hypit-ai/hypit ? reverse-engineered prompt

Build me a tool that lets an AI agent take a reference video and turn it into a full cloned version with new wording, new face, new B roll, captions, and effects.

I want it to work from a simple text based video script, and then generate a finished video from that workflow. It should also let me start from a template or just describe the kind of video I want, then have the agent create the workflow for me. Make it easy to generate lots of variants from the same base idea, and let me preview the result as I edit it.

If some parts can be done without calling an AI model, like captions, motion graphics, or code rendered visuals, do that too so it stays faster and cheaper. Use the existing project style and docs if they help, and look up current docs online if you need to. Make the whole thing feel like one command from idea to export.