Building on the subagents foundation, Brian walks through creating a custom code review subagent from scratch using Claude Code's built-in agent builder. He covers the full /agents workflow — describing what the agent should do, selecting read-only tools, choosing a model, configuring agent memory, and reviewing the generated system prompt. Along the way, he highlights practical details like using "USE PROACTIVELY" in the description to encourage automatic delegation and why letting Claude generate the initial prompt gives you a much better starting point than writing one manually.
After the subagent is created, Brian puts it to work by running a code review on the Tasks feature built in the previous lesson. He shows how custom subagents appear with colored backgrounds in the interface, walks through the review output organized by priority, and wraps up by connecting subagents to the bigger picture — when you need workers that can coordinate with each other rather than just report back, that's where agent teams come in.
/agents command and agent builder
Training and community for developers and founders building with AI.
Created by Brian Casel (that's me) — multi-time founder, product designer, teacher, and the creator behind Builder Methods.