Agent Skills: Multi-Agent Orchestration Guide

Multi-Agent Orchestration Guide

Overview

Complex tasks benefit from orchestrating multiple subagents rather than handling everything in a single agent. This guide provides patterns for task decomposition, subagent coordination, and result synthesis.

When to Use Orchestration

Appropriate for Orchestration

• Tasks requiring multiple expertise domains
• Work decomposable into independent subtasks
• Large-scale codebase analysis
• Parallelizable operations

Single Agent Suffices

• Single file modifications
• Simple bug fixes
• Clear, singular objectives

Task Decomposition Strategy

Decomposition Principles

1. Independence: Each subtask should not depend on results from other subtasks (when parallel)
2. Clarity: Each task must have clear objectives and expected deliverables
3. Right-sizing: Neither too granular nor too broad

Decomposition Process

1. Define the final goal
2. Identify required expertise domains
3. Break down into subtasks
4. Map dependencies (sequential vs parallel)
5. Determine appropriate approach for each task (subagent type, required skills)

Orchestration Patterns

Pattern 1: Parallel Analysis

Use when analysis from multiple perspectives is needed simultaneously.

Subtasks (independent):
├── Structure/architecture analysis
├── Code quality review
├── Security assessment
└── Test coverage analysis

Execution: Launch all subtasks in parallel via multiple Task tool calls
Result: Synthesize findings after all tasks complete

Pattern 2: Sequential Pipeline

Use when each stage depends on previous results.

Task 1: Exploration/Analysis → Pass findings
    ↓
Task 2: Planning → Pass plan
    ↓
Task 3: Implementation → Pass result
    ↓
Task 4: Verification

Key: Use task_id to maintain context between stages

Pattern 3: Hybrid

Use for complex tasks with mixed parallel/sequential phases.

Phase 1: Parallel Exploration
├── Explore area A
├── Explore area B
└── Explore area C
        ↓
Phase 2: Sequential Implementation (based on Phase 1 findings)
├── Implement feature
└── Verify changes

Context Management

Parallel Execution

• Each Task runs independently
• Collect and synthesize results after completion
• No context sharing between parallel tasks

Sequential Execution

• Use task_id to resume previous work
• Pass essential context explicitly in prompt
• Minimize context to necessary information only

Skill Loading Strategy

When to Load Skills

• Specialized expertise is required
• Standardized process guidance is needed
• Domain-specific best practices apply

How to Load

Include skill loading instruction in the task prompt when domain expertise would benefit the task.

Synthesis Protocol

After all tasks complete:

1. Collect Results: Gather key findings from each task
2. Identify Patterns: Remove duplicates, find common themes
3. Prioritize: Sort by importance/urgency
4. Define Actions: Specify concrete follow-up work
5. Consolidate: Produce unified report or output

Best Practices

1. Plan Before Execution: Identify required tasks upfront
2. Parallel First: Run independent tasks concurrently
3. Avoid Over-decomposition: Too many small tasks creates overhead
4. Always Synthesize: Don't end with scattered individual results
5. Minimize Context: Pass only necessary information to each task
6. Leverage Skills: Load relevant skills for specialized work