PrimeIntellect-ai/prime-agent ? reverse-engineered prompt

Build me a coding agent that can run in a project folder, make changes to files, run commands, and keep working for a long time without losing progress.

I want it to feel like a real terminal assistant, with a persistent session, the ability to start background work, pause and reattach later, and keep track of goals, notes, and reusable skills across turns. It should be able to spawn helper agents for parallel tasks, remember useful context locally, and have a way to review what it learned and make small improvements over time. A login step for choosing an API provider is fine, and I’d like commands for things like checking status, resuming sessions, running in autonomous mode, and shutting everything down cleanly.

Please make it usable from the command line, with a good setup flow, clear docs, and sensible defaults. If you need to look up current docs online, do that.