View as Markdown

CLI

Kiro CLI runs the Kiro agent in your terminal. In Kiro CLI 3.0.0 or later, its default agent harness is V3, the same unified agent harness as the Kiro IDE and Kiro Web, so your Steering, custom agents, Hooks, permissions, and MCP servers apply here too. Install it, then pick the task you need below.

Install on macOS, Linux, or
curl -fsSL https://cli.kiro.dev/install | bash

Start using Kiro CLI:

bash
cd my-project
kiro-cli

Kiro CLI 3.0.0 or later is a single standalone kiro-cli binary: the desktop app and inline shell autocomplete are retired. If you installed the desktop app, see Moving from the Kiro CLI desktop app.

Start with what you need

Start a chat
Talk to Kiro in your terminal, reference files, add images, and queue follow-up messages.
Choose an agent harness
See which harness a launch starts, pick V3 or V2 for one run or as your default, and get back to V2.
Terminal UI
Panels, themes, keyboard shortcuts, and the fullscreen and voice modes built into the terminal interface.
Manage sessions
Resume, search, and clean up local and cloud sessions, and branch a conversation with a tangent.
Automate with headless runs
Run prompts non-interactively in scripts and CI/CD pipelines.
Connect an editor over ACP
Use Kiro as the agent inside JetBrains IDEs, Zed, and other Agent Client Protocol clients.
Fix a problem
Match a symptom to the page that owns the fix.
Upgrade from CLI 2.x
Move your sessions, agent configs, permissions, Hooks, and ACP clients to V3, one checked step at a time.
Code Intelligence
Language-aware code understanding and navigation.

Which harness runs

Which harness each launch type starts, the one-run and saved choices, and the way back to V2 are covered once in Choose an agent harness.

Shared capabilities

Kiro CLI supports the capabilities below. Each linked page lists its availability by surface:

  • Specs — Define requirements, generate designs, and execute task plans
  • Workflows — Run multi-step work in the background in a V3 session; turn it on with the chat.enableWorkflows setting
  • Steering — Guide Kiro with project-specific context and conventions
  • Hooks — Automate actions before and after commands
  • MCP — Connect external tools and data sources
  • Custom agents — Build specialized agents for your workflows
  • Skills — Extend Kiro with reusable skill packages
  • Sub-agents — Delegate focused work to agents that run in parallel
  • Models — Configure which AI models to use
  • Permissions — Manage capability-based access controls
  • Cloud sessions — Keep a session running in a managed sandbox after you disconnect

Use cases

The Kiro CLI is ideal for:

  • Interactive Development: Chat with Kiro directly in your terminal for instant help
  • Headless Automation: Run prompts non-interactively in CI/CD pipelines using API key authentication
  • Custom Automation: Create specialized agents for your specific workflows
  • Team Standardization: Use team-level best practices and preferences
  • External Integrations: Connect tools and services through MCP servers
  • Editor Integration: Bring the Kiro agent into other editors over ACP
  • Workflow Optimization: Automate repetitive tasks with smart hooks
Page updated: November 3, 2026