What kind of specialized agents are available?

The skill includes specialized roles such as Architect, Coder, Tester, and Documentation for development, and Unit, Integration, E2E, and Security agents for testing.

What is the Fan-Out/Fan-In pattern in this skill?

It is a workflow where a supervisor agent decomposes a task into several sub-tasks (Fan-Out) for parallel execution by specialized agents, and then an aggregator combines those results (Fan-In) into a final report.

Can I define custom agent groups for my project?

Yes, while the skill includes pre-defined groups for Dev, Test, and Review, you can specify individual agents or groups via the command line to match your specific workflow requirements.

How does the skill handle failures during parallel execution?

The skill uses a 3-step failure policy: it first attempts a retry, then analyzes error logs to attempt a fix, and finally pauses for manual intervention if the task fails a third time.

Parallel Agent Orchestration

Name: Parallel Agent Orchestration
Author: garimto81

bygarimto81

•

生产力与工作流

Orchestrates multiple specialized Claude agents concurrently using Fan-Out/Fan-In patterns to accelerate development and testing workflows.

Parallel Agent Orchestration is a high-performance skill designed to scale Claude's capabilities by running multiple specialized agents simultaneously. It utilizes the Fan-Out/Fan-In architectural pattern to decompose complex tasks into manageable sub-tasks for parallel execution by dedicated agent groups—such as architects, coders, and testers—before aggregating the results into a single cohesive output. This skill is particularly useful for reducing development cycle times, performing comprehensive multi-layered testing, and conducting multi-faceted code reviews without sequential bottlenecks.

主要功能

01Flexible command-line interface for individual agent or group execution

02Concurrent multi-agent execution using Fan-Out/Fan-In architecture

03Tiered retry logic with error log analysis and manual intervention hooks

041 GitHub stars

05Pre-configured agent groups for Development, Testing, and Code Review

06Automated result aggregation and comprehensive success criteria validation

使用场景

01Conducting multi-perspective code reviews covering architecture, security, and logic in one pass

02Accelerating feature delivery by generating code, tests, and documentation simultaneously

03Running parallelized test suites including unit, integration, E2E, and security checks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add garimto81/claude parallel-agent-orchestration

For use in Claude.ai and ChatGPT

Download Skill