Made an Open Source Firewall for ChatGPT and other LLMs

I built an open-source gateway that sits between your app and models like OpenAI's, Gemini, or Claude. It acts like a firewall: you can define YAML policies to block prompt injections, redact PII, filter toxic responses, etc.

It's self-hosted, built with FastAPI, and easy to run with Docker. Default config blocks email addresses, try it out and see the guardrails in action.

Happy to answer questions or hear feedback.