Using the Claude model in GitHub Copilot

[likx-66]

Select Topic Area

Question

Body

I want to use the Claude model in GitHub Copilot within VS Code in a Chinese environment. How should we whitelist and allow this through our dedicated line?

[Chami-03]

From what I understand, when using GitHub Copilot in VS Code, the Claude model is accessed through GitHub’s Copilot service rather than connecting directly to Anthropic. Because of that, the network usually needs to allow GitHub related domains instead of Anthropic endpoints. In a restricted environment such as a dedicated line in China, you may need to whitelist domains like github.com, api.github.com, copilot-proxy.githubusercontent.com, and *.githubusercontent.com so that Copilot can work properly. After the network access is allowed, you can sign in to GitHub Copilot in VS Code and select the model from Copilot Chat if the Claude model is available in your Copilot plan or organization settings.

[likx-66]

Thank you for your reply. Is there any relevant official documentation?

[Chami-03]

Yes, GitHub provides official documentation about this. You can check the Copilot allowlist reference, which explains which domains and endpoints should be allowed through a firewall or restricted network for Copilot to work correctly.

There is also documentation about AI model selection in Copilot Chat, which explains how different models (including Claude in supported plans) can be selected inside VS Code.

These pages may help:

1. https://docs.github.com/copilot/reference/copilot-allowlist-reference
2. https://docs.github.com/en/copilot/how-tos/use-ai-models/change-the-chat-model

These docs explain both the network allowlist setup and how models appear in Copilot Chat once access is enabled.

[HamzaAshfaq95]

To use Claude in GitHub Copilot within VS Code in China, whitelist these key domains on your dedicated line: *.github.com, *.githubcopilot.com, copilot-proxy.githubusercontent.com, and api.githubcopilot.com over HTTPS port 443. If you use a corporate proxy, ensure it uses http:// (not https://) and install any SSL interception certificates into the OS trust store. After network access is confirmed, an org admin must enable the Claude policy in GitHub → Organization Settings → Copilot → Policies. Then in VS Code, open Copilot Chat → Manage Models → select Claude to start using it.

[suryanandanbabbar]

GitHub Copilot does not connect directly from VS Code to Anthropic’s Claude API. Instead, all requests go through GitHub’s Copilot service, which then routes the request to the appropriate model provider internally. Because of this architecture, whitelisting Anthropic endpoints alone will not enable Claude in Copilot.

To allow Copilot (including Claude-based models) in a restricted network or dedicated line environment, you should whitelist the GitHub Copilot service domains rather than Claude-specific endpoints.

Common domains that typically need to be accessible include:

github.com
api.github.com
copilot-proxy.githubusercontent.com
*.githubusercontent.com
*.githubassets.com

If your organization uses strict firewall rules or a dedicated outbound line, ensure HTTPS (port 443) access to those domains.

A few additional notes:

• Model selection (e.g., Claude vs GPT models) is handled server-side by GitHub.
• VS Code itself only communicates with GitHub’s Copilot API, not directly with Anthropic.
• If you are in a region with network restrictions, you may also need to verify that GitHub services themselves are reachable without packet filtering or TLS interception issues.

For the most accurate and up-to-date list of required domains, refer to GitHub’s Copilot network documentation:
[https://docs.github.com/en/copilot]