⚡ GateX

Your gateway to AI models — Access VS Code LLM providers via a simple HTTP API.

Turn your GitHub Copilot subscription into a local AI API server. Use Claude, GPT-4o, Gemini, and more with any OpenAI or Anthropic SDK.

✨ Features

• 🚀 Zero Configuration — Starts automatically, default port 24680
• 🔌 Dual API Format — OpenAI + Anthropic compatible endpoints
• 🌊 SSE Streaming — Real-time streaming responses
• 🔄 Smart Retry — Exponential backoff for transient failures
• 🎯 Multi-Model Support — Access Claude, GPT-4o, Gemini, and more
• ⚙️ IDE Config Generator — One-click setup for Claude Code, .env, etc.

🚀 Quick Start

1. Install & Activate

The extension starts automatically when VS Code opens. Look for the status bar item:

⚡ GateX :24680

2. Use in Your Code

OpenAI Format:

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:24680/v1",
    api_key="gatex"  # Any value works
)

response = client.chat.completions.create(
    model="claude-sonnet-4",  # Or: gpt-4o, gpt-4o-mini, etc.
    messages=[{"role": "user", "content": "Hello!"}],
    stream=True
)

for chunk in response:
    print(chunk.choices[0].delta.content or "", end="")

Anthropic Format:

from anthropic import Anthropic

client = Anthropic(
    base_url="http://localhost:24680",
    api_key="gatex"
)

message = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}],
    stream=True
)

for event in message:
    if event.type == "content_block_delta":
        print(event.delta.text, end="")

3. Configure Your IDE

Press Cmd+Shift+P (or Ctrl+Shift+P) → GateX: Configure for IDE to auto-generate configs for:

• Claude Code — Updates .claude/settings.json with dual model selection (main + fast), onboarding bypass, and telemetry opt-out
• Codex CLI — Generates ~/.codex/config.toml with GateX as model provider
• .env File — Generates both OpenAI and Anthropic environment variables
• Clipboard — Copies export or $env: commands for your terminal

The config generator uses smart merge — it only updates GateX-related keys, preserving all your other settings. Model picker shows grouped results with ⭐ recommendations per IDE.

📡 API Endpoints

⚙️ Configuration

🎯 Commands

🔧 Troubleshooting

No models available?

Make sure you have GitHub Copilot or another LLM extension installed and signed in.

Port already in use?

Set gatex.port to 0 for auto-selection, or choose a different port.

Request timeout?

Increase gatex.timeout in settings for long-running requests.

Model not found?

GateX uses fuzzy matching — try shorter names like claude-sonnet-4 or gpt-4o. Run GateX: Show Connection Info to see available models.

🏗️ Architecture

┌─────────────────────────────────────────────────┐
│  Your App / IDE / CLI                            │
│  (OpenAI SDK / Anthropic SDK / curl)             │
└─────────────┬───────────────────────────────────┘
              │ HTTP (localhost:24680)
┌─────────────▼───────────────────────────────────┐
│  GateX Extension                                 │
│  ┌─────────────────────────────────────────┐     │
│  │  server.ts — HTTP Router                │     │
│  │  • /v1/chat/completions (OpenAI)        │     │
│  │  • /v1/messages (Anthropic)             │     │
│  │  • Retry with exponential backoff       │     │
│  └──────────────┬──────────────────────────┘     │
│  ┌──────────────▼──────────────────────────┐     │
│  │  models.ts — Model Manager              │     │
│  │  • Exact + fuzzy matching               │     │
│  │  • 30s model cache                      │     │
│  └──────────────┬──────────────────────────┘     │
│  ┌──────────────▼──────────────────────────┐     │
│  │  configGenerator.ts — IDE Setup         │     │
│  │  • Claude Code (dual model + onboarding)│     │
│  │  • Codex CLI (TOML config)              │     │
│  │  • .env, clipboard                      │     │
│  │  • Smart merge (preserve existing)      │     │
│  └─────────────────────────────────────────┘     │
└─────────────┬───────────────────────────────────┘
              │ vscode.lm API (zero auth)
┌─────────────▼───────────────────────────────────┐
│  VS Code Language Model Providers                │
│  (GitHub Copilot, etc.)                          │
└─────────────────────────────────────────────────┘

📦 Source Files

Total: ~1,385 lines — lean and focused.

📄 License

MIT

🙏 Acknowledgments

Built with ❤️ using the VS Code Language Model API.