonnet model (primary coding model).
`ANTHROPIC_DEFAULT_HAIKU_MODEL`	The deployment name for the Claude Haiku model (fast operations).
`ANTHROPIC_DEFAULT_OPUS_MODEL`	The deployment name for the Claude Opus model (complex reasoning).

To persist these variables across terminal sessions, add them to your shell profile (such as ~/.bashrc or ~/.zshrc).

Authenticate with Foundry

Claude Code supports two authentication methods for Microsoft Foundry:

Microsoft Entra ID (recommended): Uses your Azure CLI credentials. Best for enterprise environments, team access, and CI/CD pipelines where you want centralized identity management without managing secrets.
API key: Uses a project API key from the Foundry portal. Best for quick testing or environments where the Azure CLI isn’t available.

Option A: Microsoft Entra ID (recommended)

Microsoft Entra ID authentication uses your Azure CLI credentials automatically. Run az login before starting Claude Code:

If your Foundry resource is in a different tenant than your default Azure CLI tenant, specify the tenant ID:

az login --tenant <tenant-id>

Verify your sign-in targets the correct subscription:
```
az account show
```

When you use Microsoft Foundry, the /login and /logout commands inside Claude Code are disabled. Authentication is handled through your Azure credentials. Claude Code detects your Azure CLI session and uses it for authentication without extra configuration.

Option B: API key

If you prefer API key authentication, set the key in your environment variables.

In the Microsoft Foundry portal, open your resource.
On the Home page, find the Project API key field.
Select Copy Project API key to copy the value. This is the key you use for the ANTHROPIC_FOUNDRY_API_KEY environment variable.
Set the environment variable in your terminal:

Bash / WSL

export ANTHROPIC_FOUNDRY_API_KEY="<your-foundry-api-key>"

PowerShell

$env:ANTHROPIC_FOUNDRY_API_KEY = "<your-foundry-api-key>"

You can also find your API key in the Foundry portal under your model deployment’s Details tab.

Configure the VS Code extension

The Claude Code VS Code extension provides a native graphical interface for Claude Code directly in your IDE.

Install the extension from the VS Code Marketplace.
Open VS Code settings (Ctrl+, or Cmd+,) and search for Claude Code: Environment Variables.

Select Edit in settings.json and add the following configuration:

{
  "Claude Code: Environment Variables": {
    "CLAUDE_CODE_USE_FOUNDRY": "1",
    "ANTHROPIC_FOUNDRY_RESOURCE": "<your-resource-name>",
    "ANTHROPIC_FOUNDRY_API_KEY": "<optional-for-non-entra-auth>"
  }
}

Select the Spark icon in the sidebar to open the Claude Code panel.
Claude Code authenticates using your Azure credentials. Run az login first if you’re using Microsoft Entra ID authentication.

The extension supports auto-accept edits mode, plan mode, extended thinking, and file management with @-mentions—all while routing through your Microsoft Foundry deployment.

Validate the configuration

Verify that Claude Code is correctly configured to use Microsoft Foundry.

Open a terminal and navigate to a project directory:
```
cd your-project
```
Launch Claude Code:
```
claude
```
Claude Code defaults to the Sonnet model for general coding. If you haven’t deployed all three models yet, you can specify a deployed model directly:
```
claude --model claude-opus-4-6
```

Run the /status command:

> /status

The output should look similar to the following example. The exact format might vary depending on your Claude Code version.

─────────────────────────────────────
Version: 2.1.52
Session name: /rename to add a name
Session ID: <your-session-ID>
cwd: C:\WINDOWS\system32
API provider: Microsoft Foundry
Microsoft Foundry Resource: <your-resource-name>
Model: Default (claude-sonnet-4-6)
Memory:
Setting sources:
─────────────────────────────────────

Confirm the following in the status output:

Field	Expected value
API provider	Microsoft Foundry
Foundry resource	Your Foundry resource name
Model	Your deployed model (for example, `claude-sonnet-4-6`)

Send a test prompt such as “Summarize this project’s structure.”
Confirm that Claude Code responds with an analysis of your project. A successful connection shows Claude Code’s interactive prompt without authentication errors. If Claude Code displays an error like “Failed to get token” or “model is not available”, see the Troubleshooting section.

Create project context with CLAUDE.md

You can give Claude Code extra instructions and guidance using CLAUDE.md files. Claude Code looks for CLAUDE.md files in the following places and merges them top-down (that is, the files load in order, with later files overriding earlier ones), giving it context about your personal preferences, project-specific details, and the current task:

~/.claude/CLAUDE.md – Global defaults across all projects
./CLAUDE.md – Repository root settings
./current-dir/CLAUDE.md – Current directory specifics

Claude Code also supports project rules (.claude/rules/*.md) and local memory (CLAUDE.local.md) for more granular control. For the full memory hierarchy, see Claude Code memory documentation. The following example shows a CLAUDE.md file for a Microsoft Agent Framework project. Create a similar file in your project root to help Claude Code understand your codebase:

# Project: Customer Service Agent

## Overview
Multi-agent system using Microsoft Agent Framework with Foundry Agent Service.

## Tech Stack
- Python 3.11+
- agent-framework (Microsoft Agent Framework Python SDK)
- Microsoft Foundry for hosted agents
- MCP tools for enterprise data access

## Architecture
- `src/agents/` - Agent definitions (triage, specialist, escalation)
- `src/tools/` - MCP tool implementations
- `src/workflows/` - Multi-agent orchestration
- `tests/` - pytest with async fixtures

## Commands
```bash
# Run locally
python -m src.main

# Test
pytest tests/ -v

# Deploy to Foundry Agent Service
az ai agent deploy --config deploy.yaml
```

## Code Patterns
Use `AzureAIAgentClient` with `AzureCliCredential`:
```python
async with AzureAIAgentClient(async_credential=AzureCliCredential()) as client:
    agent = client.create_agent(instructions="...", tools=[...])
```

## Current Sprint
- Implementing RAG grounding with Foundry IQ
- Adding Fabric connector for sales data

After you create a CLAUDE.md file, start Claude Code in your project directory to load the context:

# Start Claude Code in your project
cd your-project
claude

# Or run a one-off command
claude "explain the agent orchestration in src/workflows/"

Configure Azure RBAC

To grant team members access to your Foundry-hosted Claude models, assign one of the following built-in roles:

Role	Permissions
Foundry User	Invoke models, view deployments
Cognitive Services User	Invoke models, view deployments (legacy Azure AI Services role)

The Foundry RBAC roles were recently renamed. Foundry User, Foundry Owner, Foundry Account Owner, and Foundry Project Manager were previously named Azure AI User, Azure AI Owner, Azure AI Account Owner, and Azure AI Project Manager. You might still see the previous names in some places while the rename rolls out. The role IDs and core permissions are unchanged by the rename.

The Foundry User role is the recommended Foundry-native role. The Cognitive Services User role is a legacy role that also grants model invocation permissions at the Azure resource level. These roles include all required permissions for running Claude Code with Foundry. For more restrictive access, create a custom role scoped to the specific data actions your team needs. For guidance on defining custom roles, see Role-based access control for Microsoft Foundry.

Run Claude Code in GitHub Actions

Claude Code integrates with GitHub Actions for CI/CD automation. Store your API key in the repository’s secret store before using these workflows.

Generate tests on pull requests

Create a workflow that generates tests when agent or tool files change:

name: Generate Agent Tests
on:
  pull_request:
    paths:
      - 'src/agents/**'
      - 'src/tools/**'

jobs:
  generate-tests:
    runs-on: ubuntu-latest
    permissions:
      contents: write
      pull-requests: write
    steps:
      - uses: actions/checkout@v4
      
      - name: Run Claude Code
        uses: anthropics/claude-code-action@v1
        with:
          prompt: |
            Review the changed agent files and generate pytest tests.
            Use async fixtures for AIProjectClient mocking.
            Follow patterns in tests/conftest.py.
          allowed_tools: "edit,write,bash"
        env:
          CLAUDE_CODE_USE_FOUNDRY: "1"
          ANTHROPIC_FOUNDRY_RESOURCE: ${{ secrets.AZURE_FOUNDRY_RESOURCE }}
          ANTHROPIC_FOUNDRY_API_KEY: ${{ secrets.AZURE_FOUNDRY_API_KEY }}

Trigger PR review with @claude

Create a workflow that responds to @claude mentions in pull request comments:

name: Claude PR Assistant
on:
  issue_comment:
    types: [created]

jobs:
  respond:
    if: contains(github.event.comment.body, '@claude')
    runs-on: ubuntu-latest
    permissions:
      contents: read
      pull-requests: write
    steps:
      - uses: actions/checkout@v4
      
      - name: Claude Review
        uses: anthropics/claude-code-action@v1
        with:
          prompt: ${{ github.event.comment.body }}
          context: "PR #${{ github.event.issue.number }}"
        env:
          CLAUDE_CODE_USE_FOUNDRY: "1"
          ANTHROPIC_FOUNDRY_RESOURCE: ${{ secrets.AZURE_FOUNDRY_RESOURCE }}
          ANTHROPIC_FOUNDRY_API_KEY: ${{ secrets.AZURE_FOUNDRY_API_KEY }}

Monitor usage

Monitor Claude Code usage in the Foundry portal:

Go to Microsoft Foundry and open your project.
Navigate to Operate to view usage metrics:
- Token consumption by model
- Request latency
- Error rates and rate-limit hits

To set token limits per request, configure the ANTHROPIC_MAX_TOKENS environment variable:

The ANTHROPIC_MAX_TOKENS variable might not be supported in all Claude Code versions. Check the Claude Code documentation for the latest supported environment variables.

    export ANTHROPIC_MAX_TOKENS=100000

For configuration options that control token usage and cost, such as model-specific token limits, see the Claude Code configuration documentation.

Clean up resources

If you no longer need the Claude model deployments, delete them to free up deployment slots and quota in your resource. Claude models use Global Standard deployments with pay-per-token billing, so idle deployments don’t incur charges. However, each deployment counts against your deployment limit of 32 per resource.

Delete deployments in the Foundry portal

Use either of these navigation paths:

Select Operate from the top navigation, then select Assets > Models from the left navigation. Select the deployment you want to delete, then select Delete.
Select Build from the top navigation, then select Models from the left navigation. Select the deployment name to open its detail page, select the Details tab, then select Delete from the top panel.

Delete deployments with the Azure CLI

Run the az cognitiveservices account deployment delete command for each deployment:

az cognitiveservices account deployment delete \
  --deployment-name <deployment-name> \
  --name <resource-name> \
  --resource-group <resource-group-name>

Replace <deployment-name> with the model deployment name (such as claude-sonnet-4-6), <resource-name> with your Foundry resource name, and <resource-group-name> with the resource group that contains your Foundry resource.

Troubleshooting

Problem	Solution
Authorization failed (HTTP 401/403)	Verify that `az login` completed successfully or that the API key is set correctly. Check that your account has access to the Foundry resource.
Claude Code starts but can’t find models	Verify `ANTHROPIC_FOUNDRY_RESOURCE` matches your resource name and that the `ANTHROPIC_DEFAULT_*_MODEL` values match your deployment names.
Rate limit exceeded (HTTP 429)	Check your quotas in the Foundry portal under Operate > Quotas. Consider adjusting your token limits per the Claude Code configuration documentation.
VS Code extension not connecting	Ensure environment variables are set before launching VS Code. Try launching VS Code from the terminal after setting variables.
WSL + VS Code extension issues	The extension might check for the API key on the Windows host instead of within WSL. Set the environment variable on both the Windows host and WSL, then launch a new terminal from WSL and run `code .`
Region errors	Claude models are only available in East US 2 and Sweden Central.
”Failed to get token from azureADTokenProvider: ChainedTokenCredential authentication failed”	Sign in with `az login`, or set `ANTHROPIC_FOUNDRY_API_KEY` for API key authentication.
”The model `<model-name>` is not available on your foundry deployment”	Deploy the missing model in the Foundry portal. Claude Code requires each model role (Sonnet, Haiku, Opus) to have a corresponding deployment.
”Token tenant does not match resource tenant”	Your Azure CLI is signed in to a different tenant than your Foundry resource. Run `az login --tenant <tenant-id>` to sign in to the correct tenant.
Deployment creation fails in the Foundry portal	Verify you have Contributor or Owner role on the resource group, and your subscription has Azure Marketplace access enabled.
Claude Code prompts for Anthropic login	Verify `CLAUDE_CODE_USE_FOUNDRY=1` is set. Without this variable, Claude Code uses the default Anthropic API.

​Prerequisites

​System requirements

​Deploy a Claude model in Foundry

​Deploy model directly in Foundry

​Alternative: Deploy model using Model Router

​Install Claude Code CLI

​Native install (recommended)

​Installation with Homebrew

​Troubleshoot install location

​Configure Claude Code for Foundry

​Authenticate with Foundry

​Option A: Microsoft Entra ID (recommended)

​Option B: API key

​Bash / WSL

​PowerShell

​Configure the VS Code extension

​Validate the configuration

​Create project context with CLAUDE.md

​Configure Azure RBAC

​Run Claude Code in GitHub Actions

​Generate tests on pull requests

​Trigger PR review with @claude

​Monitor usage

​Clean up resources

​Delete deployments in the Foundry portal

​Delete deployments with the Azure CLI

​Troubleshooting

​Related content

Prerequisites

System requirements

Deploy a Claude model in Foundry

Deploy model directly in Foundry

Alternative: Deploy model using Model Router

Install Claude Code CLI

Native install (recommended)

Installation with Homebrew

Troubleshoot install location

Configure Claude Code for Foundry

Authenticate with Foundry

Option A: Microsoft Entra ID (recommended)

Option B: API key

Bash / WSL

PowerShell

Configure the VS Code extension

Validate the configuration

Create project context with CLAUDE.md

Configure Azure RBAC

Run Claude Code in GitHub Actions

Generate tests on pull requests

Trigger PR review with @claude

Monitor usage

Clean up resources

Delete deployments in the Foundry portal

Delete deployments with the Azure CLI

Troubleshooting

Related content