GitHub Copilot OpenAI-Compatible Provider for AI SDK

This package provides an OpenAI-compatible interface for the GitHub Copilot API, designed to work seamlessly with the Vercel AI SDK.

Features

• Full TypeScript support
• Seamless integration with Vercel AI SDK
• Easy to use API matching other AI SDK providers
• Flexible authentication via headers (Bearer token)
• Automatic endpoint switching: Uses /responses endpoint for Codex models, /chat/completions for others
• Smart request formatting: Automatically converts messages array to OpenAI Responses API input format for Codex models

Installation

npm install @opeoginni/github-copilot-openai-compatible

Usage

Basic Usage

import { createGithubCopilotOpenAICompatible } from '@opeoginni/github-copilot-openai-compatible';
import { generateText } from 'ai';

// Create the provider instance
const githubCopilot = createGithubCopilotOpenAICompatible({
  baseURL: 'https://api.githubcopilot.com',
  name: 'githubcopilot',
  headers: {
    Authorization: `Bearer ${process.env.COPILOT_TOKEN}`,
    "Copilot-Integration-Id": "vscode-chat", // These configs must be provided
    "User-Agent": "GitHubCopilotChat/0.26.7",
    "Editor-Version": "vscode/1.104.1",
    "Editor-Plugin-Version": "copilot-chat/0.26.7"
  },
});

// Use the chat model
const { text } = await generateText({
  model: githubCopilot.chatModel('gpt-4o'),
  prompt: 'Create a function to calculate the Fibonacci sequence',
});

console.log(text);

Using Codex Models

The provider automatically handles the different API format for Codex models:

import { createGithubCopilotOpenAICompatible } from '@opeoginni/github-copilot-openai-compatible';

const githubCopilot = createGithubCopilotOpenAICompatible({
  baseURL: 'https://api.githubcopilot.com',
  name: 'githubcopilot',
  headers: {
    "Copilot-Integration-Id": "vscode-chat", // These configs must be provided
    "User-Agent": "GitHubCopilotChat/0.26.7",
    "Editor-Version": "vscode/1.104.1",
    "Editor-Plugin-Version": "copilot-chat/0.26.7"
  },
  apiKey: process.env.COPILOT_TOKEN
});

// This will automatically use the /responses endpoint with 'item' format
const codexModel = githubCopilot.chatModel('gpt-5-codex');

const { text } = await generateText({
  model: codexModel,
  prompt: 'Write a Python function to sort a list',
});

Minimal Configuration

import { createGithubCopilotOpenAICompatible } from '@opeoginni/github-copilot-openai-compatible';

// Minimal setup - just provide the auth token and required headers
const githubCopilot = createGithubCopilotOpenAICompatible({
  baseURL: 'https://api.githubcopilot.com',
  name: 'githubcopilot',
  headers: {
    "Copilot-Integration-Id": "vscode-chat", // These configs must be provided
    "User-Agent": "GitHubCopilotChat/0.26.7",
    "Editor-Version": "vscode/1.104.1",
    "Editor-Plugin-Version": "copilot-chat/0.26.7"
  },
  apiKey: process.env.COPILOT_TOKEN
});

const model = githubCopilot.chatModel('gpt-4o');

Supported Models

Claude Models

• claude-opus-4 - Claude Opus 4
• claude-opus-41 - Claude Opus 4.1
• claude-3.5-sonnet - Claude 3.5 Sonnet
• claude-3.7-sonnet - Claude 3.7 Sonnet
• claude-3.7-sonnet-thought - Claude 3.7 Sonnet with Reasoning
• claude-sonnet-4 - Claude Sonnet 4
• claude-sonnet-4.5 - Claude Sonnet 4.5

GPT Models

• gpt-4.1 - GPT-4.1
• gpt-4o - GPT-4 Optimized
• gpt-5 - GPT-5
• gpt-5-codex - GPT-5 Codex (uses /responses endpoint)
• gpt-5-mini - GPT-5 Mini

Gemini Models

• gemini-2.0-flash-001 - Gemini 2.0 Flash
• gemini-2.5-pro - Gemini 2.5 Pro

Other Models

• grok-code-fast-1 - Grok Code Fast
• o3 - OpenAI O3
• o3-mini - OpenAI O3 Mini
• o4-mini - OpenAI O4 Mini

Plus any custom model ID supported by Github Copilot (type-safe with TypeScript)

How It Works

Endpoint Routing

The provider automatically routes requests to the correct endpoint based on the model ID:

• Codex Models (gpt-5-codex or any model containing 'codex'):

  • Uses /responses endpoint
  • Converts messages array to OpenAI Responses API input format
  • Transforms message content parts (e.g., text → input_text, image_url → input_image)
  • System messages become developer role messages

• All Other Models:

  • Uses standard /chat/completions endpoint
  • Uses standard OpenAI-compatible messages array format

This means you don't need to worry about the underlying API differences - the provider handles it automatically!

API Reference

createGithubCopilotOpenAICompatible(options)

Creates a new GitHub Copilot provider instance.

Options:

• baseURL (required): Base URL for API calls
• name (required): Provider name
• headers?: Custom headers to include in requests
• queryParams?: Optional URL query parameters
• fetch?: Custom fetch implementation
• includeUsage?: Include usage information in responses
• supportsStructuredOutputs?: Enable structured outputs support

Returns: A provider instance with chatModel() and languageModel() methods, also callable as a function.

githubCopilot

Pre-configured default provider instance with common GitHub Copilot headers. You'll need to provide authentication separately.

Getting Your GitHub Copilot API Key

To use this provider, you'll need a GitHub Copilot API token. This is typically obtained through GitHub Copilot's authentication flow in VS Code or other supported editors.

License

MIT