Claude Code Agents

A subagent in Claude Code is a specialized assistant with its own system prompt, its own set of allowed tools, and a context window separate from your main session. Instead of one general-purpose assistant juggling every task, you delegate focused work — reviewing a diff, auditing dependencies, writing a migration — to a persona that has been primed for exactly that job. Subagents are defined as Markdown files with YAML frontmatter in your project's .claude/agents directory, or in ~/.claude/agents when you want them available across every project.

Because each subagent carries its own instructions and tool permissions, Claude Code can hand off a task without polluting the main conversation's context, then fold the result back in. You can invoke a subagent explicitly by name, or let Claude route to one automatically based on its description field. Every agent in this catalog installs the same way — one command drops the Markdown definition into your project:

npx claude-code-templates@latest --agent="category/name" --yes

Reach for a subagent when a task is repeatable, benefits from a narrow set of tools, or is best handled in isolation — a code reviewer, a security auditor, or a database administrator. Browse the full list below, or explore related building blocks: slash commands for reusable prompts, MCP servers to connect external systems, and skills for packaged instructions. New to the section? Start at the Claude Code catalog overview.