wtf is google-ai-studio-proxy?

This project is a small proxy tool that connects Google AI Studio models, such as Gemini and Gemma, to Janitor AI, a third-party chat platform. It sits between the two services, handling message formatting, managing the model's safety filters, and cleanly separating the model's internal reasoning output from its final reply. To use it, you need a Google AI Studio API key and Python installed, version 3.12 on Windows or 3.10 and up on Linux and Mac. Setup involves downloading or cloning the repository, creating a Python virtual environment to keep its dependencies separate from the rest of your system, and installing the required packages with pip. Once installed, running the main script starts the proxy and prints a Cloudflare tunnel link in the terminal. That link is then pasted into Janitor AI's proxy settings, along with your Google API key and the name of the model you want to use, such as gemma-4-31b-it. The tool has a few configurable settings in its config file. One toggle enables extra prompt handling meant to reduce how often Google's content filters block generation, another controls whether the model uses its internal reasoning step before answering, since turning this off can cause empty or broken responses for some models, and a third decides whether that internal reasoning is printed to the terminal. This is a personal utility built for a specific use case (connecting to Janitor AI) rather than a general-purpose library, and the README does not mention a license, so its reuse terms are not stated. It is aimed at people who already use Janitor AI and want to plug in Google's models instead of the platform's built-in options.