What happens if one agent fails during a dispatch?

You can configure failure behavior to either 'continue', which returns results from the successful agents, or 'abort', which stops the entire process if an error occurs.

Is there a recommended limit for parallel agents?

For optimal performance, memory management, and cost-efficiency, it is recommended to dispatch 3-5 agents per single task.

Can I create my own agent groups?

Yes, the skill allows you to define custom configurations and groups in YAML, specifying unique agent names, prompt templates, and timeout settings.

How does parallel-dispatch save time?

It runs multiple specialized agents simultaneously rather than one after another, significantly reducing the total wall-clock time for complex tasks like full architectural reviews.

What output formats are supported?

The skill can return results in a 'structured' JSON format containing metadata and duration statistics, or a 'raw' format for direct text output.

Parallel Agent Dispatcher

Name: Parallel Agent Dispatcher
Author: jmagly

byjmagly

•

Productivity & Workflow

Orchestrates and launches multiple specialized AI agents concurrently to perform parallel reviews, analysis, and complex evaluations.

The Parallel Dispatch skill provides the essential infrastructure for high-concurrency agentic workflows within Claude Code. It allows users to bypass sequential bottlenecks by dispatching multiple specialized agents—such as security architects, test analysts, and technical writers—simultaneously. The skill automates the entire lifecycle of multi-agent tasks, from parsing YAML configurations and injecting shared context to handling timeouts and consolidating diverse outputs into a single, structured JSON or raw report. It is particularly effective for reducing wall-clock time during complex SDLC gates, architectural reviews, and multi-perspective content audits.

Key Features

01Concurrent multi-agent orchestration and task tool calls

02Structured result aggregation for downstream processing

0367 GitHub stars

04Pre-defined agent groups for architecture, security, and marketing

05Automated prompt templating with shared context injection

06Configurable timeout and failure recovery behaviors

Use Cases

01Automated requirement traceability analysis across complex codebases

02Simultaneous security, testability, and architectural reviews of design documents

03Multi-perspective brand compliance and legal audits for marketing content

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jmagly/ai-writing-guide parallel-dispatch

For use in Claude.ai and ChatGPT

Download Skill