Loading image...Kiro
  • CLI
  • Powers
  • Autonomous agent
  • Enterprise
  • Pricing
  • Docs
SIGN INDOWNLOADS
Loading image...Kiro
Loading image...Kiro
Product
  • About Kiro
  • CLI
  • Powers
  • Autonomous agent
  • Pricing
  • Downloads
For
  • Enterprise
  • Startups
Resources
  • Documentation
  • Blog
  • Changelog
  • FAQs
  • Report a bug
  • Suggest an idea
  • Billing support
Social
Site TermsLicenseResponsible AI PolicyLegalPrivacy PolicyCookie Preferences
  1. Docs
  2. IDE
  3. Chat
  4. Subagents

Subagents

Subagents allow Kiro to run multiple tasks in parallel, or delegate specific tasks to subagents that specialize in those tasks. Kiro will automatically launch subagents as appropriate. You can also launch subagents manually by instructing Kiro to do so via a prompt such as "Run subagents to...".

Kiro has two built-in subagents: a "context gathering" subagent used to explore a project and gather relevant context, and a "general purpose" subagent used for parallelizing all other tasks.

Subagents run in parallel; however, the main Kiro agent will wait until all subagents have completed before proceeding. Each subagent has its own context window, ensuring that the main agent context is not polluted by the subagent's execution. Subagents automatically return their results back to the main agent once they finish.

Steering files and MCP servers work in subagents exactly as they do in the main agent. However, subagents do not have access to Specs, and Hooks will not trigger in subagents.

You can significantly speed up development by leveraging subagents to perform multiple tasks simultaneously. In the following example, subagents are used to fetch and analyze several tickets in parallel. Not only is this faster than analyzing the tickets sequentially, the tool call and ticket details stay within each subagent and do not pollute the main agent's context.

Page updated: December 19, 2025
Summarization
Web tools