Skip to content

PROJECT_OVERVIEW.md

Latest commit

 

History

History
120 lines (92 loc) · 3.47 KB

PROJECT_OVERVIEW.md

File metadata and controls

120 lines (92 loc) · 3.47 KB

Copilot Ollama Bridge - Project Overview

🎯 Project Goal

Create a VS Code extension that exposes GitHub Copilot through Ollama-compatible API endpoints, solving the token extraction problem by using VS Code's official Language Model API.

📁 Project Structure

copilot-ollama-bridge/
├── package.json              # Extension manifest and dependencies
├── tsconfig.json             # TypeScript configuration
├── README.md                 # Complete documentation
├── setup.sh                  # Automated installation script
├── test-api.sh               # API testing script
├── .vscode/
│   └── launch.json           # VS Code development configuration
└── src/
    └── extension.ts          # Main extension implementation

🔧 Key Features

✅ Ollama-Compatible API

  • GET /api/tags - List available models
  • POST /api/generate - Generate text completions
  • POST /api/chat - Chat-style completions
  • 100% compatible with existing Ollama clients

✅ VS Code Integration

  • Uses official vscode.lm.selectChatModels() API
  • Automatic GitHub Copilot authentication
  • Status bar integration with server status
  • Output channel for request logging
  • Command palette controls

✅ No Token Extraction

  • No filesystem token hunting
  • No OS-specific path handling
  • No brittle storage format parsing
  • Official authentication flow through VS Code

🚀 Quick Start

  1. Setup the extension:

    cd copilot-ollama-bridge
./setup.sh

  2. Restart VS Code - extension auto-starts

  3. Configure Cline:

    • Ollama URL: http://localhost:11434
    • Model: copilot:latest

  4. Test the API:

    ./test-api.sh

🔄 Problem Solved

Before (Manual Token Extraction)

❌ Brittle filesystem token hunting
❌ OS-specific path variations  
❌ Storage format changes breaking app
❌ Authentication edge cases
❌ No user consent handling

After (VS Code Extension)

✅ Official VS Code Language Model API
✅ Automatic authentication
✅ Proper user consent dialogs
✅ Stable, supported interface
✅ Future-proof implementation

📊 Usage with Cline

  1. Install and start the extension
  2. Configure Cline with:
    • URL: http://localhost:11434
    • Model: copilot:latest
  3. Use Cline normally - it now uses Copilot through the bridge

🏗️ Architecture

Cline → HTTP Request → VS Code Extension → Language Model API → GitHub Copilot
                         ↓
                   Ollama Format ←→ VS Code Format

The extension acts as a translation layer between Ollama's API format and VS Code's Language Model API.

🎮 Commands Available

  • 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 server status (🤖 icon)
  • Output Channel: "Copilot Bridge" for detailed logs
  • Web Interface: http://localhost:11434 for documentation

✅ Benefits

  1. Reliable: Uses official, stable APIs
  2. Secure: Proper authentication flow
  3. Compatible: Drop-in Ollama replacement
  4. Maintainable: VS Code handles authentication
  5. Future-proof: Official API support

This approach completely eliminates the token extraction problem and provides a robust, officially-supported way to use GitHub Copilot with Ollama-compatible tools like Cline.