My AI agents fell into a polite, infinite argument — I only found out from the bill. So I borrowed four old ideas (deterministic simulation, saga, intuition engineering, real-time agent visibility) and built a board that replays the whole pipeline as a living flow you can scrub through time.
The proxy is honest about its layer — it shrinks ~53% of the bash output it intercepts. In Claude Code, that's roughly 4% of the input you actually pay for. A map, not an accusation.
A workflow that ships code, learns from its mistakes, and feeds the lessons back into planning — modelled explicitly as a state machine.