Software

Building the GitHub Copilot API Gateway

How a VS Code extension made Copilot, Gemini, Ollama, and other language models available through familiar APIs—and reached 57,000+ combined installs and downloads.

BySuhaib Bin Younis2 min read
  • open-source
  • vscode
  • ai
  • api-gateway

I wanted to use the language models already available in VS Code from the tools I use outside the editor. That became the GitHub Copilot API Gateway: a free, open-source VS Code extension that puts those models behind one local API.

It discovers models made available through VS Code language model providers—including GitHub Copilot, Gemini, Ollama, and other compatible extensions—and exposes them through familiar API formats. So tools such as Cursor, LangChain, Aider, and your own agents can connect to one endpoint instead of needing a separate integration for every provider.

The idea is simple: start the server in VS Code, point an API client at http://127.0.0.1:3030/v1, and choose from the models VS Code has made available. The gateway supports OpenAI-compatible requests, along with Anthropic, Google Gemini, and Llama-compatible APIs. It doesn’t replace those providers; it gives the tools that already speak these formats a way to reach them.

A very good kind of reach

At the time of writing, the extension has passed 57,000 combined installs and downloads across the two registries: 8,053 installs on the VS Code Marketplace and about 49,000 downloads on Open VSX.

The number is exciting. So is the fact that a creator in China made a YouTube video about the extension. I love that a tool I built for my own workflow found its way to someone on the other side of the world—and that they thought it was worth sharing with others.

That’s one of the best parts of open source for me: you put something useful into the world, and people you may never meet find their own way to use it, improve it, or pass it along. Thank you to everyone who installed it, shared it, opened an issue, contributed, or made a video.

Take a look