A subagent is a separate Claude you delegate a job to — "go read these forty files and tell me where logins are handled." It works in its own context window and reports back just the answer, so your main chat doesn't fill up with the messy middle. You don't set anything up: in Claude Desktop you just ask in plain language and Claude spins one up when it helps. Great for research, search, and any side-quest.
The reason a subagent helps isn’t just the extra pair of hands — it’s the fresh context. A big research dig (“read these forty files and tell me where logins are handled”) generates a mountain of intermediate reading. Done in your main chat, all of that piles up and crowds out the thread you actually care about. Hand it to a subagent and that mess stays in its window; you get back only the answer.
The trade-off is that a subagent can’t see your conversation. It starts clean, so whatever it needs to know has to be in the request you give it. That’s a feature as often as a limitation — it means the helper isn’t distracted by everything else you’ve been discussing — but it does mean a vague hand-off gets a vague result. Spell out the goal and the context up front.
Reach for one when a task is a self-contained side-quest: searching, reading widely, checking one specific thing. When the work is wide — many independent pieces at once — that’s when subagents scale up into agent teams, several running in parallel and reporting back together.
why it helps Keep your main thread clean while a helper does the digging.
examples- Delegate the messy middle (search, reading lots of files) — keep your main chat for decisions.
- A subagent can't see your conversation, so give it everything it needs in the request.
- Reach for one when a side-quest would otherwise bury the thread you actually care about.
Questions people ask
- Can a subagent see my conversation?
- No. A subagent starts with a fresh, empty context and can't see your chat — that's the point. Give it everything it needs in the request, because a vague hand-off gets a vague result.
- When should I use a subagent instead of just asking directly?
- Reach for one when a task is a self-contained side-quest — searching, reading lots of files, checking one specific thing — that would otherwise bury the thread you actually care about. The subagent does the messy middle and reports back just the answer.
- What's the difference between a subagent and an agent team?
- A subagent is one helper handling one delegated task. An agent team is many subagents running in parallel, each on a piece, then combining their results. Use a subagent for one side-quest; reach for a team when the work is wide.
- Does using a subagent keep my main context cleaner?
- Yes — that's the main benefit. All the intermediate reading happens in the subagent's own context window, so your main conversation gets only the final answer instead of the mountain of files it read to find it.
- Do I have to set up a subagent myself?
- No. In Claude Desktop you just ask for what you want in plain language — "read the whole docs folder and give me the five things I most need to know" — and Claude spins up a subagent on its own when delegating helps. There's nothing to configure.