Kiro CLI 2.0 adds headless mode, Windows, and a GA TUI

Kiro has just rolled out Kiro CLI 2.0, bringing a new headless mode for CI/CD and scripts, plus native Windows support. It also promotes its refreshed terminal UI to GA, with subagents and real-time task lists now default.

Kiro CLI 2.0 adds headless mode, Windows, and a GA TUI

TL;DR

  • Headless mode: Run Kiro CLI programmatically via API key env var; pipe inputs/outputs for CI/CD and scripts
  • Automation parity: Same tools, agents, and capabilities as interactive sessions; supports PR generation and troubleshooting workflows
  • Native Windows support: Installs and runs without WSL; works in Windows Terminal and PowerShell
  • Install command: curl -fsSL [https://cli.kiro.dev/install](https://cli.kiro.dev/install) | bash
  • TUI now GA and default: Refreshed terminal UI replaces experimental mode; includes polish and rough-edge fixes
  • Workflow features: Subagents with ctrl+g monitoring; task list updates in real time for larger tasks

Inside the broader trend of agentic developer tooling, Kiro CLI 2.0 lands as a pragmatic update shaped by terminal-heavy workflows: a programmatic “headless” mode, native Windows support, and a refreshed TUI that’s now fully GA. The through-line is flexibility—staying in an interactive terminal when that’s useful, while also making it possible to run Kiro in places where no human is present, like build systems.

Headless mode brings Kiro into CI/CD and scripts

The biggest shift in 2.0 is headless mode, which lets Kiro CLI run without an interactive session. The setup is straightforward: generate an API key, place it in an environment variable, and Kiro CLI can be executed programmatically.

From there, the CLI supports piping inputs and scripting outputs, enabling use in CI/CD pipelines, build scripts, and other automated workflows. The intent is to keep feature parity with interactive use—the same tools, agents, and capabilities, but accessible to automation. The source calls out examples like generating and publishing pull requests, or running troubleshooting workflows without local user input, and links to a longer walkthrough: a concrete example of using headless CLI.

Native Windows support (no WSL workarounds)

Kiro CLI now installs and runs natively on Windows, including use inside Windows Terminal and PowerShell. That means Kiro’s agents are available directly on Windows for tasks like working in complex codebases and debugging.

Installation details are covered in the docs, including the Windows-specific section: Install the CLI. The post also includes a single install command that applies to macOS, Linux, or Windows:

curl -fsSL [https://cli.kiro.dev/install](https://cli.kiro.dev/install) | bash

The refreshed TUI is now GA, with subagents and task lists

The updated terminal UI—previously launched experimentally—has reached GA and is now the default experience. Alongside general polish and “rough edge” fixes, the update highlights two workflow features:

  • Subagents, including a subagent monitoring view (invoked with ctrl+g) that shows traces and status across subagents, and surfaces permission approvals
  • A task list (an enhanced version of a todo list) that updates in real time as steps complete, and is used by default on larger tasks

The post frames subagents as a way to parallelize work while protecting a parent agent’s context, illustrating a designer → implementer → reviewer loop.

Defaults, fallback options, and where to learn more

Kiro CLI 2.0 and its TUI are now the default. Feedback can be submitted from inside the tool via /feedback, and it’s still possible to revert to the prior experience with kiro-cli --classic. Additional documentation is available at Kiro CLI docs, and the community Discord is linked here: Kiro community on Discord.

Source: Kiro CLI 2.0

Continue the conversation on Slack

Did this article spark your interest? Join our community of experts and enthusiasts to dive deeper, ask questions, and share your ideas.

Join our community