juanmanueldaza/n3rv ? reverse-engineered prompt

Build me a Python tool that sets up a local harness for AI coding agents in an existing project.

I want to be able to run one command in a repo and have it add the files and config needed for opencode to work with memory, skills, slash commands, and sub agents. It should create a clean project setup, inject useful context from the repo, and keep long term memory for decisions and project notes. Please include support for a background service, a simple hub for delegating tasks between agents, and a way to inspect, search, and prune memory from the command line.

Make it feel ready to use for real development, with sensible defaults for Python projects, but keep it flexible enough to support other stacks too. If you need to look up current docs or opencode details online, go ahead. Also add tests and make sure the init and update flows are safe to run more than once.