N1nEmAn/sapper-v2 ? reverse-engineered prompt

Build me the sapper v2 project, an authorized only multi agent pentest coordination tool in Python.

I want a commander style app that can spawn and manage several AI workers, let them talk to each other, share notes, and keep a persistent fact and intent history so work can survive if one session dies. It should support a simple command line flow for starting a campaign, adding roles like commander, scout, worker, and verifier, then dispatching tasks and collecting results in a way that feels like a coordinated team instead of one chat bot.

Please include a clean install script, a few example scenarios, and clear docs for how to run it locally with an AI CLI of my choice. If you need to look up current docs online, do that. Keep it flexible so the model or agent backend can be swapped out easily, and make the default experience feel polished and usable out of the box.