Skip to content

README.md

Latest commit

 

History

History
221 lines (174 loc) · 7.86 KB

README.md

File metadata and controls

221 lines (174 loc) · 7.86 KB

Copilot Ollama Bridge

A VS Code extension that exposes GitHub Copilot through Ollama-compatible API endpoints using VS Code's official Language Model API.

✨ Features

  • 🔗 Ollama-Compatible API: Drop-in replacement for Ollama
  • 🤖 GitHub Copilot Integration: Uses VS Code's official Language Model API
  • 🔐 Automatic Authentication: No manual token extraction needed
  • 🌐 OpenAI-Compatible API: Support for /v1/chat/completions endpoint
  • 🌊 Streaming Support: Real-time streaming responses with SSE
  • 📊 Real-time Status: Status bar integration with server status
  • 📝 Request Logging: Output channel with detailed request logging
  • ⚙️ Configurable: Customizable port and auto-start settings
  • 🔧 CORS Enabled: Cross-origin requests supported for web applications

📋 Requirements

  • VS Code 1.84.0 or higher
  • GitHub Copilot extension installed and authenticated
  • Active GitHub Copilot subscription

🚀 Quick Setup

  1. Install dependencies:

    cd copilot-ollama-bridge
npm install

  2. Compile TypeScript:

    npm run compile

  3. Install extension:

    • Open this folder in VS Code
    • Press F5 to launch Extension Development Host
    • Or package: npx vsce package then install the .vsix file

  4. Configure Tools:

    • For Copilot Bridge: http://localhost:11435
    • For regular Ollama: http://localhost:11434 (for embeddings, etc.)
    • Model: copilot:latest

📡 API Endpoints

GET /api/tags

List available models

curl http://localhost:11435/api/tags

POST /api/generate

Generate text completions

curl -X POST http://localhost:11435/api/generate \
  -H "Content-Type: application/json" \
  -d '{"model":"copilot:latest","prompt":"def fibonacci(n):"}'

POST /api/chat

Chat-style completions

curl -X POST http://localhost:11435/api/chat \
  -H "Content-Type: application/json" \
  -d '{"model":"copilot:latest","messages":[{"role":"user","content":"Write a function"}]}'

🌐 OpenAI-Compatible Endpoints

POST /v1/chat/completions

OpenAI-compatible chat completions with streaming support

# Non-streaming request
curl -X POST http://localhost:11435/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer dummy-token" \
  -d '{
    "model": "copilot:latest",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, how are you?"}
    ],
    "stream": false
  }'

# Streaming request
curl -X POST http://localhost:11435/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer dummy-token" \
  -d '{
    "model": "copilot:latest",
    "messages": [
      {"role": "user", "content": "Tell me a joke"}
    ],
    "stream": true,
    "stream_options": {"include_usage": true}
  }'

Features:

  • ✅ OpenAI API format compatibility
  • ✅ Streaming and non-streaming responses
  • ✅ Complex message content support (text arrays)
  • ✅ Usage statistics with stream_options.include_usage
  • ✅ CORS support for web applications
  • ✅ Works with any OpenAI-compatible client

⚙️ Configuration

Setting Default Description
copilot-bridge.port 11435 Server port (Copilot Bridge default)
copilot-bridge.autoStart true Auto-start on VS Code startup

🎮 Commands

  • Start Copilot Bridge Server - Start the API server
  • Stop Copilot Bridge Server - Stop the API server
  • Restart Copilot Bridge Server - Restart the API server

📊 Monitoring

  • Status Bar: Shows current server status and port
  • Output Channel: View detailed request logs in "Copilot Bridge" channel
  • Web Interface: Visit http://localhost:11435 for status and documentation

🔧 Usage with Tools

Cline/Claude Dev

  1. Set Ollama URL: http://localhost:11435
  2. Select model: copilot:latest

Obsidian with OpenAI-Compatible Plugins

For Obsidian plugins that expect OpenAI API endpoints:

  1. Set API URL: http://localhost:11435/v1/chat/completions
  2. Use any dummy API key (the bridge doesn't require authentication)
  3. Select model: copilot:latest or test

Other Ollama Clients

Any tool that supports Ollama can use this bridge as a drop-in replacement.

OpenAI-Compatible Applications

Any application that uses the OpenAI API format can connect to:

  • Base URL: http://localhost:11435/v1
  • API Key: Not required (use any dummy value)
  • Supported models: copilot:latest, copilot:gpt-4o, copilot:gpt-4, etc.

🔍 Troubleshooting

"No Copilot models available"

  1. Ensure GitHub Copilot extension is installed
  2. Sign in to GitHub Copilot in VS Code
  3. Verify your Copilot subscription is active
  4. Run command: GitHub Copilot: Sign In

Port Already in Use

  1. Check if Ollama is running: ps aux | grep ollama
  2. Stop Ollama: killall ollama
  3. Or change port in extension settings

API Errors

  • Check "Copilot Bridge" output channel for detailed logs
  • Verify Copilot quota hasn't been exceeded
  • Try refreshing GitHub authentication

CORS Issues with Web Applications

If you're getting CORS errors from web applications (like Obsidian):

  1. Ensure the extension is running (check status bar)
  2. Verify the correct port is being used (default: 11435)
  3. Check that the Authorization header is properly formatted
  4. Try restarting the Copilot Bridge Server

OpenAI Endpoint Not Working

  1. Verify you're using the correct endpoint: /v1/chat/completions
  2. Ensure the request includes proper Content-Type header
  3. Check that the extension server is running (not regular Ollama)
  4. Review request format matches OpenAI API specification

🏗️ Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐    ┌──────────────┐
│ Ollama Clients  │───▶│                  │───▶│                 │───▶│              │
│ (Cline, etc.)   │    │                  │    │   VS Code       │    │   GitHub     │
├─────────────────┤    │  Copilot-Ollama  │    │   Language      │    │   Copilot    │
│ OpenAI Clients  │───▶│     Bridge       │───▶│   Model API     │───▶│   Service    │
│ (Obsidian, etc.)│    │                  │    │                 │    │              │
└─────────────────┘    └──────────────────┘    └─────────────────┘    └──────────────┘
                              │
                              ▼
                       ┌──────────────────┐
                       │  Unified Response│
                       │  (Ollama/OpenAI) │
                       └──────────────────┘

The extension acts as a universal bridge, translating both Ollama and OpenAI API requests to VS Code's Language Model API calls and formatting responses back to the appropriate format.

Supported API Formats:

  • 🦙 Ollama API: /api/tags, /api/generate, /api/chat
  • 🤖 OpenAI API: /v1/chat/completions with streaming support

🔄 Why This Approach?

Aspect Manual Token Extraction VS Code Extension
Authentication ❌ Brittle, OS-dependent ✅ Official API
User Consent ❌ Bypassed ✅ Proper dialogs
Quota Limits ❌ Ignored ✅ Respected
Reliability ❌ Token hunting ✅ Stable API
Maintenance ❌ High ✅ Low

📄 License

MIT License