team

Coordinate multiple agents to decompose and execute complex multi-step tasks.

1|Updated Apr 1, 2026
One-click install
npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill team-oliverouyang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: team
Source: https://github.com/OliverOuyang/shuhe-work-skills/tree/main/skills/team
Command: npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill team-oliverouyang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinate and manage multiple autonomous agents to decompose, execute, and verify complex multi-step engineering tasks without manual orchestration, lost context, or ad-hoc process gaps.

Core Features & Use Cases

  • Automated task decomposition: Break a high-level task into file- or module-scoped subtasks with explicit dependencies and owners.
  • Team lifecycle orchestration: Create teams, spawn worker agents, persist staged state, resume or cancel runs, and perform controlled shutdowns.
  • Staged pipeline & verification: Structured stages (plan → prd → exec → verify → fix) with handoff documents and retry/limit logic for robust iteration.
  • Hybrid execution: Support for Claude teammates plus optional CLI workers (Codex/Gemini) and utilities for monitoring, outbox ingestion, and orphan cleanup.
  • Use Case: Run parallel code fixes, large refactors, or audit-and-fix workflows that require coordinated execution, verification, and safe teardown.

Quick Start

Invoke the team skill with a single-line command including optional agent count and type, for example: /team 3:executor "fix build errors across src/".

Frequently Asked Questions about team

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

FAQPage Schema
How do I orchestrate multiple Claude Code agents to handle complex refactors?

You orchestrate multiple Claude Code agents by spawning worker agents that decompose complex refactors into file-scoped subtasks with explicit dependencies, enabling parallel execution and structured verification across the codebase.

Can I use Codex CLI workers alongside Claude agents for hybrid task execution?

Yes, hybrid execution supports Claude teammates plus optional Codex CLI and Gemini CLI workers, providing utilities for monitoring, outbox ingestion, and orphan cleanup across mixed-agent task pipelines.

What is the best way to manage multi-agent task lifecycles and resume interrupted runs?

The best way to manage multi-agent task lifecycles is through staged state persistence, which stores pipeline progress across plan, prd, exec, verify, and fix stages, enabling safe resume, cancel, and controlled shutdowns.

Does multi-agent task decomposition handle build and test fix loops automatically?

Yes, multi-agent task decomposition handles build and test fix loops by applying staged verification pipelines with handoff documents and retry limit logic, ensuring robust iteration across execution and fix stages.

How do I run parallel code fixes across multiple files without losing context?

You run parallel code fixes by creating a multi-agent team that assigns file-scoped subtasks to individual workers, using staged state persistence and handoff documents to maintain context across the verification pipeline.