Skip to content

Getting Started

This page takes you from zero to a working GPT6-SOL tool-calling request in about five minutes.

1. Requirements

Requirement Version
Python 3.10 or newer
OpenAI API key with access to the GPT-6 family
An MCP server optional — only needed for tool calling

2. Install

pip install mcp-for-copilot

pip install "mcp-for-copilot[server]"

pip install "mcp-for-copilot[all]"

3. Configure

The gateway reads configuration from environment variables, and falls back to a .env file in the current working directory. Real environment variables always win over .env values.

cp .env.example .env

Then set at minimum:

LLM_API_KEY=sk-your-key-here

See Configuration for the full list.

4. First request

As a library

import asyncio
from mcp_for_copilot import Gateway

async def main() -> None:
    async with Gateway(connect_mcp=False) as gateway:
        result = await gateway.chat([{"role": "user", "content": "What is the Model Context Protocol?"}])
        print(result.content)

asyncio.run(main())

As a server

mcp-for-copilot serve --host 0.0.0.0 --port 8080

curl -s http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
        "model": "gpt-6-sol",
        "messages": [{"role": "user", "content": "What is MCP?"}]
      }' | jq -r '.choices[0].message.content'

5. Add tools

Point the gateway at any MCP server. The command below assumes a stdio server that speaks MCP:

import asyncio
from mcp_for_copilot import Gateway, allow_all_approver

async def main() -> None:
    async with Gateway(approver=allow_all_approver()) as gateway:
        result = await gateway.chat([{"role": "user", "content": "What files are in this directory?"}])
        print(result.content)

asyncio.run(main())

The MCP endpoint and allowlist come from MCP_SERVER_URL, MCP_TRANSPORT, and MCP_ALLOWED_TOOLS.

Approval is fail-closed

If you do not pass an approver, every risky tool call is denied. That is intentional. Use allow_all_approver() only in trusted, non-interactive environments.

6. Verify your setup

mcp-for-copilot status


{
  "llm": {"provider": "openai", "api_key_configured": true, "default_model": "gpt-6-sol"},
  "mcp": {"transport": "stdio", "url_configured": false, "allowed_tools": ["read_file", "..."]}
}
status never prints credential values — only booleans.

Next steps