dispatching-parallel-agents

Dispatching Parallel Agents

Overview

When you have multiple independent experiments (different datasets, different methods, different baselines), running them sequentially wastes time. Each experiment is independent and can happen in parallel.

Core principle: Dispatch one agent per independent problem domain. Let them work concurrently.

Use this skill alongside:

• amplify:experiment-execution for Phase 4 orchestration
• amplify:results-verification-protocol before claiming completion
• amplify:results-integration when consolidating outputs

When to Use

digraph when_to_use {
    "Multiple experiments?" [shape=diamond];
    "Are they independent?" [shape=diamond];
    "Single agent runs all" [shape=box];
    "One agent per experiment" [shape=box];
    "Can they work in parallel?" [shape=diamond];
    "Sequential agents" [shape=box];
    "Parallel dispatch" [shape=box];

    "Multiple experiments?" -> "Are they independent?" [label="yes"];
    "Are they independent?" -> "Single agent runs all" [label="no - related"];
    "Are they independent?" -> "Can they work in parallel?" [label="yes"];
    "Can they work in parallel?" -> "Parallel dispatch" [label="yes"];
    "Can they work in parallel?" -> "Sequential agents" [label="no - shared state"];
}

Use when:

• 3 independent experiments on different datasets
• Multiple baseline implementations to evaluate
• Each experiment can be understood without context from others
• No shared state between experiments

Don't use when:

• Experiments are related (result of one informs another)
• Need to understand full system state
• Agents would interfere with each other

The Pattern

1. Identify Independent Domains

Group experiments by what they investigate:

• Experiment A: Baseline model on dataset X
• Experiment B: Proposed method on dataset Y
• Experiment C: Ablation study on component Z

Each domain is independent - running the baseline doesn't affect the ablation study.

2. Create Focused Agent Tasks