parallel-subagents

Dispatch multiple focused subagents in one message and synthesize their summaries.

9|2|Updated Jan 29, 2026
One-click install
npx skills add https://github.com/aromanarguello/roman-skills --skill parallel-subagents-aromanarguello
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: parallel-subagents
Source: https://github.com/aromanarguello/roman-skills/tree/main/parallel-subagents
Command: npx skills add https://github.com/aromanarguello/roman-skills --skill parallel-subagents-aromanarguello

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill enables concurrent execution of multiple independent tasks by dispatching focused subagents in a single message and synthesizing their outputs, eliminating slow sequential workflows and manual result aggregation.

Core Features & Use Cases

  • Parallel research: Launch multiple Explore agents to investigate different parts of a codebase (auth, DB, API) and gather findings concurrently.
  • Quick independent tasks: Dispatch general-purpose agents to perform small, standalone implementations or fixes without a formal plan.
  • Parallel test execution: Run test suites across packages in parallel and aggregate pass/fail summaries.
  • Guidance and templates: Provides mode selection, task templates, and a synthesis pattern to unify results and surface conflicts or gaps.

Quick Start

Dispatch three focused subagents in one message to research authentication, database, and API conventions and synthesize a unified summary.

Frequently Asked Questions about parallel-subagents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run parallel research tasks across different parts of a codebase?

You can run parallel research by dispatching multiple focused subagents in a single message, assigning each a specific scope like auth, DB, or API, and synthesizing their returned summaries into consolidated findings.

What is the best way to execute independent tests concurrently without a sequential plan?

Executing independent tests concurrently is best handled by dispatching focused subagents for parallel test execution across packages, aggregating pass/fail summaries without needing a written sequential plan or shared state.

Can I dispatch multiple subagents in one message for standalone implementations?

Yes, you can dispatch multiple general-purpose subagents in a single message to perform quick standalone implementations or fixes, ensuring each agent has a single focused scope and synthesizing the results.

When should I use parallel subagent orchestration instead of a sequential workflow?

Parallel subagent orchestration is ideal when no shared state exists and tasks are independent, eliminating slow sequential workflows and manual result aggregation for research, quick implementations, or parallel test execution.

How do I synthesize conflicting findings from concurrent research agents?

You synthesize conflicting findings by applying the provided synthesis pattern to unify returned summaries from concurrent agents, surfacing conflicts or gaps to consolidate research results into a single coherent output.

Does parallel task dispatch work without a formal written plan?

Yes, parallel task dispatch works without a formal written plan by enforcing single-message dispatch of all Task tool calls, allowing quick ad-hoc tasks and parallel test execution when no shared state exists.