Skip to content

Kairo Code overview

Kairo Code is a coding workspace that can inspect repositories, edit files, run tools, coordinate specialized Agents, and present a reviewable result. You stay in control through explicit workspace scope, permission profiles, approvals, live steering, and cancellation.

The product model

Four concepts organize the experience:

ConceptWhat it means
ProjectOne or more authorized workspace roots and their shared settings.
TaskA durable conversation and execution history inside a Project.
TurnOne request and the work Kairo Code performs to answer it.
AgentThe active worker. Subagents are bounded child Tasks, not a separate execution system.

Kairo Code Desktop keeps Task state, tools, permissions, and history together in one durable workspace. Closing a window or restarting the app does not split the Task into a second history.

What a good Task looks like

Give Kairo Code an outcome, the relevant scope, and a way to verify completion:

Refactor the invoice retry path so duplicate callbacks are idempotent. Preserve the public API, add regression tests for concurrent delivery, and run the billing module test suite.

For broad or ambiguous work, ask for inspection and a plan first. For a focused change, ask Kairo Code to implement and verify it directly. You can steer either kind of turn while it runs.

Where to use it

  • Open repositories and multi-root Projects from one desktop workspace.
  • Review file changes, Browser context, Git state, approvals, and Agent activity without leaving the Task.
  • Continue, steer, stop, archive, or resume durable work from the sidebar.
  • Keep each Project inside the workspace roots and permissions you explicitly selected.

Start here

  1. Install Kairo Code.
  2. Run your first Task.
  3. Learn how Projects and Tasks define durable scope.
  4. Review permissions and approvals before enabling broader automation.

Kairo Code documentation · Private product distribution