What is Kiro?

Kiro is Amazon’s agentic AI development tool: you describe what you want in plain language, and it plans the work, edits your real files, runs commands, and shows you every change for review. It runs in five places (a desktop IDE, a terminal CLI, a browser app, an iOS app, and an open-source always-on agent called Crew) and all of them share the same signature idea, writing the plan down before writing the code.

The simple version

Kiro is a software development tool from Amazon (AWS) with an AI agent at its core.

You describe what you want in plain language (“add a search box to my recipe site”). Instead of answering with a snippet, Kiro can plan the work, write the files, run commands, and show you everything it changed so you can review it.

The official description is short: Kiro is an agentic An AI tool that can take multi-step actions on its own (planning, editing files, running commands) rather than only replying with text. IDE that helps you do your best work with specs, steering, and hooks. Don’t worry about those three words yet; the Core Concepts chapter unpacks each one gently.

One tool, five surfaces:

Surface	What it is	Best for
Kiro IDE	A desktop code editor (based on the same open-source core as VS Code)	Everyday development with full visual control
Kiro CLI	kiro-cli, an agent that lives in your terminal	Terminal lovers, servers, automation, CI
Kiro Web	A browser app at app.kiro.dev, generally available since September 1, 2026 on paid plans	Working on GitHub and GitLab repositories without installing anything
Kiro Mobile	A native iOS app (in preview via TestFlight)	Reviewing and steering agent sessions on the go
Kiro Crew	An open-source always-on agent (Apache 2.0) that runs on your own hardware	Persistent, scheduled work you reach from a dashboard or a chat app

Three analogies that make Kiro click

1. The architect, not just the builder. Most AI coding tools are like a fast builder: you shout instructions and bricks appear. Kiro prefers to act like an architect first. Before building, it writes a plan: what we’re building (requirements), how it will be built (design), and the step-by-step work list (tasks). You approve the blueprint before construction starts. This is called spec-driven development, and it’s Kiro’s signature idea.

2. A new teammate with an onboarding binder. Imagine hiring a talented developer who reads a binder on day one (your product, your tech choices, your naming conventions) and then never forgets it. That binder is Kiro’s steering: small Markdown files in your project that the agent reads every time it works.

3. A tireless junior developer who still needs code review. Kiro can work for minutes at a time without you, across many files. Like any junior teammate, it’s fast, eager, and occasionally confidently wrong. The tool is built around this reality: you can run it in a supervised mode where every edit waits for your approval, and you can roll changes back.

What problems does Kiro solve?

1. “AI code turns into chaos.” Quick AI-generated code often works in the demo and falls apart later, because nobody wrote down what it was supposed to do. Kiro’s specs keep a written record of requirements and design next to the code.

2. “I keep repeating myself.” Without memory of your conventions, every chat starts from zero. Steering files give the agent permanent project knowledge.

3. “The chores never end.” Kiro’s hooks can trigger chores such as updating docs, running tests, and fixing formatting automatically when events happen (like saving a file).

4. “My AI tool can’t see my world.” Real work involves databases, ticket systems, cloud docs, design files. Kiro connects to external tools through MCP Model Context Protocol, an open standard that lets AI tools call external services and data sources through small add-on servers. and through one-click bundles called powers.

How is Kiro different from…

Tool you know	What it does well	What Kiro adds
A normal IDE (VS Code)	Editing, debugging, extensions	An agent that plans and performs multi-file work from plain language
A chatbot (web chat AI)	Answers and snippets you paste manually	Direct access to your real files; it edits, runs, and verifies in place
A completion assistant (classic Copilot-style)	Autocompletes the line you’re typing	Whole features end-to-end: plan → files → commands → review
A classic CLI tool	Does exactly one scripted thing	Conversational, context-aware actions across your whole project

The practical difference is scope: Kiro carries a whole task from plan to reviewed changes, while a chatbot answers with text and a completion assistant finishes the line you’re typing.

Where Kiro came from (the 10-second version)

AWS launched Kiro in public preview in July 2025 and made it generally available on November 17, 2025. More than 250,000 developers used it during the preview. The IDE reached version 1.0 in June 2026, and the model lineup now spans both Anthropic (Claude) and OpenAI (GPT-5.6) models. In August 2026 AWS open-sourced Kiro Crew under Apache 2.0, and on September 1, 2026 Kiro Web reached general availability on the paid plans. The full story, including how Amazon’s earlier Q Developer CLI became Kiro CLI, is in the next chapter.

Is Kiro for you?

• Complete beginner? Yes. Kiro’s plan-first style is friendlier for beginners than freestyle AI coding, because the plan is written in English before anything happens.
• Software engineer? Yes. Specs, steering, and custom agents are designed for production teams, not demos.
• Data engineer / scientist? Yes. The same agent writes pipelines, SQL, tests, and documentation; the CLI fits remote and notebook-adjacent workflows.
• Just need quick answers? Use the cheat sheet and revision chapters; they’re built for you.