Anthropic Courses (original) (raw)

Sub-agents are one of the most practical ways to get more out of longer Claude Code sessions. They let you delegate tasks to isolated assistants that do their work separately and return just the information you need — keeping your main context window clean and your conversations focused.

In this course, you'll learn:

• How sub-agents work — what happens when Claude Code spins up a separate context window, how inputs flow in, and how summaries come back
• Creating custom sub-agents — using the /agents command to build sub-agents tailored to your workflow, from code reviewers to documentation generators
• Designing effective sub-agents — patterns that make sub-agents reliable, including structured output formats, obstacle reporting, and limiting tool access
• When to use them (and when not to) — practical guidance on where sub-agents help the most and the common anti-patterns to avoid

By the end, you'll know how to break complex work into focused pieces, build sub-agents that finish on time and report back clearly, and make the right call on when delegation is worth it.