Loopx

LoopX is a local control plane for AI agent work that runs longer than a single session, the multi-day kind. One session is easy; long work is where things break, because objectives drift, owner decisions pile up, evidence goes stale, agents hand tasks to peers, and a scheduler happily keeps spending after there is nothing useful left to do.

LoopX keeps the durable part of that in one compact layer: objective, gates, todos, scope, evidence, quota, and handoffs, so each turn stays bounded and restartable. It deliberately does not replace your agent runtime. Codex, Claude Code, Cursor, or a custom loop still executes the turns, while LoopX holds the state and asks a concrete question only when a human decision is genuinely required. The goal is a loop you can review afterward, not merely one that runs by itself.