protectai/ai-exploits ? reverse-engineered prompt

Build me a Python based security research repo that collects real world AI and ML exploit examples and scanning templates for responsibly disclosed vulnerabilities.

I want it organized by product, with separate folders for tools like Ray, MLflow, FastAPI, Gradio, H2O, BentoML, Triton, AnythingLLM, and similar AI infrastructure projects. For each one, include the kinds of proof of concept files security researchers would use, like Metasploit modules, Nuclei templates, and simple CSRF demo pages when they make sense.

Make it easy to run in Docker, so someone can build one image and get the right tooling ready to go, including Metasploit and Nuclei. Add a clear README that explains what the repo is for, how to build the container, and how to use the modules and templates in a safe research setting. Keep the tone professional and focused on awareness and responsible disclosure. If you need to, look up the current docs for the tools you’re using so the setup feels up to date.