CopilotMask - GitHub Copilot - Ollama - Proxy Service

Language: English | 中文

📖 Project Overview

CopilotMask is a GitHub Copilot BASE_URL customization tool implemented through Ollama proxy, specifically designed for VS Code's GitHub Copilot. By running this project locally or remotely, you can use custom AI API providers in GitHub Copilot, breaking through official limitations and using your own API keys.

🎯 Core Features

🔗 API Proxy Forwarding: Forward GitHub Copilot requests to custom AI service providers
🤖 Multi-Model Support: Support multiple AI models including OpenAI, Claude, DeepSeek, etc.
🧠 Reasoning Process Display: Special optimization for DeepSeek to show complete reasoning processes
📊 Powerful Logging System: Provide detailed request and response logs for easy debugging
🔄 Smart Configuration Management: Support intelligent priority configuration with .env files and environment variables

💡 Project Significance

🆓 Break Free from Subscription Limits: Use powerful AI models without GitHub Copilot paid membership
✅ Payment Alternative: Perfect solution for users who cannot easily purchase or add official payment methods
🔧 Easy Debugging: Deep understanding of Copilot's working mechanism through detailed logging system
🌐 Third-Party Relay Support: Perfect support for various third-party API relay stations
⚡ High Performance: Built on FastAPI with excellent performance

🚀 Quick Start

1. Requirements

Python 3.8+
pip package manager

2. Install Dependencies

pip install -r requirements.txt

3. Configure Environment Variables

Copy .env.example to .env file:

cp .env.example .env

Edit the .env file to configure your API keys and service addresses (see Configuration section for details).

4. Start Service

python main.py

The service will start at http://localhost:11434.

5. Configure GitHub Copilot

Open GitHub Copilot settings
Select "Manage Models"
Choose "Ollama"
Select models starting with virtual-
Start using your custom API!

🔧 Troubleshooting

Model Addition Failure： If you encounter failures when adding models in GitHub Copilot, this is typically because of VS Code Copilot. When model addition fails, backend logs show no registration requests received from VS Code Copilot, indicating that Copilot did not attempt to access the service.

Solution：

First register all available Ollama models in Copilot settings
Then unregister all models
Restart VS Code or refresh Copilot status
Retry adding the desired virtual models

This operation will reset Copilot's internal state and typically resolves model addition failures.

⚙️ Configuration Guide

🔧 Core Configuration Items

The configuration file supports intelligent priority system:

When .env file exists, use .env configuration preferentially
When .env doesn't exist, use .env.example as configuration template
System environment variables can supplement file configuration

Basic Configuration

# Environment mode
ENVIRONMENT=development

# Server configuration
SERVER_HOST=0.0.0.0          # Service binding address
SERVER_PORT=11434            # Service port (recommended to keep 11434)

# Ollama configuration
REAL_OLLAMA_BASE_URL=http://1.2.3.4:11434  # Real Ollama service address
OLLAMA_TIMEOUT=10.0          # Ollama source request timeout

Logging Configuration

LOG_LEVEL=INFO               # Log level: DEBUG, INFO, WARNING, ERROR
LOG_FILE_NAME=ollama_proxy.log  # Log file name

API Provider Configuration

OpenAI Configuration

OPENAI_API_KEY=your_openai_api_key_here
OPENAI_BASE_URL=https://api.openai.com/v1
OPENAI_MAX_TOKENS=4096
OPENAI_MODELS=gpt-4o-mini,gpt-4o,gpt-3.5-turbo
OPENAI_CAPABILITIES=completion,chat,tools,vision

Claude Configuration

CLAUDE_API_KEY=your_claude_api_key_here
CLAUDE_BASE_URL=https://api.anthropic.com
CLAUDE_MAX_TOKENS=25600
CLAUDE_MODELS=claude-sonnet-4-20250514,claude-opus-4-20250514
CLAUDE_CAPABILITIES=completion,chat,tools,vision

DeepSeek Configuration

DEEPSEEK_API_KEY=your_deepseek_api_key_here
DEEPSEEK_BASE_URL=https://api.deepseek.com
DEEPSEEK_MAX_TOKENS=30000
DEEPSEEK_MODELS=deepseek-chat,deepseek-reasoner
DEEPSEEK_CAPABILITIES=completion,chat,tools

🎛️ Capabilities Configuration (Usually No Need to Modify)

CAPABILITIES configuration supports two formats:

One-dimensional array (all models have same capabilities):
```
OPENAI_CAPABILITIES=completion,chat,tools,vision
```

Two-dimensional array (each model has different capabilities):

OPENAI_CAPABILITIES=completion,chat,tools,vision|chat,tools|completion,chat

Supported capability types:

completion: Text completion
chat: Conversational chat
tools: Tool calling
vision: Visual understanding

🏗️ Project Architecture

Core Modules

main.py: Main application entry point, FastAPI application creation and startup
config.py: Configuration management module, intelligent loading of environment variables
model_manager.py: Model manager, handles virtual model registration and client initialization
routes.py: Route handler, defines API routes and request forwarding logic
api_handlers.py: API handlers, processes API calls from different service providers
middleware.py: Middleware, handles request logging and response processing
logger.py: Logging system, provides structured logging
utils.py: Utility functions, contains various helper functions

🔄 Workflow

Request Reception: GitHub Copilot sends OpenAI format requests
Model Recognition: Detect if the model name in the request is a virtual model
Request Forwarding:
- Virtual model → Forward to corresponding remote API service provider
- Non-virtual model → Forward to Ollama source
Format Conversion: Handle conversion between different API formats (e.g., Claude)
Response Return: Convert response to OpenAI format and return to Copilot

🎨 Special Features

DeepSeek Reasoning Process Display

The project has special optimization for DeepSeek, capable of displaying complete reasoning processes, allowing you to understand AI's thinking process.

Smart Configuration System

Supports complex configuration priority management, allowing flexible use of different configurations in different environments.

📊 Logging System

The project provides a powerful logging system:

Structured logging
Detailed tracking of requests and responses
Support for different log levels
Dual output to file and console

🚧 Future Development Plans

🖥️ GUI Interface: Graphical configuration and management interface
🔧 System Service: Support running as system service
☁️ Serverless Adaptation: Adapt to serverless platforms like function computing
🌐 Forwarding Configuration Station: Create server for user_id-based configuration mapping

🤝 Contributing

Welcome to submit Issues and Pull Requests to help improve the project!

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

ℹ️ Support

If you encounter problems during use, please:

Check if the configuration file is correct
Check the log file for error information
Submit an Issue describing the problem

Enjoy the freedom and convenience brought by CopilotMask! 🎉

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
README_zh.md		README_zh.md
api_handlers.py		api_handlers.py
config.py		config.py
file_truncate.py		file_truncate.py
logger.py		logger.py
main.py		main.py
middleware.py		middleware.py
model_manager.py		model_manager.py
requirements.txt		requirements.txt
routes.py		routes.py
utils.py		utils.py

License

inklife/CopilotMask

Folders and files

Latest commit

History

Repository files navigation