The idea
Five parallel reviewers - quality, implementation, testing, simplification, docs - gate every task before it lands.
My friend Anton mentioned this to me, and it looks genuinely interesting. Ralphex is a CLI that orchestrates AI coding agents like Claude Code and codex through structured plans rather than writing code itself, running each task in a fresh session to keep context clean. It deploys five specialised reviewers in parallel - covering quality, implementation, testing, simplification and documentation - and can optionally cross-check against GPT-5 via codex. Plans get git-tracked automatically with per-task commits and a real-time web dashboard, aiming for disciplined, reviewable autonomy rather than unsupervised agent activity.