dispatching-parallel-agents

1# Dispatching Parallel Agents
2 
3## Overview
4 
5When you have multiple unrelated failures (different test files, different subsystems, different bugs), investigating them sequentially wastes time. Each investigation is independent and can happen in parallel.
6 
7**Core principle:** Dispatch one agent per independent problem domain. Let them work concurrently.
8 
9## When to Use
10 
11```dot
12digraph when_to_use {
13    "Multiple failures?" [shape=diamond];
14    "Are they independent?" [shape=diamond];
15    "Single agent investigates all" [shape=box];
16    "One agent per problem domain" [shape=box];
17    "Can they work in parallel?" [shape=diamond];
18    "Sequential agents" [shape=box];
19    "Parallel dispatch" [shape=box];
20 
21    "Multiple failures?" -> "Are they independent?" [label="yes"];
22    "Are they independent?" -> "Single agent investigates all" [label="no - related"];
23    "Are they independent?" -> "Can they work in parallel?" [label="yes"];
24    "Can they work in parallel?" -> "Parallel dispatch" [label="yes"];
25    "Can they work in parallel?" -> "Sequential agents" [label="no - shared state"];
26}
27```
28 
29**Use when:**
30- 3+ test files failing with different root causes
31- Multiple subsystems broken independently
32- Each problem can be understood without context from others
33- No shared state between investigations
34 
35**Don't use when:**
36- Failures are related (fix one might fix others)
37- Need to understand full system state
38- Agents would interfere with each other
39 
40## The Pattern
41…