What is file-based delegation?

It is a strategy where sub-agents receive instructions via files rather than long prompts, preventing the main orchestrator's context window from becoming bloated and inefficient.

Can this skill coordinate external AI models?

Yes, it includes patterns for Proxy Mode invocation, allowing you to use models like Grok or Gemini as specialized reviewers or developers within a Claude-led workflow.

When should I use sequential vs parallel execution?

Use sequential execution when one agent's output is required as input for the next; use parallel execution for independent tasks like simultaneous code reviews or browser testing.

How do I run agents in parallel?

To achieve true parallelism, you must avoid mixing tool types. Using only 'Task' tool calls in a single message signals to Claude that the operations are independent.

What is multi-agent coordination in Claude Code?

It is the process of organizing multiple specialized AI agents to work together on a single project, managing their dependencies, execution order, and shared context.

Multi-Agent Coordination

Name: Multi-Agent Coordination
Author: involvex

byinvolvex

0•

コラボレーションツール

Orchestrates multiple specialized agents through parallel and sequential workflows to handle complex development tasks efficiently.

The Multi-Agent Coordination skill provides battle-tested patterns for managing sophisticated Claude Code workflows that require more than one specialized agent. It solves the complexity of dependencies by teaching Claude when to run tasks in parallel versus sequentially, how to select the optimal agent for specific tasks, and how to delegate to sub-agents without polluting the context window. By implementing the 4-Message Pattern and file-based instructions, this skill enables users to build high-performance, multi-perspective pipelines for architecture, implementation, and validation.

主な機能

01Context window management and optimization strategies

020 GitHub stars

03Parallel and sequential workflow orchestration

04Proxy mode for external model integration

05Sub-agent delegation with file-based context isolation

06Intelligent task-based agent selection logic

ユースケース

01Running simultaneous design, testing, and code review validations

02Managing multi-phase implementation pipelines from planning to deployment

03Delegating large-scale architectural planning to specialized sub-agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add involvex/involvex-claude-marketplace multi-agent-coordination

For use in Claude.ai and ChatGPT

Download Skill