Aether

Supercharge your GitHub Copilot with 20+ AI providers

Install · Quick Start · Features · Providers

A powerful VS Code extension that provides model support for GitHub Copilot Chat, seamlessly integrating 20+ AI providers including ZhipuAI, MiniMax, MoonshotAI, DeepSeek, Codex (OpenAI), Chutes, OpenCode, Blackbox, Vercel AI, Cline, and custom OpenAI/Anthropic compatible models.

Aether now includes an MCP (Model Context Protocol) Bridge that exposes VS Code Language Model tools to external AI clients:

• What it does: Exposes vscode.lm.tools to MCP clients like Claude Code, Claude Desktop, Cursor, and Windsurf
• Auto-Start: Enabled by default (aether.mcpBridge.http.enabled)
• Port: Runs on http://localhost:59684 (configurable via aether.mcpBridge.http.port)
• Features:
  • Activity Bar panel with tool management (checkboxes to expose/hide tools)
  • Tools grouped by extension with exposure controls
  • HTTP MCP Server using StreamableHTTP transport
  • Dynamic tool discovery with configurable polling
  • Rich tooltips with metadata and input schemas
  • Status bar integration showing server status and tool count

Quick Start:

1. Launch Extension Development Host (F5)
2. Find Aether MCP Bridge in the Activity Bar
3. Click to view and manage tools
4. Run "Aether: Copy MCP Config to Clipboard"
5. Paste into your MCP client config (e.g., .mcp.json)

Example MCP Client Config:

{
  "mcpServers": {
    "aether-mcp-bridge": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "http://localhost:59684/mcp", "--allow-http"]
    }
  }
}

• Add unlimited accounts for each AI provider
• Quick switch between accounts with Ctrl+Shift+Q / Cmd+Shift+Q
• Visual account status in the status bar
• Secure credential storage using VS Code Secret Storage

• Auto-switch when hitting rate limits or quota exhaustion
• Intelligent retry with exponential backoff strategy
• Real-time quota monitoring and usage statistics
• Seamless failover without interrupting your workflow

Example usage in Copilot Chat:

@workspace #zhipuWebSearch What are the latest features in TypeScript 5.5?

Enable in Settings:

{
    "chp.fimCompletion.enabled": true,
    "chp.nesCompletion.enabled": true
}

Keybindings:

# Clone the repository
git clone https://github.com/OEvortex/aether.git
cd aether

# Install dependencies
npm install

# Build the extension
npm run compile

# Package as .vsix
npm run package

# Install the packaged extension
code --install-extension aether-*.vsix

1. Open GitHub Copilot Chat
2. Click the model dropdown
3. Select a model from your configured provider (e.g., ⦿ ZhipuAI > glm-4.5)

Cmd+Shift+P → "Aether: Settings"
→ Select a provider
→ Add accounts with API keys

Cmd+Shift+P → "Aether: Settings"
→ Select a provider
→ Toggle "Load Balance" for automatic account switching

Follow these simple steps to add and manage providers using the Settings page:

Press Cmd+Shift+P (macOS) or Ctrl+Shift+P (Windows/Linux) and type:

Aether: Settings

Click on the provider you want to configure (e.g., ZhipuAI, MiniMax, MoonshotAI, etc.)

Enter your API key and configure provider settings:

• API Key: Your provider's API key
• Base URL: Custom API endpoint (optional)
• Additional Settings: Provider-specific settings exposed by the extension manifest, such as endpoint, SDK mode, and other provider options

Toggle the "Load Balance" switch to enable automatic account switching when rate limits are hit.

• Add Multiple Accounts: Add multiple API keys per provider for load balancing
• Edit Settings: Click the edit icon to modify provider details and provider-specific settings
• Delete Account: Remove accounts you no longer need
• Switch Account: Use Ctrl+Shift+Q / Cmd+Shift+Q for quick switching
• Load Balance: Automatically distribute requests across accounts
• Quota Tracking: Monitor usage and remaining quota in real-time

Add your own OpenAI or Anthropic compatible models:

1. Run Aether: Compatible Provider Settings
2. Click "Add Model"
3. Configure your model:

{
    "id": "my-custom-model",
    "name": "My Custom Model",
    "baseUrl": "https://api.example.com/v1",
    "apiKey": "your-api-key",
    "model": "model-name",
    "maxInputTokens": 128000,
    "maxOutputTokens": 8192,
    "sdkMode": "openai",
    "capabilities": {
        "toolCalling": true,
        "imageInput": false
    }
}

# Install dependencies
npm install

# Build in development mode
npm run compile:dev

# Build for production
npm run compile

# Watch mode
npm run watch

# Run linting
npm run lint

# Format code
npm run format

# Package extension
npm run package

copilot-helper/
├── src/
│   ├── extension.ts          # Entry point
│   ├── accounts/             # Multi-account management
│   ├── copilot/              # Core Copilot integration
│   ├── providers/            # AI provider implementations
│   │   ├── providerRegistry.ts  # Provider registry
│   │   ├── zhipu/            # ZhipuAI provider
│   │   ├── minimax/          # MiniMax provider
│   │   ├── moonshot/         # MoonshotAI provider
│   │   ├── codex/            # OpenAI Codex
│   │   └── ...               # Other providers
│   ├── tools/                # Web search tools
│   ├── types/                # TypeScript definitions
│   ├── ui/                   # Settings pages
│   └── utils/                # Shared utilities
├── dist/                     # Compiled output
├── package.json              # Extension manifest
└── tsconfig.json             # TypeScript config

Special thanks to these amazing projects:

LLMux	GCMP

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