anxmeshhh/verdict ? reverse-engineered prompt

Build me a Python tool called Verdict that helps me verify code changes before they get reviewed.

I want it to be a self hosted CLI that reads a code diff plus a short statement of intent, then checks whether the change actually does what it says by creating scenarios, turning them into runnable tests, and executing those tests in a Docker sandbox. It should feel like a neutral checker, not another coding assistant.

Please make the basic flow simple, with commands like init, check, model, and install hook, so someone can set it up once per project and then run verification on each change. It should work with either a local Ollama model or an OpenAI compatible cloud provider, and it should be okay to switch later. If useful, look up current docs online for the LLM and Docker pieces.

Also add the safety net where git push can be blocked if the change looks risky or can not be verified, plus some way to view logs or a report when something fails.