Skip to main content
The proxy server translates requests from Anthropic’s Messages API format to OpenAI’s Chat Completions format and forwards them to GitHub Copilot’s API. This allows Claude Code to use GitHub Copilot as its model provider.

​
Server architecture

The proxy server is built on Node.js’s native HTTP server and runs on port 18080 by default (configurable via COPILOT_PROXY_PORT). 
import { createServer } from "node:http"

const PORT = parseInt(process.env.COPILOT_PROXY_PORT || "18080", 10)
const COPILOT_API_BASE = "https://api.githubcopilot.com"
const USER_AGENT = "claude-code-copilot-provider/1.0.0"

const server = createServer((req, res) => handleRequest(req, res, token))
server.listen(PORT)
The server handles CORS automatically for all endpoints: 
res.setHeader("Access-Control-Allow-Origin", "*")
res.setHeader("Access-Control-Allow-Methods", "GET, POST, OPTIONS")
res.setHeader("Access-Control-Allow-Headers", "*")

​
API endpoints

​
POST /v1/messages

The main endpoint for chat completions. Accepts Anthropic Messages API format and returns responses in the same format.
​
model
string
required
The Anthropic model name (e.g., claude-opus-4-6, claude-sonnet-4-5). The proxy automatically maps this to the corresponding Copilot model.
​
messages
array
required
Array of message objects with role and content fields. Supports user and assistant roles.
​
system
string | array
System prompt as a string or array of content blocks.
​
max_tokens
integer
default:4096
Maximum number of tokens to generate in the response.
​
temperature
number
Sampling temperature between 0 and 1. Higher values make output more random.
​
top_p
number
Nucleus sampling parameter. Alternative to temperature.
​
stream
boolean
default:false
Whether to stream the response using Server-Sent Events.
​
tools
array
Array of tool definitions in Anthropic format. The proxy translates these to OpenAI function calling format.
​
stop_sequences
array
Array of strings that will stop generation when encountered.

​
Example request

{
  "model": "claude-sonnet-4-5",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "What is the capital of France?"
    }
  ],
  "stream": false
}

​
Example response (non-streaming)

​
id
string
Unique identifier for the message.
​
type
string
Always "message" for successful responses.
​
role
string
Always "assistant" for responses.
​
model
string
The model name from the request.
​
content
array
Array of content blocks. Each block has a type field (e.g., "text", "tool_use").
​
stop_reason
string
Reason the model stopped: "end_turn", "tool_use", or "max_tokens".
​
usage
object
Token usage statistics with input_tokens and output_tokens fields.
{
  "id": "msg_1234567890",
  "type": "message",
  "role": "assistant",
  "model": "claude-sonnet-4-5",
  "content": [
    {
      "type": "text",
      "text": "The capital of France is Paris."
    }
  ],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 12,
    "output_tokens": 8,
    "cache_creation_input_tokens": 0,
    "cache_read_input_tokens": 0
  }
}

​
Streaming response

When stream: true, the endpoint returns Server-Sent Events: 
event: message_start
data: {"type":"message_start","message":{...}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"The"}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":" capital"}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn"},"usage":{"output_tokens":8}}

event: message_stop
data: {"type":"message_stop"}

​
GET /health

Health check endpoint that returns the server status.

​
Example response

{
  "status": "ok",
  "provider": "github-copilot"
}

​
GET /models

Returns a list of available models in OpenAI format.

​
Example response

{
  "data": [
    { "id": "claude-opus-4-6", "object": "model" },
    { "id": "claude-sonnet-4-5-20250929", "object": "model" },
    { "id": "claude-sonnet-4-20250514", "object": "model" },
    { "id": "claude-opus-4-5-20251101", "object": "model" },
    { "id": "claude-haiku-4-5", "object": "model" }
  ]
}

​
POST /count_tokens

Estimates token count for a request. Uses a rough heuristic of ~4 characters per token.

​
Example request

{
  "messages": [
    { "role": "user", "content": "Hello world" }
  ],
  "system": "You are a helpful assistant."
}

​
Example response

{
  "input_tokens": 15
}

​
Authentication

The proxy loads the GitHub OAuth token from the auth file at startup: 
const AUTH_FILE = process.env.COPILOT_AUTH_FILE || 
  join(homedir(), ".claude-copilot-auth.json")

function loadAuth() {
  const data = JSON.parse(readFileSync(AUTH_FILE, "utf-8"))
  return data.access_token
}
All requests to GitHub Copilot include: 
const headers = {
  "Content-Type": "application/json",
  "Authorization": `Bearer ${token}`,
  "User-Agent": USER_AGENT,
  "Openai-Intent": "conversation-edits",
  "x-initiator": "user"
}

​
Error handling

The proxy translates GitHub Copilot errors to Anthropic’s error format: 
res.end(JSON.stringify({
  type: "error",
  error: {
    type: copilotRes.status === 401 ? "authentication_error"
      : copilotRes.status === 429 ? "rate_limit_error"
      : copilotRes.status === 403 ? "permission_error"
      : "api_error",
    message: `Copilot API error (${copilotRes.status}): ${errorText}`
  }
}))
The proxy automatically handles token refresh and retry logic for transient errors.

​
Web search support

When a request includes the web_search_20250305 tool, the proxy:
  1. Adds a web_search function tool to the Copilot request
  2. Executes searches using Brave Search API or DuckDuckGo
  3. Feeds results back to the model
  4. Repeats until no more searches are needed
if (wsConfig.hasWebSearch) {
  const { contentBlocks, searchCount } = await handleWebSearchLoop(
    openaiReq, token, wsConfig.maxUses
  )
}
See Message translation for details on tool translation.

​
Configuration

​
COPILOT_PROXY_PORT
integer
default:18080
Port to run the proxy server on.
​
COPILOT_AUTH_FILE
string
default:"~/.claude-copilot-auth.json"
Path to the authentication file.
​
BRAVE_API_KEY
string
API key for Brave Search (optional, improves web search quality).
​
WEB_SEARCH_MAX_RESULTS
integer
default:5
Maximum number of search results to return per query.
​
DEBUG_STREAM
string
Set to "1" to enable debug logging for streaming responses.