Agentic Workflow Pattern FAQs

Question 1

Why does this skill use background tasks for agents?

Accepted Answer

By running agents in the background and avoiding full transcript injection, the skill keeps the main conversation context minimal, preventing performance degradation and token limits on long-running tasks.

Question 2

How can I tell if an agent is stuck or making progress?

Accepted Answer

You can monitor system reminders for token usage and tool calls, or poll the output directory using shell commands to check for file size growth or new timestamps.

Question 3

What is the Agentic Workflow Pattern?

Accepted Answer

It is a structured methodology for Claude Code that delegates complex tasks across specialized sub-agents (Research, Plan, Validate, Implement, Review) to ensure high-quality software delivery.

Question 4

Does this workflow support Test-Driven Development?

Accepted Answer

Yes, the Implementation Agent is explicitly instructed to follow a TDD approach, writing failing tests first before implementing the logic to satisfy them.

Question 5

How do sub-agents communicate with each other?

Accepted Answer

This skill uses a file-based handoff system. Each agent writes its findings or plans to specific markdown files within the .maestro/cache/agents/ directory, which subsequent agents then read as context.

Agentic Workflow Pattern

主要功能

使用场景

Agentic Workflow Pattern

主要功能

使用场景