cohen-liel/hivemind ? reverse-engineered prompt

Build me Hivemind, an AI coding team that takes one plain English request and turns it into finished, tested code without me babysitting it.

I want a Python backend that can plan work, split a feature into tasks, run different agent roles in parallel when it makes sense, review the changes, retry when something fails, and keep track of project memory so future tasks get smarter. It should feel like a real engineering team with a PM, architect, developer, reviewer, and QA working together. Also build a simple dashboard so I can watch what the agents are doing, see the task flow, and check progress from a browser.

Make it work with Claude Code style agent workflows if needed, and keep the experience focused on starting from one prompt and ending with committed code. If you need current docs or best practices, look them up online.