dispatching-parallel-agentslisted

# Dispatching Parallel Agents ## Contents - [Overview](#overview) - [When to Use](#when-to-use) - [Pattern](#pattern) - [1. Identify Independent Domains](#1-identify-independent-domains) - [2. Create Focused Agent Tasks](#2-create-focused-agent-tasks) - [3. Dispatch in Parallel](#3-dispatch-in-parallel) - [4. Review and Integrate](#4-review-and-integrate) - [Agent Prompt Structure](#agent-prompt-structure) - [Common Mistakes](#common-mistakes) - [When Not to Use](#when-not-to-use) - [Key Benefits](#key-benefits) - [Verification](#verification) ## Overview Delegate work to specialized agents with an isolated context. Construct precise instructions and context so each agent focuses on its own task. An agent must never inherit the session context or history. Construct only exactly what it needs. This preserves your own context for coordination work. When there are several unrelated failures (different test files, different subsystems, different bugs), investigating them sequentially is a waste of time. Each investigation is independent and can proceed in parallel. Core principle: dispatch one agent per independent problem domain. Have them work at the same time. ## When to Use ```dot digraph when_to_use { "Multiple failures?" [shape=diamond]; "Independent?" [shape=diamond]; "Single agent investigates all" [shape=box]; "One agent per problem domain" [shape=box]; "Can work in parallel?" [shape=diamond]; "Sequential agents" [shape=box]; "