subagent

To run parallel background agents for coding tasks, you spawn subagents with a specific label, objective, and role like coder. This delegates specialized work to run concurrently without blocking your main conversation.

Autonomous task orchestration coordinates background agents to execute tasks in parallel. You need it when managing multi-role workflows, allowing specialized agents to handle planning, research, or coding simultaneously for faster outcomes.

Yes, you can send messages and read outputs from a running background agent. The orchestration lifecycle defines communication protocols including status checks, mid-run notifications, and read operations to integrate results into the parent task.

To stop or abort a background task running in parallel, use the abort operation defined in the subagent lifecycle. This terminates the autonomous agent and halts its delegated workflow within the orchestration environment.

You can assign researcher, coder, and planner roles to autonomous agents for parallel workflows. These roles tailor tool access and responsibilities to match the specific objectives of information gathering, code changes, or planning.

No, parallel agent orchestration does not block the main conversation thread. Subagents are spawned to execute tasks concurrently in the background, allowing the main task to continue while delegated work runs independently.