About Kiro

The Origin of Kiro

Kiro in Japanese means "circuit", "path" or "route", a name that symbolizes our vision for the software development process. Just like a carefully designed circuit, we believe excellent software should be composed of clear structures and efficient pathways.

At the current crossroads of software development, Kiro represents the perfect fusion of human creativity and AI automation. Unlike other AI coding assistants that only focus on completing code snippets, Kiro emphasizes guiding the entire development process through structured specification files.

What is Kiro?

Kiro is an experimental, intelligent agent-driven integrated development environment (IDE) launched by AWS. The term "intelligent agent" is crucial, it means that Kiro doesn't just respond to prompts, but can execute autonomous, goal-oriented operations. You only need to describe what you want to build, and Kiro will actively explore your codebase, open relevant files, and modify them accordingly to meet your needs.

Traditional AI Coding Assistants

  • Only handle individual code snippets and files
  • Lack understanding of overall project structure
  • Require extensive manual guidance and correction

Kiro Intelligent Agent

  • Understands and modifies multiple files and project structures
  • Provides end-to-end feature development capabilities
  • Executes structured plans and tasks

Kiro's Core Advantages

Specification-Driven Development

Kiro's specification-driven development approach creates a single source of truth for your project through three key files:

  • requirements.md - Captures what needs to be built using user stories and EARS-formatted acceptance criteria, ensuring everyone understands the purpose of the feature.
  • design.md - Outlines the technical architecture, including components, data models, and interfaces, serving as a blueprint for implementation.
  • tasks.md - Breaks down the work into a list of coding tasks, properly sequenced according to dependencies, providing a clear implementation plan.

Intelligent Agent Hooks

Kiro's agent hooks work like an experienced developer, capable of catching things you might miss or completing boilerplate tasks in the background. These event-driven automations trigger agent tasks when you save, create, or delete files:

  • When you save a React component, hooks update the test files.
  • When you modify an API endpoint, hooks refresh the README file.
  • When you prepare to commit code, security hooks scan for leaked credentials.

Multimodal Context Integration

Kiro seamlessly handles various inputs, including files, codebases, documentation, images, repository maps, git diffs, terminal outputs, current issues, URLs, and external documents, building a comprehensive multimodal understanding of the project. This enables Kiro to provide context-aware coding assistance and specification development that perfectly matches the current state of the project.

How Kiro Works

Kiro is built on several core principles that make it powerful:

  1. Intelligent Agent Reasoning Loop: It uses a structured cycle of planning, reasoning, taking action, and evaluating results. This cycle enables it to handle multi-step tasks.
  2. Context Awareness: It integrates closely with your local environment using the MCP protocol and local LSP (Language Server Protocol) tools.
  3. Security and Privacy: All operations are transparent, and code execution is done locally. Your data and code are not sent to the cloud unless you explicitly allow it.
  4. FastAPI and Amazon Q Integration: Kiro can be extended using custom MCP servers (such as Git Repo Research server) and integrates with Amazon Q CLI to provide real-time code analysis and problem-solving.

Kiro's Vision

Our vision is to address the fundamental challenges that make building software products so difficult—from ensuring design consistency across teams and resolving conflicting requirements, to eliminating technical debt, enhancing the rigor of code reviews, and preserving institutional knowledge when senior engineers depart.

The way humans and machines coordinate to build software is still chaotic and fragmented, but we're working to change that. Specification-driven development is an important step in that direction.