Azure Foundry on Github Copilot/CLI on VSCODE?

[irl231]

🏷️ Discussion Type

Question

💬 Feature/Topic Area

VS Code

Body

I know how to use the Azure Foundry on Copilot CLI by adding these on .zshrc

export COPILOT_PROVIDER_TYPE="azure"
export COPILOT_PROVIDER_WIRE_API="responses"
# Azure OpenAI endpoint
export COPILOT_PROVIDER_BASE_URL="https://irl321-agent.openai.azure.com/" 
export COPILOT_PROVIDER_API_KEY="xxxxxxxxxx"
export COPILOT_PROVIDER_MAX_PROMPT_TOKENS="272000"
export COPILOT_PROVIDER_MAX_OUTPUT_TOKENS="128000"
# Model Name
export COPILOT_MODEL="gpt-5.3-codex-1"

but how can I use it on VSCODE? on Copilot ext,. CLI/GUI? because i don't know how..

[Julianv3534]

What you have configured there just works for the Copilot Cli / agent, it is not automatically configured for the official github copilot extension, I reccomend you to install an extension which let you use custom providers for example continue.dev, just go to the marketplace and search continue, and then configure as you like, there are more other options like Cline or roo code, but I didn't try them.

Once you installed continue, go to the config file ~/.continue/config.json

and configure

{
  "models": [
    {
      "title": "example",
      "provider": "azure",
      "model": "gpt-5.3-codex-1",
      "apiBase": "https://irl321-agent.openai.azure.com/",
      "apiKey": "YOUR_KEY"
    }
  ]
}

[JayeshWani7]

Currently, those environment variables only work for the Copilot CLI / unofficial OpenAI-compatible flows because they override the provider endpoint manually.

For VS Code GitHub Copilot extension, Microsoft does not officially support custom Azure OpenAI endpoints yet. The extension is tightly connected to GitHub’s own backend/auth flow, so you can’t directly point it to:

https://<your-resource>.openai.azure.com/

inside the normal Copilot extension settings.

What you can do

Option 1 — Use Copilot CLI (works)

What you already configured in .zshrc is valid for:

• Copilot CLI
• OpenAI-compatible wrappers
• Some community forks/tools

After exporting the vars:

source ~/.zshrc

you can run:

copilot

or related CLI commands using your Azure endpoint.

---

Option 2 — Use VS Code with another extension

Instead of the official GitHub Copilot extension, use extensions that support custom OpenAI/Azure providers, for example:

• Continue
• Cline
• Roo Code
• CodeGPT
• Twinny

These usually allow:

• Azure OpenAI Base URL
• API Key
• Deployment/model name

directly in settings.

Example config (varies by extension):

{
  "provider": "azure",
  "apiBase": "https://<resource>.openai.azure.com/",
  "apiKey": "...",
  "deployment": "gpt-5.3-codex-1"
}

---

Option 3 — Build/Open a proxy

Some people create an OpenAI-compatible proxy server and route VS Code AI extensions through it, but this is more advanced.

---