copilot2api

A lightweight Go proxy that exposes GitHub Copilot as OpenAI-compatible, Anthropic-compatible, Gemini-compatible, and AmpCode-compatible API endpoints.

Features

OpenAI API Compatible: /v1/chat/completions, /v1/models, /v1/embeddings, /v1/responses
Anthropic API Compatible: /v1/messages, /v1/messages/count_tokens
Gemini API Compatible: /v1beta/models, /v1beta/models/{model}:generateContent, etc.
AmpCode Compatible: /amp/v1/* routes
Streaming Support: Full SSE streaming for OpenAI and Anthropic formats
Auto Authentication: GitHub Device Flow OAuth with automatic token refresh
Prompt Caching: cache_control fields are passed through transparently
1M Context: anthropic-beta: context-1m-* header auto-appends -1m to model ID
Thinking Mode: thinking.type: "enabled" is rewritten to "adaptive" for Claude Code compatibility
Usage Monitoring: Built-in /usage endpoint (Control Plane) for quota tracking
Multi-Account (fork feature): One process manages multiple Copilot accounts with a Control Plane API

Quick Start

Docker

docker run -it --rm \
  -p 127.0.0.1:7777:7777 \
  -p 127.0.0.1:7778:7778 \
  -v ~/.config/copilot2api:/root/.config/copilot2api \
  -e API_TOKEN=your-api-token-here \
  -e ADMIN_TOKEN=your-admin-token-here \
  ghcr.io/senorsen/copilot2api:latest

Docker Compose

services:
  copilot2api:
    image: ghcr.io/senorsen/copilot2api:latest
    ports:
      - "127.0.0.1:7777:7777"
      - "127.0.0.1:7778:7778"
    volumes:
      - ${HOME}/.config/copilot2api:/root/.config/copilot2api
    environment:
      API_TOKEN: your-api-token-here
      ADMIN_TOKEN: your-admin-token-here

Binary

# Linux x64
curl -L -o copilot2api \
  https://github.com/Senorsen/copilot2api/releases/latest/download/copilot2api-linux-amd64
chmod +x copilot2api

API_TOKEN=your-api-token-here ADMIN_TOKEN=your-admin-token-here ./copilot2api

Multi-Account Architecture (fork feature)

This fork manages multiple GitHub Copilot accounts in a single process. Each account has its own credentials stored under a UUID v4 directory:

~/.config/copilot2api/
├── 550e8400-e29b-41d4-a716-446655440000/
│   └── credentials.json   ← auto-created after device flow
├── 6ba7b810-9dad-11d1-80b4-00c04fd430c8/
│   └── credentials.json
└── ...

Accounts are managed via the Control Plane API (port 7778). The Data Plane API (port 7777) routes requests to a specific account by account ID in the URL.

Data Plane API — Port 7777

All inference requests go through this port. The URL includes the account ID:

/api/{account_id}/v1/...

Endpoints

Path	Method	Description
`/api/{account_id}/v1/messages`	POST	Anthropic Messages API
`/api/{account_id}/v1/messages/count_tokens`	POST	Anthropic token counting
`/api/{account_id}/v1/chat/completions`	POST	OpenAI Chat Completions
`/api/{account_id}/v1/models`	GET	List available models
`/api/{account_id}/v1/*`	ANY	Other paths are proxied as-is
`/gw/api/v1/messages`	POST	Gateway: Anthropic Messages (load-balanced)
`/gw/api/v1/chat/completions`	POST	Gateway: OpenAI Chat Completions (load-balanced)

Gateway Routes (Load-Balanced)

The /gw/api/... routes act as a load-balancing gateway — they automatically select from all logged-in accounts on each request, with retry on failure.

Routes:

Path	Description
`/gw/api/v1/messages`	Anthropic Messages API (load-balanced)
`/gw/api/v1/chat/completions`	OpenAI Chat Completions (load-balanced)

Features:

Random account selection across all registered accounts.
IP affinity: Within a 5-minute window, the same (client IP, model) pair preferentially reuses the same account.
- Anthropic: Affinity only applies when both the incoming request and the stored session carry cache_control.
- OpenAI: Affinity always applies.
Auto-retry: On failure, the gateway automatically switches to a different account.
GW_EXCLUDE environment variable: Comma-separated list of account_id values to exclude from gateway routing (e.g. GW_EXCLUDE=uuid1,uuid2).

Authentication: Same as the Data Plane — use API_TOKEN via Authorization: Bearer or x-api-key.

Authentication

Set API_TOKEN environment variable. Requests must include one of:

Authorization: Bearer your-api-token-here
x-api-key: your-api-token-here

If API_TOKEN is empty, authentication is skipped (dev mode).

Example

# Anthropic Messages via account ID
curl http://localhost:7777/api/550e8400-e29b-41d4-a716-446655440000/v1/messages \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer your-api-token-here" \
  -d '{"model":"claude-sonnet-4.6","messages":[{"role":"user","content":"Hello!"}],"max_tokens":100}'

# OpenAI Chat Completions via account ID
curl http://localhost:7777/api/550e8400-e29b-41d4-a716-446655440000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer your-api-token-here" \
  -d '{"model":"gpt-4o","messages":[{"role":"user","content":"Hello!"}]}'

Control Plane API — Port 7778

Account management runs on a separate port. All endpoints require:

Authorization: Bearer your-admin-token-here

Set ADMIN_TOKEN environment variable. If empty, authentication is skipped (dev mode).

Endpoints

Add a new account

POST /accounts/login

Initiates GitHub Device Flow. Returns a progress_id and the device code URL.

Optional body:

{ "username_suffix": "@yourorg.com" }

Use username_suffix to restrict which GitHub account can complete the flow.

Response:

{
  "progress_id": "...",
  "user_code": "XXXX-XXXX",
  "verification_uri": "https://github.com/login/device"
}

Poll login progress

GET /accounts/{progress_id}/status

Poll until status changes from "pending". Possible statuses:

Status	Description
`pending`	Waiting for user to complete device code flow
`completed`	Login successful, returns `account_id` and `github_username`
`expired`	Device code timed out, need to start a new login
`error`	Login failed (e.g. username suffix mismatch), includes `error` message

Success:

{
  "status": "completed",
  "account_id": "550e8400-e29b-41d4-a716-446655440000",
  "github_username": "yourname"
}

Error (e.g. username suffix mismatch):

{
  "status": "error",
  "error": "username \"john\" does not end with required suffix \"_microsoft\"",
  "github_username": "john"
}

Expired:

{
  "status": "expired"
}

List all accounts

GET /accounts

Returns all registered accounts with their GitHub username and account ID.

Delete an account

DELETE /accounts/{account_id}

Removes the account and its credentials from disk.

Usage / Quota

GET /usage

Returns quota usage across all registered accounts.

Usage Statistics (opt-in)

Enable with COPILOT2API_STATS_ENABLED=true. Records per-request token usage to JSONL files.

GET /dashboard

Interactive HTML dashboard with stacked bar charts, estimated costs (via LiteLLM pricing data), time range selectors, combinable model/account/reasoning-effort dimensions and filters, and auto-refresh. Requires ADMIN_TOKEN entered in the UI (stored in localStorage).

GET /usage?start=YYYY-MM-DD&end=YYYY-MM-DD[&account_id=...][&model=...][&reasoning_effort=...]

Returns aggregated daily token stats (input, output, cached) per model/account/reasoning effort. Reasoning effort is the client-requested value when provided explicitly, or a classification derived from a supplied thinking budget; requests without either are reported as unspecified. Usage stats cover recorded OpenAI-compatible and Anthropic traffic; native Gemini traffic is not currently recorded. Requires ADMIN_TOKEN via Authorization: Bearer header.

GET /usage/accounts

Lists accounts that have usage data. Requires ADMIN_TOKEN.

GET /usage/pricing

Returns cached LiteLLM model pricing JSON. Automatically fetched on startup and refreshed daily at 3:00 AM UTC. Requires ADMIN_TOKEN.

Environment Variables

Variable	Description	Default
`API_TOKEN`	Data plane auth token (optional)	—
`ADMIN_TOKEN`	Control plane auth token (optional)	—
`GW_EXCLUDE`	Comma-separated `account_id` list to exclude from gateway routing	—
`COPILOT2API_TOKEN_DIR`	Credentials storage directory	`~/.config/copilot2api`
`COPILOT2API_HOST`	Server host	`127.0.0.1`
`COPILOT2API_PORT`	Data plane port	`7777`
`COPILOT2API_ADMIN_PORT`	Control plane port	`7778`
`COPILOT2API_DEBUG`	Enable debug logging	`false`
`COPILOT2API_STATS_ENABLED`	Enable usage stats recording, dashboard, and pricing	`false`
`COPILOT2API_STATS_DIR`	Stats data directory	`~/.config/copilot2api/stats`

Usage with Claude Code

Add to ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "http://127.0.0.1:7777/api/your-account-id-here",
    "ANTHROPIC_API_KEY": "your-api-token-here",
    "ANTHROPIC_MODEL": "claude-opus-4.6",
    "ANTHROPIC_SMALL_FAST_MODEL": "claude-haiku-4.5",
    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC": "1"
  }
}

1M Context Window

When Claude Code sends the anthropic-beta: context-1m-... header, copilot2api automatically appends -1m to the model ID (e.g. claude-opus-4.6 → claude-opus-4.6-1m). Select Opus (1M) via /model in Claude Code to activate it.

Thinking Mode

thinking.type: "enabled" is automatically rewritten to "adaptive" for compatibility with Claude Code's extended thinking requests.

Usage with Codex

Add to ~/.codex/config.toml:

model = "gpt-5.3-codex"
model_provider = "copilot2api"
model_reasoning_effort = "high"
web_search = "disabled"
requires_openai_auth = false

[model_providers.copilot2api]
name = "copilot2api"
base_url = "http://127.0.0.1:7777/api/your-account-id-here/v1"
wire_api = "responses"
api_key = "your-api-token-here"

Usage with Gemini CLI

Add to ~/.gemini/.env:

GOOGLE_GEMINI_BASE_URL=http://127.0.0.1:7777/api/your-account-id-here
GEMINI_API_KEY=your-api-token-here
GEMINI_MODEL=claude-opus-4.6-1m

Docker Image

The image is built FROM scratch — a single static binary plus CA certificates. No shell, no OS packages, no CVEs from base layers.

EXPOSE 7777  # Data Plane
EXPOSE 7778  # Control Plane

Development

go test ./...              # Run tests
go build -o copilot2api .  # Build

License

MIT

Name		Name	Last commit message	Last commit date
Latest commit History 120 Commits
.github/workflows		.github/workflows
anthropic		anthropic
auth		auth
control		control
debug		debug
dist		dist
gateway		gateway
gemini		gemini
internal		internal
migrations		migrations
proxy		proxy
stats		stats
storage		storage
.gitignore		.gitignore
AGENTS.md		AGENTS.md
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
Dockerfile		Dockerfile
Dockerfile.ci		Dockerfile.ci
Dockerfile.simple		Dockerfile.simple
LICENSE		LICENSE
README.md		README.md
go.mod		go.mod
go.sum		go.sum
main.go		main.go

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Repository files navigation

copilot2api

Features

Quick Start

Docker

Binary

Multi-Account Architecture (fork feature)

Data Plane API — Port 7777

Endpoints

Gateway Routes (Load-Balanced)

Authentication

Example

Control Plane API — Port 7778

Endpoints

Add a new account

Poll login progress

List all accounts

Delete an account

Usage / Quota

Usage Statistics (opt-in)

Environment Variables

Usage with Claude Code

1M Context Window

Thinking Mode

Usage with Codex

Usage with Gemini CLI

Docker Image

Development

License

About

Releases

Packages

Contributors

Languages

Uh oh!

Folders and files

Latest commit

History

Repository files navigation

copilot2api

Features

Quick Start

Docker

Binary

Multi-Account Architecture (fork feature)

Data Plane API — Port 7777

Endpoints

Gateway Routes (Load-Balanced)

Authentication

Example

Control Plane API — Port 7778

Endpoints

Add a new account

Poll login progress

List all accounts

Delete an account

Usage / Quota

Usage Statistics (opt-in)

Environment Variables

Usage with Claude Code

1M Context Window

Thinking Mode

Usage with Codex

Usage with Gemini CLI

Docker Image

Development

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages