team

Coordinates multi-agent tasks with tmq-based worker panes and shared team state.

Updated Apr 20, 2026
One-click install
npx skills add https://github.com/hotaq/Sprite_harmess --skill team-hotaq
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: team
Source: https://github.com/hotaq/Sprite_harmess/tree/main/.codex/skills/team
Command: npx skills add https://github.com/hotaq/Sprite_harmess --skill team-hotaq

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you run coordinated multi-agent work when a task is too large, long-running, or parallelizable for a single agent thread. It provides a durable tmux-based workflow for launching workers, tracking shared state, and avoiding fragile ad hoc coordination.

Core Features & Use Cases

  • Team orchestration: Start and manage multiple Codex or Claude worker sessions with shared task state.
  • Stateful coordination: Use mailbox, inbox, heartbeat, and task files to track progress without losing context.
  • Verification and recovery: Monitor status, inspect pane evidence, handle stalled workers, and shut down cleanly when work is complete.
  • Use Case: A developer needs one lane to implement code, another to verify tests, and a third to inspect edge cases while keeping the run recoverable and auditable.

Quick Start

Ask the assistant to launch an omx team for your task, then keep monitoring status and evidence until the workers finish and the team can be shut down cleanly.

Frequently Asked Questions about team

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

FAQPage Schema
How do I run parallel coding agents in tmux for complex development tasks?

You can orchestrate parallel multi-agent task execution by launching coordinated Codex or Claude worker sessions in tmux-backed panes with shared team state for durable, recoverable workflows.

What is multi-agent orchestration for parallel software development?

Multi-agent orchestration coordinates multiple AI worker sessions using mailbox synchronization, heartbeat monitoring, and task files to track shared state and progress without losing context.

How do I manage shared state and communication between multiple AI agent workers?

You manage shared state using mailbox, inbox, heartbeat, and task files to synchronize context and track progress across parallel agent workers running in separate terminal panes.

Can I use Codex and Claude sessions together in a coordinated multi-agent workflow?

Yes, you can start and manage multiple Codex or Claude worker sessions simultaneously, assigning each a dedicated lane such as implementation, verification, or edge case inspection.

How do I monitor agent worker status and handle stalled workers during long-running tasks?

You monitor status, inspect pane evidence, and handle stalled workers by checking startup evidence, mailbox synchronization, and terminal-completion checks to ensure durable coordination and recovery.

What is the best way to shut down parallel agent teams cleanly after task completion?

The best way to shut down involves monitoring terminal-completion checks and applying graceful shutdown semantics to close tmux-backed worker panes only after all agents finish their tasks.