Welcome to LoopX¶

LoopX is the local control plane for long-running AI agent work. It keeps objectives, gates, todos, evidence, quota, and handoffs stable while Codex, Claude Code, OpenCode, Cursor, or a custom runner executes bounded turns.

Choose Your Path¶

Start using LoopX

Install the CLI, connect a project, inspect the current gate, and start a real goal from your agent.

Getting started
Understand the control plane

Learn how goals, user gates, agent todos, quota, evidence, and handoffs fit into one durable state kernel.

Concepts
Operate a long task

Use status, quota, review packets, and the local dashboard without making the browser the source of truth.

Operations
Build or extend LoopX

Work from the developer guide, testing policy, protocol references, and public/private boundary checks.

Development

Quick Start¶

curl -fsSL https://raw.githubusercontent.com/huangruiteng/loopx/main/scripts/install-from-github.sh | bash
export PATH="$HOME/.local/bin:$PATH"
loopx doctor

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

Then start real work from your agent:

/loopx <complex task>

Core Commands¶

Need	Command
Check installation	`loopx doctor`
Inspect current state	`loopx status`
Decide whether a turn may run	`loopx quota should-run --goal-id <goal-id>`
Manage user and agent todos	`loopx todo --help`
Build a handoff packet	`loopx review-packet --goal-id <goal-id>`
Serve local dashboard data	`loopx serve-status --global-registry --port 8766`

Source Of Truth¶

The docs site is a published read model over repository Markdown. The canonical source remains the Markdown in docs/, while project-local runtime state stays ignored and private.