criticic/reagent ? reverse-engineered prompt

Build me a command line tool that can take a binary file and a goal, then automatically analyze it with multiple AI agents and return a structured report. I want it to do the kind of reverse engineering workflow a human would do, like quick triage, static analysis, and runtime debugging, then verify its guesses instead of just making claims.

It should support a simple command like analyze a binary and give it a goal such as finding a password check, recovering a flag, or understanding a protocol. I also want an interactive terminal mode for exploring the analysis live. Use a main orchestrator that can hand work off to specialist agents, keep track of observations and hypotheses, and store confirmed findings with evidence.

If it helps, look up current docs online for the reverse engineering tools or model APIs you need. Also include a few sample crackme challenges so I can test it end to end.