What is the difference between the task tool and agent teams?

Use the Task tool (star topology) for simple, centralized work, and Agent Teams (mesh topology) for complex, cross-cutting concerns where agents need to message peers directly.

Can I test my agent across different difficulty levels simultaneously?

Yes, the scenario-orchestrator feature allows you to run a single skill across simple, medium, and complex scenarios in parallel with progressive difficulty scaling.

How does the skill prevent infinite loops in autonomous agents?

It implements strictly recommended safeguards including step limits (typically 5-15 steps) and sliding-window memory management to prevent context overflow and runaway execution.

Which agent frameworks are supported by this skill?

This skill provides patterns and implementation guidance for CrewAI, AutoGen, OpenAI Agents SDK, LangGraph, and the Microsoft Agent Framework.

Agent Orchestration

Name: Agent Orchestration
Author: yonatangross

byyonatangross

•

116

•

Data Science & ML

Builds and coordinates sophisticated AI agent systems using industry-standard patterns like ReAct loops, multi-agent supervision, and framework-specific implementations.

This skill provides a comprehensive toolkit for architecting autonomous agentic workflows within Claude Code. It offers production-ready patterns for single-agent reasoning loops (ReAct), multi-agent coordination topologies (star vs. mesh), and seamless integration with popular frameworks like CrewAI, AutoGen, and LangGraph. Whether you are building complex multi-step reasoning systems, implementing self-correction loops, or orchestrating parallel scenarios with varying difficulty levels, this skill provides the architectural guidance and implementation templates needed to create robust, scalable AI agents.

Key Features

01Implements autonomous ReAct and Plan-and-Execute reasoning loops with self-correction.

02Includes built-in memory management and error isolation to prevent infinite loops and context overflow.

03Provides multi-scenario orchestration for testing skills across progressive difficulty levels.

04Facilitates multi-agent coordination via supervisor routing, fan-out/fan-in, and agent debate patterns.

05116 GitHub stars

06Supports major agent frameworks including CrewAI, AutoGen, and OpenAI Agents SDK.

Use Cases

01Designing a complex support system where a supervisor agent routes queries to specialized billing, technical, and account agents.

02Benchmarking different AI agent frameworks to determine which best fits specific project requirements.

03Developing a self-healing coding assistant that plans a task, executes it, and re-plans based on execution errors.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add yonatangross/orchestkit agent-orchestration

For use in Claude.ai and ChatGPT

Download Skill