How does the Supervisor pattern work?

A central coordinator agent analyzes the task, creates a routing plan, and delegates specific steps to specialist agents in a determined order.

How is conflict resolution handled between agents?

Conflicts are resolved either by comparing agent confidence scores or by using an LLM as an impartial arbitrator to decide the best path forward.

When should I use Agent Teams vs. standard subagents?

Use Agent Teams for complex, cross-cutting tasks requiring high peer communication; use standard subagents for simpler, sequential, or cheaper tasks.

What are Claude Code Agent Teams?

Introduced in CC 2.1.33, Agent Teams allow for a mesh topology where agents can message each other directly and share a common task list.

What is the Fan-Out/Fan-In pattern?

It is a workflow where a task is split among multiple specialized agents running in parallel (fan-out) and then their results are merged into a single output (fan-in).

Multi-Agent Orchestration

Name: Multi-Agent Orchestration
Author: yonatangross

byyonatangross

•

Productivity & Workflow

Coordinates specialized AI agents using advanced patterns like fan-out/fan-in and supervisor architectures to solve complex development tasks.

This skill provides a robust framework for managing multiple AI agents within Claude Code, enabling developers to break down complex tasks into specialized sub-tasks. It implements production-ready orchestration patterns including parallel fan-out/fan-in, central supervisor routing, and the latest Claude Code Agent Teams for peer-to-peer mesh communication. Whether you need to synthesize findings from a security and performance audit or manage a large-scale feature implementation across backend and frontend agents, this skill handles conflict resolution, synthesis, and inter-agent communication efficiently.

Key Features

0173 GitHub stars

02Conflict resolution via confidence scoring or LLM arbitration

03Fan-Out/Fan-In parallel execution patterns for specialists

04Centralized Supervisor routing for complex task delegation

05Automated synthesis of multi-agent outputs into coherent summaries

06Native Claude Code Agent Teams integration (v2.1.33+)

Use Cases

01Performing concurrent security, performance, and architecture audits

02Managing complex feature implementations across multiple codebases

03Coordinating peer-to-peer agent communication for collaborative research

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit multi-agent-orchestration

For use in Claude.ai and ChatGPT

Download Skill