Subagents

Use one when a side task would flood the main conversation
Project agents live in .claude/agents/; personal agents in ~/.claude/agents/
Ask Claude to create the file, or write the markdown file yourself
Use /fork for a background subagent that inherits the current conversation
Use /background (or /bg) to detach the whole session
Use /tasks or claude agents to monitor background work

Narration

Listen or continue with the transcript

Auto-play audio on each slide

Transcript

Subagents: specialized workers

---

Subagents are one of Claude Code's most powerful features. They let you create specialized AI assistants for specific tasks within your project.

Think of subagents as team members with specific roles. You might have a subagent dedicated to code review, another focused on documentation, another specialized in testing. Each subagent can have its own instructions, its own context, its own way of working.

You can define subagents in project or personal agent files, for example under .claude/agents/ or ~/.claude/agents/. You can also ask Claude to draft the agent file for you.

There are now several ways to run side work. Use /fork when you want a background subagent that inherits the current conversation and reports back later. Use /background, or /bg, when you want the whole current session to keep running while your terminal is free. Use /tasks or claude agents to inspect background work.

What makes subagents valuable is specialization. Instead of asking the general Claude Code assistant to do everything, you can delegate specific tasks to agents that are configured for exactly that purpose. Your code review agent knows to look for specific issues, your testing agent knows your testing patterns, your documentation agent knows your doc standards.

It's like having a team of AI specialists, each expert in their domain, all working together on your codebase.