LoopX

// Overview

LoopX describes itself as the open, provider-neutral, stateful control plane for long-horizon agents. It does not replace an agent harness — it sits above Codex, Claude Code, Cursor and others and holds the state they lose between turns: the objective, the gates it has to pass, the todo list, the evidence collected so far, the remaining quota, and the handoffs between whoever picks the work up next.

The core is a durable state kernel over goals, todos, claims, leases and evidence. Around it sit quota-aware scheduling and cross-turn continuation (so a run that is interrupted resumes instead of restarting), typed handoffs for peer coordination between agents, evidence-backed recovery with a safe fallback path, and a personal workspace dashboard for watching long-running work.

It installs as a Python package and connects to an existing project directory, with the Python core paired with Node components. LoopX is Apache-2.0 licensed from v0.4.8 onward; earlier releases were published under MIT terms.

// What it does

• Durable state kernel for goals, todos, claims, leases and evidence that survives across sessions
• Quota-aware scheduling and cross-turn continuation so an interrupted run resumes rather than restarts
• Typed handoffs for coordinating peer agents on the same objective
• Evidence-backed recovery with a governed fallback path when a step fails
• Personal workspace dashboard for monitoring long-horizon work in progress
• Provider-neutral: runs on top of Codex, Claude Code and other agent runtimes rather than replacing them

// Getting started

Install the package, install its workflow skills, then connect LoopX to the project you want an agent to work on.

python3 -m pip install --upgrade loopx
loopx workflow-skills --install
loopx doctor

cd /path/to/your-project
loopx connect
loopx status

loopx start-goal --guided --project . --goal-text "Your long-running objective"

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Keep a multi-day refactor or migration on track across many separate agent sessions
• Give two or more coding agents a shared objective with explicit handoffs instead of duplicated work
• Recover a long agent run after an interruption without losing the evidence already gathered
• Stay portable across harnesses when you do not want the state of your work locked into one vendor's tool

// How LoopX compares

LoopX alongside other open-source multi-agent systems tools AI/TLDR tracks, ranked by GitHub stars.