LoopX is a lightweight state kernel and control plane for governing long-running AI agent teams.
The problem LoopX solves is that long-running agent work requires durable state management across multiple turns, tool invocations, and agent handoffs—beyond what chat memory and timers can provide. Objectives change, decisions must be made by humans, evidence becomes stale, and schedulers risk spending quota without progress. LoopX keeps this control state in a compact, local-first layer that remains agent-runtime agnostic. It works by maintaining durable goals, executable todos, evidence logs, quota awareness, and verifiable handoffs while the actual work executes in any coding agent runtime such as Codex, Claude Code, or Cursor. The tool validates state transitions through operators like claim, gate, monitor, and writeback, functioning as an agent-native Kanban board where registered agents act as peers without requiring a durable leader identity.
LoopX suits teams running long-running discovery, research, engineering, or operations loops where work spans multiple sessions and agent handoffs. It is designed for projects where human judgment must remain in the loop and work must be reviewable and restartable. The tool is provider-neutral and does not replace the agent runtime; it governs the state layer that lets work continue across runs. Choose it if you need to keep objectives stable, track evidence, manage quota spending, and hand work between agents without losing context or control.
The maintainers respond to new issues and pull requests within a day. Work in the issue tracker centers on enhancement requests and workflow-audit concerns.
