How I Run 10 AI Agents in Parallel with Claude Code

Three research agents running in parallel. Codebase analysis dropped from 10 minutes to 3. I'm a solo developer, but for the first time it felt like I had a team. Claude Code subagents are specialized AI assistants that run in their own context windows inside your main Claude session. Each one has its own system prompt, its own tool access, its own model configuration. You can run up to 10 of them simultaneously. The main session delegates tasks to them, and they report back when done. I've been using subagents for about two months now. This is what I've learned about making them actually useful, not just a novelty. The Built-in Agents You Didn't Know About Claude Code ships with six agents out of the box. The one I use most is Explore, which runs on Haiku for fast, read-only codebase search. It fires automatically when Claude needs to understand your project structure. No risk of accidental modifications because it literally can't write files. The Plan agent handles research in plan m