LoopX

LoopX is an open-source control plane for AI agents working on multi-day coding, research and personal projects. It preserves the goal and decisions between sessions so work can resume after a pause or handoff.

• Persistent goals: Keep tasks, ownership, run history and handoffs across sessions, including when you change the runtime or host.
• Human approval gates: Hold actions for review while independent tasks continue within their assigned scope; publishing, production writes and final ownership remain with you.
• Agent teams: Let agents share a goal while keeping separate task ownership and explicit handoffs between implementation and review.
• Review evidence: Carry validation results and review decisions into the next run so another person or agent can inspect the work.
• Personal Workspace: View goals, tasks, chat and outputs together to track progress and resolve decisions.

LoopX works with Codex, Claude Code, DeepSeek Harness, OpenCode, TraeX, Pi and custom runners. It's local-first: LoopX preserves control state, while the connected runtime supplies the model and tools. Model execution depends on that runtime and its provider, so local control doesn't mean inference runs locally or works offline.

The project uses the Apache 2.0 license. Its Mac app targets Apple Silicon and requires Python. Background execution needs an available host and runtime.