Agents in Claude Code: how they work and how to configure your own

Claude Code runs agents in parallel contexts — each one sees only its own task and doesn't affect neighboring ones. The analysis shows how isolation works and why it solves the chaos problem in long projects. Persistent agents are configured via files in ~/.claude/agents/: you can create your own for code review or routine edits. The article includes an example on a real task — with steps that are easy to repe