cyberark/FuzzyAI ? reverse-engineered prompt

Build me a tool for testing LLM APIs for jailbreaks and unsafe responses.

I want a command line app that can take a target model or my own REST endpoint, run a set of fuzzing attacks against it, and show me which prompts cause risky or policy breaking outputs. It should support local models through Ollama as well as cloud models like OpenAI, Anthropic, Gemini, Azure, Bedrock, AI21, DeepSeek, and similar providers. I also want a simple web UI for running tests and reviewing results, plus a few example notebooks or sample runs so I can understand how to use it quickly.

Please make it easy to install, easy to extend with new models or attacks later, and include some example config and target prompts. If you need to look up current docs online while wiring it up, go ahead.