How to Use Sub Agents in Claude Code
This article explains the concept of Sub Agents in Claude Code, which are specialized AI assistants dedicated to handling specific tasks within a Claude Code session. Sub Agents have their own independent context, system prompts, and restricted tool access.
Why it matters
Sub Agents in Claude Code allow for more efficient and specialized AI-powered workflows by delegating tasks to dedicated assistants.
Key Points
- 1Sub Agents are specialized AI assistants that handle specific tasks within a Claude Code session
- 2Each Sub Agent has its own independent context window, dedicated system prompt, and restricted tool access
- 3Sub Agents help with context preservation, specialization, tool restriction, and reusability
- 4Claude Code includes built-in Sub Agents like Explore, Plan, General-purpose, and Bash
Details
The article describes Sub Agents as 'team members' of the main Claude Code 'general manager'. Sub Agents operate independently, with their own context and system prompts, to perform tasks like code review, test writing, and security analysis. This helps preserve the main context, enable specialization, apply the principle of least privilege, and make Sub Agents reusable across projects. The article also covers how to create and manage custom Sub Agents using the /agents command or by writing a Markdown file directly.
No comments yet
Be the first to comment