LangChain GitHub Copilot Chat

This package provides a LangChain integration for GitHub Copilot, allowing you to use Copilot's models (including GPT-4o, Claude 3.5 Sonnet, etc.) as standard LangChain BaseChatModel components.

Unlike other integrations, this package mimics the official VS Code Copilot Chat extension behavior, providing access to the full suite of models available to Copilot subscribers.

🚀 Features

• Real Copilot API: Connects to api.githubcopilot.com using official VS Code headers.
• Easy Auth: Built-in GitHub Device Flow for acquiring a valid Copilot Token.
• Model Discovery: Dynamic fetching of all models authorized for your account.
• LangChain Native: Full support for Streaming, Tool Calling, and Async operations.

📦 Installation

pip install -U langchain-githubcopilot-chat

🔐 Authentication

To use GitHub Copilot, you need a valid Copilot Token. You can obtain one interactively using the built-in helper:

from langchain_githubcopilot_chat import get_vscode_token

# This will prompt you to visit a GitHub URL and enter a code
token = get_vscode_token()
print(f"Your Token: {token}")

For custom output handling (e.g., in GUI applications), pass a callback:

from langchain_githubcopilot_chat import get_copilot_token

def on_message(msg):
    # Handle status messages (e.g., display in UI)
    print(f"[Copilot] {msg}")

token = get_vscode_token(callback=on_message)

Alternatively, set it as an environment variable:

export GITHUB_TOKEN="your_copilot_token_here"

🛠 Usage

Chat Models

Access any model supported by Copilot (e.g., gpt-4o, gpt-4o-mini, claude-3.5-sonnet).

from langchain_githubcopilot_chat import ChatGithubCopilot

# Initialize with a specific model
llm = ChatGithubCopilot(
    model="gpt-4o", 
    temperature=0.7
)

# Simple invocation
response = llm.invoke("Explain Quantum Entanglement in one sentence.")
print(response.content)

# Streaming
for chunk in llm.stream("Write a short poem about coding."):
    print(chunk.content, end="", flush=True)

Discovery Available Models

GitHub Copilot periodically updates its available models. You can list what's currently available for your token:

from langchain_githubcopilot_chat import get_available_models

models = get_available_models()
for model in models:
    print(f"ID: {model['id']} - Name: {model.get('name')}")

Embeddings

Use Copilot's embedding models for RAG or semantic search:

from langchain_githubcopilot_chat import GithubcopilotChatEmbeddings

embeddings = GithubcopilotChatEmbeddings(model="text-embedding-3-small")
vector = embeddings.embed_query("GitHub Copilot is awesome!")

📖 Advanced: Tool Calling

from pydantic import BaseModel, Field
from langchain_githubcopilot_chat import ChatGithubCopilot

class GetWeather(BaseModel):
    """Get the current weather in a given location."""
    location: str = Field(..., description="The city and state, e.g. San Francisco, CA")

llm = ChatGithubCopilot(model="gpt-4o")
llm_with_tools = llm.bind_tools([GetWeather])

ai_msg = llm_with_tools.invoke("What's the weather like in Tokyo?")
print(ai_msg.tool_calls)

⚖️ Disclaimer

This project is an independent community integration and is not affiliated with, endorsed by, or supported by GitHub, Inc. Usage of this package must comply with GitHub's Terms of Service.