ls-liteharness

Orchestrate multi-CLI agent terminal sessions with naming and message routing.

Updated Aug 5, 2026
One-click install
npx skills add https://github.com/ahostbr/liteharness --skill ls-liteharness
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ls-liteharness
Source: https://github.com/ahostbr/liteharness/tree/main/liteharness/catalog/skills/ls-liteharness
Command: npx skills add https://github.com/ahostbr/liteharness --skill ls-liteharness

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

LiteHarness eliminates the friction of manually spawning, identifying, messaging, and controlling coding-agent terminal sessions across multiple CLIs by providing a consistent orchestration layer.

Core Features & Use Cases

  • Session startup and self-registration: Ensures every spawned agent activates the correct inbox watcher (per CLI surface) and can receive messages reliably.
  • Deterministic agent naming + discovery: Generates stable two-word names from agent UUIDs to make monitoring and routing easier when managing many sessions.
  • PTY daemon and headed UIAutomation modes: Supports headless programmatic control for fast workflows, plus headed terminal control when visible UI automation is required.
  • Inbox-based messaging + lifecycle commands: Uses a Maildir-style inbox for message delivery and supports safe task rotation with /clear (while blocking /exit for safety).

Quick Start

Ask the system to spawn a headless agent and register it: "Spawn a new LiteHarness-controlled agent in headless PTY mode with model haiku, named Worker, that runs the prompt 'run the tests'."

Frequently Asked Questions about ls-liteharness

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate multiple Claude Code terminal sessions programmatically?

You can orchestrate multiple Claude Code terminal sessions by spawning, naming, and routing messages to specific agent instances using a PTY daemon for headless control or UIAutomation for headed terminal control.

Can I manage agent lifecycle and clear context without accidentally terminating the terminal session?

Yes, agent lifecycle management supports safe task rotation using the /clear command while explicitly blocking the /exit command to prevent accidental termination of active terminal sessions.

What is the best way to route messages to specific coding agent instances across different CLIs?

The best way to route messages is via a Maildir-style inbox system that delivers messages to deterministic two-word names generated from agent UUIDs for reliable discovery and routing.

Does this agent orchestration approach work with coding CLIs other than Claude Code?

Yes, the orchestration layer applies to Claude Code, Codex, Copilot, Pi, and related coding CLIs by applying per-CLI watcher setup and validated agent identifiers for safe PTY control paths.

Why do I need sentinel startup rules for terminal automation and agent session management?

Sentinel startup and monitor restart rules are required to ensure every spawned agent activates the correct inbox watcher per CLI surface and can reliably receive routed messages without missing initialization.

How do I spawn a headless coding agent and register it for automated message routing?

You spawn a headless agent in PTY mode with a specific name and prompt, which triggers self-registration to activate the inbox watcher and enable reliable message delivery for the session.